|
Ruby
2.0.0p353(2013-11-22revision43784)
|
00001 /********************************************************************** 00002 00003 parse.y - 00004 00005 $Author: nagachika $ 00006 created at: Fri May 28 18:02:42 JST 1993 00007 00008 Copyright (C) 1993-2007 Yukihiro Matsumoto 00009 00010 **********************************************************************/ 00011 00012 %{ 00013 00014 #ifndef PARSER_DEBUG 00015 #define PARSER_DEBUG 0 00016 #endif 00017 #define YYDEBUG 1 00018 #define YYERROR_VERBOSE 1 00019 #define YYSTACK_USE_ALLOCA 0 00020 00021 #include "ruby/ruby.h" 00022 #include "ruby/st.h" 00023 #include "ruby/encoding.h" 00024 #include "internal.h" 00025 #include "node.h" 00026 #include "parse.h" 00027 #include "id.h" 00028 #include "regenc.h" 00029 #include <stdio.h> 00030 #include <errno.h> 00031 #include <ctype.h> 00032 #include "probes.h" 00033 00034 #define numberof(array) (int)(sizeof(array) / sizeof((array)[0])) 00035 00036 #define YYMALLOC(size) rb_parser_malloc(parser, (size)) 00037 #define YYREALLOC(ptr, size) rb_parser_realloc(parser, (ptr), (size)) 00038 #define YYCALLOC(nelem, size) rb_parser_calloc(parser, (nelem), (size)) 00039 #define YYFREE(ptr) rb_parser_free(parser, (ptr)) 00040 #define malloc YYMALLOC 00041 #define realloc YYREALLOC 00042 #define calloc YYCALLOC 00043 #define free YYFREE 00044 00045 #ifndef RIPPER 00046 static ID register_symid(ID, const char *, long, rb_encoding *); 00047 static ID register_symid_str(ID, VALUE); 00048 #define REGISTER_SYMID(id, name) register_symid((id), (name), strlen(name), enc) 00049 #include "id.c" 00050 #endif 00051 00052 #define is_notop_id(id) ((id)>tLAST_OP_ID) 00053 #define is_local_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_LOCAL) 00054 #define is_global_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_GLOBAL) 00055 #define is_instance_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_INSTANCE) 00056 #define is_attrset_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_ATTRSET) 00057 #define is_const_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_CONST) 00058 #define is_class_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_CLASS) 00059 #define is_junk_id(id) (is_notop_id(id)&&((id)&ID_SCOPE_MASK)==ID_JUNK) 00060 #define id_type(id) (is_notop_id(id) ? (int)((id)&ID_SCOPE_MASK) : -1) 00061 00062 #define is_asgn_or_id(id) ((is_notop_id(id)) && \ 00063 (((id)&ID_SCOPE_MASK) == ID_GLOBAL || \ 00064 ((id)&ID_SCOPE_MASK) == ID_INSTANCE || \ 00065 ((id)&ID_SCOPE_MASK) == ID_CLASS)) 00066 00067 enum lex_state_bits { 00068 EXPR_BEG_bit, /* ignore newline, +/- is a sign. */ 00069 EXPR_END_bit, /* newline significant, +/- is an operator. */ 00070 EXPR_ENDARG_bit, /* ditto, and unbound braces. */ 00071 EXPR_ENDFN_bit, /* ditto, and unbound braces. */ 00072 EXPR_ARG_bit, /* newline significant, +/- is an operator. */ 00073 EXPR_CMDARG_bit, /* newline significant, +/- is an operator. */ 00074 EXPR_MID_bit, /* newline significant, +/- is an operator. */ 00075 EXPR_FNAME_bit, /* ignore newline, no reserved words. */ 00076 EXPR_DOT_bit, /* right after `.' or `::', no reserved words. */ 00077 EXPR_CLASS_bit, /* immediate after `class', no here document. */ 00078 EXPR_VALUE_bit, /* alike EXPR_BEG but label is disallowed. */ 00079 EXPR_MAX_STATE 00080 }; 00081 /* examine combinations */ 00082 enum lex_state_e { 00083 #define DEF_EXPR(n) EXPR_##n = (1 << EXPR_##n##_bit) 00084 DEF_EXPR(BEG), 00085 DEF_EXPR(END), 00086 DEF_EXPR(ENDARG), 00087 DEF_EXPR(ENDFN), 00088 DEF_EXPR(ARG), 00089 DEF_EXPR(CMDARG), 00090 DEF_EXPR(MID), 00091 DEF_EXPR(FNAME), 00092 DEF_EXPR(DOT), 00093 DEF_EXPR(CLASS), 00094 DEF_EXPR(VALUE), 00095 EXPR_BEG_ANY = (EXPR_BEG | EXPR_VALUE | EXPR_MID | EXPR_CLASS), 00096 EXPR_ARG_ANY = (EXPR_ARG | EXPR_CMDARG), 00097 EXPR_END_ANY = (EXPR_END | EXPR_ENDARG | EXPR_ENDFN) 00098 }; 00099 #define IS_lex_state_for(x, ls) ((x) & (ls)) 00100 #define IS_lex_state(ls) IS_lex_state_for(lex_state, (ls)) 00101 00102 #if PARSER_DEBUG 00103 static const char *lex_state_name(enum lex_state_e state); 00104 #endif 00105 00106 typedef VALUE stack_type; 00107 00108 # define BITSTACK_PUSH(stack, n) ((stack) = ((stack)<<1)|((n)&1)) 00109 # define BITSTACK_POP(stack) ((stack) = (stack) >> 1) 00110 # define BITSTACK_LEXPOP(stack) ((stack) = ((stack) >> 1) | ((stack) & 1)) 00111 # define BITSTACK_SET_P(stack) ((stack)&1) 00112 00113 #define COND_PUSH(n) BITSTACK_PUSH(cond_stack, (n)) 00114 #define COND_POP() BITSTACK_POP(cond_stack) 00115 #define COND_LEXPOP() BITSTACK_LEXPOP(cond_stack) 00116 #define COND_P() BITSTACK_SET_P(cond_stack) 00117 00118 #define CMDARG_PUSH(n) BITSTACK_PUSH(cmdarg_stack, (n)) 00119 #define CMDARG_POP() BITSTACK_POP(cmdarg_stack) 00120 #define CMDARG_LEXPOP() BITSTACK_LEXPOP(cmdarg_stack) 00121 #define CMDARG_P() BITSTACK_SET_P(cmdarg_stack) 00122 00123 struct vtable { 00124 ID *tbl; 00125 int pos; 00126 int capa; 00127 struct vtable *prev; 00128 }; 00129 00130 struct local_vars { 00131 struct vtable *args; 00132 struct vtable *vars; 00133 struct vtable *used; 00134 struct local_vars *prev; 00135 }; 00136 00137 #define DVARS_INHERIT ((void*)1) 00138 #define DVARS_TOPSCOPE NULL 00139 #define DVARS_SPECIAL_P(tbl) (!POINTER_P(tbl)) 00140 #define POINTER_P(val) ((VALUE)(val) & ~(VALUE)3) 00141 00142 static int 00143 vtable_size(const struct vtable *tbl) 00144 { 00145 if (POINTER_P(tbl)) { 00146 return tbl->pos; 00147 } 00148 else { 00149 return 0; 00150 } 00151 } 00152 00153 #define VTBL_DEBUG 0 00154 00155 static struct vtable * 00156 vtable_alloc(struct vtable *prev) 00157 { 00158 struct vtable *tbl = ALLOC(struct vtable); 00159 tbl->pos = 0; 00160 tbl->capa = 8; 00161 tbl->tbl = ALLOC_N(ID, tbl->capa); 00162 tbl->prev = prev; 00163 if (VTBL_DEBUG) printf("vtable_alloc: %p\n", (void *)tbl); 00164 return tbl; 00165 } 00166 00167 static void 00168 vtable_free(struct vtable *tbl) 00169 { 00170 if (VTBL_DEBUG)printf("vtable_free: %p\n", (void *)tbl); 00171 if (POINTER_P(tbl)) { 00172 if (tbl->tbl) { 00173 xfree(tbl->tbl); 00174 } 00175 xfree(tbl); 00176 } 00177 } 00178 00179 static void 00180 vtable_add(struct vtable *tbl, ID id) 00181 { 00182 if (!POINTER_P(tbl)) { 00183 rb_bug("vtable_add: vtable is not allocated (%p)", (void *)tbl); 00184 } 00185 if (VTBL_DEBUG) printf("vtable_add: %p, %s\n", (void *)tbl, rb_id2name(id)); 00186 00187 if (tbl->pos == tbl->capa) { 00188 tbl->capa = tbl->capa * 2; 00189 REALLOC_N(tbl->tbl, ID, tbl->capa); 00190 } 00191 tbl->tbl[tbl->pos++] = id; 00192 } 00193 00194 static int 00195 vtable_included(const struct vtable * tbl, ID id) 00196 { 00197 int i; 00198 00199 if (POINTER_P(tbl)) { 00200 for (i = 0; i < tbl->pos; i++) { 00201 if (tbl->tbl[i] == id) { 00202 return i+1; 00203 } 00204 } 00205 } 00206 return 0; 00207 } 00208 00209 00210 #ifndef RIPPER 00211 typedef struct token_info { 00212 const char *token; 00213 int linenum; 00214 int column; 00215 int nonspc; 00216 struct token_info *next; 00217 } token_info; 00218 #endif 00219 00220 /* 00221 Structure of Lexer Buffer: 00222 00223 lex_pbeg tokp lex_p lex_pend 00224 | | | | 00225 |-----------+--------------+------------| 00226 |<------------>| 00227 token 00228 */ 00229 struct parser_params { 00230 int is_ripper; 00231 NODE *heap; 00232 00233 YYSTYPE *parser_yylval; 00234 VALUE eofp; 00235 00236 NODE *parser_lex_strterm; 00237 enum lex_state_e parser_lex_state; 00238 stack_type parser_cond_stack; 00239 stack_type parser_cmdarg_stack; 00240 int parser_class_nest; 00241 int parser_paren_nest; 00242 int parser_lpar_beg; 00243 int parser_in_single; 00244 int parser_in_def; 00245 int parser_brace_nest; 00246 int parser_compile_for_eval; 00247 VALUE parser_cur_mid; 00248 int parser_in_defined; 00249 char *parser_tokenbuf; 00250 int parser_tokidx; 00251 int parser_toksiz; 00252 int parser_tokline; 00253 VALUE parser_lex_input; 00254 VALUE parser_lex_lastline; 00255 VALUE parser_lex_nextline; 00256 const char *parser_lex_pbeg; 00257 const char *parser_lex_p; 00258 const char *parser_lex_pend; 00259 int parser_heredoc_end; 00260 int parser_command_start; 00261 NODE *parser_deferred_nodes; 00262 long parser_lex_gets_ptr; 00263 VALUE (*parser_lex_gets)(struct parser_params*,VALUE); 00264 struct local_vars *parser_lvtbl; 00265 int parser_ruby__end__seen; 00266 int line_count; 00267 int has_shebang; 00268 char *parser_ruby_sourcefile; /* current source file */ 00269 int parser_ruby_sourceline; /* current line no. */ 00270 rb_encoding *enc; 00271 00272 int parser_yydebug; 00273 00274 #ifndef RIPPER 00275 /* Ruby core only */ 00276 NODE *parser_eval_tree_begin; 00277 NODE *parser_eval_tree; 00278 VALUE debug_lines; 00279 VALUE coverage; 00280 int nerr; 00281 00282 int parser_token_info_enabled; 00283 token_info *parser_token_info; 00284 #else 00285 /* Ripper only */ 00286 VALUE parser_ruby_sourcefile_string; 00287 const char *tokp; 00288 VALUE delayed; 00289 int delayed_line; 00290 int delayed_col; 00291 00292 VALUE value; 00293 VALUE result; 00294 VALUE parsing_thread; 00295 int toplevel_p; 00296 #endif 00297 }; 00298 00299 #define STR_NEW(p,n) rb_enc_str_new((p),(n),current_enc) 00300 #define STR_NEW0() rb_enc_str_new(0,0,current_enc) 00301 #define STR_NEW2(p) rb_enc_str_new((p),strlen(p),current_enc) 00302 #define STR_NEW3(p,n,e,func) parser_str_new((p),(n),(e),(func),current_enc) 00303 #define ENC_SINGLE(cr) ((cr)==ENC_CODERANGE_7BIT) 00304 #define TOK_INTERN(mb) rb_intern3(tok(), toklen(), current_enc) 00305 00306 static int parser_yyerror(struct parser_params*, const char*); 00307 #define yyerror(msg) parser_yyerror(parser, (msg)) 00308 00309 #define lex_strterm (parser->parser_lex_strterm) 00310 #define lex_state (parser->parser_lex_state) 00311 #define cond_stack (parser->parser_cond_stack) 00312 #define cmdarg_stack (parser->parser_cmdarg_stack) 00313 #define class_nest (parser->parser_class_nest) 00314 #define paren_nest (parser->parser_paren_nest) 00315 #define lpar_beg (parser->parser_lpar_beg) 00316 #define brace_nest (parser->parser_brace_nest) 00317 #define in_single (parser->parser_in_single) 00318 #define in_def (parser->parser_in_def) 00319 #define compile_for_eval (parser->parser_compile_for_eval) 00320 #define cur_mid (parser->parser_cur_mid) 00321 #define in_defined (parser->parser_in_defined) 00322 #define tokenbuf (parser->parser_tokenbuf) 00323 #define tokidx (parser->parser_tokidx) 00324 #define toksiz (parser->parser_toksiz) 00325 #define tokline (parser->parser_tokline) 00326 #define lex_input (parser->parser_lex_input) 00327 #define lex_lastline (parser->parser_lex_lastline) 00328 #define lex_nextline (parser->parser_lex_nextline) 00329 #define lex_pbeg (parser->parser_lex_pbeg) 00330 #define lex_p (parser->parser_lex_p) 00331 #define lex_pend (parser->parser_lex_pend) 00332 #define heredoc_end (parser->parser_heredoc_end) 00333 #define command_start (parser->parser_command_start) 00334 #define deferred_nodes (parser->parser_deferred_nodes) 00335 #define lex_gets_ptr (parser->parser_lex_gets_ptr) 00336 #define lex_gets (parser->parser_lex_gets) 00337 #define lvtbl (parser->parser_lvtbl) 00338 #define ruby__end__seen (parser->parser_ruby__end__seen) 00339 #define ruby_sourceline (parser->parser_ruby_sourceline) 00340 #define ruby_sourcefile (parser->parser_ruby_sourcefile) 00341 #define current_enc (parser->enc) 00342 #define yydebug (parser->parser_yydebug) 00343 #ifdef RIPPER 00344 #else 00345 #define ruby_eval_tree (parser->parser_eval_tree) 00346 #define ruby_eval_tree_begin (parser->parser_eval_tree_begin) 00347 #define ruby_debug_lines (parser->debug_lines) 00348 #define ruby_coverage (parser->coverage) 00349 #endif 00350 00351 #if YYPURE 00352 static int yylex(void*, void*); 00353 #else 00354 static int yylex(void*); 00355 #endif 00356 00357 #ifndef RIPPER 00358 #define yyparse ruby_yyparse 00359 00360 static NODE* node_newnode(struct parser_params *, enum node_type, VALUE, VALUE, VALUE); 00361 #define rb_node_newnode(type, a1, a2, a3) node_newnode(parser, (type), (a1), (a2), (a3)) 00362 00363 static NODE *cond_gen(struct parser_params*,NODE*); 00364 #define cond(node) cond_gen(parser, (node)) 00365 static NODE *logop_gen(struct parser_params*,enum node_type,NODE*,NODE*); 00366 #define logop(type,node1,node2) logop_gen(parser, (type), (node1), (node2)) 00367 00368 static NODE *newline_node(NODE*); 00369 static void fixpos(NODE*,NODE*); 00370 00371 static int value_expr_gen(struct parser_params*,NODE*); 00372 static void void_expr_gen(struct parser_params*,NODE*); 00373 static NODE *remove_begin(NODE*); 00374 #define value_expr(node) value_expr_gen(parser, (node) = remove_begin(node)) 00375 #define void_expr0(node) void_expr_gen(parser, (node)) 00376 #define void_expr(node) void_expr0((node) = remove_begin(node)) 00377 static void void_stmts_gen(struct parser_params*,NODE*); 00378 #define void_stmts(node) void_stmts_gen(parser, (node)) 00379 static void reduce_nodes_gen(struct parser_params*,NODE**); 00380 #define reduce_nodes(n) reduce_nodes_gen(parser,(n)) 00381 static void block_dup_check_gen(struct parser_params*,NODE*,NODE*); 00382 #define block_dup_check(n1,n2) block_dup_check_gen(parser,(n1),(n2)) 00383 00384 static NODE *block_append_gen(struct parser_params*,NODE*,NODE*); 00385 #define block_append(h,t) block_append_gen(parser,(h),(t)) 00386 static NODE *list_append_gen(struct parser_params*,NODE*,NODE*); 00387 #define list_append(l,i) list_append_gen(parser,(l),(i)) 00388 static NODE *list_concat_gen(struct parser_params*,NODE*,NODE*); 00389 #define list_concat(h,t) list_concat_gen(parser,(h),(t)) 00390 static NODE *arg_append_gen(struct parser_params*,NODE*,NODE*); 00391 #define arg_append(h,t) arg_append_gen(parser,(h),(t)) 00392 static NODE *arg_concat_gen(struct parser_params*,NODE*,NODE*); 00393 #define arg_concat(h,t) arg_concat_gen(parser,(h),(t)) 00394 static NODE *literal_concat_gen(struct parser_params*,NODE*,NODE*); 00395 #define literal_concat(h,t) literal_concat_gen(parser,(h),(t)) 00396 static int literal_concat0(struct parser_params *, VALUE, VALUE); 00397 static NODE *new_evstr_gen(struct parser_params*,NODE*); 00398 #define new_evstr(n) new_evstr_gen(parser,(n)) 00399 static NODE *evstr2dstr_gen(struct parser_params*,NODE*); 00400 #define evstr2dstr(n) evstr2dstr_gen(parser,(n)) 00401 static NODE *splat_array(NODE*); 00402 00403 static NODE *call_bin_op_gen(struct parser_params*,NODE*,ID,NODE*); 00404 #define call_bin_op(recv,id,arg1) call_bin_op_gen(parser, (recv),(id),(arg1)) 00405 static NODE *call_uni_op_gen(struct parser_params*,NODE*,ID); 00406 #define call_uni_op(recv,id) call_uni_op_gen(parser, (recv),(id)) 00407 00408 static NODE *new_args_gen(struct parser_params*,NODE*,NODE*,ID,NODE*,NODE*); 00409 #define new_args(f,o,r,p,t) new_args_gen(parser, (f),(o),(r),(p),(t)) 00410 static NODE *new_args_tail_gen(struct parser_params*,NODE*,ID,ID); 00411 #define new_args_tail(k,kr,b) new_args_tail_gen(parser, (k),(kr),(b)) 00412 00413 static NODE *negate_lit(NODE*); 00414 static NODE *ret_args_gen(struct parser_params*,NODE*); 00415 #define ret_args(node) ret_args_gen(parser, (node)) 00416 static NODE *arg_blk_pass(NODE*,NODE*); 00417 static NODE *new_yield_gen(struct parser_params*,NODE*); 00418 #define new_yield(node) new_yield_gen(parser, (node)) 00419 static NODE *dsym_node_gen(struct parser_params*,NODE*); 00420 #define dsym_node(node) dsym_node_gen(parser, (node)) 00421 00422 static NODE *gettable_gen(struct parser_params*,ID); 00423 #define gettable(id) gettable_gen(parser,(id)) 00424 static NODE *assignable_gen(struct parser_params*,ID,NODE*); 00425 #define assignable(id,node) assignable_gen(parser, (id), (node)) 00426 00427 static NODE *aryset_gen(struct parser_params*,NODE*,NODE*); 00428 #define aryset(node1,node2) aryset_gen(parser, (node1), (node2)) 00429 static NODE *attrset_gen(struct parser_params*,NODE*,ID); 00430 #define attrset(node,id) attrset_gen(parser, (node), (id)) 00431 00432 static void rb_backref_error_gen(struct parser_params*,NODE*); 00433 #define rb_backref_error(n) rb_backref_error_gen(parser,(n)) 00434 static NODE *node_assign_gen(struct parser_params*,NODE*,NODE*); 00435 #define node_assign(node1, node2) node_assign_gen(parser, (node1), (node2)) 00436 00437 static NODE *new_op_assign_gen(struct parser_params *parser, NODE *lhs, ID op, NODE *rhs); 00438 static NODE *new_attr_op_assign_gen(struct parser_params *parser, NODE *lhs, ID attr, ID op, NODE *rhs); 00439 #define new_attr_op_assign(lhs, type, attr, op, rhs) new_attr_op_assign_gen(parser, (lhs), (attr), (op), (rhs)) 00440 static NODE *new_const_op_assign_gen(struct parser_params *parser, NODE *lhs, ID op, NODE *rhs); 00441 #define new_const_op_assign(lhs, op, rhs) new_const_op_assign_gen(parser, (lhs), (op), (rhs)) 00442 00443 static NODE *match_op_gen(struct parser_params*,NODE*,NODE*); 00444 #define match_op(node1,node2) match_op_gen(parser, (node1), (node2)) 00445 00446 static ID *local_tbl_gen(struct parser_params*); 00447 #define local_tbl() local_tbl_gen(parser) 00448 00449 static void fixup_nodes(NODE **); 00450 00451 static VALUE reg_compile_gen(struct parser_params*, VALUE, int); 00452 #define reg_compile(str,options) reg_compile_gen(parser, (str), (options)) 00453 static void reg_fragment_setenc_gen(struct parser_params*, VALUE, int); 00454 #define reg_fragment_setenc(str,options) reg_fragment_setenc_gen(parser, (str), (options)) 00455 static int reg_fragment_check_gen(struct parser_params*, VALUE, int); 00456 #define reg_fragment_check(str,options) reg_fragment_check_gen(parser, (str), (options)) 00457 static NODE *reg_named_capture_assign_gen(struct parser_params* parser, VALUE regexp, NODE *match); 00458 #define reg_named_capture_assign(regexp,match) reg_named_capture_assign_gen(parser,(regexp),(match)) 00459 00460 #define get_id(id) (id) 00461 #define get_value(val) (val) 00462 #else 00463 #define value_expr(node) ((void)(node)) 00464 #define remove_begin(node) (node) 00465 #define rb_dvar_defined(id) 0 00466 #define rb_local_defined(id) 0 00467 static ID ripper_get_id(VALUE); 00468 #define get_id(id) ripper_get_id(id) 00469 static VALUE ripper_get_value(VALUE); 00470 #define get_value(val) ripper_get_value(val) 00471 static VALUE assignable_gen(struct parser_params*,VALUE); 00472 #define assignable(lhs,node) assignable_gen(parser, (lhs)) 00473 static int id_is_var_gen(struct parser_params *parser, ID id); 00474 #define id_is_var(id) id_is_var_gen(parser, (id)) 00475 00476 #define node_assign(node1, node2) dispatch2(assign, (node1), (node2)) 00477 00478 static VALUE new_op_assign_gen(struct parser_params *parser, VALUE lhs, VALUE op, VALUE rhs); 00479 static VALUE new_attr_op_assign_gen(struct parser_params *parser, VALUE lhs, VALUE type, VALUE attr, VALUE op, VALUE rhs); 00480 #define new_attr_op_assign(lhs, type, attr, op, rhs) new_attr_op_assign_gen(parser, (lhs), (type), (attr), (op), (rhs)) 00481 00482 #endif /* !RIPPER */ 00483 00484 #define new_op_assign(lhs, op, rhs) new_op_assign_gen(parser, (lhs), (op), (rhs)) 00485 00486 static ID formal_argument_gen(struct parser_params*, ID); 00487 #define formal_argument(id) formal_argument_gen(parser, (id)) 00488 static ID shadowing_lvar_gen(struct parser_params*,ID); 00489 #define shadowing_lvar(name) shadowing_lvar_gen(parser, (name)) 00490 static void new_bv_gen(struct parser_params*,ID); 00491 #define new_bv(id) new_bv_gen(parser, (id)) 00492 00493 static void local_push_gen(struct parser_params*,int); 00494 #define local_push(top) local_push_gen(parser,(top)) 00495 static void local_pop_gen(struct parser_params*); 00496 #define local_pop() local_pop_gen(parser) 00497 static int local_var_gen(struct parser_params*, ID); 00498 #define local_var(id) local_var_gen(parser, (id)) 00499 static int arg_var_gen(struct parser_params*, ID); 00500 #define arg_var(id) arg_var_gen(parser, (id)) 00501 static int local_id_gen(struct parser_params*, ID); 00502 #define local_id(id) local_id_gen(parser, (id)) 00503 static ID internal_id_gen(struct parser_params*); 00504 #define internal_id() internal_id_gen(parser) 00505 00506 static const struct vtable *dyna_push_gen(struct parser_params *); 00507 #define dyna_push() dyna_push_gen(parser) 00508 static void dyna_pop_gen(struct parser_params*, const struct vtable *); 00509 #define dyna_pop(node) dyna_pop_gen(parser, (node)) 00510 static int dyna_in_block_gen(struct parser_params*); 00511 #define dyna_in_block() dyna_in_block_gen(parser) 00512 #define dyna_var(id) local_var(id) 00513 static int dvar_defined_gen(struct parser_params*,ID,int); 00514 #define dvar_defined(id) dvar_defined_gen(parser, (id), 0) 00515 #define dvar_defined_get(id) dvar_defined_gen(parser, (id), 1) 00516 static int dvar_curr_gen(struct parser_params*,ID); 00517 #define dvar_curr(id) dvar_curr_gen(parser, (id)) 00518 00519 static int lvar_defined_gen(struct parser_params*, ID); 00520 #define lvar_defined(id) lvar_defined_gen(parser, (id)) 00521 00522 #define RE_OPTION_ONCE (1<<16) 00523 #define RE_OPTION_ENCODING_SHIFT 8 00524 #define RE_OPTION_ENCODING(e) (((e)&0xff)<<RE_OPTION_ENCODING_SHIFT) 00525 #define RE_OPTION_ENCODING_IDX(o) (((o)>>RE_OPTION_ENCODING_SHIFT)&0xff) 00526 #define RE_OPTION_ENCODING_NONE(o) ((o)&RE_OPTION_ARG_ENCODING_NONE) 00527 #define RE_OPTION_MASK 0xff 00528 #define RE_OPTION_ARG_ENCODING_NONE 32 00529 00530 #define NODE_STRTERM NODE_ZARRAY /* nothing to gc */ 00531 #define NODE_HEREDOC NODE_ARRAY /* 1, 3 to gc */ 00532 #define SIGN_EXTEND(x,n) (((1<<(n)-1)^((x)&~(~0<<(n))))-(1<<(n)-1)) 00533 #define nd_func u1.id 00534 #if SIZEOF_SHORT == 2 00535 #define nd_term(node) ((signed short)(node)->u2.id) 00536 #else 00537 #define nd_term(node) SIGN_EXTEND((node)->u2.id, CHAR_BIT*2) 00538 #endif 00539 #define nd_paren(node) (char)((node)->u2.id >> CHAR_BIT*2) 00540 #define nd_nest u3.cnt 00541 00542 /****** Ripper *******/ 00543 00544 #ifdef RIPPER 00545 #define RIPPER_VERSION "0.1.0" 00546 00547 #include "eventids1.c" 00548 #include "eventids2.c" 00549 00550 static VALUE ripper_dispatch0(struct parser_params*,ID); 00551 static VALUE ripper_dispatch1(struct parser_params*,ID,VALUE); 00552 static VALUE ripper_dispatch2(struct parser_params*,ID,VALUE,VALUE); 00553 static VALUE ripper_dispatch3(struct parser_params*,ID,VALUE,VALUE,VALUE); 00554 static VALUE ripper_dispatch4(struct parser_params*,ID,VALUE,VALUE,VALUE,VALUE); 00555 static VALUE ripper_dispatch5(struct parser_params*,ID,VALUE,VALUE,VALUE,VALUE,VALUE); 00556 static VALUE ripper_dispatch7(struct parser_params*,ID,VALUE,VALUE,VALUE,VALUE,VALUE,VALUE,VALUE); 00557 00558 #define dispatch0(n) ripper_dispatch0(parser, TOKEN_PASTE(ripper_id_, n)) 00559 #define dispatch1(n,a) ripper_dispatch1(parser, TOKEN_PASTE(ripper_id_, n), (a)) 00560 #define dispatch2(n,a,b) ripper_dispatch2(parser, TOKEN_PASTE(ripper_id_, n), (a), (b)) 00561 #define dispatch3(n,a,b,c) ripper_dispatch3(parser, TOKEN_PASTE(ripper_id_, n), (a), (b), (c)) 00562 #define dispatch4(n,a,b,c,d) ripper_dispatch4(parser, TOKEN_PASTE(ripper_id_, n), (a), (b), (c), (d)) 00563 #define dispatch5(n,a,b,c,d,e) ripper_dispatch5(parser, TOKEN_PASTE(ripper_id_, n), (a), (b), (c), (d), (e)) 00564 #define dispatch7(n,a,b,c,d,e,f,g) ripper_dispatch7(parser, TOKEN_PASTE(ripper_id_, n), (a), (b), (c), (d), (e), (f), (g)) 00565 00566 #define yyparse ripper_yyparse 00567 00568 #define ripper_intern(s) ID2SYM(rb_intern(s)) 00569 static VALUE ripper_id2sym(ID); 00570 #ifdef __GNUC__ 00571 #define ripper_id2sym(id) ((id) < 256 && rb_ispunct(id) ? \ 00572 ID2SYM(id) : ripper_id2sym(id)) 00573 #endif 00574 00575 #define arg_new() dispatch0(args_new) 00576 #define arg_add(l,a) dispatch2(args_add, (l), (a)) 00577 #define arg_add_star(l,a) dispatch2(args_add_star, (l), (a)) 00578 #define arg_add_block(l,b) dispatch2(args_add_block, (l), (b)) 00579 #define arg_add_optblock(l,b) ((b)==Qundef? (l) : dispatch2(args_add_block, (l), (b))) 00580 #define bare_assoc(v) dispatch1(bare_assoc_hash, (v)) 00581 #define arg_add_assocs(l,b) arg_add((l), bare_assoc(b)) 00582 00583 #define args2mrhs(a) dispatch1(mrhs_new_from_args, (a)) 00584 #define mrhs_new() dispatch0(mrhs_new) 00585 #define mrhs_add(l,a) dispatch2(mrhs_add, (l), (a)) 00586 #define mrhs_add_star(l,a) dispatch2(mrhs_add_star, (l), (a)) 00587 00588 #define mlhs_new() dispatch0(mlhs_new) 00589 #define mlhs_add(l,a) dispatch2(mlhs_add, (l), (a)) 00590 #define mlhs_add_star(l,a) dispatch2(mlhs_add_star, (l), (a)) 00591 00592 #define params_new(pars, opts, rest, pars2, kws, kwrest, blk) \ 00593 dispatch7(params, (pars), (opts), (rest), (pars2), (kws), (kwrest), (blk)) 00594 00595 #define blockvar_new(p,v) dispatch2(block_var, (p), (v)) 00596 #define blockvar_add_star(l,a) dispatch2(block_var_add_star, (l), (a)) 00597 #define blockvar_add_block(l,a) dispatch2(block_var_add_block, (l), (a)) 00598 00599 #define method_optarg(m,a) ((a)==Qundef ? (m) : dispatch2(method_add_arg,(m),(a))) 00600 #define method_arg(m,a) dispatch2(method_add_arg,(m),(a)) 00601 #define method_add_block(m,b) dispatch2(method_add_block, (m), (b)) 00602 00603 #define escape_Qundef(x) ((x)==Qundef ? Qnil : (x)) 00604 00605 static inline VALUE 00606 new_args_gen(struct parser_params *parser, VALUE f, VALUE o, VALUE r, VALUE p, VALUE tail) 00607 { 00608 NODE *t = (NODE *)tail; 00609 VALUE k = t->u1.value, kr = t->u2.value, b = t->u3.value; 00610 return params_new(f, o, r, p, k, kr, escape_Qundef(b)); 00611 } 00612 #define new_args(f,o,r,p,t) new_args_gen(parser, (f),(o),(r),(p),(t)) 00613 00614 static inline VALUE 00615 new_args_tail_gen(struct parser_params *parser, VALUE k, VALUE kr, VALUE b) 00616 { 00617 return (VALUE)rb_node_newnode(NODE_MEMO, k, kr, b); 00618 } 00619 #define new_args_tail(k,kr,b) new_args_tail_gen(parser, (k),(kr),(b)) 00620 00621 #define FIXME 0 00622 00623 #endif /* RIPPER */ 00624 00625 #ifndef RIPPER 00626 # define Qnone 0 00627 # define ifndef_ripper(x) (x) 00628 #else 00629 # define Qnone Qnil 00630 # define ifndef_ripper(x) 00631 #endif 00632 00633 #ifndef RIPPER 00634 # define rb_warn0(fmt) rb_compile_warn(ruby_sourcefile, ruby_sourceline, (fmt)) 00635 # define rb_warnI(fmt,a) rb_compile_warn(ruby_sourcefile, ruby_sourceline, (fmt), (a)) 00636 # define rb_warnS(fmt,a) rb_compile_warn(ruby_sourcefile, ruby_sourceline, (fmt), (a)) 00637 # define rb_warn4S(file,line,fmt,a) rb_compile_warn((file), (line), (fmt), (a)) 00638 # define rb_warning0(fmt) rb_compile_warning(ruby_sourcefile, ruby_sourceline, (fmt)) 00639 # define rb_warningS(fmt,a) rb_compile_warning(ruby_sourcefile, ruby_sourceline, (fmt), (a)) 00640 #else 00641 # define rb_warn0(fmt) ripper_warn0(parser, (fmt)) 00642 # define rb_warnI(fmt,a) ripper_warnI(parser, (fmt), (a)) 00643 # define rb_warnS(fmt,a) ripper_warnS(parser, (fmt), (a)) 00644 # define rb_warn4S(file,line,fmt,a) ripper_warnS(parser, (fmt), (a)) 00645 # define rb_warning0(fmt) ripper_warning0(parser, (fmt)) 00646 # define rb_warningS(fmt,a) ripper_warningS(parser, (fmt), (a)) 00647 static void ripper_warn0(struct parser_params*, const char*); 00648 static void ripper_warnI(struct parser_params*, const char*, int); 00649 static void ripper_warnS(struct parser_params*, const char*, const char*); 00650 static void ripper_warning0(struct parser_params*, const char*); 00651 static void ripper_warningS(struct parser_params*, const char*, const char*); 00652 #endif 00653 00654 #ifdef RIPPER 00655 static void ripper_compile_error(struct parser_params*, const char *fmt, ...); 00656 # define rb_compile_error ripper_compile_error 00657 # define compile_error ripper_compile_error 00658 # define PARSER_ARG parser, 00659 #else 00660 # define rb_compile_error rb_compile_error_with_enc 00661 # define compile_error parser->nerr++,rb_compile_error_with_enc 00662 # define PARSER_ARG ruby_sourcefile, ruby_sourceline, current_enc, 00663 #endif 00664 00665 /* Older versions of Yacc set YYMAXDEPTH to a very low value by default (150, 00666 for instance). This is too low for Ruby to parse some files, such as 00667 date/format.rb, therefore bump the value up to at least Bison's default. */ 00668 #ifdef OLD_YACC 00669 #ifndef YYMAXDEPTH 00670 #define YYMAXDEPTH 10000 00671 #endif 00672 #endif 00673 00674 #ifndef RIPPER 00675 static void token_info_push(struct parser_params*, const char *token); 00676 static void token_info_pop(struct parser_params*, const char *token); 00677 #define token_info_push(token) (RTEST(ruby_verbose) ? token_info_push(parser, (token)) : (void)0) 00678 #define token_info_pop(token) (RTEST(ruby_verbose) ? token_info_pop(parser, (token)) : (void)0) 00679 #else 00680 #define token_info_push(token) /* nothing */ 00681 #define token_info_pop(token) /* nothing */ 00682 #endif 00683 %} 00684 00685 %pure-parser 00686 %lex-param {struct parser_params *parser} 00687 %parse-param {struct parser_params *parser} 00688 00689 %union { 00690 VALUE val; 00691 NODE *node; 00692 ID id; 00693 int num; 00694 const struct vtable *vars; 00695 } 00696 00697 /*%%%*/ 00698 %token 00699 /*% 00700 %token <val> 00701 %*/ 00702 keyword_class 00703 keyword_module 00704 keyword_def 00705 keyword_undef 00706 keyword_begin 00707 keyword_rescue 00708 keyword_ensure 00709 keyword_end 00710 keyword_if 00711 keyword_unless 00712 keyword_then 00713 keyword_elsif 00714 keyword_else 00715 keyword_case 00716 keyword_when 00717 keyword_while 00718 keyword_until 00719 keyword_for 00720 keyword_break 00721 keyword_next 00722 keyword_redo 00723 keyword_retry 00724 keyword_in 00725 keyword_do 00726 keyword_do_cond 00727 keyword_do_block 00728 keyword_do_LAMBDA 00729 keyword_return 00730 keyword_yield 00731 keyword_super 00732 keyword_self 00733 keyword_nil 00734 keyword_true 00735 keyword_false 00736 keyword_and 00737 keyword_or 00738 keyword_not 00739 modifier_if 00740 modifier_unless 00741 modifier_while 00742 modifier_until 00743 modifier_rescue 00744 keyword_alias 00745 keyword_defined 00746 keyword_BEGIN 00747 keyword_END 00748 keyword__LINE__ 00749 keyword__FILE__ 00750 keyword__ENCODING__ 00751 00752 %token <id> tIDENTIFIER tFID tGVAR tIVAR tCONSTANT tCVAR tLABEL 00753 %token <node> tINTEGER tFLOAT tSTRING_CONTENT tCHAR 00754 %token <node> tNTH_REF tBACK_REF 00755 %token <num> tREGEXP_END 00756 00757 %type <node> singleton strings string string1 xstring regexp 00758 %type <node> string_contents xstring_contents regexp_contents string_content 00759 %type <node> words symbols symbol_list qwords qsymbols word_list qword_list qsym_list word 00760 %type <node> literal numeric dsym cpath 00761 %type <node> top_compstmt top_stmts top_stmt 00762 %type <node> bodystmt compstmt stmts stmt_or_begin stmt expr arg primary command command_call method_call 00763 %type <node> expr_value arg_value primary_value fcall 00764 %type <node> if_tail opt_else case_body cases opt_rescue exc_list exc_var opt_ensure 00765 %type <node> args call_args opt_call_args 00766 %type <node> paren_args opt_paren_args args_tail opt_args_tail block_args_tail opt_block_args_tail 00767 %type <node> command_args aref_args opt_block_arg block_arg var_ref var_lhs 00768 %type <node> command_asgn mrhs superclass block_call block_command 00769 %type <node> f_block_optarg f_block_opt 00770 %type <node> f_arglist f_args f_arg f_arg_item f_optarg f_marg f_marg_list f_margs 00771 %type <node> assoc_list assocs assoc undef_list backref string_dvar for_var 00772 %type <node> block_param opt_block_param block_param_def f_opt 00773 %type <node> f_kwarg f_kw f_block_kwarg f_block_kw 00774 %type <node> bv_decls opt_bv_decl bvar 00775 %type <node> lambda f_larglist lambda_body 00776 %type <node> brace_block cmd_brace_block do_block lhs none fitem 00777 %type <node> mlhs mlhs_head mlhs_basic mlhs_item mlhs_node mlhs_post mlhs_inner 00778 %type <id> fsym keyword_variable user_variable sym symbol operation operation2 operation3 00779 %type <id> cname fname op f_rest_arg f_block_arg opt_f_block_arg f_norm_arg f_bad_arg 00780 %type <id> f_kwrest 00781 /*%%%*/ 00782 /*% 00783 %type <val> program reswords then do dot_or_colon 00784 %*/ 00785 %token END_OF_INPUT 0 "end-of-input" 00786 %token tUPLUS RUBY_TOKEN(UPLUS) "unary+" 00787 %token tUMINUS RUBY_TOKEN(UMINUS) "unary-" 00788 %token tPOW RUBY_TOKEN(POW) "**" 00789 %token tCMP RUBY_TOKEN(CMP) "<=>" 00790 %token tEQ RUBY_TOKEN(EQ) "==" 00791 %token tEQQ RUBY_TOKEN(EQQ) "===" 00792 %token tNEQ RUBY_TOKEN(NEQ) "!=" 00793 %token tGEQ RUBY_TOKEN(GEQ) ">=" 00794 %token tLEQ RUBY_TOKEN(LEQ) "<=" 00795 %token tANDOP "&&" 00796 %token tOROP "||" 00797 %token tMATCH RUBY_TOKEN(MATCH) "=~" 00798 %token tNMATCH RUBY_TOKEN(NMATCH) "!~" 00799 %token tDOT2 RUBY_TOKEN(DOT2) ".." 00800 %token tDOT3 RUBY_TOKEN(DOT3) "..." 00801 %token tAREF RUBY_TOKEN(AREF) "[]" 00802 %token tASET RUBY_TOKEN(ASET) "[]=" 00803 %token tLSHFT RUBY_TOKEN(LSHFT) "<<" 00804 %token tRSHFT RUBY_TOKEN(RSHFT) ">>" 00805 %token tCOLON2 "::" 00806 %token tCOLON3 ":: at EXPR_BEG" 00807 %token <id> tOP_ASGN /* +=, -= etc. */ 00808 %token tASSOC "=>" 00809 %token tLPAREN "(" 00810 %token tLPAREN_ARG "( arg" 00811 %token tRPAREN ")" 00812 %token tLBRACK "[" 00813 %token tLBRACE "{" 00814 %token tLBRACE_ARG "{ arg" 00815 %token tSTAR "*" 00816 %token tDSTAR "**arg" 00817 %token tAMPER "&" 00818 %token tLAMBDA "->" 00819 %token tSYMBEG tSTRING_BEG tXSTRING_BEG tREGEXP_BEG tWORDS_BEG tQWORDS_BEG tSYMBOLS_BEG tQSYMBOLS_BEG 00820 %token tSTRING_DBEG tSTRING_DEND tSTRING_DVAR tSTRING_END tLAMBEG 00821 00822 /* 00823 * precedence table 00824 */ 00825 00826 %nonassoc tLOWEST 00827 %nonassoc tLBRACE_ARG 00828 00829 %nonassoc modifier_if modifier_unless modifier_while modifier_until 00830 %left keyword_or keyword_and 00831 %right keyword_not 00832 %nonassoc keyword_defined 00833 %right '=' tOP_ASGN 00834 %left modifier_rescue 00835 %right '?' ':' 00836 %nonassoc tDOT2 tDOT3 00837 %left tOROP 00838 %left tANDOP 00839 %nonassoc tCMP tEQ tEQQ tNEQ tMATCH tNMATCH 00840 %left '>' tGEQ '<' tLEQ 00841 %left '|' '^' 00842 %left '&' 00843 %left tLSHFT tRSHFT 00844 %left '+' '-' 00845 %left '*' '/' '%' 00846 %right tUMINUS_NUM tUMINUS 00847 %right tPOW 00848 %right '!' '~' tUPLUS 00849 00850 %token tLAST_TOKEN 00851 00852 %% 00853 program : { 00854 lex_state = EXPR_BEG; 00855 /*%%%*/ 00856 local_push(compile_for_eval || rb_parse_in_main()); 00857 /*% 00858 local_push(0); 00859 %*/ 00860 } 00861 top_compstmt 00862 { 00863 /*%%%*/ 00864 if ($2 && !compile_for_eval) { 00865 /* last expression should not be void */ 00866 if (nd_type($2) != NODE_BLOCK) void_expr($2); 00867 else { 00868 NODE *node = $2; 00869 while (node->nd_next) { 00870 node = node->nd_next; 00871 } 00872 void_expr(node->nd_head); 00873 } 00874 } 00875 ruby_eval_tree = NEW_SCOPE(0, block_append(ruby_eval_tree, $2)); 00876 /*% 00877 $$ = $2; 00878 parser->result = dispatch1(program, $$); 00879 %*/ 00880 local_pop(); 00881 } 00882 ; 00883 00884 top_compstmt : top_stmts opt_terms 00885 { 00886 /*%%%*/ 00887 void_stmts($1); 00888 fixup_nodes(&deferred_nodes); 00889 /*% 00890 %*/ 00891 $$ = $1; 00892 } 00893 ; 00894 00895 top_stmts : none 00896 { 00897 /*%%%*/ 00898 $$ = NEW_BEGIN(0); 00899 /*% 00900 $$ = dispatch2(stmts_add, dispatch0(stmts_new), 00901 dispatch0(void_stmt)); 00902 %*/ 00903 } 00904 | top_stmt 00905 { 00906 /*%%%*/ 00907 $$ = newline_node($1); 00908 /*% 00909 $$ = dispatch2(stmts_add, dispatch0(stmts_new), $1); 00910 %*/ 00911 } 00912 | top_stmts terms top_stmt 00913 { 00914 /*%%%*/ 00915 $$ = block_append($1, newline_node($3)); 00916 /*% 00917 $$ = dispatch2(stmts_add, $1, $3); 00918 %*/ 00919 } 00920 | error top_stmt 00921 { 00922 $$ = remove_begin($2); 00923 } 00924 ; 00925 00926 top_stmt : stmt 00927 | keyword_BEGIN 00928 { 00929 /*%%%*/ 00930 /* local_push(0); */ 00931 /*% 00932 %*/ 00933 } 00934 '{' top_compstmt '}' 00935 { 00936 /*%%%*/ 00937 ruby_eval_tree_begin = block_append(ruby_eval_tree_begin, 00938 $4); 00939 /* NEW_PREEXE($4)); */ 00940 /* local_pop(); */ 00941 $$ = NEW_BEGIN(0); 00942 /*% 00943 $$ = dispatch1(BEGIN, $4); 00944 %*/ 00945 } 00946 ; 00947 00948 bodystmt : compstmt 00949 opt_rescue 00950 opt_else 00951 opt_ensure 00952 { 00953 /*%%%*/ 00954 $$ = $1; 00955 if ($2) { 00956 $$ = NEW_RESCUE($1, $2, $3); 00957 } 00958 else if ($3) { 00959 rb_warn0("else without rescue is useless"); 00960 $$ = block_append($$, $3); 00961 } 00962 if ($4) { 00963 if ($$) { 00964 $$ = NEW_ENSURE($$, $4); 00965 } 00966 else { 00967 $$ = block_append($4, NEW_NIL()); 00968 } 00969 } 00970 fixpos($$, $1); 00971 /*% 00972 $$ = dispatch4(bodystmt, 00973 escape_Qundef($1), 00974 escape_Qundef($2), 00975 escape_Qundef($3), 00976 escape_Qundef($4)); 00977 %*/ 00978 } 00979 ; 00980 00981 compstmt : stmts opt_terms 00982 { 00983 /*%%%*/ 00984 void_stmts($1); 00985 fixup_nodes(&deferred_nodes); 00986 /*% 00987 %*/ 00988 $$ = $1; 00989 } 00990 ; 00991 00992 stmts : none 00993 { 00994 /*%%%*/ 00995 $$ = NEW_BEGIN(0); 00996 /*% 00997 $$ = dispatch2(stmts_add, dispatch0(stmts_new), 00998 dispatch0(void_stmt)); 00999 %*/ 01000 } 01001 | stmt_or_begin 01002 { 01003 /*%%%*/ 01004 $$ = newline_node($1); 01005 /*% 01006 $$ = dispatch2(stmts_add, dispatch0(stmts_new), $1); 01007 %*/ 01008 } 01009 | stmts terms stmt_or_begin 01010 { 01011 /*%%%*/ 01012 $$ = block_append($1, newline_node($3)); 01013 /*% 01014 $$ = dispatch2(stmts_add, $1, $3); 01015 %*/ 01016 } 01017 | error stmt 01018 { 01019 $$ = remove_begin($2); 01020 } 01021 ; 01022 01023 stmt_or_begin : stmt 01024 { 01025 $$ = $1; 01026 } 01027 | keyword_BEGIN 01028 { 01029 yyerror("BEGIN is permitted only at toplevel"); 01030 /*%%%*/ 01031 /* local_push(0); */ 01032 /*% 01033 %*/ 01034 } 01035 '{' top_compstmt '}' 01036 { 01037 /*%%%*/ 01038 ruby_eval_tree_begin = block_append(ruby_eval_tree_begin, 01039 $4); 01040 /* NEW_PREEXE($4)); */ 01041 /* local_pop(); */ 01042 $$ = NEW_BEGIN(0); 01043 /*% 01044 $$ = dispatch1(BEGIN, $4); 01045 %*/ 01046 } 01047 01048 stmt : keyword_alias fitem {lex_state = EXPR_FNAME;} fitem 01049 { 01050 /*%%%*/ 01051 $$ = NEW_ALIAS($2, $4); 01052 /*% 01053 $$ = dispatch2(alias, $2, $4); 01054 %*/ 01055 } 01056 | keyword_alias tGVAR tGVAR 01057 { 01058 /*%%%*/ 01059 $$ = NEW_VALIAS($2, $3); 01060 /*% 01061 $$ = dispatch2(var_alias, $2, $3); 01062 %*/ 01063 } 01064 | keyword_alias tGVAR tBACK_REF 01065 { 01066 /*%%%*/ 01067 char buf[2]; 01068 buf[0] = '$'; 01069 buf[1] = (char)$3->nd_nth; 01070 $$ = NEW_VALIAS($2, rb_intern2(buf, 2)); 01071 /*% 01072 $$ = dispatch2(var_alias, $2, $3); 01073 %*/ 01074 } 01075 | keyword_alias tGVAR tNTH_REF 01076 { 01077 /*%%%*/ 01078 yyerror("can't make alias for the number variables"); 01079 $$ = NEW_BEGIN(0); 01080 /*% 01081 $$ = dispatch2(var_alias, $2, $3); 01082 $$ = dispatch1(alias_error, $$); 01083 %*/ 01084 } 01085 | keyword_undef undef_list 01086 { 01087 /*%%%*/ 01088 $$ = $2; 01089 /*% 01090 $$ = dispatch1(undef, $2); 01091 %*/ 01092 } 01093 | stmt modifier_if expr_value 01094 { 01095 /*%%%*/ 01096 $$ = NEW_IF(cond($3), remove_begin($1), 0); 01097 fixpos($$, $3); 01098 /*% 01099 $$ = dispatch2(if_mod, $3, $1); 01100 %*/ 01101 } 01102 | stmt modifier_unless expr_value 01103 { 01104 /*%%%*/ 01105 $$ = NEW_UNLESS(cond($3), remove_begin($1), 0); 01106 fixpos($$, $3); 01107 /*% 01108 $$ = dispatch2(unless_mod, $3, $1); 01109 %*/ 01110 } 01111 | stmt modifier_while expr_value 01112 { 01113 /*%%%*/ 01114 if ($1 && nd_type($1) == NODE_BEGIN) { 01115 $$ = NEW_WHILE(cond($3), $1->nd_body, 0); 01116 } 01117 else { 01118 $$ = NEW_WHILE(cond($3), $1, 1); 01119 } 01120 /*% 01121 $$ = dispatch2(while_mod, $3, $1); 01122 %*/ 01123 } 01124 | stmt modifier_until expr_value 01125 { 01126 /*%%%*/ 01127 if ($1 && nd_type($1) == NODE_BEGIN) { 01128 $$ = NEW_UNTIL(cond($3), $1->nd_body, 0); 01129 } 01130 else { 01131 $$ = NEW_UNTIL(cond($3), $1, 1); 01132 } 01133 /*% 01134 $$ = dispatch2(until_mod, $3, $1); 01135 %*/ 01136 } 01137 | stmt modifier_rescue stmt 01138 { 01139 /*%%%*/ 01140 NODE *resq = NEW_RESBODY(0, remove_begin($3), 0); 01141 $$ = NEW_RESCUE(remove_begin($1), resq, 0); 01142 /*% 01143 $$ = dispatch2(rescue_mod, $1, $3); 01144 %*/ 01145 } 01146 | keyword_END '{' compstmt '}' 01147 { 01148 if (in_def || in_single) { 01149 rb_warn0("END in method; use at_exit"); 01150 } 01151 /*%%%*/ 01152 $$ = NEW_POSTEXE(NEW_NODE( 01153 NODE_SCOPE, 0 /* tbl */, $3 /* body */, 0 /* args */)); 01154 /*% 01155 $$ = dispatch1(END, $3); 01156 %*/ 01157 } 01158 | command_asgn 01159 | mlhs '=' command_call 01160 { 01161 /*%%%*/ 01162 value_expr($3); 01163 $1->nd_value = $3; 01164 $$ = $1; 01165 /*% 01166 $$ = dispatch2(massign, $1, $3); 01167 %*/ 01168 } 01169 | var_lhs tOP_ASGN command_call 01170 { 01171 value_expr($3); 01172 $$ = new_op_assign($1, $2, $3); 01173 } 01174 | primary_value '[' opt_call_args rbracket tOP_ASGN command_call 01175 { 01176 /*%%%*/ 01177 NODE *args; 01178 01179 value_expr($6); 01180 if (!$3) $3 = NEW_ZARRAY(); 01181 args = arg_concat($3, $6); 01182 if ($5 == tOROP) { 01183 $5 = 0; 01184 } 01185 else if ($5 == tANDOP) { 01186 $5 = 1; 01187 } 01188 $$ = NEW_OP_ASGN1($1, $5, args); 01189 fixpos($$, $1); 01190 /*% 01191 $$ = dispatch2(aref_field, $1, escape_Qundef($3)); 01192 $$ = dispatch3(opassign, $$, $5, $6); 01193 %*/ 01194 } 01195 | primary_value '.' tIDENTIFIER tOP_ASGN command_call 01196 { 01197 value_expr($5); 01198 $$ = new_attr_op_assign($1, ripper_id2sym('.'), $3, $4, $5); 01199 } 01200 | primary_value '.' tCONSTANT tOP_ASGN command_call 01201 { 01202 value_expr($5); 01203 $$ = new_attr_op_assign($1, ripper_id2sym('.'), $3, $4, $5); 01204 } 01205 | primary_value tCOLON2 tCONSTANT tOP_ASGN command_call 01206 { 01207 /*%%%*/ 01208 $$ = NEW_COLON2($1, $3); 01209 $$ = new_const_op_assign($$, $4, $5); 01210 /*% 01211 $$ = dispatch2(const_path_field, $1, $3); 01212 $$ = dispatch3(opassign, $$, $4, $5); 01213 %*/ 01214 } 01215 | primary_value tCOLON2 tIDENTIFIER tOP_ASGN command_call 01216 { 01217 value_expr($5); 01218 $$ = new_attr_op_assign($1, ripper_intern("::"), $3, $4, $5); 01219 } 01220 | backref tOP_ASGN command_call 01221 { 01222 /*%%%*/ 01223 rb_backref_error($1); 01224 $$ = NEW_BEGIN(0); 01225 /*% 01226 $$ = dispatch2(assign, dispatch1(var_field, $1), $3); 01227 $$ = dispatch1(assign_error, $$); 01228 %*/ 01229 } 01230 | lhs '=' mrhs 01231 { 01232 /*%%%*/ 01233 value_expr($3); 01234 $$ = node_assign($1, $3); 01235 /*% 01236 $$ = dispatch2(assign, $1, $3); 01237 %*/ 01238 } 01239 | mlhs '=' arg_value 01240 { 01241 /*%%%*/ 01242 $1->nd_value = $3; 01243 $$ = $1; 01244 /*% 01245 $$ = dispatch2(massign, $1, $3); 01246 %*/ 01247 } 01248 | mlhs '=' mrhs 01249 { 01250 /*%%%*/ 01251 $1->nd_value = $3; 01252 $$ = $1; 01253 /*% 01254 $$ = dispatch2(massign, $1, $3); 01255 %*/ 01256 } 01257 | expr 01258 ; 01259 01260 command_asgn : lhs '=' command_call 01261 { 01262 /*%%%*/ 01263 value_expr($3); 01264 $$ = node_assign($1, $3); 01265 /*% 01266 $$ = dispatch2(assign, $1, $3); 01267 %*/ 01268 } 01269 | lhs '=' command_asgn 01270 { 01271 /*%%%*/ 01272 value_expr($3); 01273 $$ = node_assign($1, $3); 01274 /*% 01275 $$ = dispatch2(assign, $1, $3); 01276 %*/ 01277 } 01278 ; 01279 01280 01281 expr : command_call 01282 | expr keyword_and expr 01283 { 01284 /*%%%*/ 01285 $$ = logop(NODE_AND, $1, $3); 01286 /*% 01287 $$ = dispatch3(binary, $1, ripper_intern("and"), $3); 01288 %*/ 01289 } 01290 | expr keyword_or expr 01291 { 01292 /*%%%*/ 01293 $$ = logop(NODE_OR, $1, $3); 01294 /*% 01295 $$ = dispatch3(binary, $1, ripper_intern("or"), $3); 01296 %*/ 01297 } 01298 | keyword_not opt_nl expr 01299 { 01300 /*%%%*/ 01301 $$ = call_uni_op(cond($3), '!'); 01302 /*% 01303 $$ = dispatch2(unary, ripper_intern("not"), $3); 01304 %*/ 01305 } 01306 | '!' command_call 01307 { 01308 /*%%%*/ 01309 $$ = call_uni_op(cond($2), '!'); 01310 /*% 01311 $$ = dispatch2(unary, ripper_id2sym('!'), $2); 01312 %*/ 01313 } 01314 | arg 01315 ; 01316 01317 expr_value : expr 01318 { 01319 /*%%%*/ 01320 value_expr($1); 01321 $$ = $1; 01322 if (!$$) $$ = NEW_NIL(); 01323 /*% 01324 $$ = $1; 01325 %*/ 01326 } 01327 ; 01328 01329 command_call : command 01330 | block_command 01331 ; 01332 01333 block_command : block_call 01334 | block_call dot_or_colon operation2 command_args 01335 { 01336 /*%%%*/ 01337 $$ = NEW_CALL($1, $3, $4); 01338 /*% 01339 $$ = dispatch3(call, $1, $2, $3); 01340 $$ = method_arg($$, $4); 01341 %*/ 01342 } 01343 ; 01344 01345 cmd_brace_block : tLBRACE_ARG 01346 { 01347 $<vars>1 = dyna_push(); 01348 /*%%%*/ 01349 $<num>$ = ruby_sourceline; 01350 /*% 01351 %*/ 01352 } 01353 opt_block_param 01354 compstmt 01355 '}' 01356 { 01357 /*%%%*/ 01358 $$ = NEW_ITER($3,$4); 01359 nd_set_line($$, $<num>2); 01360 /*% 01361 $$ = dispatch2(brace_block, escape_Qundef($3), $4); 01362 %*/ 01363 dyna_pop($<vars>1); 01364 } 01365 ; 01366 01367 fcall : operation 01368 { 01369 /*%%%*/ 01370 $$ = NEW_FCALL($1, 0); 01371 nd_set_line($$, tokline); 01372 /*% 01373 %*/ 01374 } 01375 ; 01376 01377 command : fcall command_args %prec tLOWEST 01378 { 01379 /*%%%*/ 01380 $$ = $1; 01381 $$->nd_args = $2; 01382 /*% 01383 $$ = dispatch2(command, $1, $2); 01384 %*/ 01385 } 01386 | fcall command_args cmd_brace_block 01387 { 01388 /*%%%*/ 01389 block_dup_check($2,$3); 01390 $1->nd_args = $2; 01391 $3->nd_iter = $1; 01392 $$ = $3; 01393 fixpos($$, $1); 01394 /*% 01395 $$ = dispatch2(command, $1, $2); 01396 $$ = method_add_block($$, $3); 01397 %*/ 01398 } 01399 | primary_value '.' operation2 command_args %prec tLOWEST 01400 { 01401 /*%%%*/ 01402 $$ = NEW_CALL($1, $3, $4); 01403 fixpos($$, $1); 01404 /*% 01405 $$ = dispatch4(command_call, $1, ripper_id2sym('.'), $3, $4); 01406 %*/ 01407 } 01408 | primary_value '.' operation2 command_args cmd_brace_block 01409 { 01410 /*%%%*/ 01411 block_dup_check($4,$5); 01412 $5->nd_iter = NEW_CALL($1, $3, $4); 01413 $$ = $5; 01414 fixpos($$, $1); 01415 /*% 01416 $$ = dispatch4(command_call, $1, ripper_id2sym('.'), $3, $4); 01417 $$ = method_add_block($$, $5); 01418 %*/ 01419 } 01420 | primary_value tCOLON2 operation2 command_args %prec tLOWEST 01421 { 01422 /*%%%*/ 01423 $$ = NEW_CALL($1, $3, $4); 01424 fixpos($$, $1); 01425 /*% 01426 $$ = dispatch4(command_call, $1, ripper_intern("::"), $3, $4); 01427 %*/ 01428 } 01429 | primary_value tCOLON2 operation2 command_args cmd_brace_block 01430 { 01431 /*%%%*/ 01432 block_dup_check($4,$5); 01433 $5->nd_iter = NEW_CALL($1, $3, $4); 01434 $$ = $5; 01435 fixpos($$, $1); 01436 /*% 01437 $$ = dispatch4(command_call, $1, ripper_intern("::"), $3, $4); 01438 $$ = method_add_block($$, $5); 01439 %*/ 01440 } 01441 | keyword_super command_args 01442 { 01443 /*%%%*/ 01444 $$ = NEW_SUPER($2); 01445 fixpos($$, $2); 01446 /*% 01447 $$ = dispatch1(super, $2); 01448 %*/ 01449 } 01450 | keyword_yield command_args 01451 { 01452 /*%%%*/ 01453 $$ = new_yield($2); 01454 fixpos($$, $2); 01455 /*% 01456 $$ = dispatch1(yield, $2); 01457 %*/ 01458 } 01459 | keyword_return call_args 01460 { 01461 /*%%%*/ 01462 $$ = NEW_RETURN(ret_args($2)); 01463 /*% 01464 $$ = dispatch1(return, $2); 01465 %*/ 01466 } 01467 | keyword_break call_args 01468 { 01469 /*%%%*/ 01470 $$ = NEW_BREAK(ret_args($2)); 01471 /*% 01472 $$ = dispatch1(break, $2); 01473 %*/ 01474 } 01475 | keyword_next call_args 01476 { 01477 /*%%%*/ 01478 $$ = NEW_NEXT(ret_args($2)); 01479 /*% 01480 $$ = dispatch1(next, $2); 01481 %*/ 01482 } 01483 ; 01484 01485 mlhs : mlhs_basic 01486 | tLPAREN mlhs_inner rparen 01487 { 01488 /*%%%*/ 01489 $$ = $2; 01490 /*% 01491 $$ = dispatch1(mlhs_paren, $2); 01492 %*/ 01493 } 01494 ; 01495 01496 mlhs_inner : mlhs_basic 01497 | tLPAREN mlhs_inner rparen 01498 { 01499 /*%%%*/ 01500 $$ = NEW_MASGN(NEW_LIST($2), 0); 01501 /*% 01502 $$ = dispatch1(mlhs_paren, $2); 01503 %*/ 01504 } 01505 ; 01506 01507 mlhs_basic : mlhs_head 01508 { 01509 /*%%%*/ 01510 $$ = NEW_MASGN($1, 0); 01511 /*% 01512 $$ = $1; 01513 %*/ 01514 } 01515 | mlhs_head mlhs_item 01516 { 01517 /*%%%*/ 01518 $$ = NEW_MASGN(list_append($1,$2), 0); 01519 /*% 01520 $$ = mlhs_add($1, $2); 01521 %*/ 01522 } 01523 | mlhs_head tSTAR mlhs_node 01524 { 01525 /*%%%*/ 01526 $$ = NEW_MASGN($1, $3); 01527 /*% 01528 $$ = mlhs_add_star($1, $3); 01529 %*/ 01530 } 01531 | mlhs_head tSTAR mlhs_node ',' mlhs_post 01532 { 01533 /*%%%*/ 01534 $$ = NEW_MASGN($1, NEW_POSTARG($3,$5)); 01535 /*% 01536 $1 = mlhs_add_star($1, $3); 01537 $$ = mlhs_add($1, $5); 01538 %*/ 01539 } 01540 | mlhs_head tSTAR 01541 { 01542 /*%%%*/ 01543 $$ = NEW_MASGN($1, -1); 01544 /*% 01545 $$ = mlhs_add_star($1, Qnil); 01546 %*/ 01547 } 01548 | mlhs_head tSTAR ',' mlhs_post 01549 { 01550 /*%%%*/ 01551 $$ = NEW_MASGN($1, NEW_POSTARG(-1, $4)); 01552 /*% 01553 $1 = mlhs_add_star($1, Qnil); 01554 $$ = mlhs_add($1, $4); 01555 %*/ 01556 } 01557 | tSTAR mlhs_node 01558 { 01559 /*%%%*/ 01560 $$ = NEW_MASGN(0, $2); 01561 /*% 01562 $$ = mlhs_add_star(mlhs_new(), $2); 01563 %*/ 01564 } 01565 | tSTAR mlhs_node ',' mlhs_post 01566 { 01567 /*%%%*/ 01568 $$ = NEW_MASGN(0, NEW_POSTARG($2,$4)); 01569 /*% 01570 $2 = mlhs_add_star(mlhs_new(), $2); 01571 $$ = mlhs_add($2, $4); 01572 %*/ 01573 } 01574 | tSTAR 01575 { 01576 /*%%%*/ 01577 $$ = NEW_MASGN(0, -1); 01578 /*% 01579 $$ = mlhs_add_star(mlhs_new(), Qnil); 01580 %*/ 01581 } 01582 | tSTAR ',' mlhs_post 01583 { 01584 /*%%%*/ 01585 $$ = NEW_MASGN(0, NEW_POSTARG(-1, $3)); 01586 /*% 01587 $$ = mlhs_add_star(mlhs_new(), Qnil); 01588 $$ = mlhs_add($$, $3); 01589 %*/ 01590 } 01591 ; 01592 01593 mlhs_item : mlhs_node 01594 | tLPAREN mlhs_inner rparen 01595 { 01596 /*%%%*/ 01597 $$ = $2; 01598 /*% 01599 $$ = dispatch1(mlhs_paren, $2); 01600 %*/ 01601 } 01602 ; 01603 01604 mlhs_head : mlhs_item ',' 01605 { 01606 /*%%%*/ 01607 $$ = NEW_LIST($1); 01608 /*% 01609 $$ = mlhs_add(mlhs_new(), $1); 01610 %*/ 01611 } 01612 | mlhs_head mlhs_item ',' 01613 { 01614 /*%%%*/ 01615 $$ = list_append($1, $2); 01616 /*% 01617 $$ = mlhs_add($1, $2); 01618 %*/ 01619 } 01620 ; 01621 01622 mlhs_post : mlhs_item 01623 { 01624 /*%%%*/ 01625 $$ = NEW_LIST($1); 01626 /*% 01627 $$ = mlhs_add(mlhs_new(), $1); 01628 %*/ 01629 } 01630 | mlhs_post ',' mlhs_item 01631 { 01632 /*%%%*/ 01633 $$ = list_append($1, $3); 01634 /*% 01635 $$ = mlhs_add($1, $3); 01636 %*/ 01637 } 01638 ; 01639 01640 mlhs_node : user_variable 01641 { 01642 $$ = assignable($1, 0); 01643 } 01644 | keyword_variable 01645 { 01646 $$ = assignable($1, 0); 01647 } 01648 | primary_value '[' opt_call_args rbracket 01649 { 01650 /*%%%*/ 01651 $$ = aryset($1, $3); 01652 /*% 01653 $$ = dispatch2(aref_field, $1, escape_Qundef($3)); 01654 %*/ 01655 } 01656 | primary_value '.' tIDENTIFIER 01657 { 01658 /*%%%*/ 01659 $$ = attrset($1, $3); 01660 /*% 01661 $$ = dispatch3(field, $1, ripper_id2sym('.'), $3); 01662 %*/ 01663 } 01664 | primary_value tCOLON2 tIDENTIFIER 01665 { 01666 /*%%%*/ 01667 $$ = attrset($1, $3); 01668 /*% 01669 $$ = dispatch2(const_path_field, $1, $3); 01670 %*/ 01671 } 01672 | primary_value '.' tCONSTANT 01673 { 01674 /*%%%*/ 01675 $$ = attrset($1, $3); 01676 /*% 01677 $$ = dispatch3(field, $1, ripper_id2sym('.'), $3); 01678 %*/ 01679 } 01680 | primary_value tCOLON2 tCONSTANT 01681 { 01682 /*%%%*/ 01683 if (in_def || in_single) 01684 yyerror("dynamic constant assignment"); 01685 $$ = NEW_CDECL(0, 0, NEW_COLON2($1, $3)); 01686 /*% 01687 if (in_def || in_single) 01688 yyerror("dynamic constant assignment"); 01689 $$ = dispatch2(const_path_field, $1, $3); 01690 %*/ 01691 } 01692 | tCOLON3 tCONSTANT 01693 { 01694 /*%%%*/ 01695 if (in_def || in_single) 01696 yyerror("dynamic constant assignment"); 01697 $$ = NEW_CDECL(0, 0, NEW_COLON3($2)); 01698 /*% 01699 $$ = dispatch1(top_const_field, $2); 01700 %*/ 01701 } 01702 | backref 01703 { 01704 /*%%%*/ 01705 rb_backref_error($1); 01706 $$ = NEW_BEGIN(0); 01707 /*% 01708 $$ = dispatch1(var_field, $1); 01709 $$ = dispatch1(assign_error, $$); 01710 %*/ 01711 } 01712 ; 01713 01714 lhs : user_variable 01715 { 01716 $$ = assignable($1, 0); 01717 /*%%%*/ 01718 if (!$$) $$ = NEW_BEGIN(0); 01719 /*% 01720 $$ = dispatch1(var_field, $$); 01721 %*/ 01722 } 01723 | keyword_variable 01724 { 01725 $$ = assignable($1, 0); 01726 /*%%%*/ 01727 if (!$$) $$ = NEW_BEGIN(0); 01728 /*% 01729 $$ = dispatch1(var_field, $$); 01730 %*/ 01731 } 01732 | primary_value '[' opt_call_args rbracket 01733 { 01734 /*%%%*/ 01735 $$ = aryset($1, $3); 01736 /*% 01737 $$ = dispatch2(aref_field, $1, escape_Qundef($3)); 01738 %*/ 01739 } 01740 | primary_value '.' tIDENTIFIER 01741 { 01742 /*%%%*/ 01743 $$ = attrset($1, $3); 01744 /*% 01745 $$ = dispatch3(field, $1, ripper_id2sym('.'), $3); 01746 %*/ 01747 } 01748 | primary_value tCOLON2 tIDENTIFIER 01749 { 01750 /*%%%*/ 01751 $$ = attrset($1, $3); 01752 /*% 01753 $$ = dispatch3(field, $1, ripper_intern("::"), $3); 01754 %*/ 01755 } 01756 | primary_value '.' tCONSTANT 01757 { 01758 /*%%%*/ 01759 $$ = attrset($1, $3); 01760 /*% 01761 $$ = dispatch3(field, $1, ripper_id2sym('.'), $3); 01762 %*/ 01763 } 01764 | primary_value tCOLON2 tCONSTANT 01765 { 01766 /*%%%*/ 01767 if (in_def || in_single) 01768 yyerror("dynamic constant assignment"); 01769 $$ = NEW_CDECL(0, 0, NEW_COLON2($1, $3)); 01770 /*% 01771 $$ = dispatch2(const_path_field, $1, $3); 01772 if (in_def || in_single) { 01773 $$ = dispatch1(assign_error, $$); 01774 } 01775 %*/ 01776 } 01777 | tCOLON3 tCONSTANT 01778 { 01779 /*%%%*/ 01780 if (in_def || in_single) 01781 yyerror("dynamic constant assignment"); 01782 $$ = NEW_CDECL(0, 0, NEW_COLON3($2)); 01783 /*% 01784 $$ = dispatch1(top_const_field, $2); 01785 if (in_def || in_single) { 01786 $$ = dispatch1(assign_error, $$); 01787 } 01788 %*/ 01789 } 01790 | backref 01791 { 01792 /*%%%*/ 01793 rb_backref_error($1); 01794 $$ = NEW_BEGIN(0); 01795 /*% 01796 $$ = dispatch1(assign_error, $1); 01797 %*/ 01798 } 01799 ; 01800 01801 cname : tIDENTIFIER 01802 { 01803 /*%%%*/ 01804 yyerror("class/module name must be CONSTANT"); 01805 /*% 01806 $$ = dispatch1(class_name_error, $1); 01807 %*/ 01808 } 01809 | tCONSTANT 01810 ; 01811 01812 cpath : tCOLON3 cname 01813 { 01814 /*%%%*/ 01815 $$ = NEW_COLON3($2); 01816 /*% 01817 $$ = dispatch1(top_const_ref, $2); 01818 %*/ 01819 } 01820 | cname 01821 { 01822 /*%%%*/ 01823 $$ = NEW_COLON2(0, $$); 01824 /*% 01825 $$ = dispatch1(const_ref, $1); 01826 %*/ 01827 } 01828 | primary_value tCOLON2 cname 01829 { 01830 /*%%%*/ 01831 $$ = NEW_COLON2($1, $3); 01832 /*% 01833 $$ = dispatch2(const_path_ref, $1, $3); 01834 %*/ 01835 } 01836 ; 01837 01838 fname : tIDENTIFIER 01839 | tCONSTANT 01840 | tFID 01841 | op 01842 { 01843 lex_state = EXPR_ENDFN; 01844 $$ = $1; 01845 } 01846 | reswords 01847 { 01848 lex_state = EXPR_ENDFN; 01849 /*%%%*/ 01850 $$ = $<id>1; 01851 /*% 01852 $$ = $1; 01853 %*/ 01854 } 01855 ; 01856 01857 fsym : fname 01858 | symbol 01859 ; 01860 01861 fitem : fsym 01862 { 01863 /*%%%*/ 01864 $$ = NEW_LIT(ID2SYM($1)); 01865 /*% 01866 $$ = dispatch1(symbol_literal, $1); 01867 %*/ 01868 } 01869 | dsym 01870 ; 01871 01872 undef_list : fitem 01873 { 01874 /*%%%*/ 01875 $$ = NEW_UNDEF($1); 01876 /*% 01877 $$ = rb_ary_new3(1, $1); 01878 %*/ 01879 } 01880 | undef_list ',' {lex_state = EXPR_FNAME;} fitem 01881 { 01882 /*%%%*/ 01883 $$ = block_append($1, NEW_UNDEF($4)); 01884 /*% 01885 rb_ary_push($1, $4); 01886 %*/ 01887 } 01888 ; 01889 01890 op : '|' { ifndef_ripper($$ = '|'); } 01891 | '^' { ifndef_ripper($$ = '^'); } 01892 | '&' { ifndef_ripper($$ = '&'); } 01893 | tCMP { ifndef_ripper($$ = tCMP); } 01894 | tEQ { ifndef_ripper($$ = tEQ); } 01895 | tEQQ { ifndef_ripper($$ = tEQQ); } 01896 | tMATCH { ifndef_ripper($$ = tMATCH); } 01897 | tNMATCH { ifndef_ripper($$ = tNMATCH); } 01898 | '>' { ifndef_ripper($$ = '>'); } 01899 | tGEQ { ifndef_ripper($$ = tGEQ); } 01900 | '<' { ifndef_ripper($$ = '<'); } 01901 | tLEQ { ifndef_ripper($$ = tLEQ); } 01902 | tNEQ { ifndef_ripper($$ = tNEQ); } 01903 | tLSHFT { ifndef_ripper($$ = tLSHFT); } 01904 | tRSHFT { ifndef_ripper($$ = tRSHFT); } 01905 | '+' { ifndef_ripper($$ = '+'); } 01906 | '-' { ifndef_ripper($$ = '-'); } 01907 | '*' { ifndef_ripper($$ = '*'); } 01908 | tSTAR { ifndef_ripper($$ = '*'); } 01909 | '/' { ifndef_ripper($$ = '/'); } 01910 | '%' { ifndef_ripper($$ = '%'); } 01911 | tPOW { ifndef_ripper($$ = tPOW); } 01912 | tDSTAR { ifndef_ripper($$ = tDSTAR); } 01913 | '!' { ifndef_ripper($$ = '!'); } 01914 | '~' { ifndef_ripper($$ = '~'); } 01915 | tUPLUS { ifndef_ripper($$ = tUPLUS); } 01916 | tUMINUS { ifndef_ripper($$ = tUMINUS); } 01917 | tAREF { ifndef_ripper($$ = tAREF); } 01918 | tASET { ifndef_ripper($$ = tASET); } 01919 | '`' { ifndef_ripper($$ = '`'); } 01920 ; 01921 01922 reswords : keyword__LINE__ | keyword__FILE__ | keyword__ENCODING__ 01923 | keyword_BEGIN | keyword_END 01924 | keyword_alias | keyword_and | keyword_begin 01925 | keyword_break | keyword_case | keyword_class | keyword_def 01926 | keyword_defined | keyword_do | keyword_else | keyword_elsif 01927 | keyword_end | keyword_ensure | keyword_false 01928 | keyword_for | keyword_in | keyword_module | keyword_next 01929 | keyword_nil | keyword_not | keyword_or | keyword_redo 01930 | keyword_rescue | keyword_retry | keyword_return | keyword_self 01931 | keyword_super | keyword_then | keyword_true | keyword_undef 01932 | keyword_when | keyword_yield | keyword_if | keyword_unless 01933 | keyword_while | keyword_until 01934 ; 01935 01936 arg : lhs '=' arg 01937 { 01938 /*%%%*/ 01939 value_expr($3); 01940 $$ = node_assign($1, $3); 01941 /*% 01942 $$ = dispatch2(assign, $1, $3); 01943 %*/ 01944 } 01945 | lhs '=' arg modifier_rescue arg 01946 { 01947 /*%%%*/ 01948 value_expr($3); 01949 $3 = NEW_RESCUE($3, NEW_RESBODY(0,$5,0), 0); 01950 $$ = node_assign($1, $3); 01951 /*% 01952 $$ = dispatch2(assign, $1, dispatch2(rescue_mod, $3, $5)); 01953 %*/ 01954 } 01955 | var_lhs tOP_ASGN arg 01956 { 01957 value_expr($3); 01958 $$ = new_op_assign($1, $2, $3); 01959 } 01960 | var_lhs tOP_ASGN arg modifier_rescue arg 01961 { 01962 /*%%%*/ 01963 value_expr($3); 01964 $3 = NEW_RESCUE($3, NEW_RESBODY(0,$5,0), 0); 01965 /*% 01966 $3 = dispatch2(rescue_mod, $3, $5); 01967 %*/ 01968 $$ = new_op_assign($1, $2, $3); 01969 } 01970 | primary_value '[' opt_call_args rbracket tOP_ASGN arg 01971 { 01972 /*%%%*/ 01973 NODE *args; 01974 01975 value_expr($6); 01976 if (!$3) $3 = NEW_ZARRAY(); 01977 if (nd_type($3) == NODE_BLOCK_PASS) { 01978 args = NEW_ARGSCAT($3, $6); 01979 } 01980 else { 01981 args = arg_concat($3, $6); 01982 } 01983 if ($5 == tOROP) { 01984 $5 = 0; 01985 } 01986 else if ($5 == tANDOP) { 01987 $5 = 1; 01988 } 01989 $$ = NEW_OP_ASGN1($1, $5, args); 01990 fixpos($$, $1); 01991 /*% 01992 $1 = dispatch2(aref_field, $1, escape_Qundef($3)); 01993 $$ = dispatch3(opassign, $1, $5, $6); 01994 %*/ 01995 } 01996 | primary_value '.' tIDENTIFIER tOP_ASGN arg 01997 { 01998 value_expr($5); 01999 $$ = new_attr_op_assign($1, ripper_id2sym('.'), $3, $4, $5); 02000 } 02001 | primary_value '.' tCONSTANT tOP_ASGN arg 02002 { 02003 value_expr($5); 02004 $$ = new_attr_op_assign($1, ripper_id2sym('.'), $3, $4, $5); 02005 } 02006 | primary_value tCOLON2 tIDENTIFIER tOP_ASGN arg 02007 { 02008 value_expr($5); 02009 $$ = new_attr_op_assign($1, ripper_intern("::"), $3, $4, $5); 02010 } 02011 | primary_value tCOLON2 tCONSTANT tOP_ASGN arg 02012 { 02013 /*%%%*/ 02014 $$ = NEW_COLON2($1, $3); 02015 $$ = new_const_op_assign($$, $4, $5); 02016 /*% 02017 $$ = dispatch2(const_path_field, $1, $3); 02018 $$ = dispatch3(opassign, $$, $4, $5); 02019 %*/ 02020 } 02021 | tCOLON3 tCONSTANT tOP_ASGN arg 02022 { 02023 /*%%%*/ 02024 $$ = NEW_COLON3($2); 02025 $$ = new_const_op_assign($$, $3, $4); 02026 /*% 02027 $$ = dispatch1(top_const_field, $2); 02028 $$ = dispatch3(opassign, $$, $3, $4); 02029 %*/ 02030 } 02031 | backref tOP_ASGN arg 02032 { 02033 /*%%%*/ 02034 rb_backref_error($1); 02035 $$ = NEW_BEGIN(0); 02036 /*% 02037 $$ = dispatch1(var_field, $1); 02038 $$ = dispatch3(opassign, $$, $2, $3); 02039 $$ = dispatch1(assign_error, $$); 02040 %*/ 02041 } 02042 | arg tDOT2 arg 02043 { 02044 /*%%%*/ 02045 value_expr($1); 02046 value_expr($3); 02047 $$ = NEW_DOT2($1, $3); 02048 if (nd_type($1) == NODE_LIT && FIXNUM_P($1->nd_lit) && 02049 nd_type($3) == NODE_LIT && FIXNUM_P($3->nd_lit)) { 02050 deferred_nodes = list_append(deferred_nodes, $$); 02051 } 02052 /*% 02053 $$ = dispatch2(dot2, $1, $3); 02054 %*/ 02055 } 02056 | arg tDOT3 arg 02057 { 02058 /*%%%*/ 02059 value_expr($1); 02060 value_expr($3); 02061 $$ = NEW_DOT3($1, $3); 02062 if (nd_type($1) == NODE_LIT && FIXNUM_P($1->nd_lit) && 02063 nd_type($3) == NODE_LIT && FIXNUM_P($3->nd_lit)) { 02064 deferred_nodes = list_append(deferred_nodes, $$); 02065 } 02066 /*% 02067 $$ = dispatch2(dot3, $1, $3); 02068 %*/ 02069 } 02070 | arg '+' arg 02071 { 02072 /*%%%*/ 02073 $$ = call_bin_op($1, '+', $3); 02074 /*% 02075 $$ = dispatch3(binary, $1, ID2SYM('+'), $3); 02076 %*/ 02077 } 02078 | arg '-' arg 02079 { 02080 /*%%%*/ 02081 $$ = call_bin_op($1, '-', $3); 02082 /*% 02083 $$ = dispatch3(binary, $1, ID2SYM('-'), $3); 02084 %*/ 02085 } 02086 | arg '*' arg 02087 { 02088 /*%%%*/ 02089 $$ = call_bin_op($1, '*', $3); 02090 /*% 02091 $$ = dispatch3(binary, $1, ID2SYM('*'), $3); 02092 %*/ 02093 } 02094 | arg '/' arg 02095 { 02096 /*%%%*/ 02097 $$ = call_bin_op($1, '/', $3); 02098 /*% 02099 $$ = dispatch3(binary, $1, ID2SYM('/'), $3); 02100 %*/ 02101 } 02102 | arg '%' arg 02103 { 02104 /*%%%*/ 02105 $$ = call_bin_op($1, '%', $3); 02106 /*% 02107 $$ = dispatch3(binary, $1, ID2SYM('%'), $3); 02108 %*/ 02109 } 02110 | arg tPOW arg 02111 { 02112 /*%%%*/ 02113 $$ = call_bin_op($1, tPOW, $3); 02114 /*% 02115 $$ = dispatch3(binary, $1, ripper_intern("**"), $3); 02116 %*/ 02117 } 02118 | tUMINUS_NUM tINTEGER tPOW arg 02119 { 02120 /*%%%*/ 02121 $$ = NEW_CALL(call_bin_op($2, tPOW, $4), tUMINUS, 0); 02122 /*% 02123 $$ = dispatch3(binary, $2, ripper_intern("**"), $4); 02124 $$ = dispatch2(unary, ripper_intern("-@"), $$); 02125 %*/ 02126 } 02127 | tUMINUS_NUM tFLOAT tPOW arg 02128 { 02129 /*%%%*/ 02130 $$ = NEW_CALL(call_bin_op($2, tPOW, $4), tUMINUS, 0); 02131 /*% 02132 $$ = dispatch3(binary, $2, ripper_intern("**"), $4); 02133 $$ = dispatch2(unary, ripper_intern("-@"), $$); 02134 %*/ 02135 } 02136 | tUPLUS arg 02137 { 02138 /*%%%*/ 02139 $$ = call_uni_op($2, tUPLUS); 02140 /*% 02141 $$ = dispatch2(unary, ripper_intern("+@"), $2); 02142 %*/ 02143 } 02144 | tUMINUS arg 02145 { 02146 /*%%%*/ 02147 $$ = call_uni_op($2, tUMINUS); 02148 /*% 02149 $$ = dispatch2(unary, ripper_intern("-@"), $2); 02150 %*/ 02151 } 02152 | arg '|' arg 02153 { 02154 /*%%%*/ 02155 $$ = call_bin_op($1, '|', $3); 02156 /*% 02157 $$ = dispatch3(binary, $1, ID2SYM('|'), $3); 02158 %*/ 02159 } 02160 | arg '^' arg 02161 { 02162 /*%%%*/ 02163 $$ = call_bin_op($1, '^', $3); 02164 /*% 02165 $$ = dispatch3(binary, $1, ID2SYM('^'), $3); 02166 %*/ 02167 } 02168 | arg '&' arg 02169 { 02170 /*%%%*/ 02171 $$ = call_bin_op($1, '&', $3); 02172 /*% 02173 $$ = dispatch3(binary, $1, ID2SYM('&'), $3); 02174 %*/ 02175 } 02176 | arg tCMP arg 02177 { 02178 /*%%%*/ 02179 $$ = call_bin_op($1, tCMP, $3); 02180 /*% 02181 $$ = dispatch3(binary, $1, ripper_intern("<=>"), $3); 02182 %*/ 02183 } 02184 | arg '>' arg 02185 { 02186 /*%%%*/ 02187 $$ = call_bin_op($1, '>', $3); 02188 /*% 02189 $$ = dispatch3(binary, $1, ID2SYM('>'), $3); 02190 %*/ 02191 } 02192 | arg tGEQ arg 02193 { 02194 /*%%%*/ 02195 $$ = call_bin_op($1, tGEQ, $3); 02196 /*% 02197 $$ = dispatch3(binary, $1, ripper_intern(">="), $3); 02198 %*/ 02199 } 02200 | arg '<' arg 02201 { 02202 /*%%%*/ 02203 $$ = call_bin_op($1, '<', $3); 02204 /*% 02205 $$ = dispatch3(binary, $1, ID2SYM('<'), $3); 02206 %*/ 02207 } 02208 | arg tLEQ arg 02209 { 02210 /*%%%*/ 02211 $$ = call_bin_op($1, tLEQ, $3); 02212 /*% 02213 $$ = dispatch3(binary, $1, ripper_intern("<="), $3); 02214 %*/ 02215 } 02216 | arg tEQ arg 02217 { 02218 /*%%%*/ 02219 $$ = call_bin_op($1, tEQ, $3); 02220 /*% 02221 $$ = dispatch3(binary, $1, ripper_intern("=="), $3); 02222 %*/ 02223 } 02224 | arg tEQQ arg 02225 { 02226 /*%%%*/ 02227 $$ = call_bin_op($1, tEQQ, $3); 02228 /*% 02229 $$ = dispatch3(binary, $1, ripper_intern("==="), $3); 02230 %*/ 02231 } 02232 | arg tNEQ arg 02233 { 02234 /*%%%*/ 02235 $$ = call_bin_op($1, tNEQ, $3); 02236 /*% 02237 $$ = dispatch3(binary, $1, ripper_intern("!="), $3); 02238 %*/ 02239 } 02240 | arg tMATCH arg 02241 { 02242 /*%%%*/ 02243 $$ = match_op($1, $3); 02244 if (nd_type($1) == NODE_LIT && RB_TYPE_P($1->nd_lit, T_REGEXP)) { 02245 $$ = reg_named_capture_assign($1->nd_lit, $$); 02246 } 02247 /*% 02248 $$ = dispatch3(binary, $1, ripper_intern("=~"), $3); 02249 %*/ 02250 } 02251 | arg tNMATCH arg 02252 { 02253 /*%%%*/ 02254 $$ = call_bin_op($1, tNMATCH, $3); 02255 /*% 02256 $$ = dispatch3(binary, $1, ripper_intern("!~"), $3); 02257 %*/ 02258 } 02259 | '!' arg 02260 { 02261 /*%%%*/ 02262 $$ = call_uni_op(cond($2), '!'); 02263 /*% 02264 $$ = dispatch2(unary, ID2SYM('!'), $2); 02265 %*/ 02266 } 02267 | '~' arg 02268 { 02269 /*%%%*/ 02270 $$ = call_uni_op($2, '~'); 02271 /*% 02272 $$ = dispatch2(unary, ID2SYM('~'), $2); 02273 %*/ 02274 } 02275 | arg tLSHFT arg 02276 { 02277 /*%%%*/ 02278 $$ = call_bin_op($1, tLSHFT, $3); 02279 /*% 02280 $$ = dispatch3(binary, $1, ripper_intern("<<"), $3); 02281 %*/ 02282 } 02283 | arg tRSHFT arg 02284 { 02285 /*%%%*/ 02286 $$ = call_bin_op($1, tRSHFT, $3); 02287 /*% 02288 $$ = dispatch3(binary, $1, ripper_intern(">>"), $3); 02289 %*/ 02290 } 02291 | arg tANDOP arg 02292 { 02293 /*%%%*/ 02294 $$ = logop(NODE_AND, $1, $3); 02295 /*% 02296 $$ = dispatch3(binary, $1, ripper_intern("&&"), $3); 02297 %*/ 02298 } 02299 | arg tOROP arg 02300 { 02301 /*%%%*/ 02302 $$ = logop(NODE_OR, $1, $3); 02303 /*% 02304 $$ = dispatch3(binary, $1, ripper_intern("||"), $3); 02305 %*/ 02306 } 02307 | keyword_defined opt_nl {in_defined = 1;} arg 02308 { 02309 /*%%%*/ 02310 in_defined = 0; 02311 $$ = NEW_DEFINED($4); 02312 /*% 02313 in_defined = 0; 02314 $$ = dispatch1(defined, $4); 02315 %*/ 02316 } 02317 | arg '?' arg opt_nl ':' arg 02318 { 02319 /*%%%*/ 02320 value_expr($1); 02321 $$ = NEW_IF(cond($1), $3, $6); 02322 fixpos($$, $1); 02323 /*% 02324 $$ = dispatch3(ifop, $1, $3, $6); 02325 %*/ 02326 } 02327 | primary 02328 { 02329 $$ = $1; 02330 } 02331 ; 02332 02333 arg_value : arg 02334 { 02335 /*%%%*/ 02336 value_expr($1); 02337 $$ = $1; 02338 if (!$$) $$ = NEW_NIL(); 02339 /*% 02340 $$ = $1; 02341 %*/ 02342 } 02343 ; 02344 02345 aref_args : none 02346 | args trailer 02347 { 02348 $$ = $1; 02349 } 02350 | args ',' assocs trailer 02351 { 02352 /*%%%*/ 02353 $$ = arg_append($1, NEW_HASH($3)); 02354 /*% 02355 $$ = arg_add_assocs($1, $3); 02356 %*/ 02357 } 02358 | assocs trailer 02359 { 02360 /*%%%*/ 02361 $$ = NEW_LIST(NEW_HASH($1)); 02362 /*% 02363 $$ = arg_add_assocs(arg_new(), $1); 02364 %*/ 02365 } 02366 ; 02367 02368 paren_args : '(' opt_call_args rparen 02369 { 02370 /*%%%*/ 02371 $$ = $2; 02372 /*% 02373 $$ = dispatch1(arg_paren, escape_Qundef($2)); 02374 %*/ 02375 } 02376 ; 02377 02378 opt_paren_args : none 02379 | paren_args 02380 ; 02381 02382 opt_call_args : none 02383 | call_args 02384 | args ',' 02385 { 02386 $$ = $1; 02387 } 02388 | args ',' assocs ',' 02389 { 02390 /*%%%*/ 02391 $$ = arg_append($1, NEW_HASH($3)); 02392 /*% 02393 $$ = arg_add_assocs($1, $3); 02394 %*/ 02395 } 02396 | assocs ',' 02397 { 02398 /*%%%*/ 02399 $$ = NEW_LIST(NEW_HASH($1)); 02400 /*% 02401 $$ = arg_add_assocs(arg_new(), $1); 02402 %*/ 02403 } 02404 ; 02405 02406 call_args : command 02407 { 02408 /*%%%*/ 02409 value_expr($1); 02410 $$ = NEW_LIST($1); 02411 /*% 02412 $$ = arg_add(arg_new(), $1); 02413 %*/ 02414 } 02415 | args opt_block_arg 02416 { 02417 /*%%%*/ 02418 $$ = arg_blk_pass($1, $2); 02419 /*% 02420 $$ = arg_add_optblock($1, $2); 02421 %*/ 02422 } 02423 | assocs opt_block_arg 02424 { 02425 /*%%%*/ 02426 $$ = NEW_LIST(NEW_HASH($1)); 02427 $$ = arg_blk_pass($$, $2); 02428 /*% 02429 $$ = arg_add_assocs(arg_new(), $1); 02430 $$ = arg_add_optblock($$, $2); 02431 %*/ 02432 } 02433 | args ',' assocs opt_block_arg 02434 { 02435 /*%%%*/ 02436 $$ = arg_append($1, NEW_HASH($3)); 02437 $$ = arg_blk_pass($$, $4); 02438 /*% 02439 $$ = arg_add_optblock(arg_add_assocs($1, $3), $4); 02440 %*/ 02441 } 02442 | block_arg 02443 /*%c%*/ 02444 /*%c 02445 { 02446 $$ = arg_add_block(arg_new(), $1); 02447 } 02448 %*/ 02449 ; 02450 02451 command_args : { 02452 $<val>$ = cmdarg_stack; 02453 CMDARG_PUSH(1); 02454 } 02455 call_args 02456 { 02457 /* CMDARG_POP() */ 02458 cmdarg_stack = $<val>1; 02459 $$ = $2; 02460 } 02461 ; 02462 02463 block_arg : tAMPER arg_value 02464 { 02465 /*%%%*/ 02466 $$ = NEW_BLOCK_PASS($2); 02467 /*% 02468 $$ = $2; 02469 %*/ 02470 } 02471 ; 02472 02473 opt_block_arg : ',' block_arg 02474 { 02475 $$ = $2; 02476 } 02477 | none 02478 { 02479 $$ = 0; 02480 } 02481 ; 02482 02483 args : arg_value 02484 { 02485 /*%%%*/ 02486 $$ = NEW_LIST($1); 02487 /*% 02488 $$ = arg_add(arg_new(), $1); 02489 %*/ 02490 } 02491 | tSTAR arg_value 02492 { 02493 /*%%%*/ 02494 $$ = NEW_SPLAT($2); 02495 /*% 02496 $$ = arg_add_star(arg_new(), $2); 02497 %*/ 02498 } 02499 | args ',' arg_value 02500 { 02501 /*%%%*/ 02502 NODE *n1; 02503 if ((n1 = splat_array($1)) != 0) { 02504 $$ = list_append(n1, $3); 02505 } 02506 else { 02507 $$ = arg_append($1, $3); 02508 } 02509 /*% 02510 $$ = arg_add($1, $3); 02511 %*/ 02512 } 02513 | args ',' tSTAR arg_value 02514 { 02515 /*%%%*/ 02516 NODE *n1; 02517 if ((nd_type($4) == NODE_ARRAY) && (n1 = splat_array($1)) != 0) { 02518 $$ = list_concat(n1, $4); 02519 } 02520 else { 02521 $$ = arg_concat($1, $4); 02522 } 02523 /*% 02524 $$ = arg_add_star($1, $4); 02525 %*/ 02526 } 02527 ; 02528 02529 mrhs : args ',' arg_value 02530 { 02531 /*%%%*/ 02532 NODE *n1; 02533 if ((n1 = splat_array($1)) != 0) { 02534 $$ = list_append(n1, $3); 02535 } 02536 else { 02537 $$ = arg_append($1, $3); 02538 } 02539 /*% 02540 $$ = mrhs_add(args2mrhs($1), $3); 02541 %*/ 02542 } 02543 | args ',' tSTAR arg_value 02544 { 02545 /*%%%*/ 02546 NODE *n1; 02547 if (nd_type($4) == NODE_ARRAY && 02548 (n1 = splat_array($1)) != 0) { 02549 $$ = list_concat(n1, $4); 02550 } 02551 else { 02552 $$ = arg_concat($1, $4); 02553 } 02554 /*% 02555 $$ = mrhs_add_star(args2mrhs($1), $4); 02556 %*/ 02557 } 02558 | tSTAR arg_value 02559 { 02560 /*%%%*/ 02561 $$ = NEW_SPLAT($2); 02562 /*% 02563 $$ = mrhs_add_star(mrhs_new(), $2); 02564 %*/ 02565 } 02566 ; 02567 02568 primary : literal 02569 | strings 02570 | xstring 02571 | regexp 02572 | words 02573 | qwords 02574 | symbols 02575 | qsymbols 02576 | var_ref 02577 | backref 02578 | tFID 02579 { 02580 /*%%%*/ 02581 $$ = NEW_FCALL($1, 0); 02582 /*% 02583 $$ = method_arg(dispatch1(fcall, $1), arg_new()); 02584 %*/ 02585 } 02586 | k_begin 02587 { 02588 $<val>1 = cmdarg_stack; 02589 cmdarg_stack = 0; 02590 /*%%%*/ 02591 $<num>$ = ruby_sourceline; 02592 /*% 02593 %*/ 02594 } 02595 bodystmt 02596 k_end 02597 { 02598 cmdarg_stack = $<val>1; 02599 /*%%%*/ 02600 if ($3 == NULL) { 02601 $$ = NEW_NIL(); 02602 } 02603 else { 02604 if (nd_type($3) == NODE_RESCUE || 02605 nd_type($3) == NODE_ENSURE) 02606 nd_set_line($3, $<num>2); 02607 $$ = NEW_BEGIN($3); 02608 } 02609 nd_set_line($$, $<num>2); 02610 /*% 02611 $$ = dispatch1(begin, $3); 02612 %*/ 02613 } 02614 | tLPAREN_ARG {lex_state = EXPR_ENDARG;} rparen 02615 { 02616 /*%%%*/ 02617 $$ = 0; 02618 /*% 02619 $$ = dispatch1(paren, 0); 02620 %*/ 02621 } 02622 | tLPAREN_ARG expr {lex_state = EXPR_ENDARG;} rparen 02623 { 02624 /*%%%*/ 02625 $$ = $2; 02626 /*% 02627 $$ = dispatch1(paren, $2); 02628 %*/ 02629 } 02630 | tLPAREN compstmt ')' 02631 { 02632 /*%%%*/ 02633 $$ = $2; 02634 /*% 02635 $$ = dispatch1(paren, $2); 02636 %*/ 02637 } 02638 | primary_value tCOLON2 tCONSTANT 02639 { 02640 /*%%%*/ 02641 $$ = NEW_COLON2($1, $3); 02642 /*% 02643 $$ = dispatch2(const_path_ref, $1, $3); 02644 %*/ 02645 } 02646 | tCOLON3 tCONSTANT 02647 { 02648 /*%%%*/ 02649 $$ = NEW_COLON3($2); 02650 /*% 02651 $$ = dispatch1(top_const_ref, $2); 02652 %*/ 02653 } 02654 | tLBRACK aref_args ']' 02655 { 02656 /*%%%*/ 02657 if ($2 == 0) { 02658 $$ = NEW_ZARRAY(); /* zero length array*/ 02659 } 02660 else { 02661 $$ = $2; 02662 } 02663 /*% 02664 $$ = dispatch1(array, escape_Qundef($2)); 02665 %*/ 02666 } 02667 | tLBRACE assoc_list '}' 02668 { 02669 /*%%%*/ 02670 $$ = NEW_HASH($2); 02671 /*% 02672 $$ = dispatch1(hash, escape_Qundef($2)); 02673 %*/ 02674 } 02675 | keyword_return 02676 { 02677 /*%%%*/ 02678 $$ = NEW_RETURN(0); 02679 /*% 02680 $$ = dispatch0(return0); 02681 %*/ 02682 } 02683 | keyword_yield '(' call_args rparen 02684 { 02685 /*%%%*/ 02686 $$ = new_yield($3); 02687 /*% 02688 $$ = dispatch1(yield, dispatch1(paren, $3)); 02689 %*/ 02690 } 02691 | keyword_yield '(' rparen 02692 { 02693 /*%%%*/ 02694 $$ = NEW_YIELD(0); 02695 /*% 02696 $$ = dispatch1(yield, dispatch1(paren, arg_new())); 02697 %*/ 02698 } 02699 | keyword_yield 02700 { 02701 /*%%%*/ 02702 $$ = NEW_YIELD(0); 02703 /*% 02704 $$ = dispatch0(yield0); 02705 %*/ 02706 } 02707 | keyword_defined opt_nl '(' {in_defined = 1;} expr rparen 02708 { 02709 /*%%%*/ 02710 in_defined = 0; 02711 $$ = NEW_DEFINED($5); 02712 /*% 02713 in_defined = 0; 02714 $$ = dispatch1(defined, $5); 02715 %*/ 02716 } 02717 | keyword_not '(' expr rparen 02718 { 02719 /*%%%*/ 02720 $$ = call_uni_op(cond($3), '!'); 02721 /*% 02722 $$ = dispatch2(unary, ripper_intern("not"), $3); 02723 %*/ 02724 } 02725 | keyword_not '(' rparen 02726 { 02727 /*%%%*/ 02728 $$ = call_uni_op(cond(NEW_NIL()), '!'); 02729 /*% 02730 $$ = dispatch2(unary, ripper_intern("not"), Qnil); 02731 %*/ 02732 } 02733 | fcall brace_block 02734 { 02735 /*%%%*/ 02736 $2->nd_iter = $1; 02737 $$ = $2; 02738 /*% 02739 $$ = method_arg(dispatch1(fcall, $1), arg_new()); 02740 $$ = method_add_block($$, $2); 02741 %*/ 02742 } 02743 | method_call 02744 | method_call brace_block 02745 { 02746 /*%%%*/ 02747 block_dup_check($1->nd_args, $2); 02748 $2->nd_iter = $1; 02749 $$ = $2; 02750 /*% 02751 $$ = method_add_block($1, $2); 02752 %*/ 02753 } 02754 | tLAMBDA lambda 02755 { 02756 $$ = $2; 02757 } 02758 | k_if expr_value then 02759 compstmt 02760 if_tail 02761 k_end 02762 { 02763 /*%%%*/ 02764 $$ = NEW_IF(cond($2), $4, $5); 02765 fixpos($$, $2); 02766 /*% 02767 $$ = dispatch3(if, $2, $4, escape_Qundef($5)); 02768 %*/ 02769 } 02770 | k_unless expr_value then 02771 compstmt 02772 opt_else 02773 k_end 02774 { 02775 /*%%%*/ 02776 $$ = NEW_UNLESS(cond($2), $4, $5); 02777 fixpos($$, $2); 02778 /*% 02779 $$ = dispatch3(unless, $2, $4, escape_Qundef($5)); 02780 %*/ 02781 } 02782 | k_while {COND_PUSH(1);} expr_value do {COND_POP();} 02783 compstmt 02784 k_end 02785 { 02786 /*%%%*/ 02787 $$ = NEW_WHILE(cond($3), $6, 1); 02788 fixpos($$, $3); 02789 /*% 02790 $$ = dispatch2(while, $3, $6); 02791 %*/ 02792 } 02793 | k_until {COND_PUSH(1);} expr_value do {COND_POP();} 02794 compstmt 02795 k_end 02796 { 02797 /*%%%*/ 02798 $$ = NEW_UNTIL(cond($3), $6, 1); 02799 fixpos($$, $3); 02800 /*% 02801 $$ = dispatch2(until, $3, $6); 02802 %*/ 02803 } 02804 | k_case expr_value opt_terms 02805 case_body 02806 k_end 02807 { 02808 /*%%%*/ 02809 $$ = NEW_CASE($2, $4); 02810 fixpos($$, $2); 02811 /*% 02812 $$ = dispatch2(case, $2, $4); 02813 %*/ 02814 } 02815 | k_case opt_terms case_body k_end 02816 { 02817 /*%%%*/ 02818 $$ = NEW_CASE(0, $3); 02819 /*% 02820 $$ = dispatch2(case, Qnil, $3); 02821 %*/ 02822 } 02823 | k_for for_var keyword_in 02824 {COND_PUSH(1);} 02825 expr_value do 02826 {COND_POP();} 02827 compstmt 02828 k_end 02829 { 02830 /*%%%*/ 02831 /* 02832 * for a, b, c in e 02833 * #=> 02834 * e.each{|*x| a, b, c = x 02835 * 02836 * for a in e 02837 * #=> 02838 * e.each{|x| a, = x} 02839 */ 02840 ID id = internal_id(); 02841 ID *tbl = ALLOC_N(ID, 2); 02842 NODE *m = NEW_ARGS_AUX(0, 0); 02843 NODE *args, *scope; 02844 02845 if (nd_type($2) == NODE_MASGN) { 02846 /* if args.length == 1 && args[0].kind_of?(Array) 02847 * args = args[0] 02848 * end 02849 */ 02850 NODE *one = NEW_LIST(NEW_LIT(INT2FIX(1))); 02851 NODE *zero = NEW_LIST(NEW_LIT(INT2FIX(0))); 02852 m->nd_next = block_append( 02853 NEW_IF( 02854 NEW_NODE(NODE_AND, 02855 NEW_CALL(NEW_CALL(NEW_DVAR(id), idLength, 0), 02856 idEq, one), 02857 NEW_CALL(NEW_CALL(NEW_DVAR(id), idAREF, zero), 02858 rb_intern("kind_of?"), NEW_LIST(NEW_LIT(rb_cArray))), 02859 0), 02860 NEW_DASGN_CURR(id, 02861 NEW_CALL(NEW_DVAR(id), idAREF, zero)), 02862 0), 02863 node_assign($2, NEW_DVAR(id))); 02864 02865 args = new_args(m, 0, id, 0, new_args_tail(0, 0, 0)); 02866 } 02867 else { 02868 if (nd_type($2) == NODE_LASGN || 02869 nd_type($2) == NODE_DASGN || 02870 nd_type($2) == NODE_DASGN_CURR) { 02871 $2->nd_value = NEW_DVAR(id); 02872 m->nd_plen = 1; 02873 m->nd_next = $2; 02874 args = new_args(m, 0, 0, 0, new_args_tail(0, 0, 0)); 02875 } 02876 else { 02877 m->nd_next = node_assign(NEW_MASGN(NEW_LIST($2), 0), NEW_DVAR(id)); 02878 args = new_args(m, 0, id, 0, new_args_tail(0, 0, 0)); 02879 } 02880 } 02881 scope = NEW_NODE(NODE_SCOPE, tbl, $8, args); 02882 tbl[0] = 1; tbl[1] = id; 02883 $$ = NEW_FOR(0, $5, scope); 02884 fixpos($$, $2); 02885 /*% 02886 $$ = dispatch3(for, $2, $5, $8); 02887 %*/ 02888 } 02889 | k_class cpath superclass 02890 { 02891 if (in_def || in_single) 02892 yyerror("class definition in method body"); 02893 local_push(0); 02894 /*%%%*/ 02895 $<num>$ = ruby_sourceline; 02896 /*% 02897 %*/ 02898 } 02899 bodystmt 02900 k_end 02901 { 02902 /*%%%*/ 02903 $$ = NEW_CLASS($2, $5, $3); 02904 nd_set_line($$, $<num>4); 02905 /*% 02906 $$ = dispatch3(class, $2, $3, $5); 02907 %*/ 02908 local_pop(); 02909 } 02910 | k_class tLSHFT expr 02911 { 02912 $<num>$ = in_def; 02913 in_def = 0; 02914 } 02915 term 02916 { 02917 $<num>$ = in_single; 02918 in_single = 0; 02919 local_push(0); 02920 } 02921 bodystmt 02922 k_end 02923 { 02924 /*%%%*/ 02925 $$ = NEW_SCLASS($3, $7); 02926 fixpos($$, $3); 02927 /*% 02928 $$ = dispatch2(sclass, $3, $7); 02929 %*/ 02930 local_pop(); 02931 in_def = $<num>4; 02932 in_single = $<num>6; 02933 } 02934 | k_module cpath 02935 { 02936 if (in_def || in_single) 02937 yyerror("module definition in method body"); 02938 local_push(0); 02939 /*%%%*/ 02940 $<num>$ = ruby_sourceline; 02941 /*% 02942 %*/ 02943 } 02944 bodystmt 02945 k_end 02946 { 02947 /*%%%*/ 02948 $$ = NEW_MODULE($2, $4); 02949 nd_set_line($$, $<num>3); 02950 /*% 02951 $$ = dispatch2(module, $2, $4); 02952 %*/ 02953 local_pop(); 02954 } 02955 | k_def fname 02956 { 02957 $<id>$ = cur_mid; 02958 cur_mid = $2; 02959 in_def++; 02960 local_push(0); 02961 } 02962 f_arglist 02963 bodystmt 02964 k_end 02965 { 02966 /*%%%*/ 02967 NODE *body = remove_begin($5); 02968 reduce_nodes(&body); 02969 $$ = NEW_DEFN($2, $4, body, NOEX_PRIVATE); 02970 nd_set_line($$, $<num>1); 02971 /*% 02972 $$ = dispatch3(def, $2, $4, $5); 02973 %*/ 02974 local_pop(); 02975 in_def--; 02976 cur_mid = $<id>3; 02977 } 02978 | k_def singleton dot_or_colon {lex_state = EXPR_FNAME;} fname 02979 { 02980 in_single++; 02981 lex_state = EXPR_ENDFN; /* force for args */ 02982 local_push(0); 02983 } 02984 f_arglist 02985 bodystmt 02986 k_end 02987 { 02988 /*%%%*/ 02989 NODE *body = remove_begin($8); 02990 reduce_nodes(&body); 02991 $$ = NEW_DEFS($2, $5, $7, body); 02992 nd_set_line($$, $<num>1); 02993 /*% 02994 $$ = dispatch5(defs, $2, $3, $5, $7, $8); 02995 %*/ 02996 local_pop(); 02997 in_single--; 02998 } 02999 | keyword_break 03000 { 03001 /*%%%*/ 03002 $$ = NEW_BREAK(0); 03003 /*% 03004 $$ = dispatch1(break, arg_new()); 03005 %*/ 03006 } 03007 | keyword_next 03008 { 03009 /*%%%*/ 03010 $$ = NEW_NEXT(0); 03011 /*% 03012 $$ = dispatch1(next, arg_new()); 03013 %*/ 03014 } 03015 | keyword_redo 03016 { 03017 /*%%%*/ 03018 $$ = NEW_REDO(); 03019 /*% 03020 $$ = dispatch0(redo); 03021 %*/ 03022 } 03023 | keyword_retry 03024 { 03025 /*%%%*/ 03026 $$ = NEW_RETRY(); 03027 /*% 03028 $$ = dispatch0(retry); 03029 %*/ 03030 } 03031 ; 03032 03033 primary_value : primary 03034 { 03035 /*%%%*/ 03036 value_expr($1); 03037 $$ = $1; 03038 if (!$$) $$ = NEW_NIL(); 03039 /*% 03040 $$ = $1; 03041 %*/ 03042 } 03043 ; 03044 03045 k_begin : keyword_begin 03046 { 03047 token_info_push("begin"); 03048 } 03049 ; 03050 03051 k_if : keyword_if 03052 { 03053 token_info_push("if"); 03054 } 03055 ; 03056 03057 k_unless : keyword_unless 03058 { 03059 token_info_push("unless"); 03060 } 03061 ; 03062 03063 k_while : keyword_while 03064 { 03065 token_info_push("while"); 03066 } 03067 ; 03068 03069 k_until : keyword_until 03070 { 03071 token_info_push("until"); 03072 } 03073 ; 03074 03075 k_case : keyword_case 03076 { 03077 token_info_push("case"); 03078 } 03079 ; 03080 03081 k_for : keyword_for 03082 { 03083 token_info_push("for"); 03084 } 03085 ; 03086 03087 k_class : keyword_class 03088 { 03089 token_info_push("class"); 03090 } 03091 ; 03092 03093 k_module : keyword_module 03094 { 03095 token_info_push("module"); 03096 } 03097 ; 03098 03099 k_def : keyword_def 03100 { 03101 token_info_push("def"); 03102 /*%%%*/ 03103 $<num>$ = ruby_sourceline; 03104 /*% 03105 %*/ 03106 } 03107 ; 03108 03109 k_end : keyword_end 03110 { 03111 token_info_pop("end"); 03112 } 03113 ; 03114 03115 then : term 03116 /*%c%*/ 03117 /*%c 03118 { $$ = Qnil; } 03119 %*/ 03120 | keyword_then 03121 | term keyword_then 03122 /*%c%*/ 03123 /*%c 03124 { $$ = $2; } 03125 %*/ 03126 ; 03127 03128 do : term 03129 /*%c%*/ 03130 /*%c 03131 { $$ = Qnil; } 03132 %*/ 03133 | keyword_do_cond 03134 ; 03135 03136 if_tail : opt_else 03137 | keyword_elsif expr_value then 03138 compstmt 03139 if_tail 03140 { 03141 /*%%%*/ 03142 $$ = NEW_IF(cond($2), $4, $5); 03143 fixpos($$, $2); 03144 /*% 03145 $$ = dispatch3(elsif, $2, $4, escape_Qundef($5)); 03146 %*/ 03147 } 03148 ; 03149 03150 opt_else : none 03151 | keyword_else compstmt 03152 { 03153 /*%%%*/ 03154 $$ = $2; 03155 /*% 03156 $$ = dispatch1(else, $2); 03157 %*/ 03158 } 03159 ; 03160 03161 for_var : lhs 03162 | mlhs 03163 ; 03164 03165 f_marg : f_norm_arg 03166 { 03167 $$ = assignable($1, 0); 03168 /*%%%*/ 03169 /*% 03170 $$ = dispatch1(mlhs_paren, $$); 03171 %*/ 03172 } 03173 | tLPAREN f_margs rparen 03174 { 03175 /*%%%*/ 03176 $$ = $2; 03177 /*% 03178 $$ = dispatch1(mlhs_paren, $2); 03179 %*/ 03180 } 03181 ; 03182 03183 f_marg_list : f_marg 03184 { 03185 /*%%%*/ 03186 $$ = NEW_LIST($1); 03187 /*% 03188 $$ = mlhs_add(mlhs_new(), $1); 03189 %*/ 03190 } 03191 | f_marg_list ',' f_marg 03192 { 03193 /*%%%*/ 03194 $$ = list_append($1, $3); 03195 /*% 03196 $$ = mlhs_add($1, $3); 03197 %*/ 03198 } 03199 ; 03200 03201 f_margs : f_marg_list 03202 { 03203 /*%%%*/ 03204 $$ = NEW_MASGN($1, 0); 03205 /*% 03206 $$ = $1; 03207 %*/ 03208 } 03209 | f_marg_list ',' tSTAR f_norm_arg 03210 { 03211 $$ = assignable($4, 0); 03212 /*%%%*/ 03213 $$ = NEW_MASGN($1, $$); 03214 /*% 03215 $$ = mlhs_add_star($1, $$); 03216 %*/ 03217 } 03218 | f_marg_list ',' tSTAR f_norm_arg ',' f_marg_list 03219 { 03220 $$ = assignable($4, 0); 03221 /*%%%*/ 03222 $$ = NEW_MASGN($1, NEW_POSTARG($$, $6)); 03223 /*% 03224 $$ = mlhs_add_star($1, $$); 03225 %*/ 03226 } 03227 | f_marg_list ',' tSTAR 03228 { 03229 /*%%%*/ 03230 $$ = NEW_MASGN($1, -1); 03231 /*% 03232 $$ = mlhs_add_star($1, Qnil); 03233 %*/ 03234 } 03235 | f_marg_list ',' tSTAR ',' f_marg_list 03236 { 03237 /*%%%*/ 03238 $$ = NEW_MASGN($1, NEW_POSTARG(-1, $5)); 03239 /*% 03240 $$ = mlhs_add_star($1, $5); 03241 %*/ 03242 } 03243 | tSTAR f_norm_arg 03244 { 03245 $$ = assignable($2, 0); 03246 /*%%%*/ 03247 $$ = NEW_MASGN(0, $$); 03248 /*% 03249 $$ = mlhs_add_star(mlhs_new(), $$); 03250 %*/ 03251 } 03252 | tSTAR f_norm_arg ',' f_marg_list 03253 { 03254 $$ = assignable($2, 0); 03255 /*%%%*/ 03256 $$ = NEW_MASGN(0, NEW_POSTARG($$, $4)); 03257 /*% 03258 #if 0 03259 TODO: Check me 03260 #endif 03261 $$ = mlhs_add_star($$, $4); 03262 %*/ 03263 } 03264 | tSTAR 03265 { 03266 /*%%%*/ 03267 $$ = NEW_MASGN(0, -1); 03268 /*% 03269 $$ = mlhs_add_star(mlhs_new(), Qnil); 03270 %*/ 03271 } 03272 | tSTAR ',' f_marg_list 03273 { 03274 /*%%%*/ 03275 $$ = NEW_MASGN(0, NEW_POSTARG(-1, $3)); 03276 /*% 03277 $$ = mlhs_add_star(mlhs_new(), Qnil); 03278 %*/ 03279 } 03280 ; 03281 03282 03283 block_args_tail : f_block_kwarg ',' f_kwrest opt_f_block_arg 03284 { 03285 $$ = new_args_tail($1, $3, $4); 03286 } 03287 | f_block_kwarg opt_f_block_arg 03288 { 03289 $$ = new_args_tail($1, Qnone, $2); 03290 } 03291 | f_kwrest opt_f_block_arg 03292 { 03293 $$ = new_args_tail(Qnone, $1, $2); 03294 } 03295 | f_block_arg 03296 { 03297 $$ = new_args_tail(Qnone, Qnone, $1); 03298 } 03299 ; 03300 03301 opt_block_args_tail : ',' block_args_tail 03302 { 03303 $$ = $2; 03304 } 03305 | /* none */ 03306 { 03307 $$ = new_args_tail(Qnone, Qnone, Qnone); 03308 } 03309 ; 03310 03311 block_param : f_arg ',' f_block_optarg ',' f_rest_arg opt_block_args_tail 03312 { 03313 $$ = new_args($1, $3, $5, Qnone, $6); 03314 } 03315 | f_arg ',' f_block_optarg ',' f_rest_arg ',' f_arg opt_block_args_tail 03316 { 03317 $$ = new_args($1, $3, $5, $7, $8); 03318 } 03319 | f_arg ',' f_block_optarg opt_block_args_tail 03320 { 03321 $$ = new_args($1, $3, Qnone, Qnone, $4); 03322 } 03323 | f_arg ',' f_block_optarg ',' f_arg opt_block_args_tail 03324 { 03325 $$ = new_args($1, $3, Qnone, $5, $6); 03326 } 03327 | f_arg ',' f_rest_arg opt_block_args_tail 03328 { 03329 $$ = new_args($1, Qnone, $3, Qnone, $4); 03330 } 03331 | f_arg ',' 03332 { 03333 $$ = new_args($1, Qnone, 1, Qnone, new_args_tail(Qnone, Qnone, Qnone)); 03334 /*%%%*/ 03335 /*% 03336 dispatch1(excessed_comma, $$); 03337 %*/ 03338 } 03339 | f_arg ',' f_rest_arg ',' f_arg opt_block_args_tail 03340 { 03341 $$ = new_args($1, Qnone, $3, $5, $6); 03342 } 03343 | f_arg opt_block_args_tail 03344 { 03345 $$ = new_args($1, Qnone, Qnone, Qnone, $2); 03346 } 03347 | f_block_optarg ',' f_rest_arg opt_block_args_tail 03348 { 03349 $$ = new_args(Qnone, $1, $3, Qnone, $4); 03350 } 03351 | f_block_optarg ',' f_rest_arg ',' f_arg opt_block_args_tail 03352 { 03353 $$ = new_args(Qnone, $1, $3, $5, $6); 03354 } 03355 | f_block_optarg opt_block_args_tail 03356 { 03357 $$ = new_args(Qnone, $1, Qnone, Qnone, $2); 03358 } 03359 | f_block_optarg ',' f_arg opt_block_args_tail 03360 { 03361 $$ = new_args(Qnone, $1, Qnone, $3, $4); 03362 } 03363 | f_rest_arg opt_block_args_tail 03364 { 03365 $$ = new_args(Qnone, Qnone, $1, Qnone, $2); 03366 } 03367 | f_rest_arg ',' f_arg opt_block_args_tail 03368 { 03369 $$ = new_args(Qnone, Qnone, $1, $3, $4); 03370 } 03371 | block_args_tail 03372 { 03373 $$ = new_args(Qnone, Qnone, Qnone, Qnone, $1); 03374 } 03375 ; 03376 03377 opt_block_param : none 03378 | block_param_def 03379 { 03380 command_start = TRUE; 03381 } 03382 ; 03383 03384 block_param_def : '|' opt_bv_decl '|' 03385 { 03386 /*%%%*/ 03387 $$ = 0; 03388 /*% 03389 $$ = blockvar_new(params_new(Qnil,Qnil,Qnil,Qnil,Qnil,Qnil,Qnil), 03390 escape_Qundef($2)); 03391 %*/ 03392 } 03393 | tOROP 03394 { 03395 /*%%%*/ 03396 $$ = 0; 03397 /*% 03398 $$ = blockvar_new(params_new(Qnil,Qnil,Qnil,Qnil,Qnil,Qnil,Qnil), 03399 Qnil); 03400 %*/ 03401 } 03402 | '|' block_param opt_bv_decl '|' 03403 { 03404 /*%%%*/ 03405 $$ = $2; 03406 /*% 03407 $$ = blockvar_new(escape_Qundef($2), escape_Qundef($3)); 03408 %*/ 03409 } 03410 ; 03411 03412 03413 opt_bv_decl : opt_nl 03414 { 03415 $$ = 0; 03416 } 03417 | opt_nl ';' bv_decls opt_nl 03418 { 03419 /*%%%*/ 03420 $$ = 0; 03421 /*% 03422 $$ = $3; 03423 %*/ 03424 } 03425 ; 03426 03427 bv_decls : bvar 03428 /*%c%*/ 03429 /*%c 03430 { 03431 $$ = rb_ary_new3(1, $1); 03432 } 03433 %*/ 03434 | bv_decls ',' bvar 03435 /*%c%*/ 03436 /*%c 03437 { 03438 rb_ary_push($1, $3); 03439 } 03440 %*/ 03441 ; 03442 03443 bvar : tIDENTIFIER 03444 { 03445 new_bv(get_id($1)); 03446 /*%%%*/ 03447 /*% 03448 $$ = get_value($1); 03449 %*/ 03450 } 03451 | f_bad_arg 03452 { 03453 $$ = 0; 03454 } 03455 ; 03456 03457 lambda : { 03458 $<vars>$ = dyna_push(); 03459 } 03460 { 03461 $<num>$ = lpar_beg; 03462 lpar_beg = ++paren_nest; 03463 } 03464 f_larglist 03465 { 03466 $<num>$ = ruby_sourceline; 03467 } 03468 lambda_body 03469 { 03470 lpar_beg = $<num>2; 03471 /*%%%*/ 03472 $$ = NEW_LAMBDA($3, $5); 03473 nd_set_line($$, $<num>4); 03474 /*% 03475 $$ = dispatch2(lambda, $3, $5); 03476 %*/ 03477 dyna_pop($<vars>1); 03478 } 03479 ; 03480 03481 f_larglist : '(' f_args opt_bv_decl ')' 03482 { 03483 /*%%%*/ 03484 $$ = $2; 03485 /*% 03486 $$ = dispatch1(paren, $2); 03487 %*/ 03488 } 03489 | f_args 03490 { 03491 /*%%%*/ 03492 $$ = $1; 03493 /*% 03494 $$ = $1; 03495 %*/ 03496 } 03497 ; 03498 03499 lambda_body : tLAMBEG compstmt '}' 03500 { 03501 $$ = $2; 03502 } 03503 | keyword_do_LAMBDA compstmt keyword_end 03504 { 03505 $$ = $2; 03506 } 03507 ; 03508 03509 do_block : keyword_do_block 03510 { 03511 $<vars>1 = dyna_push(); 03512 /*%%%*/ 03513 $<num>$ = ruby_sourceline; 03514 /*% %*/ 03515 } 03516 opt_block_param 03517 compstmt 03518 keyword_end 03519 { 03520 /*%%%*/ 03521 $$ = NEW_ITER($3,$4); 03522 nd_set_line($$, $<num>2); 03523 /*% 03524 $$ = dispatch2(do_block, escape_Qundef($3), $4); 03525 %*/ 03526 dyna_pop($<vars>1); 03527 } 03528 ; 03529 03530 block_call : command do_block 03531 { 03532 /*%%%*/ 03533 if (nd_type($1) == NODE_YIELD) { 03534 compile_error(PARSER_ARG "block given to yield"); 03535 } 03536 else { 03537 block_dup_check($1->nd_args, $2); 03538 } 03539 $2->nd_iter = $1; 03540 $$ = $2; 03541 fixpos($$, $1); 03542 /*% 03543 $$ = method_add_block($1, $2); 03544 %*/ 03545 } 03546 | block_call dot_or_colon operation2 opt_paren_args 03547 { 03548 /*%%%*/ 03549 $$ = NEW_CALL($1, $3, $4); 03550 /*% 03551 $$ = dispatch3(call, $1, $2, $3); 03552 $$ = method_optarg($$, $4); 03553 %*/ 03554 } 03555 | block_call dot_or_colon operation2 opt_paren_args brace_block 03556 { 03557 /*%%%*/ 03558 block_dup_check($4, $5); 03559 $5->nd_iter = NEW_CALL($1, $3, $4); 03560 $$ = $5; 03561 fixpos($$, $1); 03562 /*% 03563 $$ = dispatch4(command_call, $1, $2, $3, $4); 03564 $$ = method_add_block($$, $5); 03565 %*/ 03566 } 03567 | block_call dot_or_colon operation2 command_args do_block 03568 { 03569 /*%%%*/ 03570 block_dup_check($4, $5); 03571 $5->nd_iter = NEW_CALL($1, $3, $4); 03572 $$ = $5; 03573 fixpos($$, $1); 03574 /*% 03575 $$ = dispatch4(command_call, $1, $2, $3, $4); 03576 $$ = method_add_block($$, $5); 03577 %*/ 03578 } 03579 ; 03580 03581 method_call : fcall paren_args 03582 { 03583 /*%%%*/ 03584 $$ = $1; 03585 $$->nd_args = $2; 03586 /*% 03587 $$ = method_arg(dispatch1(fcall, $1), $2); 03588 %*/ 03589 } 03590 | primary_value '.' operation2 03591 { 03592 /*%%%*/ 03593 $<num>$ = ruby_sourceline; 03594 /*% %*/ 03595 } 03596 opt_paren_args 03597 { 03598 /*%%%*/ 03599 $$ = NEW_CALL($1, $3, $5); 03600 nd_set_line($$, $<num>4); 03601 /*% 03602 $$ = dispatch3(call, $1, ripper_id2sym('.'), $3); 03603 $$ = method_optarg($$, $5); 03604 %*/ 03605 } 03606 | primary_value tCOLON2 operation2 03607 { 03608 /*%%%*/ 03609 $<num>$ = ruby_sourceline; 03610 /*% %*/ 03611 } 03612 paren_args 03613 { 03614 /*%%%*/ 03615 $$ = NEW_CALL($1, $3, $5); 03616 nd_set_line($$, $<num>4); 03617 /*% 03618 $$ = dispatch3(call, $1, ripper_id2sym('.'), $3); 03619 $$ = method_optarg($$, $5); 03620 %*/ 03621 } 03622 | primary_value tCOLON2 operation3 03623 { 03624 /*%%%*/ 03625 $$ = NEW_CALL($1, $3, 0); 03626 /*% 03627 $$ = dispatch3(call, $1, ripper_intern("::"), $3); 03628 %*/ 03629 } 03630 | primary_value '.' 03631 { 03632 /*%%%*/ 03633 $<num>$ = ruby_sourceline; 03634 /*% %*/ 03635 } 03636 paren_args 03637 { 03638 /*%%%*/ 03639 $$ = NEW_CALL($1, rb_intern("call"), $4); 03640 nd_set_line($$, $<num>3); 03641 /*% 03642 $$ = dispatch3(call, $1, ripper_id2sym('.'), 03643 ripper_intern("call")); 03644 $$ = method_optarg($$, $4); 03645 %*/ 03646 } 03647 | primary_value tCOLON2 03648 { 03649 /*%%%*/ 03650 $<num>$ = ruby_sourceline; 03651 /*% %*/ 03652 } 03653 paren_args 03654 { 03655 /*%%%*/ 03656 $$ = NEW_CALL($1, rb_intern("call"), $4); 03657 nd_set_line($$, $<num>3); 03658 /*% 03659 $$ = dispatch3(call, $1, ripper_intern("::"), 03660 ripper_intern("call")); 03661 $$ = method_optarg($$, $4); 03662 %*/ 03663 } 03664 | keyword_super paren_args 03665 { 03666 /*%%%*/ 03667 $$ = NEW_SUPER($2); 03668 /*% 03669 $$ = dispatch1(super, $2); 03670 %*/ 03671 } 03672 | keyword_super 03673 { 03674 /*%%%*/ 03675 $$ = NEW_ZSUPER(); 03676 /*% 03677 $$ = dispatch0(zsuper); 03678 %*/ 03679 } 03680 | primary_value '[' opt_call_args rbracket 03681 { 03682 /*%%%*/ 03683 if ($1 && nd_type($1) == NODE_SELF) 03684 $$ = NEW_FCALL(tAREF, $3); 03685 else 03686 $$ = NEW_CALL($1, tAREF, $3); 03687 fixpos($$, $1); 03688 /*% 03689 $$ = dispatch2(aref, $1, escape_Qundef($3)); 03690 %*/ 03691 } 03692 ; 03693 03694 brace_block : '{' 03695 { 03696 $<vars>1 = dyna_push(); 03697 /*%%%*/ 03698 $<num>$ = ruby_sourceline; 03699 /*% 03700 %*/ 03701 } 03702 opt_block_param 03703 compstmt '}' 03704 { 03705 /*%%%*/ 03706 $$ = NEW_ITER($3,$4); 03707 nd_set_line($$, $<num>2); 03708 /*% 03709 $$ = dispatch2(brace_block, escape_Qundef($3), $4); 03710 %*/ 03711 dyna_pop($<vars>1); 03712 } 03713 | keyword_do 03714 { 03715 $<vars>1 = dyna_push(); 03716 /*%%%*/ 03717 $<num>$ = ruby_sourceline; 03718 /*% 03719 %*/ 03720 } 03721 opt_block_param 03722 compstmt keyword_end 03723 { 03724 /*%%%*/ 03725 $$ = NEW_ITER($3,$4); 03726 nd_set_line($$, $<num>2); 03727 /*% 03728 $$ = dispatch2(do_block, escape_Qundef($3), $4); 03729 %*/ 03730 dyna_pop($<vars>1); 03731 } 03732 ; 03733 03734 case_body : keyword_when args then 03735 compstmt 03736 cases 03737 { 03738 /*%%%*/ 03739 $$ = NEW_WHEN($2, $4, $5); 03740 /*% 03741 $$ = dispatch3(when, $2, $4, escape_Qundef($5)); 03742 %*/ 03743 } 03744 ; 03745 03746 cases : opt_else 03747 | case_body 03748 ; 03749 03750 opt_rescue : keyword_rescue exc_list exc_var then 03751 compstmt 03752 opt_rescue 03753 { 03754 /*%%%*/ 03755 if ($3) { 03756 $3 = node_assign($3, NEW_ERRINFO()); 03757 $5 = block_append($3, $5); 03758 } 03759 $$ = NEW_RESBODY($2, $5, $6); 03760 fixpos($$, $2?$2:$5); 03761 /*% 03762 $$ = dispatch4(rescue, 03763 escape_Qundef($2), 03764 escape_Qundef($3), 03765 escape_Qundef($5), 03766 escape_Qundef($6)); 03767 %*/ 03768 } 03769 | none 03770 ; 03771 03772 exc_list : arg_value 03773 { 03774 /*%%%*/ 03775 $$ = NEW_LIST($1); 03776 /*% 03777 $$ = rb_ary_new3(1, $1); 03778 %*/ 03779 } 03780 | mrhs 03781 { 03782 /*%%%*/ 03783 if (!($$ = splat_array($1))) $$ = $1; 03784 /*% 03785 $$ = $1; 03786 %*/ 03787 } 03788 | none 03789 ; 03790 03791 exc_var : tASSOC lhs 03792 { 03793 $$ = $2; 03794 } 03795 | none 03796 ; 03797 03798 opt_ensure : keyword_ensure compstmt 03799 { 03800 /*%%%*/ 03801 $$ = $2; 03802 /*% 03803 $$ = dispatch1(ensure, $2); 03804 %*/ 03805 } 03806 | none 03807 ; 03808 03809 literal : numeric 03810 | symbol 03811 { 03812 /*%%%*/ 03813 $$ = NEW_LIT(ID2SYM($1)); 03814 /*% 03815 $$ = dispatch1(symbol_literal, $1); 03816 %*/ 03817 } 03818 | dsym 03819 ; 03820 03821 strings : string 03822 { 03823 /*%%%*/ 03824 NODE *node = $1; 03825 if (!node) { 03826 node = NEW_STR(STR_NEW0()); 03827 } 03828 else { 03829 node = evstr2dstr(node); 03830 } 03831 $$ = node; 03832 /*% 03833 $$ = $1; 03834 %*/ 03835 } 03836 ; 03837 03838 string : tCHAR 03839 | string1 03840 | string string1 03841 { 03842 /*%%%*/ 03843 $$ = literal_concat($1, $2); 03844 /*% 03845 $$ = dispatch2(string_concat, $1, $2); 03846 %*/ 03847 } 03848 ; 03849 03850 string1 : tSTRING_BEG string_contents tSTRING_END 03851 { 03852 /*%%%*/ 03853 $$ = $2; 03854 /*% 03855 $$ = dispatch1(string_literal, $2); 03856 %*/ 03857 } 03858 ; 03859 03860 xstring : tXSTRING_BEG xstring_contents tSTRING_END 03861 { 03862 /*%%%*/ 03863 NODE *node = $2; 03864 if (!node) { 03865 node = NEW_XSTR(STR_NEW0()); 03866 } 03867 else { 03868 switch (nd_type(node)) { 03869 case NODE_STR: 03870 nd_set_type(node, NODE_XSTR); 03871 break; 03872 case NODE_DSTR: 03873 nd_set_type(node, NODE_DXSTR); 03874 break; 03875 default: 03876 node = NEW_NODE(NODE_DXSTR, Qnil, 1, NEW_LIST(node)); 03877 break; 03878 } 03879 } 03880 $$ = node; 03881 /*% 03882 $$ = dispatch1(xstring_literal, $2); 03883 %*/ 03884 } 03885 ; 03886 03887 regexp : tREGEXP_BEG regexp_contents tREGEXP_END 03888 { 03889 /*%%%*/ 03890 int options = $3; 03891 NODE *node = $2; 03892 NODE *list, *prev; 03893 if (!node) { 03894 node = NEW_LIT(reg_compile(STR_NEW0(), options)); 03895 } 03896 else switch (nd_type(node)) { 03897 case NODE_STR: 03898 { 03899 VALUE src = node->nd_lit; 03900 nd_set_type(node, NODE_LIT); 03901 node->nd_lit = reg_compile(src, options); 03902 } 03903 break; 03904 default: 03905 node = NEW_NODE(NODE_DSTR, STR_NEW0(), 1, NEW_LIST(node)); 03906 case NODE_DSTR: 03907 if (options & RE_OPTION_ONCE) { 03908 nd_set_type(node, NODE_DREGX_ONCE); 03909 } 03910 else { 03911 nd_set_type(node, NODE_DREGX); 03912 } 03913 node->nd_cflag = options & RE_OPTION_MASK; 03914 if (!NIL_P(node->nd_lit)) reg_fragment_check(node->nd_lit, options); 03915 for (list = (prev = node)->nd_next; list; list = list->nd_next) { 03916 if (nd_type(list->nd_head) == NODE_STR) { 03917 VALUE tail = list->nd_head->nd_lit; 03918 if (reg_fragment_check(tail, options) && prev && !NIL_P(prev->nd_lit)) { 03919 VALUE lit = prev == node ? prev->nd_lit : prev->nd_head->nd_lit; 03920 if (!literal_concat0(parser, lit, tail)) { 03921 node = 0; 03922 break; 03923 } 03924 rb_str_resize(tail, 0); 03925 prev->nd_next = list->nd_next; 03926 rb_gc_force_recycle((VALUE)list->nd_head); 03927 rb_gc_force_recycle((VALUE)list); 03928 list = prev; 03929 } 03930 else { 03931 prev = list; 03932 } 03933 } 03934 else { 03935 prev = 0; 03936 } 03937 } 03938 if (!node->nd_next) { 03939 VALUE src = node->nd_lit; 03940 nd_set_type(node, NODE_LIT); 03941 node->nd_lit = reg_compile(src, options); 03942 } 03943 break; 03944 } 03945 $$ = node; 03946 /*% 03947 $$ = dispatch2(regexp_literal, $2, $3); 03948 %*/ 03949 } 03950 ; 03951 03952 words : tWORDS_BEG ' ' tSTRING_END 03953 { 03954 /*%%%*/ 03955 $$ = NEW_ZARRAY(); 03956 /*% 03957 $$ = dispatch0(words_new); 03958 $$ = dispatch1(array, $$); 03959 %*/ 03960 } 03961 | tWORDS_BEG word_list tSTRING_END 03962 { 03963 /*%%%*/ 03964 $$ = $2; 03965 /*% 03966 $$ = dispatch1(array, $2); 03967 %*/ 03968 } 03969 ; 03970 03971 word_list : /* none */ 03972 { 03973 /*%%%*/ 03974 $$ = 0; 03975 /*% 03976 $$ = dispatch0(words_new); 03977 %*/ 03978 } 03979 | word_list word ' ' 03980 { 03981 /*%%%*/ 03982 $$ = list_append($1, evstr2dstr($2)); 03983 /*% 03984 $$ = dispatch2(words_add, $1, $2); 03985 %*/ 03986 } 03987 ; 03988 03989 word : string_content 03990 /*%c%*/ 03991 /*%c 03992 { 03993 $$ = dispatch0(word_new); 03994 $$ = dispatch2(word_add, $$, $1); 03995 } 03996 %*/ 03997 | word string_content 03998 { 03999 /*%%%*/ 04000 $$ = literal_concat($1, $2); 04001 /*% 04002 $$ = dispatch2(word_add, $1, $2); 04003 %*/ 04004 } 04005 ; 04006 04007 symbols : tSYMBOLS_BEG ' ' tSTRING_END 04008 { 04009 /*%%%*/ 04010 $$ = NEW_ZARRAY(); 04011 /*% 04012 $$ = dispatch0(symbols_new); 04013 $$ = dispatch1(array, $$); 04014 %*/ 04015 } 04016 | tSYMBOLS_BEG symbol_list tSTRING_END 04017 { 04018 /*%%%*/ 04019 $$ = $2; 04020 /*% 04021 $$ = dispatch1(array, $2); 04022 %*/ 04023 } 04024 ; 04025 04026 symbol_list : /* none */ 04027 { 04028 /*%%%*/ 04029 $$ = 0; 04030 /*% 04031 $$ = dispatch0(symbols_new); 04032 %*/ 04033 } 04034 | symbol_list word ' ' 04035 { 04036 /*%%%*/ 04037 $2 = evstr2dstr($2); 04038 nd_set_type($2, NODE_DSYM); 04039 $$ = list_append($1, $2); 04040 /*% 04041 $$ = dispatch2(symbols_add, $1, $2); 04042 %*/ 04043 } 04044 ; 04045 04046 qwords : tQWORDS_BEG ' ' tSTRING_END 04047 { 04048 /*%%%*/ 04049 $$ = NEW_ZARRAY(); 04050 /*% 04051 $$ = dispatch0(qwords_new); 04052 $$ = dispatch1(array, $$); 04053 %*/ 04054 } 04055 | tQWORDS_BEG qword_list tSTRING_END 04056 { 04057 /*%%%*/ 04058 $$ = $2; 04059 /*% 04060 $$ = dispatch1(array, $2); 04061 %*/ 04062 } 04063 ; 04064 04065 qsymbols : tQSYMBOLS_BEG ' ' tSTRING_END 04066 { 04067 /*%%%*/ 04068 $$ = NEW_ZARRAY(); 04069 /*% 04070 $$ = dispatch0(qsymbols_new); 04071 $$ = dispatch1(array, $$); 04072 %*/ 04073 } 04074 | tQSYMBOLS_BEG qsym_list tSTRING_END 04075 { 04076 /*%%%*/ 04077 $$ = $2; 04078 /*% 04079 $$ = dispatch1(array, $2); 04080 %*/ 04081 } 04082 ; 04083 04084 qword_list : /* none */ 04085 { 04086 /*%%%*/ 04087 $$ = 0; 04088 /*% 04089 $$ = dispatch0(qwords_new); 04090 %*/ 04091 } 04092 | qword_list tSTRING_CONTENT ' ' 04093 { 04094 /*%%%*/ 04095 $$ = list_append($1, $2); 04096 /*% 04097 $$ = dispatch2(qwords_add, $1, $2); 04098 %*/ 04099 } 04100 ; 04101 04102 qsym_list : /* none */ 04103 { 04104 /*%%%*/ 04105 $$ = 0; 04106 /*% 04107 $$ = dispatch0(qsymbols_new); 04108 %*/ 04109 } 04110 | qsym_list tSTRING_CONTENT ' ' 04111 { 04112 /*%%%*/ 04113 VALUE lit; 04114 lit = $2->nd_lit; 04115 $2->nd_lit = ID2SYM(rb_intern_str(lit)); 04116 nd_set_type($2, NODE_LIT); 04117 $$ = list_append($1, $2); 04118 /*% 04119 $$ = dispatch2(qsymbols_add, $1, $2); 04120 %*/ 04121 } 04122 ; 04123 04124 string_contents : /* none */ 04125 { 04126 /*%%%*/ 04127 $$ = 0; 04128 /*% 04129 $$ = dispatch0(string_content); 04130 %*/ 04131 } 04132 | string_contents string_content 04133 { 04134 /*%%%*/ 04135 $$ = literal_concat($1, $2); 04136 /*% 04137 $$ = dispatch2(string_add, $1, $2); 04138 %*/ 04139 } 04140 ; 04141 04142 xstring_contents: /* none */ 04143 { 04144 /*%%%*/ 04145 $$ = 0; 04146 /*% 04147 $$ = dispatch0(xstring_new); 04148 %*/ 04149 } 04150 | xstring_contents string_content 04151 { 04152 /*%%%*/ 04153 $$ = literal_concat($1, $2); 04154 /*% 04155 $$ = dispatch2(xstring_add, $1, $2); 04156 %*/ 04157 } 04158 ; 04159 04160 regexp_contents: /* none */ 04161 { 04162 /*%%%*/ 04163 $$ = 0; 04164 /*% 04165 $$ = dispatch0(regexp_new); 04166 %*/ 04167 } 04168 | regexp_contents string_content 04169 { 04170 /*%%%*/ 04171 NODE *head = $1, *tail = $2; 04172 if (!head) { 04173 $$ = tail; 04174 } 04175 else if (!tail) { 04176 $$ = head; 04177 } 04178 else { 04179 switch (nd_type(head)) { 04180 case NODE_STR: 04181 nd_set_type(head, NODE_DSTR); 04182 break; 04183 case NODE_DSTR: 04184 break; 04185 default: 04186 head = list_append(NEW_DSTR(Qnil), head); 04187 break; 04188 } 04189 $$ = list_append(head, tail); 04190 } 04191 /*% 04192 $$ = dispatch2(regexp_add, $1, $2); 04193 %*/ 04194 } 04195 ; 04196 04197 string_content : tSTRING_CONTENT 04198 | tSTRING_DVAR 04199 { 04200 $<node>$ = lex_strterm; 04201 lex_strterm = 0; 04202 lex_state = EXPR_BEG; 04203 } 04204 string_dvar 04205 { 04206 /*%%%*/ 04207 lex_strterm = $<node>2; 04208 $$ = NEW_EVSTR($3); 04209 /*% 04210 lex_strterm = $<node>2; 04211 $$ = dispatch1(string_dvar, $3); 04212 %*/ 04213 } 04214 | tSTRING_DBEG 04215 { 04216 $<val>1 = cond_stack; 04217 $<val>$ = cmdarg_stack; 04218 cond_stack = 0; 04219 cmdarg_stack = 0; 04220 } 04221 { 04222 $<node>$ = lex_strterm; 04223 lex_strterm = 0; 04224 lex_state = EXPR_BEG; 04225 } 04226 { 04227 $<num>$ = brace_nest; 04228 brace_nest = 0; 04229 } 04230 compstmt tSTRING_DEND 04231 { 04232 cond_stack = $<val>1; 04233 cmdarg_stack = $<val>2; 04234 lex_strterm = $<node>3; 04235 brace_nest = $<num>4; 04236 /*%%%*/ 04237 if ($5) $5->flags &= ~NODE_FL_NEWLINE; 04238 $$ = new_evstr($5); 04239 /*% 04240 $$ = dispatch1(string_embexpr, $5); 04241 %*/ 04242 } 04243 ; 04244 04245 string_dvar : tGVAR 04246 { 04247 /*%%%*/ 04248 $$ = NEW_GVAR($1); 04249 /*% 04250 $$ = dispatch1(var_ref, $1); 04251 %*/ 04252 } 04253 | tIVAR 04254 { 04255 /*%%%*/ 04256 $$ = NEW_IVAR($1); 04257 /*% 04258 $$ = dispatch1(var_ref, $1); 04259 %*/ 04260 } 04261 | tCVAR 04262 { 04263 /*%%%*/ 04264 $$ = NEW_CVAR($1); 04265 /*% 04266 $$ = dispatch1(var_ref, $1); 04267 %*/ 04268 } 04269 | backref 04270 ; 04271 04272 symbol : tSYMBEG sym 04273 { 04274 lex_state = EXPR_END; 04275 /*%%%*/ 04276 $$ = $2; 04277 /*% 04278 $$ = dispatch1(symbol, $2); 04279 %*/ 04280 } 04281 ; 04282 04283 sym : fname 04284 | tIVAR 04285 | tGVAR 04286 | tCVAR 04287 ; 04288 04289 dsym : tSYMBEG xstring_contents tSTRING_END 04290 { 04291 lex_state = EXPR_END; 04292 /*%%%*/ 04293 $$ = dsym_node($2); 04294 /*% 04295 $$ = dispatch1(dyna_symbol, $2); 04296 %*/ 04297 } 04298 ; 04299 04300 numeric : tINTEGER 04301 | tFLOAT 04302 | tUMINUS_NUM tINTEGER %prec tLOWEST 04303 { 04304 /*%%%*/ 04305 $$ = negate_lit($2); 04306 /*% 04307 $$ = dispatch2(unary, ripper_intern("-@"), $2); 04308 %*/ 04309 } 04310 | tUMINUS_NUM tFLOAT %prec tLOWEST 04311 { 04312 /*%%%*/ 04313 $$ = negate_lit($2); 04314 /*% 04315 $$ = dispatch2(unary, ripper_intern("-@"), $2); 04316 %*/ 04317 } 04318 ; 04319 04320 user_variable : tIDENTIFIER 04321 | tIVAR 04322 | tGVAR 04323 | tCONSTANT 04324 | tCVAR 04325 ; 04326 04327 keyword_variable: keyword_nil {ifndef_ripper($$ = keyword_nil);} 04328 | keyword_self {ifndef_ripper($$ = keyword_self);} 04329 | keyword_true {ifndef_ripper($$ = keyword_true);} 04330 | keyword_false {ifndef_ripper($$ = keyword_false);} 04331 | keyword__FILE__ {ifndef_ripper($$ = keyword__FILE__);} 04332 | keyword__LINE__ {ifndef_ripper($$ = keyword__LINE__);} 04333 | keyword__ENCODING__ {ifndef_ripper($$ = keyword__ENCODING__);} 04334 ; 04335 04336 var_ref : user_variable 04337 { 04338 /*%%%*/ 04339 if (!($$ = gettable($1))) $$ = NEW_BEGIN(0); 04340 /*% 04341 if (id_is_var(get_id($1))) { 04342 $$ = dispatch1(var_ref, $1); 04343 } 04344 else { 04345 $$ = dispatch1(vcall, $1); 04346 } 04347 %*/ 04348 } 04349 | keyword_variable 04350 { 04351 /*%%%*/ 04352 if (!($$ = gettable($1))) $$ = NEW_BEGIN(0); 04353 /*% 04354 $$ = dispatch1(var_ref, $1); 04355 %*/ 04356 } 04357 ; 04358 04359 var_lhs : user_variable 04360 { 04361 $$ = assignable($1, 0); 04362 /*%%%*/ 04363 /*% 04364 $$ = dispatch1(var_field, $$); 04365 %*/ 04366 } 04367 | keyword_variable 04368 { 04369 $$ = assignable($1, 0); 04370 /*%%%*/ 04371 /*% 04372 $$ = dispatch1(var_field, $$); 04373 %*/ 04374 } 04375 ; 04376 04377 backref : tNTH_REF 04378 | tBACK_REF 04379 ; 04380 04381 superclass : term 04382 { 04383 /*%%%*/ 04384 $$ = 0; 04385 /*% 04386 $$ = Qnil; 04387 %*/ 04388 } 04389 | '<' 04390 { 04391 lex_state = EXPR_BEG; 04392 command_start = TRUE; 04393 } 04394 expr_value term 04395 { 04396 $$ = $3; 04397 } 04398 | error term 04399 { 04400 /*%%%*/ 04401 yyerrok; 04402 $$ = 0; 04403 /*% 04404 yyerrok; 04405 $$ = Qnil; 04406 %*/ 04407 } 04408 ; 04409 04410 f_arglist : '(' f_args rparen 04411 { 04412 /*%%%*/ 04413 $$ = $2; 04414 /*% 04415 $$ = dispatch1(paren, $2); 04416 %*/ 04417 lex_state = EXPR_BEG; 04418 command_start = TRUE; 04419 } 04420 | f_args term 04421 { 04422 $$ = $1; 04423 lex_state = EXPR_BEG; 04424 command_start = TRUE; 04425 } 04426 ; 04427 04428 args_tail : f_kwarg ',' f_kwrest opt_f_block_arg 04429 { 04430 $$ = new_args_tail($1, $3, $4); 04431 } 04432 | f_kwarg opt_f_block_arg 04433 { 04434 $$ = new_args_tail($1, Qnone, $2); 04435 } 04436 | f_kwrest opt_f_block_arg 04437 { 04438 $$ = new_args_tail(Qnone, $1, $2); 04439 } 04440 | f_block_arg 04441 { 04442 $$ = new_args_tail(Qnone, Qnone, $1); 04443 } 04444 ; 04445 04446 opt_args_tail : ',' args_tail 04447 { 04448 $$ = $2; 04449 } 04450 | /* none */ 04451 { 04452 $$ = new_args_tail(Qnone, Qnone, Qnone); 04453 } 04454 ; 04455 04456 f_args : f_arg ',' f_optarg ',' f_rest_arg opt_args_tail 04457 { 04458 $$ = new_args($1, $3, $5, Qnone, $6); 04459 } 04460 | f_arg ',' f_optarg ',' f_rest_arg ',' f_arg opt_args_tail 04461 { 04462 $$ = new_args($1, $3, $5, $7, $8); 04463 } 04464 | f_arg ',' f_optarg opt_args_tail 04465 { 04466 $$ = new_args($1, $3, Qnone, Qnone, $4); 04467 } 04468 | f_arg ',' f_optarg ',' f_arg opt_args_tail 04469 { 04470 $$ = new_args($1, $3, Qnone, $5, $6); 04471 } 04472 | f_arg ',' f_rest_arg opt_args_tail 04473 { 04474 $$ = new_args($1, Qnone, $3, Qnone, $4); 04475 } 04476 | f_arg ',' f_rest_arg ',' f_arg opt_args_tail 04477 { 04478 $$ = new_args($1, Qnone, $3, $5, $6); 04479 } 04480 | f_arg opt_args_tail 04481 { 04482 $$ = new_args($1, Qnone, Qnone, Qnone, $2); 04483 } 04484 | f_optarg ',' f_rest_arg opt_args_tail 04485 { 04486 $$ = new_args(Qnone, $1, $3, Qnone, $4); 04487 } 04488 | f_optarg ',' f_rest_arg ',' f_arg opt_args_tail 04489 { 04490 $$ = new_args(Qnone, $1, $3, $5, $6); 04491 } 04492 | f_optarg opt_args_tail 04493 { 04494 $$ = new_args(Qnone, $1, Qnone, Qnone, $2); 04495 } 04496 | f_optarg ',' f_arg opt_args_tail 04497 { 04498 $$ = new_args(Qnone, $1, Qnone, $3, $4); 04499 } 04500 | f_rest_arg opt_args_tail 04501 { 04502 $$ = new_args(Qnone, Qnone, $1, Qnone, $2); 04503 } 04504 | f_rest_arg ',' f_arg opt_args_tail 04505 { 04506 $$ = new_args(Qnone, Qnone, $1, $3, $4); 04507 } 04508 | args_tail 04509 { 04510 $$ = new_args(Qnone, Qnone, Qnone, Qnone, $1); 04511 } 04512 | /* none */ 04513 { 04514 $$ = new_args_tail(Qnone, Qnone, Qnone); 04515 $$ = new_args(Qnone, Qnone, Qnone, Qnone, $$); 04516 } 04517 ; 04518 04519 f_bad_arg : tCONSTANT 04520 { 04521 /*%%%*/ 04522 yyerror("formal argument cannot be a constant"); 04523 $$ = 0; 04524 /*% 04525 $$ = dispatch1(param_error, $1); 04526 %*/ 04527 } 04528 | tIVAR 04529 { 04530 /*%%%*/ 04531 yyerror("formal argument cannot be an instance variable"); 04532 $$ = 0; 04533 /*% 04534 $$ = dispatch1(param_error, $1); 04535 %*/ 04536 } 04537 | tGVAR 04538 { 04539 /*%%%*/ 04540 yyerror("formal argument cannot be a global variable"); 04541 $$ = 0; 04542 /*% 04543 $$ = dispatch1(param_error, $1); 04544 %*/ 04545 } 04546 | tCVAR 04547 { 04548 /*%%%*/ 04549 yyerror("formal argument cannot be a class variable"); 04550 $$ = 0; 04551 /*% 04552 $$ = dispatch1(param_error, $1); 04553 %*/ 04554 } 04555 ; 04556 04557 f_norm_arg : f_bad_arg 04558 | tIDENTIFIER 04559 { 04560 formal_argument(get_id($1)); 04561 $$ = $1; 04562 } 04563 ; 04564 04565 f_arg_item : f_norm_arg 04566 { 04567 arg_var(get_id($1)); 04568 /*%%%*/ 04569 $$ = NEW_ARGS_AUX($1, 1); 04570 /*% 04571 $$ = get_value($1); 04572 %*/ 04573 } 04574 | tLPAREN f_margs rparen 04575 { 04576 ID tid = internal_id(); 04577 arg_var(tid); 04578 /*%%%*/ 04579 if (dyna_in_block()) { 04580 $2->nd_value = NEW_DVAR(tid); 04581 } 04582 else { 04583 $2->nd_value = NEW_LVAR(tid); 04584 } 04585 $$ = NEW_ARGS_AUX(tid, 1); 04586 $$->nd_next = $2; 04587 /*% 04588 $$ = dispatch1(mlhs_paren, $2); 04589 %*/ 04590 } 04591 ; 04592 04593 f_arg : f_arg_item 04594 /*%c%*/ 04595 /*%c 04596 { 04597 $$ = rb_ary_new3(1, $1); 04598 } 04599 c%*/ 04600 | f_arg ',' f_arg_item 04601 { 04602 /*%%%*/ 04603 $$ = $1; 04604 $$->nd_plen++; 04605 $$->nd_next = block_append($$->nd_next, $3->nd_next); 04606 rb_gc_force_recycle((VALUE)$3); 04607 /*% 04608 $$ = rb_ary_push($1, $3); 04609 %*/ 04610 } 04611 ; 04612 04613 f_kw : tLABEL arg_value 04614 { 04615 arg_var(formal_argument(get_id($1))); 04616 $$ = assignable($1, $2); 04617 /*%%%*/ 04618 $$ = NEW_KW_ARG(0, $$); 04619 /*% 04620 $$ = rb_assoc_new($$, $2); 04621 %*/ 04622 } 04623 ; 04624 04625 f_block_kw : tLABEL primary_value 04626 { 04627 arg_var(formal_argument(get_id($1))); 04628 $$ = assignable($1, $2); 04629 /*%%%*/ 04630 $$ = NEW_KW_ARG(0, $$); 04631 /*% 04632 $$ = rb_assoc_new($$, $2); 04633 %*/ 04634 } 04635 ; 04636 04637 f_block_kwarg : f_block_kw 04638 { 04639 /*%%%*/ 04640 $$ = $1; 04641 /*% 04642 $$ = rb_ary_new3(1, $1); 04643 %*/ 04644 } 04645 | f_block_kwarg ',' f_block_kw 04646 { 04647 /*%%%*/ 04648 NODE *kws = $1; 04649 04650 while (kws->nd_next) { 04651 kws = kws->nd_next; 04652 } 04653 kws->nd_next = $3; 04654 $$ = $1; 04655 /*% 04656 $$ = rb_ary_push($1, $3); 04657 %*/ 04658 } 04659 ; 04660 04661 04662 f_kwarg : f_kw 04663 { 04664 /*%%%*/ 04665 $$ = $1; 04666 /*% 04667 $$ = rb_ary_new3(1, $1); 04668 %*/ 04669 } 04670 | f_kwarg ',' f_kw 04671 { 04672 /*%%%*/ 04673 NODE *kws = $1; 04674 04675 while (kws->nd_next) { 04676 kws = kws->nd_next; 04677 } 04678 kws->nd_next = $3; 04679 $$ = $1; 04680 /*% 04681 $$ = rb_ary_push($1, $3); 04682 %*/ 04683 } 04684 ; 04685 04686 kwrest_mark : tPOW 04687 | tDSTAR 04688 ; 04689 04690 f_kwrest : kwrest_mark tIDENTIFIER 04691 { 04692 shadowing_lvar(get_id($2)); 04693 $$ = $2; 04694 } 04695 | kwrest_mark 04696 { 04697 $$ = internal_id(); 04698 } 04699 ; 04700 04701 f_opt : tIDENTIFIER '=' arg_value 04702 { 04703 arg_var(formal_argument(get_id($1))); 04704 $$ = assignable($1, $3); 04705 /*%%%*/ 04706 $$ = NEW_OPT_ARG(0, $$); 04707 /*% 04708 $$ = rb_assoc_new($$, $3); 04709 %*/ 04710 } 04711 ; 04712 04713 f_block_opt : tIDENTIFIER '=' primary_value 04714 { 04715 arg_var(formal_argument(get_id($1))); 04716 $$ = assignable($1, $3); 04717 /*%%%*/ 04718 $$ = NEW_OPT_ARG(0, $$); 04719 /*% 04720 $$ = rb_assoc_new($$, $3); 04721 %*/ 04722 } 04723 ; 04724 04725 f_block_optarg : f_block_opt 04726 { 04727 /*%%%*/ 04728 $$ = $1; 04729 /*% 04730 $$ = rb_ary_new3(1, $1); 04731 %*/ 04732 } 04733 | f_block_optarg ',' f_block_opt 04734 { 04735 /*%%%*/ 04736 NODE *opts = $1; 04737 04738 while (opts->nd_next) { 04739 opts = opts->nd_next; 04740 } 04741 opts->nd_next = $3; 04742 $$ = $1; 04743 /*% 04744 $$ = rb_ary_push($1, $3); 04745 %*/ 04746 } 04747 ; 04748 04749 f_optarg : f_opt 04750 { 04751 /*%%%*/ 04752 $$ = $1; 04753 /*% 04754 $$ = rb_ary_new3(1, $1); 04755 %*/ 04756 } 04757 | f_optarg ',' f_opt 04758 { 04759 /*%%%*/ 04760 NODE *opts = $1; 04761 04762 while (opts->nd_next) { 04763 opts = opts->nd_next; 04764 } 04765 opts->nd_next = $3; 04766 $$ = $1; 04767 /*% 04768 $$ = rb_ary_push($1, $3); 04769 %*/ 04770 } 04771 ; 04772 04773 restarg_mark : '*' 04774 | tSTAR 04775 ; 04776 04777 f_rest_arg : restarg_mark tIDENTIFIER 04778 { 04779 /*%%%*/ 04780 if (!is_local_id($2)) 04781 yyerror("rest argument must be local variable"); 04782 /*% %*/ 04783 arg_var(shadowing_lvar(get_id($2))); 04784 /*%%%*/ 04785 $$ = $2; 04786 /*% 04787 $$ = dispatch1(rest_param, $2); 04788 %*/ 04789 } 04790 | restarg_mark 04791 { 04792 /*%%%*/ 04793 $$ = internal_id(); 04794 arg_var($$); 04795 /*% 04796 $$ = dispatch1(rest_param, Qnil); 04797 %*/ 04798 } 04799 ; 04800 04801 blkarg_mark : '&' 04802 | tAMPER 04803 ; 04804 04805 f_block_arg : blkarg_mark tIDENTIFIER 04806 { 04807 /*%%%*/ 04808 if (!is_local_id($2)) 04809 yyerror("block argument must be local variable"); 04810 else if (!dyna_in_block() && local_id($2)) 04811 yyerror("duplicated block argument name"); 04812 /*% %*/ 04813 arg_var(shadowing_lvar(get_id($2))); 04814 /*%%%*/ 04815 $$ = $2; 04816 /*% 04817 $$ = dispatch1(blockarg, $2); 04818 %*/ 04819 } 04820 ; 04821 04822 opt_f_block_arg : ',' f_block_arg 04823 { 04824 $$ = $2; 04825 } 04826 | none 04827 { 04828 /*%%%*/ 04829 $$ = 0; 04830 /*% 04831 $$ = Qundef; 04832 %*/ 04833 } 04834 ; 04835 04836 singleton : var_ref 04837 { 04838 /*%%%*/ 04839 value_expr($1); 04840 $$ = $1; 04841 if (!$$) $$ = NEW_NIL(); 04842 /*% 04843 $$ = $1; 04844 %*/ 04845 } 04846 | '(' {lex_state = EXPR_BEG;} expr rparen 04847 { 04848 /*%%%*/ 04849 if ($3 == 0) { 04850 yyerror("can't define singleton method for ()."); 04851 } 04852 else { 04853 switch (nd_type($3)) { 04854 case NODE_STR: 04855 case NODE_DSTR: 04856 case NODE_XSTR: 04857 case NODE_DXSTR: 04858 case NODE_DREGX: 04859 case NODE_LIT: 04860 case NODE_ARRAY: 04861 case NODE_ZARRAY: 04862 yyerror("can't define singleton method for literals"); 04863 default: 04864 value_expr($3); 04865 break; 04866 } 04867 } 04868 $$ = $3; 04869 /*% 04870 $$ = dispatch1(paren, $3); 04871 %*/ 04872 } 04873 ; 04874 04875 assoc_list : none 04876 | assocs trailer 04877 { 04878 /*%%%*/ 04879 $$ = $1; 04880 /*% 04881 $$ = dispatch1(assoclist_from_args, $1); 04882 %*/ 04883 } 04884 ; 04885 04886 assocs : assoc 04887 /*%c%*/ 04888 /*%c 04889 { 04890 $$ = rb_ary_new3(1, $1); 04891 } 04892 %*/ 04893 | assocs ',' assoc 04894 { 04895 /*%%%*/ 04896 $$ = list_concat($1, $3); 04897 /*% 04898 $$ = rb_ary_push($1, $3); 04899 %*/ 04900 } 04901 ; 04902 04903 assoc : arg_value tASSOC arg_value 04904 { 04905 /*%%%*/ 04906 $$ = list_append(NEW_LIST($1), $3); 04907 /*% 04908 $$ = dispatch2(assoc_new, $1, $3); 04909 %*/ 04910 } 04911 | tLABEL arg_value 04912 { 04913 /*%%%*/ 04914 $$ = list_append(NEW_LIST(NEW_LIT(ID2SYM($1))), $2); 04915 /*% 04916 $$ = dispatch2(assoc_new, $1, $2); 04917 %*/ 04918 } 04919 | tDSTAR arg_value 04920 { 04921 /*%%%*/ 04922 $$ = list_append(NEW_LIST(0), $2); 04923 /*% 04924 $$ = dispatch1(assoc_splat, $2); 04925 %*/ 04926 } 04927 ; 04928 04929 ; 04930 04931 operation : tIDENTIFIER 04932 | tCONSTANT 04933 | tFID 04934 ; 04935 04936 operation2 : tIDENTIFIER 04937 | tCONSTANT 04938 | tFID 04939 | op 04940 ; 04941 04942 operation3 : tIDENTIFIER 04943 | tFID 04944 | op 04945 ; 04946 04947 dot_or_colon : '.' 04948 /*%c%*/ 04949 /*%c 04950 { $$ = $<val>1; } 04951 %*/ 04952 | tCOLON2 04953 /*%c%*/ 04954 /*%c 04955 { $$ = $<val>1; } 04956 %*/ 04957 ; 04958 04959 opt_terms : /* none */ 04960 | terms 04961 ; 04962 04963 opt_nl : /* none */ 04964 | '\n' 04965 ; 04966 04967 rparen : opt_nl ')' 04968 ; 04969 04970 rbracket : opt_nl ']' 04971 ; 04972 04973 trailer : /* none */ 04974 | '\n' 04975 | ',' 04976 ; 04977 04978 term : ';' {yyerrok;} 04979 | '\n' 04980 ; 04981 04982 terms : term 04983 | terms ';' {yyerrok;} 04984 ; 04985 04986 none : /* none */ 04987 { 04988 /*%%%*/ 04989 $$ = 0; 04990 /*% 04991 $$ = Qundef; 04992 %*/ 04993 } 04994 ; 04995 %% 04996 # undef parser 04997 # undef yylex 04998 # undef yylval 04999 # define yylval (*((YYSTYPE*)(parser->parser_yylval))) 05000 05001 static int parser_regx_options(struct parser_params*); 05002 static int parser_tokadd_string(struct parser_params*,int,int,int,long*,rb_encoding**); 05003 static void parser_tokaddmbc(struct parser_params *parser, int c, rb_encoding *enc); 05004 static int parser_parse_string(struct parser_params*,NODE*); 05005 static int parser_here_document(struct parser_params*,NODE*); 05006 05007 05008 # define nextc() parser_nextc(parser) 05009 # define pushback(c) parser_pushback(parser, (c)) 05010 # define newtok() parser_newtok(parser) 05011 # define tokspace(n) parser_tokspace(parser, (n)) 05012 # define tokadd(c) parser_tokadd(parser, (c)) 05013 # define tok_hex(numlen) parser_tok_hex(parser, (numlen)) 05014 # define read_escape(flags,e) parser_read_escape(parser, (flags), (e)) 05015 # define tokadd_escape(e) parser_tokadd_escape(parser, (e)) 05016 # define regx_options() parser_regx_options(parser) 05017 # define tokadd_string(f,t,p,n,e) parser_tokadd_string(parser,(f),(t),(p),(n),(e)) 05018 # define parse_string(n) parser_parse_string(parser,(n)) 05019 # define tokaddmbc(c, enc) parser_tokaddmbc(parser, (c), (enc)) 05020 # define here_document(n) parser_here_document(parser,(n)) 05021 # define heredoc_identifier() parser_heredoc_identifier(parser) 05022 # define heredoc_restore(n) parser_heredoc_restore(parser,(n)) 05023 # define whole_match_p(e,l,i) parser_whole_match_p(parser,(e),(l),(i)) 05024 05025 #ifndef RIPPER 05026 # define set_yylval_str(x) (yylval.node = NEW_STR(x)) 05027 # define set_yylval_num(x) (yylval.num = (x)) 05028 # define set_yylval_id(x) (yylval.id = (x)) 05029 # define set_yylval_name(x) (yylval.id = (x)) 05030 # define set_yylval_literal(x) (yylval.node = NEW_LIT(x)) 05031 # define set_yylval_node(x) (yylval.node = (x)) 05032 # define yylval_id() (yylval.id) 05033 #else 05034 static inline VALUE 05035 ripper_yylval_id(ID x) 05036 { 05037 return (VALUE)NEW_LASGN(x, ID2SYM(x)); 05038 } 05039 # define set_yylval_str(x) (void)(x) 05040 # define set_yylval_num(x) (void)(x) 05041 # define set_yylval_id(x) (void)(x) 05042 # define set_yylval_name(x) (void)(yylval.val = ripper_yylval_id(x)) 05043 # define set_yylval_literal(x) (void)(x) 05044 # define set_yylval_node(x) (void)(x) 05045 # define yylval_id() yylval.id 05046 #endif 05047 05048 #ifndef RIPPER 05049 #define ripper_flush(p) (void)(p) 05050 #else 05051 #define ripper_flush(p) ((p)->tokp = (p)->parser_lex_p) 05052 05053 #define yylval_rval (*(RB_TYPE_P(yylval.val, T_NODE) ? &yylval.node->nd_rval : &yylval.val)) 05054 05055 static int 05056 ripper_has_scan_event(struct parser_params *parser) 05057 { 05058 05059 if (lex_p < parser->tokp) rb_raise(rb_eRuntimeError, "lex_p < tokp"); 05060 return lex_p > parser->tokp; 05061 } 05062 05063 static VALUE 05064 ripper_scan_event_val(struct parser_params *parser, int t) 05065 { 05066 VALUE str = STR_NEW(parser->tokp, lex_p - parser->tokp); 05067 VALUE rval = ripper_dispatch1(parser, ripper_token2eventid(t), str); 05068 ripper_flush(parser); 05069 return rval; 05070 } 05071 05072 static void 05073 ripper_dispatch_scan_event(struct parser_params *parser, int t) 05074 { 05075 if (!ripper_has_scan_event(parser)) return; 05076 yylval_rval = ripper_scan_event_val(parser, t); 05077 } 05078 05079 static void 05080 ripper_dispatch_ignored_scan_event(struct parser_params *parser, int t) 05081 { 05082 if (!ripper_has_scan_event(parser)) return; 05083 (void)ripper_scan_event_val(parser, t); 05084 } 05085 05086 static void 05087 ripper_dispatch_delayed_token(struct parser_params *parser, int t) 05088 { 05089 int saved_line = ruby_sourceline; 05090 const char *saved_tokp = parser->tokp; 05091 05092 ruby_sourceline = parser->delayed_line; 05093 parser->tokp = lex_pbeg + parser->delayed_col; 05094 yylval_rval = ripper_dispatch1(parser, ripper_token2eventid(t), parser->delayed); 05095 parser->delayed = Qnil; 05096 ruby_sourceline = saved_line; 05097 parser->tokp = saved_tokp; 05098 } 05099 #endif /* RIPPER */ 05100 05101 #include "ruby/regex.h" 05102 #include "ruby/util.h" 05103 05104 /* We remove any previous definition of `SIGN_EXTEND_CHAR', 05105 since ours (we hope) works properly with all combinations of 05106 machines, compilers, `char' and `unsigned char' argument types. 05107 (Per Bothner suggested the basic approach.) */ 05108 #undef SIGN_EXTEND_CHAR 05109 #if __STDC__ 05110 # define SIGN_EXTEND_CHAR(c) ((signed char)(c)) 05111 #else /* not __STDC__ */ 05112 /* As in Harbison and Steele. */ 05113 # define SIGN_EXTEND_CHAR(c) ((((unsigned char)(c)) ^ 128) - 128) 05114 #endif 05115 05116 #define parser_encoding_name() (current_enc->name) 05117 #define parser_mbclen() mbclen((lex_p-1),lex_pend,current_enc) 05118 #define parser_precise_mbclen() rb_enc_precise_mbclen((lex_p-1),lex_pend,current_enc) 05119 #define is_identchar(p,e,enc) (rb_enc_isalnum(*(p),(enc)) || (*(p)) == '_' || !ISASCII(*(p))) 05120 #define parser_is_identchar() (!parser->eofp && is_identchar((lex_p-1),lex_pend,current_enc)) 05121 05122 #define parser_isascii() ISASCII(*(lex_p-1)) 05123 05124 #ifndef RIPPER 05125 static int 05126 token_info_get_column(struct parser_params *parser, const char *token) 05127 { 05128 int column = 1; 05129 const char *p, *pend = lex_p - strlen(token); 05130 for (p = lex_pbeg; p < pend; p++) { 05131 if (*p == '\t') { 05132 column = (((column - 1) / 8) + 1) * 8; 05133 } 05134 column++; 05135 } 05136 return column; 05137 } 05138 05139 static int 05140 token_info_has_nonspaces(struct parser_params *parser, const char *token) 05141 { 05142 const char *p, *pend = lex_p - strlen(token); 05143 for (p = lex_pbeg; p < pend; p++) { 05144 if (*p != ' ' && *p != '\t') { 05145 return 1; 05146 } 05147 } 05148 return 0; 05149 } 05150 05151 #undef token_info_push 05152 static void 05153 token_info_push(struct parser_params *parser, const char *token) 05154 { 05155 token_info *ptinfo; 05156 05157 if (!parser->parser_token_info_enabled) return; 05158 ptinfo = ALLOC(token_info); 05159 ptinfo->token = token; 05160 ptinfo->linenum = ruby_sourceline; 05161 ptinfo->column = token_info_get_column(parser, token); 05162 ptinfo->nonspc = token_info_has_nonspaces(parser, token); 05163 ptinfo->next = parser->parser_token_info; 05164 05165 parser->parser_token_info = ptinfo; 05166 } 05167 05168 #undef token_info_pop 05169 static void 05170 token_info_pop(struct parser_params *parser, const char *token) 05171 { 05172 int linenum; 05173 token_info *ptinfo = parser->parser_token_info; 05174 05175 if (!ptinfo) return; 05176 parser->parser_token_info = ptinfo->next; 05177 if (token_info_get_column(parser, token) == ptinfo->column) { /* OK */ 05178 goto finish; 05179 } 05180 linenum = ruby_sourceline; 05181 if (linenum == ptinfo->linenum) { /* SKIP */ 05182 goto finish; 05183 } 05184 if (token_info_has_nonspaces(parser, token) || ptinfo->nonspc) { /* SKIP */ 05185 goto finish; 05186 } 05187 if (parser->parser_token_info_enabled) { 05188 rb_compile_warn(ruby_sourcefile, linenum, 05189 "mismatched indentations at '%s' with '%s' at %d", 05190 token, ptinfo->token, ptinfo->linenum); 05191 } 05192 05193 finish: 05194 xfree(ptinfo); 05195 } 05196 #endif /* RIPPER */ 05197 05198 static int 05199 parser_yyerror(struct parser_params *parser, const char *msg) 05200 { 05201 #ifndef RIPPER 05202 const int max_line_margin = 30; 05203 const char *p, *pe; 05204 char *buf; 05205 long len; 05206 int i; 05207 05208 compile_error(PARSER_ARG "%s", msg); 05209 p = lex_p; 05210 while (lex_pbeg <= p) { 05211 if (*p == '\n') break; 05212 p--; 05213 } 05214 p++; 05215 05216 pe = lex_p; 05217 while (pe < lex_pend) { 05218 if (*pe == '\n') break; 05219 pe++; 05220 } 05221 05222 len = pe - p; 05223 if (len > 4) { 05224 char *p2; 05225 const char *pre = "", *post = ""; 05226 05227 if (len > max_line_margin * 2 + 10) { 05228 if (lex_p - p > max_line_margin) { 05229 p = rb_enc_prev_char(p, lex_p - max_line_margin, pe, rb_enc_get(lex_lastline)); 05230 pre = "..."; 05231 } 05232 if (pe - lex_p > max_line_margin) { 05233 pe = rb_enc_prev_char(lex_p, lex_p + max_line_margin, pe, rb_enc_get(lex_lastline)); 05234 post = "..."; 05235 } 05236 len = pe - p; 05237 } 05238 buf = ALLOCA_N(char, len+2); 05239 MEMCPY(buf, p, char, len); 05240 buf[len] = '\0'; 05241 rb_compile_error_append("%s%s%s", pre, buf, post); 05242 05243 i = (int)(lex_p - p); 05244 p2 = buf; pe = buf + len; 05245 05246 while (p2 < pe) { 05247 if (*p2 != '\t') *p2 = ' '; 05248 p2++; 05249 } 05250 buf[i] = '^'; 05251 buf[i+1] = '\0'; 05252 rb_compile_error_append("%s%s", pre, buf); 05253 } 05254 #else 05255 dispatch1(parse_error, STR_NEW2(msg)); 05256 #endif /* !RIPPER */ 05257 return 0; 05258 } 05259 05260 static void parser_prepare(struct parser_params *parser); 05261 05262 #ifndef RIPPER 05263 static VALUE 05264 debug_lines(const char *f) 05265 { 05266 ID script_lines; 05267 CONST_ID(script_lines, "SCRIPT_LINES__"); 05268 if (rb_const_defined_at(rb_cObject, script_lines)) { 05269 VALUE hash = rb_const_get_at(rb_cObject, script_lines); 05270 if (RB_TYPE_P(hash, T_HASH)) { 05271 VALUE fname = rb_external_str_new_with_enc(f, strlen(f), rb_filesystem_encoding()); 05272 VALUE lines = rb_ary_new(); 05273 rb_hash_aset(hash, fname, lines); 05274 return lines; 05275 } 05276 } 05277 return 0; 05278 } 05279 05280 static VALUE 05281 coverage(const char *f, int n) 05282 { 05283 VALUE coverages = rb_get_coverages(); 05284 if (RTEST(coverages) && RBASIC(coverages)->klass == 0) { 05285 VALUE fname = rb_external_str_new_with_enc(f, strlen(f), rb_filesystem_encoding()); 05286 VALUE lines = rb_ary_new2(n); 05287 int i; 05288 RBASIC(lines)->klass = 0; 05289 for (i = 0; i < n; i++) RARRAY_PTR(lines)[i] = Qnil; 05290 RARRAY(lines)->as.heap.len = n; 05291 rb_hash_aset(coverages, fname, lines); 05292 return lines; 05293 } 05294 return 0; 05295 } 05296 05297 static int 05298 e_option_supplied(struct parser_params *parser) 05299 { 05300 return strcmp(ruby_sourcefile, "-e") == 0; 05301 } 05302 05303 static VALUE 05304 yycompile0(VALUE arg) 05305 { 05306 int n; 05307 NODE *tree; 05308 struct parser_params *parser = (struct parser_params *)arg; 05309 05310 if (!compile_for_eval && rb_safe_level() == 0) { 05311 ruby_debug_lines = debug_lines(ruby_sourcefile); 05312 if (ruby_debug_lines && ruby_sourceline > 0) { 05313 VALUE str = STR_NEW0(); 05314 n = ruby_sourceline; 05315 do { 05316 rb_ary_push(ruby_debug_lines, str); 05317 } while (--n); 05318 } 05319 05320 if (!e_option_supplied(parser)) { 05321 ruby_coverage = coverage(ruby_sourcefile, ruby_sourceline); 05322 } 05323 } 05324 05325 parser_prepare(parser); 05326 deferred_nodes = 0; 05327 #ifndef RIPPER 05328 parser->parser_token_info_enabled = !compile_for_eval && RTEST(ruby_verbose); 05329 #endif 05330 #ifndef RIPPER 05331 if (RUBY_DTRACE_PARSE_BEGIN_ENABLED()) { 05332 RUBY_DTRACE_PARSE_BEGIN(parser->parser_ruby_sourcefile, 05333 parser->parser_ruby_sourceline); 05334 } 05335 #endif 05336 n = yyparse((void*)parser); 05337 #ifndef RIPPER 05338 if (RUBY_DTRACE_PARSE_END_ENABLED()) { 05339 RUBY_DTRACE_PARSE_END(parser->parser_ruby_sourcefile, 05340 parser->parser_ruby_sourceline); 05341 } 05342 #endif 05343 ruby_debug_lines = 0; 05344 ruby_coverage = 0; 05345 compile_for_eval = 0; 05346 05347 lex_strterm = 0; 05348 lex_p = lex_pbeg = lex_pend = 0; 05349 lex_lastline = lex_nextline = 0; 05350 if (parser->nerr) { 05351 return 0; 05352 } 05353 tree = ruby_eval_tree; 05354 if (!tree) { 05355 tree = NEW_NIL(); 05356 } 05357 else if (ruby_eval_tree_begin) { 05358 tree->nd_body = NEW_PRELUDE(ruby_eval_tree_begin, tree->nd_body); 05359 } 05360 return (VALUE)tree; 05361 } 05362 05363 static NODE* 05364 yycompile(struct parser_params *parser, const char *f, int line) 05365 { 05366 ruby_sourcefile = ruby_strdup(f); 05367 ruby_sourceline = line - 1; 05368 return (NODE *)rb_suppress_tracing(yycompile0, (VALUE)parser); 05369 } 05370 #endif /* !RIPPER */ 05371 05372 static rb_encoding * 05373 must_be_ascii_compatible(VALUE s) 05374 { 05375 rb_encoding *enc = rb_enc_get(s); 05376 if (!rb_enc_asciicompat(enc)) { 05377 rb_raise(rb_eArgError, "invalid source encoding"); 05378 } 05379 return enc; 05380 } 05381 05382 static VALUE 05383 lex_get_str(struct parser_params *parser, VALUE s) 05384 { 05385 char *beg, *end, *pend; 05386 rb_encoding *enc = must_be_ascii_compatible(s); 05387 05388 beg = RSTRING_PTR(s); 05389 if (lex_gets_ptr) { 05390 if (RSTRING_LEN(s) == lex_gets_ptr) return Qnil; 05391 beg += lex_gets_ptr; 05392 } 05393 pend = RSTRING_PTR(s) + RSTRING_LEN(s); 05394 end = beg; 05395 while (end < pend) { 05396 if (*end++ == '\n') break; 05397 } 05398 lex_gets_ptr = end - RSTRING_PTR(s); 05399 return rb_enc_str_new(beg, end - beg, enc); 05400 } 05401 05402 static VALUE 05403 lex_getline(struct parser_params *parser) 05404 { 05405 VALUE line = (*parser->parser_lex_gets)(parser, parser->parser_lex_input); 05406 if (NIL_P(line)) return line; 05407 must_be_ascii_compatible(line); 05408 #ifndef RIPPER 05409 if (ruby_debug_lines) { 05410 rb_enc_associate(line, current_enc); 05411 rb_ary_push(ruby_debug_lines, line); 05412 } 05413 if (ruby_coverage) { 05414 rb_ary_push(ruby_coverage, Qnil); 05415 } 05416 #endif 05417 return line; 05418 } 05419 05420 #ifdef RIPPER 05421 static rb_data_type_t parser_data_type; 05422 #else 05423 static const rb_data_type_t parser_data_type; 05424 05425 static NODE* 05426 parser_compile_string(volatile VALUE vparser, const char *f, VALUE s, int line) 05427 { 05428 struct parser_params *parser; 05429 NODE *node; 05430 05431 TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, parser); 05432 lex_gets = lex_get_str; 05433 lex_gets_ptr = 0; 05434 lex_input = s; 05435 lex_pbeg = lex_p = lex_pend = 0; 05436 compile_for_eval = rb_parse_in_eval(); 05437 05438 node = yycompile(parser, f, line); 05439 RB_GC_GUARD(vparser); /* prohibit tail call optimization */ 05440 05441 return node; 05442 } 05443 05444 NODE* 05445 rb_compile_string(const char *f, VALUE s, int line) 05446 { 05447 must_be_ascii_compatible(s); 05448 return parser_compile_string(rb_parser_new(), f, s, line); 05449 } 05450 05451 NODE* 05452 rb_parser_compile_string(volatile VALUE vparser, const char *f, VALUE s, int line) 05453 { 05454 must_be_ascii_compatible(s); 05455 return parser_compile_string(vparser, f, s, line); 05456 } 05457 05458 NODE* 05459 rb_compile_cstr(const char *f, const char *s, int len, int line) 05460 { 05461 VALUE str = rb_str_new(s, len); 05462 return parser_compile_string(rb_parser_new(), f, str, line); 05463 } 05464 05465 NODE* 05466 rb_parser_compile_cstr(volatile VALUE vparser, const char *f, const char *s, int len, int line) 05467 { 05468 VALUE str = rb_str_new(s, len); 05469 return parser_compile_string(vparser, f, str, line); 05470 } 05471 05472 static VALUE 05473 lex_io_gets(struct parser_params *parser, VALUE io) 05474 { 05475 return rb_io_gets(io); 05476 } 05477 05478 NODE* 05479 rb_compile_file(const char *f, VALUE file, int start) 05480 { 05481 VALUE volatile vparser = rb_parser_new(); 05482 05483 return rb_parser_compile_file(vparser, f, file, start); 05484 } 05485 05486 NODE* 05487 rb_parser_compile_file(volatile VALUE vparser, const char *f, VALUE file, int start) 05488 { 05489 struct parser_params *parser; 05490 NODE *node; 05491 05492 TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, parser); 05493 lex_gets = lex_io_gets; 05494 lex_input = file; 05495 lex_pbeg = lex_p = lex_pend = 0; 05496 compile_for_eval = rb_parse_in_eval(); 05497 05498 node = yycompile(parser, f, start); 05499 RB_GC_GUARD(vparser); /* prohibit tail call optimization */ 05500 05501 return node; 05502 } 05503 #endif /* !RIPPER */ 05504 05505 #define STR_FUNC_ESCAPE 0x01 05506 #define STR_FUNC_EXPAND 0x02 05507 #define STR_FUNC_REGEXP 0x04 05508 #define STR_FUNC_QWORDS 0x08 05509 #define STR_FUNC_SYMBOL 0x10 05510 #define STR_FUNC_INDENT 0x20 05511 05512 enum string_type { 05513 str_squote = (0), 05514 str_dquote = (STR_FUNC_EXPAND), 05515 str_xquote = (STR_FUNC_EXPAND), 05516 str_regexp = (STR_FUNC_REGEXP|STR_FUNC_ESCAPE|STR_FUNC_EXPAND), 05517 str_sword = (STR_FUNC_QWORDS), 05518 str_dword = (STR_FUNC_QWORDS|STR_FUNC_EXPAND), 05519 str_ssym = (STR_FUNC_SYMBOL), 05520 str_dsym = (STR_FUNC_SYMBOL|STR_FUNC_EXPAND) 05521 }; 05522 05523 static VALUE 05524 parser_str_new(const char *p, long n, rb_encoding *enc, int func, rb_encoding *enc0) 05525 { 05526 VALUE str; 05527 05528 str = rb_enc_str_new(p, n, enc); 05529 if (!(func & STR_FUNC_REGEXP) && rb_enc_asciicompat(enc)) { 05530 if (rb_enc_str_coderange(str) == ENC_CODERANGE_7BIT) { 05531 } 05532 else if (enc0 == rb_usascii_encoding() && enc != rb_utf8_encoding()) { 05533 rb_enc_associate(str, rb_ascii8bit_encoding()); 05534 } 05535 } 05536 05537 return str; 05538 } 05539 05540 #define lex_goto_eol(parser) ((parser)->parser_lex_p = (parser)->parser_lex_pend) 05541 #define lex_eol_p() (lex_p >= lex_pend) 05542 #define peek(c) peek_n((c), 0) 05543 #define peek_n(c,n) (lex_p+(n) < lex_pend && (c) == (unsigned char)lex_p[n]) 05544 05545 static inline int 05546 parser_nextc(struct parser_params *parser) 05547 { 05548 int c; 05549 05550 if (lex_p == lex_pend) { 05551 VALUE v = lex_nextline; 05552 lex_nextline = 0; 05553 if (!v) { 05554 if (parser->eofp) 05555 return -1; 05556 05557 if (!lex_input || NIL_P(v = lex_getline(parser))) { 05558 parser->eofp = Qtrue; 05559 lex_goto_eol(parser); 05560 return -1; 05561 } 05562 } 05563 { 05564 #ifdef RIPPER 05565 if (parser->tokp < lex_pend) { 05566 if (NIL_P(parser->delayed)) { 05567 parser->delayed = rb_str_buf_new(1024); 05568 rb_enc_associate(parser->delayed, current_enc); 05569 rb_str_buf_cat(parser->delayed, 05570 parser->tokp, lex_pend - parser->tokp); 05571 parser->delayed_line = ruby_sourceline; 05572 parser->delayed_col = (int)(parser->tokp - lex_pbeg); 05573 } 05574 else { 05575 rb_str_buf_cat(parser->delayed, 05576 parser->tokp, lex_pend - parser->tokp); 05577 } 05578 } 05579 #endif 05580 if (heredoc_end > 0) { 05581 ruby_sourceline = heredoc_end; 05582 heredoc_end = 0; 05583 } 05584 ruby_sourceline++; 05585 parser->line_count++; 05586 lex_pbeg = lex_p = RSTRING_PTR(v); 05587 lex_pend = lex_p + RSTRING_LEN(v); 05588 ripper_flush(parser); 05589 lex_lastline = v; 05590 } 05591 } 05592 c = (unsigned char)*lex_p++; 05593 if (c == '\r' && peek('\n')) { 05594 lex_p++; 05595 c = '\n'; 05596 } 05597 05598 return c; 05599 } 05600 05601 static void 05602 parser_pushback(struct parser_params *parser, int c) 05603 { 05604 if (c == -1) return; 05605 lex_p--; 05606 if (lex_p > lex_pbeg && lex_p[0] == '\n' && lex_p[-1] == '\r') { 05607 lex_p--; 05608 } 05609 } 05610 05611 #define was_bol() (lex_p == lex_pbeg + 1) 05612 05613 #define tokfix() (tokenbuf[tokidx]='\0') 05614 #define tok() tokenbuf 05615 #define toklen() tokidx 05616 #define toklast() (tokidx>0?tokenbuf[tokidx-1]:0) 05617 05618 static char* 05619 parser_newtok(struct parser_params *parser) 05620 { 05621 tokidx = 0; 05622 tokline = ruby_sourceline; 05623 if (!tokenbuf) { 05624 toksiz = 60; 05625 tokenbuf = ALLOC_N(char, 60); 05626 } 05627 if (toksiz > 4096) { 05628 toksiz = 60; 05629 REALLOC_N(tokenbuf, char, 60); 05630 } 05631 return tokenbuf; 05632 } 05633 05634 static char * 05635 parser_tokspace(struct parser_params *parser, int n) 05636 { 05637 tokidx += n; 05638 05639 if (tokidx >= toksiz) { 05640 do {toksiz *= 2;} while (toksiz < tokidx); 05641 REALLOC_N(tokenbuf, char, toksiz); 05642 } 05643 return &tokenbuf[tokidx-n]; 05644 } 05645 05646 static void 05647 parser_tokadd(struct parser_params *parser, int c) 05648 { 05649 tokenbuf[tokidx++] = (char)c; 05650 if (tokidx >= toksiz) { 05651 toksiz *= 2; 05652 REALLOC_N(tokenbuf, char, toksiz); 05653 } 05654 } 05655 05656 static int 05657 parser_tok_hex(struct parser_params *parser, size_t *numlen) 05658 { 05659 int c; 05660 05661 c = scan_hex(lex_p, 2, numlen); 05662 if (!*numlen) { 05663 yyerror("invalid hex escape"); 05664 return 0; 05665 } 05666 lex_p += *numlen; 05667 return c; 05668 } 05669 05670 #define tokcopy(n) memcpy(tokspace(n), lex_p - (n), (n)) 05671 05672 /* return value is for ?\u3042 */ 05673 static int 05674 parser_tokadd_utf8(struct parser_params *parser, rb_encoding **encp, 05675 int string_literal, int symbol_literal, int regexp_literal) 05676 { 05677 /* 05678 * If string_literal is true, then we allow multiple codepoints 05679 * in \u{}, and add the codepoints to the current token. 05680 * Otherwise we're parsing a character literal and return a single 05681 * codepoint without adding it 05682 */ 05683 05684 int codepoint; 05685 size_t numlen; 05686 05687 if (regexp_literal) { tokadd('\\'); tokadd('u'); } 05688 05689 if (peek('{')) { /* handle \u{...} form */ 05690 do { 05691 if (regexp_literal) { tokadd(*lex_p); } 05692 nextc(); 05693 codepoint = scan_hex(lex_p, 6, &numlen); 05694 if (numlen == 0) { 05695 yyerror("invalid Unicode escape"); 05696 return 0; 05697 } 05698 if (codepoint > 0x10ffff) { 05699 yyerror("invalid Unicode codepoint (too large)"); 05700 return 0; 05701 } 05702 lex_p += numlen; 05703 if (regexp_literal) { 05704 tokcopy((int)numlen); 05705 } 05706 else if (codepoint >= 0x80) { 05707 *encp = rb_utf8_encoding(); 05708 if (string_literal) tokaddmbc(codepoint, *encp); 05709 } 05710 else if (string_literal) { 05711 tokadd(codepoint); 05712 } 05713 } while (string_literal && (peek(' ') || peek('\t'))); 05714 05715 if (!peek('}')) { 05716 yyerror("unterminated Unicode escape"); 05717 return 0; 05718 } 05719 05720 if (regexp_literal) { tokadd('}'); } 05721 nextc(); 05722 } 05723 else { /* handle \uxxxx form */ 05724 codepoint = scan_hex(lex_p, 4, &numlen); 05725 if (numlen < 4) { 05726 yyerror("invalid Unicode escape"); 05727 return 0; 05728 } 05729 lex_p += 4; 05730 if (regexp_literal) { 05731 tokcopy(4); 05732 } 05733 else if (codepoint >= 0x80) { 05734 *encp = rb_utf8_encoding(); 05735 if (string_literal) tokaddmbc(codepoint, *encp); 05736 } 05737 else if (string_literal) { 05738 tokadd(codepoint); 05739 } 05740 } 05741 05742 return codepoint; 05743 } 05744 05745 #define ESCAPE_CONTROL 1 05746 #define ESCAPE_META 2 05747 05748 static int 05749 parser_read_escape(struct parser_params *parser, int flags, 05750 rb_encoding **encp) 05751 { 05752 int c; 05753 size_t numlen; 05754 05755 switch (c = nextc()) { 05756 case '\\': /* Backslash */ 05757 return c; 05758 05759 case 'n': /* newline */ 05760 return '\n'; 05761 05762 case 't': /* horizontal tab */ 05763 return '\t'; 05764 05765 case 'r': /* carriage-return */ 05766 return '\r'; 05767 05768 case 'f': /* form-feed */ 05769 return '\f'; 05770 05771 case 'v': /* vertical tab */ 05772 return '\13'; 05773 05774 case 'a': /* alarm(bell) */ 05775 return '\007'; 05776 05777 case 'e': /* escape */ 05778 return 033; 05779 05780 case '0': case '1': case '2': case '3': /* octal constant */ 05781 case '4': case '5': case '6': case '7': 05782 pushback(c); 05783 c = scan_oct(lex_p, 3, &numlen); 05784 lex_p += numlen; 05785 return c; 05786 05787 case 'x': /* hex constant */ 05788 c = tok_hex(&numlen); 05789 if (numlen == 0) return 0; 05790 return c; 05791 05792 case 'b': /* backspace */ 05793 return '\010'; 05794 05795 case 's': /* space */ 05796 return ' '; 05797 05798 case 'M': 05799 if (flags & ESCAPE_META) goto eof; 05800 if ((c = nextc()) != '-') { 05801 pushback(c); 05802 goto eof; 05803 } 05804 if ((c = nextc()) == '\\') { 05805 if (peek('u')) goto eof; 05806 return read_escape(flags|ESCAPE_META, encp) | 0x80; 05807 } 05808 else if (c == -1 || !ISASCII(c)) goto eof; 05809 else { 05810 return ((c & 0xff) | 0x80); 05811 } 05812 05813 case 'C': 05814 if ((c = nextc()) != '-') { 05815 pushback(c); 05816 goto eof; 05817 } 05818 case 'c': 05819 if (flags & ESCAPE_CONTROL) goto eof; 05820 if ((c = nextc())== '\\') { 05821 if (peek('u')) goto eof; 05822 c = read_escape(flags|ESCAPE_CONTROL, encp); 05823 } 05824 else if (c == '?') 05825 return 0177; 05826 else if (c == -1 || !ISASCII(c)) goto eof; 05827 return c & 0x9f; 05828 05829 eof: 05830 case -1: 05831 yyerror("Invalid escape character syntax"); 05832 return '\0'; 05833 05834 default: 05835 return c; 05836 } 05837 } 05838 05839 static void 05840 parser_tokaddmbc(struct parser_params *parser, int c, rb_encoding *enc) 05841 { 05842 int len = rb_enc_codelen(c, enc); 05843 rb_enc_mbcput(c, tokspace(len), enc); 05844 } 05845 05846 static int 05847 parser_tokadd_escape(struct parser_params *parser, rb_encoding **encp) 05848 { 05849 int c; 05850 int flags = 0; 05851 size_t numlen; 05852 05853 first: 05854 switch (c = nextc()) { 05855 case '\n': 05856 return 0; /* just ignore */ 05857 05858 case '0': case '1': case '2': case '3': /* octal constant */ 05859 case '4': case '5': case '6': case '7': 05860 { 05861 ruby_scan_oct(--lex_p, 3, &numlen); 05862 if (numlen == 0) goto eof; 05863 lex_p += numlen; 05864 tokcopy((int)numlen + 1); 05865 } 05866 return 0; 05867 05868 case 'x': /* hex constant */ 05869 { 05870 tok_hex(&numlen); 05871 if (numlen == 0) return -1; 05872 tokcopy((int)numlen + 2); 05873 } 05874 return 0; 05875 05876 case 'M': 05877 if (flags & ESCAPE_META) goto eof; 05878 if ((c = nextc()) != '-') { 05879 pushback(c); 05880 goto eof; 05881 } 05882 tokcopy(3); 05883 flags |= ESCAPE_META; 05884 goto escaped; 05885 05886 case 'C': 05887 if (flags & ESCAPE_CONTROL) goto eof; 05888 if ((c = nextc()) != '-') { 05889 pushback(c); 05890 goto eof; 05891 } 05892 tokcopy(3); 05893 goto escaped; 05894 05895 case 'c': 05896 if (flags & ESCAPE_CONTROL) goto eof; 05897 tokcopy(2); 05898 flags |= ESCAPE_CONTROL; 05899 escaped: 05900 if ((c = nextc()) == '\\') { 05901 goto first; 05902 } 05903 else if (c == -1) goto eof; 05904 tokadd(c); 05905 return 0; 05906 05907 eof: 05908 case -1: 05909 yyerror("Invalid escape character syntax"); 05910 return -1; 05911 05912 default: 05913 tokadd('\\'); 05914 tokadd(c); 05915 } 05916 return 0; 05917 } 05918 05919 static int 05920 parser_regx_options(struct parser_params *parser) 05921 { 05922 int kcode = 0; 05923 int kopt = 0; 05924 int options = 0; 05925 int c, opt, kc; 05926 05927 newtok(); 05928 while (c = nextc(), ISALPHA(c)) { 05929 if (c == 'o') { 05930 options |= RE_OPTION_ONCE; 05931 } 05932 else if (rb_char_to_option_kcode(c, &opt, &kc)) { 05933 if (kc >= 0) { 05934 if (kc != rb_ascii8bit_encindex()) kcode = c; 05935 kopt = opt; 05936 } 05937 else { 05938 options |= opt; 05939 } 05940 } 05941 else { 05942 tokadd(c); 05943 } 05944 } 05945 options |= kopt; 05946 pushback(c); 05947 if (toklen()) { 05948 tokfix(); 05949 compile_error(PARSER_ARG "unknown regexp option%s - %s", 05950 toklen() > 1 ? "s" : "", tok()); 05951 } 05952 return options | RE_OPTION_ENCODING(kcode); 05953 } 05954 05955 static void 05956 dispose_string(VALUE str) 05957 { 05958 rb_str_free(str); 05959 rb_gc_force_recycle(str); 05960 } 05961 05962 static int 05963 parser_tokadd_mbchar(struct parser_params *parser, int c) 05964 { 05965 int len = parser_precise_mbclen(); 05966 if (!MBCLEN_CHARFOUND_P(len)) { 05967 compile_error(PARSER_ARG "invalid multibyte char (%s)", parser_encoding_name()); 05968 return -1; 05969 } 05970 tokadd(c); 05971 lex_p += --len; 05972 if (len > 0) tokcopy(len); 05973 return c; 05974 } 05975 05976 #define tokadd_mbchar(c) parser_tokadd_mbchar(parser, (c)) 05977 05978 static inline int 05979 simple_re_meta(int c) 05980 { 05981 switch (c) { 05982 case '$': case '*': case '+': case '.': 05983 case '?': case '^': case '|': 05984 case ')': case ']': case '}': case '>': 05985 return TRUE; 05986 default: 05987 return FALSE; 05988 } 05989 } 05990 05991 static int 05992 parser_tokadd_string(struct parser_params *parser, 05993 int func, int term, int paren, long *nest, 05994 rb_encoding **encp) 05995 { 05996 int c; 05997 int has_nonascii = 0; 05998 rb_encoding *enc = *encp; 05999 char *errbuf = 0; 06000 static const char mixed_msg[] = "%s mixed within %s source"; 06001 06002 #define mixed_error(enc1, enc2) if (!errbuf) { \ 06003 size_t len = sizeof(mixed_msg) - 4; \ 06004 len += strlen(rb_enc_name(enc1)); \ 06005 len += strlen(rb_enc_name(enc2)); \ 06006 errbuf = ALLOCA_N(char, len); \ 06007 snprintf(errbuf, len, mixed_msg, \ 06008 rb_enc_name(enc1), \ 06009 rb_enc_name(enc2)); \ 06010 yyerror(errbuf); \ 06011 } 06012 #define mixed_escape(beg, enc1, enc2) do { \ 06013 const char *pos = lex_p; \ 06014 lex_p = (beg); \ 06015 mixed_error((enc1), (enc2)); \ 06016 lex_p = pos; \ 06017 } while (0) 06018 06019 while ((c = nextc()) != -1) { 06020 if (paren && c == paren) { 06021 ++*nest; 06022 } 06023 else if (c == term) { 06024 if (!nest || !*nest) { 06025 pushback(c); 06026 break; 06027 } 06028 --*nest; 06029 } 06030 else if ((func & STR_FUNC_EXPAND) && c == '#' && lex_p < lex_pend) { 06031 int c2 = *lex_p; 06032 if (c2 == '$' || c2 == '@' || c2 == '{') { 06033 pushback(c); 06034 break; 06035 } 06036 } 06037 else if (c == '\\') { 06038 const char *beg = lex_p - 1; 06039 c = nextc(); 06040 switch (c) { 06041 case '\n': 06042 if (func & STR_FUNC_QWORDS) break; 06043 if (func & STR_FUNC_EXPAND) continue; 06044 tokadd('\\'); 06045 break; 06046 06047 case '\\': 06048 if (func & STR_FUNC_ESCAPE) tokadd(c); 06049 break; 06050 06051 case 'u': 06052 if ((func & STR_FUNC_EXPAND) == 0) { 06053 tokadd('\\'); 06054 break; 06055 } 06056 parser_tokadd_utf8(parser, &enc, 1, 06057 func & STR_FUNC_SYMBOL, 06058 func & STR_FUNC_REGEXP); 06059 if (has_nonascii && enc != *encp) { 06060 mixed_escape(beg, enc, *encp); 06061 } 06062 continue; 06063 06064 default: 06065 if (c == -1) return -1; 06066 if (!ISASCII(c)) { 06067 if ((func & STR_FUNC_EXPAND) == 0) tokadd('\\'); 06068 goto non_ascii; 06069 } 06070 if (func & STR_FUNC_REGEXP) { 06071 if (c == term && !simple_re_meta(c)) { 06072 tokadd(c); 06073 continue; 06074 } 06075 pushback(c); 06076 if ((c = tokadd_escape(&enc)) < 0) 06077 return -1; 06078 if (has_nonascii && enc != *encp) { 06079 mixed_escape(beg, enc, *encp); 06080 } 06081 continue; 06082 } 06083 else if (func & STR_FUNC_EXPAND) { 06084 pushback(c); 06085 if (func & STR_FUNC_ESCAPE) tokadd('\\'); 06086 c = read_escape(0, &enc); 06087 } 06088 else if ((func & STR_FUNC_QWORDS) && ISSPACE(c)) { 06089 /* ignore backslashed spaces in %w */ 06090 } 06091 else if (c != term && !(paren && c == paren)) { 06092 tokadd('\\'); 06093 pushback(c); 06094 continue; 06095 } 06096 } 06097 } 06098 else if (!parser_isascii()) { 06099 non_ascii: 06100 has_nonascii = 1; 06101 if (enc != *encp) { 06102 mixed_error(enc, *encp); 06103 continue; 06104 } 06105 if (tokadd_mbchar(c) == -1) return -1; 06106 continue; 06107 } 06108 else if ((func & STR_FUNC_QWORDS) && ISSPACE(c)) { 06109 pushback(c); 06110 break; 06111 } 06112 if (c & 0x80) { 06113 has_nonascii = 1; 06114 if (enc != *encp) { 06115 mixed_error(enc, *encp); 06116 continue; 06117 } 06118 } 06119 tokadd(c); 06120 } 06121 *encp = enc; 06122 return c; 06123 } 06124 06125 #define NEW_STRTERM(func, term, paren) \ 06126 rb_node_newnode(NODE_STRTERM, (func), (term) | ((paren) << (CHAR_BIT * 2)), 0) 06127 06128 #ifdef RIPPER 06129 static void 06130 ripper_flush_string_content(struct parser_params *parser, rb_encoding *enc) 06131 { 06132 if (!NIL_P(parser->delayed)) { 06133 ptrdiff_t len = lex_p - parser->tokp; 06134 if (len > 0) { 06135 rb_enc_str_buf_cat(parser->delayed, parser->tokp, len, enc); 06136 } 06137 ripper_dispatch_delayed_token(parser, tSTRING_CONTENT); 06138 parser->tokp = lex_p; 06139 } 06140 } 06141 06142 #define flush_string_content(enc) ripper_flush_string_content(parser, (enc)) 06143 #else 06144 #define flush_string_content(enc) ((void)(enc)) 06145 #endif 06146 06147 RUBY_FUNC_EXPORTED const unsigned int ruby_global_name_punct_bits[(0x7e - 0x20 + 31) / 32]; 06148 /* this can be shared with ripper, since it's independent from struct 06149 * parser_params. */ 06150 #ifndef RIPPER 06151 #define BIT(c, idx) (((c) / 32 - 1 == idx) ? (1U << ((c) % 32)) : 0) 06152 #define SPECIAL_PUNCT(idx) ( \ 06153 BIT('~', idx) | BIT('*', idx) | BIT('$', idx) | BIT('?', idx) | \ 06154 BIT('!', idx) | BIT('@', idx) | BIT('/', idx) | BIT('\\', idx) | \ 06155 BIT(';', idx) | BIT(',', idx) | BIT('.', idx) | BIT('=', idx) | \ 06156 BIT(':', idx) | BIT('<', idx) | BIT('>', idx) | BIT('\"', idx) | \ 06157 BIT('&', idx) | BIT('`', idx) | BIT('\'', idx) | BIT('+', idx) | \ 06158 BIT('0', idx)) 06159 const unsigned int ruby_global_name_punct_bits[] = { 06160 SPECIAL_PUNCT(0), 06161 SPECIAL_PUNCT(1), 06162 SPECIAL_PUNCT(2), 06163 }; 06164 #undef BIT 06165 #undef SPECIAL_PUNCT 06166 #endif 06167 06168 static inline int 06169 is_global_name_punct(const char c) 06170 { 06171 if (c <= 0x20 || 0x7e < c) return 0; 06172 return (ruby_global_name_punct_bits[(c - 0x20) / 32] >> (c % 32)) & 1; 06173 } 06174 06175 static int 06176 parser_peek_variable_name(struct parser_params *parser) 06177 { 06178 int c; 06179 const char *p = lex_p; 06180 06181 if (p + 1 >= lex_pend) return 0; 06182 c = *p++; 06183 switch (c) { 06184 case '$': 06185 if ((c = *p) == '-') { 06186 if (++p >= lex_pend) return 0; 06187 c = *p; 06188 } 06189 else if (is_global_name_punct(c) || ISDIGIT(c)) { 06190 return tSTRING_DVAR; 06191 } 06192 break; 06193 case '@': 06194 if ((c = *p) == '@') { 06195 if (++p >= lex_pend) return 0; 06196 c = *p; 06197 } 06198 break; 06199 case '{': 06200 lex_p = p; 06201 command_start = TRUE; 06202 return tSTRING_DBEG; 06203 default: 06204 return 0; 06205 } 06206 if (!ISASCII(c) || c == '_' || ISALPHA(c)) 06207 return tSTRING_DVAR; 06208 return 0; 06209 } 06210 06211 static int 06212 parser_parse_string(struct parser_params *parser, NODE *quote) 06213 { 06214 int func = (int)quote->nd_func; 06215 int term = nd_term(quote); 06216 int paren = nd_paren(quote); 06217 int c, space = 0; 06218 rb_encoding *enc = current_enc; 06219 06220 if (func == -1) return tSTRING_END; 06221 c = nextc(); 06222 if ((func & STR_FUNC_QWORDS) && ISSPACE(c)) { 06223 do {c = nextc();} while (ISSPACE(c)); 06224 space = 1; 06225 } 06226 if (c == term && !quote->nd_nest) { 06227 if (func & STR_FUNC_QWORDS) { 06228 quote->nd_func = -1; 06229 return ' '; 06230 } 06231 if (!(func & STR_FUNC_REGEXP)) return tSTRING_END; 06232 set_yylval_num(regx_options()); 06233 return tREGEXP_END; 06234 } 06235 if (space) { 06236 pushback(c); 06237 return ' '; 06238 } 06239 newtok(); 06240 if ((func & STR_FUNC_EXPAND) && c == '#') { 06241 int t = parser_peek_variable_name(parser); 06242 if (t) return t; 06243 tokadd('#'); 06244 c = nextc(); 06245 } 06246 pushback(c); 06247 if (tokadd_string(func, term, paren, "e->nd_nest, 06248 &enc) == -1) { 06249 ruby_sourceline = nd_line(quote); 06250 if (func & STR_FUNC_REGEXP) { 06251 if (parser->eofp) 06252 compile_error(PARSER_ARG "unterminated regexp meets end of file"); 06253 return tREGEXP_END; 06254 } 06255 else { 06256 if (parser->eofp) 06257 compile_error(PARSER_ARG "unterminated string meets end of file"); 06258 return tSTRING_END; 06259 } 06260 } 06261 06262 tokfix(); 06263 set_yylval_str(STR_NEW3(tok(), toklen(), enc, func)); 06264 flush_string_content(enc); 06265 06266 return tSTRING_CONTENT; 06267 } 06268 06269 static int 06270 parser_heredoc_identifier(struct parser_params *parser) 06271 { 06272 int c = nextc(), term, func = 0; 06273 long len; 06274 06275 if (c == '-') { 06276 c = nextc(); 06277 func = STR_FUNC_INDENT; 06278 } 06279 switch (c) { 06280 case '\'': 06281 func |= str_squote; goto quoted; 06282 case '"': 06283 func |= str_dquote; goto quoted; 06284 case '`': 06285 func |= str_xquote; 06286 quoted: 06287 newtok(); 06288 tokadd(func); 06289 term = c; 06290 while ((c = nextc()) != -1 && c != term) { 06291 if (tokadd_mbchar(c) == -1) return 0; 06292 } 06293 if (c == -1) { 06294 compile_error(PARSER_ARG "unterminated here document identifier"); 06295 return 0; 06296 } 06297 break; 06298 06299 default: 06300 if (!parser_is_identchar()) { 06301 pushback(c); 06302 if (func & STR_FUNC_INDENT) { 06303 pushback('-'); 06304 } 06305 return 0; 06306 } 06307 newtok(); 06308 term = '"'; 06309 tokadd(func |= str_dquote); 06310 do { 06311 if (tokadd_mbchar(c) == -1) return 0; 06312 } while ((c = nextc()) != -1 && parser_is_identchar()); 06313 pushback(c); 06314 break; 06315 } 06316 06317 tokfix(); 06318 #ifdef RIPPER 06319 ripper_dispatch_scan_event(parser, tHEREDOC_BEG); 06320 #endif 06321 len = lex_p - lex_pbeg; 06322 lex_goto_eol(parser); 06323 lex_strterm = rb_node_newnode(NODE_HEREDOC, 06324 STR_NEW(tok(), toklen()), /* nd_lit */ 06325 len, /* nd_nth */ 06326 lex_lastline); /* nd_orig */ 06327 nd_set_line(lex_strterm, ruby_sourceline); 06328 ripper_flush(parser); 06329 return term == '`' ? tXSTRING_BEG : tSTRING_BEG; 06330 } 06331 06332 static void 06333 parser_heredoc_restore(struct parser_params *parser, NODE *here) 06334 { 06335 VALUE line; 06336 06337 line = here->nd_orig; 06338 lex_lastline = line; 06339 lex_pbeg = RSTRING_PTR(line); 06340 lex_pend = lex_pbeg + RSTRING_LEN(line); 06341 lex_p = lex_pbeg + here->nd_nth; 06342 heredoc_end = ruby_sourceline; 06343 ruby_sourceline = nd_line(here); 06344 dispose_string(here->nd_lit); 06345 rb_gc_force_recycle((VALUE)here); 06346 ripper_flush(parser); 06347 } 06348 06349 static int 06350 parser_whole_match_p(struct parser_params *parser, 06351 const char *eos, long len, int indent) 06352 { 06353 const char *p = lex_pbeg; 06354 long n; 06355 06356 if (indent) { 06357 while (*p && ISSPACE(*p)) p++; 06358 } 06359 n = lex_pend - (p + len); 06360 if (n < 0 || (n > 0 && p[len] != '\n' && p[len] != '\r')) return FALSE; 06361 return strncmp(eos, p, len) == 0; 06362 } 06363 06364 #ifdef RIPPER 06365 static void 06366 ripper_dispatch_heredoc_end(struct parser_params *parser) 06367 { 06368 if (!NIL_P(parser->delayed)) 06369 ripper_dispatch_delayed_token(parser, tSTRING_CONTENT); 06370 lex_goto_eol(parser); 06371 ripper_dispatch_ignored_scan_event(parser, tHEREDOC_END); 06372 } 06373 06374 #define dispatch_heredoc_end() ripper_dispatch_heredoc_end(parser) 06375 #else 06376 #define dispatch_heredoc_end() ((void)0) 06377 #endif 06378 06379 static int 06380 parser_here_document(struct parser_params *parser, NODE *here) 06381 { 06382 int c, func, indent = 0; 06383 const char *eos, *p, *pend; 06384 long len; 06385 VALUE str = 0; 06386 rb_encoding *enc = current_enc; 06387 06388 eos = RSTRING_PTR(here->nd_lit); 06389 len = RSTRING_LEN(here->nd_lit) - 1; 06390 indent = (func = *eos++) & STR_FUNC_INDENT; 06391 06392 if ((c = nextc()) == -1) { 06393 error: 06394 compile_error(PARSER_ARG "can't find string \"%s\" anywhere before EOF", eos); 06395 #ifdef RIPPER 06396 if (NIL_P(parser->delayed)) { 06397 ripper_dispatch_scan_event(parser, tSTRING_CONTENT); 06398 } 06399 else { 06400 if (str || 06401 ((len = lex_p - parser->tokp) > 0 && 06402 (str = STR_NEW3(parser->tokp, len, enc, func), 1))) { 06403 rb_str_append(parser->delayed, str); 06404 } 06405 ripper_dispatch_delayed_token(parser, tSTRING_CONTENT); 06406 } 06407 lex_goto_eol(parser); 06408 #endif 06409 restore: 06410 heredoc_restore(lex_strterm); 06411 lex_strterm = 0; 06412 return 0; 06413 } 06414 if (was_bol() && whole_match_p(eos, len, indent)) { 06415 dispatch_heredoc_end(); 06416 heredoc_restore(lex_strterm); 06417 return tSTRING_END; 06418 } 06419 06420 if (!(func & STR_FUNC_EXPAND)) { 06421 do { 06422 p = RSTRING_PTR(lex_lastline); 06423 pend = lex_pend; 06424 if (pend > p) { 06425 switch (pend[-1]) { 06426 case '\n': 06427 if (--pend == p || pend[-1] != '\r') { 06428 pend++; 06429 break; 06430 } 06431 case '\r': 06432 --pend; 06433 } 06434 } 06435 if (str) 06436 rb_str_cat(str, p, pend - p); 06437 else 06438 str = STR_NEW(p, pend - p); 06439 if (pend < lex_pend) rb_str_cat(str, "\n", 1); 06440 lex_goto_eol(parser); 06441 if (nextc() == -1) { 06442 if (str) dispose_string(str); 06443 goto error; 06444 } 06445 } while (!whole_match_p(eos, len, indent)); 06446 } 06447 else { 06448 /* int mb = ENC_CODERANGE_7BIT, *mbp = &mb;*/ 06449 newtok(); 06450 if (c == '#') { 06451 int t = parser_peek_variable_name(parser); 06452 if (t) return t; 06453 tokadd('#'); 06454 c = nextc(); 06455 } 06456 do { 06457 pushback(c); 06458 if ((c = tokadd_string(func, '\n', 0, NULL, &enc)) == -1) { 06459 if (parser->eofp) goto error; 06460 goto restore; 06461 } 06462 if (c != '\n') { 06463 set_yylval_str(STR_NEW3(tok(), toklen(), enc, func)); 06464 flush_string_content(enc); 06465 return tSTRING_CONTENT; 06466 } 06467 tokadd(nextc()); 06468 /* if (mbp && mb == ENC_CODERANGE_UNKNOWN) mbp = 0;*/ 06469 if ((c = nextc()) == -1) goto error; 06470 } while (!whole_match_p(eos, len, indent)); 06471 str = STR_NEW3(tok(), toklen(), enc, func); 06472 } 06473 dispatch_heredoc_end(); 06474 heredoc_restore(lex_strterm); 06475 lex_strterm = NEW_STRTERM(-1, 0, 0); 06476 set_yylval_str(str); 06477 return tSTRING_CONTENT; 06478 } 06479 06480 #include "lex.c" 06481 06482 static void 06483 arg_ambiguous_gen(struct parser_params *parser) 06484 { 06485 #ifndef RIPPER 06486 rb_warning0("ambiguous first argument; put parentheses or even spaces"); 06487 #else 06488 dispatch0(arg_ambiguous); 06489 #endif 06490 } 06491 #define arg_ambiguous() (arg_ambiguous_gen(parser), 1) 06492 06493 static ID 06494 formal_argument_gen(struct parser_params *parser, ID lhs) 06495 { 06496 #ifndef RIPPER 06497 if (!is_local_id(lhs)) 06498 yyerror("formal argument must be local variable"); 06499 #endif 06500 shadowing_lvar(lhs); 06501 return lhs; 06502 } 06503 06504 static int 06505 lvar_defined_gen(struct parser_params *parser, ID id) 06506 { 06507 return (dyna_in_block() && dvar_defined_get(id)) || local_id(id); 06508 } 06509 06510 /* emacsen -*- hack */ 06511 static long 06512 parser_encode_length(struct parser_params *parser, const char *name, long len) 06513 { 06514 long nlen; 06515 06516 if (len > 5 && name[nlen = len - 5] == '-') { 06517 if (rb_memcicmp(name + nlen + 1, "unix", 4) == 0) 06518 return nlen; 06519 } 06520 if (len > 4 && name[nlen = len - 4] == '-') { 06521 if (rb_memcicmp(name + nlen + 1, "dos", 3) == 0) 06522 return nlen; 06523 if (rb_memcicmp(name + nlen + 1, "mac", 3) == 0 && 06524 !(len == 8 && rb_memcicmp(name, "utf8-mac", len) == 0)) 06525 /* exclude UTF8-MAC because the encoding named "UTF8" doesn't exist in Ruby */ 06526 return nlen; 06527 } 06528 return len; 06529 } 06530 06531 static void 06532 parser_set_encode(struct parser_params *parser, const char *name) 06533 { 06534 int idx = rb_enc_find_index(name); 06535 rb_encoding *enc; 06536 VALUE excargs[3]; 06537 06538 if (idx < 0) { 06539 excargs[1] = rb_sprintf("unknown encoding name: %s", name); 06540 error: 06541 excargs[0] = rb_eArgError; 06542 excargs[2] = rb_make_backtrace(); 06543 rb_ary_unshift(excargs[2], rb_sprintf("%s:%d", ruby_sourcefile, ruby_sourceline)); 06544 rb_exc_raise(rb_make_exception(3, excargs)); 06545 } 06546 enc = rb_enc_from_index(idx); 06547 if (!rb_enc_asciicompat(enc)) { 06548 excargs[1] = rb_sprintf("%s is not ASCII compatible", rb_enc_name(enc)); 06549 goto error; 06550 } 06551 parser->enc = enc; 06552 #ifndef RIPPER 06553 if (ruby_debug_lines) { 06554 long i, n = RARRAY_LEN(ruby_debug_lines); 06555 const VALUE *p = RARRAY_PTR(ruby_debug_lines); 06556 for (i = 0; i < n; ++i) { 06557 rb_enc_associate_index(*p, idx); 06558 } 06559 } 06560 #endif 06561 } 06562 06563 static int 06564 comment_at_top(struct parser_params *parser) 06565 { 06566 const char *p = lex_pbeg, *pend = lex_p - 1; 06567 if (parser->line_count != (parser->has_shebang ? 2 : 1)) return 0; 06568 while (p < pend) { 06569 if (!ISSPACE(*p)) return 0; 06570 p++; 06571 } 06572 return 1; 06573 } 06574 06575 #ifndef RIPPER 06576 typedef long (*rb_magic_comment_length_t)(struct parser_params *parser, const char *name, long len); 06577 typedef void (*rb_magic_comment_setter_t)(struct parser_params *parser, const char *name, const char *val); 06578 06579 static void 06580 magic_comment_encoding(struct parser_params *parser, const char *name, const char *val) 06581 { 06582 if (!comment_at_top(parser)) { 06583 return; 06584 } 06585 parser_set_encode(parser, val); 06586 } 06587 06588 static void 06589 parser_set_token_info(struct parser_params *parser, const char *name, const char *val) 06590 { 06591 int *p = &parser->parser_token_info_enabled; 06592 06593 switch (*val) { 06594 case 't': case 'T': 06595 if (strcasecmp(val, "true") == 0) { 06596 *p = TRUE; 06597 return; 06598 } 06599 break; 06600 case 'f': case 'F': 06601 if (strcasecmp(val, "false") == 0) { 06602 *p = FALSE; 06603 return; 06604 } 06605 break; 06606 } 06607 rb_compile_warning(ruby_sourcefile, ruby_sourceline, "invalid value for %s: %s", name, val); 06608 } 06609 06610 struct magic_comment { 06611 const char *name; 06612 rb_magic_comment_setter_t func; 06613 rb_magic_comment_length_t length; 06614 }; 06615 06616 static const struct magic_comment magic_comments[] = { 06617 {"coding", magic_comment_encoding, parser_encode_length}, 06618 {"encoding", magic_comment_encoding, parser_encode_length}, 06619 {"warn_indent", parser_set_token_info}, 06620 }; 06621 #endif 06622 06623 static const char * 06624 magic_comment_marker(const char *str, long len) 06625 { 06626 long i = 2; 06627 06628 while (i < len) { 06629 switch (str[i]) { 06630 case '-': 06631 if (str[i-1] == '*' && str[i-2] == '-') { 06632 return str + i + 1; 06633 } 06634 i += 2; 06635 break; 06636 case '*': 06637 if (i + 1 >= len) return 0; 06638 if (str[i+1] != '-') { 06639 i += 4; 06640 } 06641 else if (str[i-1] != '-') { 06642 i += 2; 06643 } 06644 else { 06645 return str + i + 2; 06646 } 06647 break; 06648 default: 06649 i += 3; 06650 break; 06651 } 06652 } 06653 return 0; 06654 } 06655 06656 static int 06657 parser_magic_comment(struct parser_params *parser, const char *str, long len) 06658 { 06659 VALUE name = 0, val = 0; 06660 const char *beg, *end, *vbeg, *vend; 06661 #define str_copy(_s, _p, _n) ((_s) \ 06662 ? (void)(rb_str_resize((_s), (_n)), \ 06663 MEMCPY(RSTRING_PTR(_s), (_p), char, (_n)), (_s)) \ 06664 : (void)((_s) = STR_NEW((_p), (_n)))) 06665 06666 if (len <= 7) return FALSE; 06667 if (!(beg = magic_comment_marker(str, len))) return FALSE; 06668 if (!(end = magic_comment_marker(beg, str + len - beg))) return FALSE; 06669 str = beg; 06670 len = end - beg - 3; 06671 06672 /* %r"([^\\s\'\":;]+)\\s*:\\s*(\"(?:\\\\.|[^\"])*\"|[^\"\\s;]+)[\\s;]*" */ 06673 while (len > 0) { 06674 #ifndef RIPPER 06675 const struct magic_comment *p = magic_comments; 06676 #endif 06677 char *s; 06678 int i; 06679 long n = 0; 06680 06681 for (; len > 0 && *str; str++, --len) { 06682 switch (*str) { 06683 case '\'': case '"': case ':': case ';': 06684 continue; 06685 } 06686 if (!ISSPACE(*str)) break; 06687 } 06688 for (beg = str; len > 0; str++, --len) { 06689 switch (*str) { 06690 case '\'': case '"': case ':': case ';': 06691 break; 06692 default: 06693 if (ISSPACE(*str)) break; 06694 continue; 06695 } 06696 break; 06697 } 06698 for (end = str; len > 0 && ISSPACE(*str); str++, --len); 06699 if (!len) break; 06700 if (*str != ':') continue; 06701 06702 do str++; while (--len > 0 && ISSPACE(*str)); 06703 if (!len) break; 06704 if (*str == '"') { 06705 for (vbeg = ++str; --len > 0 && *str != '"'; str++) { 06706 if (*str == '\\') { 06707 --len; 06708 ++str; 06709 } 06710 } 06711 vend = str; 06712 if (len) { 06713 --len; 06714 ++str; 06715 } 06716 } 06717 else { 06718 for (vbeg = str; len > 0 && *str != '"' && *str != ';' && !ISSPACE(*str); --len, str++); 06719 vend = str; 06720 } 06721 while (len > 0 && (*str == ';' || ISSPACE(*str))) --len, str++; 06722 06723 n = end - beg; 06724 str_copy(name, beg, n); 06725 s = RSTRING_PTR(name); 06726 for (i = 0; i < n; ++i) { 06727 if (s[i] == '-') s[i] = '_'; 06728 } 06729 #ifndef RIPPER 06730 do { 06731 if (STRNCASECMP(p->name, s, n) == 0) { 06732 n = vend - vbeg; 06733 if (p->length) { 06734 n = (*p->length)(parser, vbeg, n); 06735 } 06736 str_copy(val, vbeg, n); 06737 (*p->func)(parser, s, RSTRING_PTR(val)); 06738 break; 06739 } 06740 } while (++p < magic_comments + numberof(magic_comments)); 06741 #else 06742 str_copy(val, vbeg, vend - vbeg); 06743 dispatch2(magic_comment, name, val); 06744 #endif 06745 } 06746 06747 return TRUE; 06748 } 06749 06750 static void 06751 set_file_encoding(struct parser_params *parser, const char *str, const char *send) 06752 { 06753 int sep = 0; 06754 const char *beg = str; 06755 VALUE s; 06756 06757 for (;;) { 06758 if (send - str <= 6) return; 06759 switch (str[6]) { 06760 case 'C': case 'c': str += 6; continue; 06761 case 'O': case 'o': str += 5; continue; 06762 case 'D': case 'd': str += 4; continue; 06763 case 'I': case 'i': str += 3; continue; 06764 case 'N': case 'n': str += 2; continue; 06765 case 'G': case 'g': str += 1; continue; 06766 case '=': case ':': 06767 sep = 1; 06768 str += 6; 06769 break; 06770 default: 06771 str += 6; 06772 if (ISSPACE(*str)) break; 06773 continue; 06774 } 06775 if (STRNCASECMP(str-6, "coding", 6) == 0) break; 06776 } 06777 for (;;) { 06778 do { 06779 if (++str >= send) return; 06780 } while (ISSPACE(*str)); 06781 if (sep) break; 06782 if (*str != '=' && *str != ':') return; 06783 sep = 1; 06784 str++; 06785 } 06786 beg = str; 06787 while ((*str == '-' || *str == '_' || ISALNUM(*str)) && ++str < send); 06788 s = rb_str_new(beg, parser_encode_length(parser, beg, str - beg)); 06789 parser_set_encode(parser, RSTRING_PTR(s)); 06790 rb_str_resize(s, 0); 06791 } 06792 06793 static void 06794 parser_prepare(struct parser_params *parser) 06795 { 06796 int c = nextc(); 06797 switch (c) { 06798 case '#': 06799 if (peek('!')) parser->has_shebang = 1; 06800 break; 06801 case 0xef: /* UTF-8 BOM marker */ 06802 if (lex_pend - lex_p >= 2 && 06803 (unsigned char)lex_p[0] == 0xbb && 06804 (unsigned char)lex_p[1] == 0xbf) { 06805 parser->enc = rb_utf8_encoding(); 06806 lex_p += 2; 06807 lex_pbeg = lex_p; 06808 return; 06809 } 06810 break; 06811 case EOF: 06812 return; 06813 } 06814 pushback(c); 06815 parser->enc = rb_enc_get(lex_lastline); 06816 } 06817 06818 #define IS_ARG() IS_lex_state(EXPR_ARG_ANY) 06819 #define IS_END() IS_lex_state(EXPR_END_ANY) 06820 #define IS_BEG() IS_lex_state(EXPR_BEG_ANY) 06821 #define IS_SPCARG(c) (IS_ARG() && space_seen && !ISSPACE(c)) 06822 #define IS_LABEL_POSSIBLE() ((IS_lex_state(EXPR_BEG | EXPR_ENDFN) && !cmd_state) || IS_ARG()) 06823 #define IS_LABEL_SUFFIX(n) (peek_n(':',(n)) && !peek_n(':', (n)+1)) 06824 #define IS_AFTER_OPERATOR() IS_lex_state(EXPR_FNAME | EXPR_DOT) 06825 06826 #ifndef RIPPER 06827 #define ambiguous_operator(op, syn) ( \ 06828 rb_warning0("`"op"' after local variable is interpreted as binary operator"), \ 06829 rb_warning0("even though it seems like "syn"")) 06830 #else 06831 #define ambiguous_operator(op, syn) dispatch2(operator_ambiguous, ripper_intern(op), rb_str_new_cstr(syn)) 06832 #endif 06833 #define warn_balanced(op, syn) ((void) \ 06834 (!IS_lex_state_for(last_state, EXPR_CLASS|EXPR_DOT|EXPR_FNAME|EXPR_ENDFN|EXPR_ENDARG) && \ 06835 space_seen && !ISSPACE(c) && \ 06836 (ambiguous_operator(op, syn), 0))) 06837 06838 static int 06839 parser_yylex(struct parser_params *parser) 06840 { 06841 register int c; 06842 int space_seen = 0; 06843 int cmd_state; 06844 enum lex_state_e last_state; 06845 rb_encoding *enc; 06846 int mb; 06847 #ifdef RIPPER 06848 int fallthru = FALSE; 06849 #endif 06850 06851 if (lex_strterm) { 06852 int token; 06853 if (nd_type(lex_strterm) == NODE_HEREDOC) { 06854 token = here_document(lex_strterm); 06855 if (token == tSTRING_END) { 06856 lex_strterm = 0; 06857 lex_state = EXPR_END; 06858 } 06859 } 06860 else { 06861 token = parse_string(lex_strterm); 06862 if (token == tSTRING_END || token == tREGEXP_END) { 06863 rb_gc_force_recycle((VALUE)lex_strterm); 06864 lex_strterm = 0; 06865 lex_state = EXPR_END; 06866 } 06867 } 06868 return token; 06869 } 06870 cmd_state = command_start; 06871 command_start = FALSE; 06872 retry: 06873 last_state = lex_state; 06874 switch (c = nextc()) { 06875 case '\0': /* NUL */ 06876 case '\004': /* ^D */ 06877 case '\032': /* ^Z */ 06878 case -1: /* end of script. */ 06879 return 0; 06880 06881 /* white spaces */ 06882 case ' ': case '\t': case '\f': case '\r': 06883 case '\13': /* '\v' */ 06884 space_seen = 1; 06885 #ifdef RIPPER 06886 while ((c = nextc())) { 06887 switch (c) { 06888 case ' ': case '\t': case '\f': case '\r': 06889 case '\13': /* '\v' */ 06890 break; 06891 default: 06892 goto outofloop; 06893 } 06894 } 06895 outofloop: 06896 pushback(c); 06897 ripper_dispatch_scan_event(parser, tSP); 06898 #endif 06899 goto retry; 06900 06901 case '#': /* it's a comment */ 06902 /* no magic_comment in shebang line */ 06903 if (!parser_magic_comment(parser, lex_p, lex_pend - lex_p)) { 06904 if (comment_at_top(parser)) { 06905 set_file_encoding(parser, lex_p, lex_pend); 06906 } 06907 } 06908 lex_p = lex_pend; 06909 #ifdef RIPPER 06910 ripper_dispatch_scan_event(parser, tCOMMENT); 06911 fallthru = TRUE; 06912 #endif 06913 /* fall through */ 06914 case '\n': 06915 if (IS_lex_state(EXPR_BEG | EXPR_VALUE | EXPR_CLASS | EXPR_FNAME | EXPR_DOT)) { 06916 #ifdef RIPPER 06917 if (!fallthru) { 06918 ripper_dispatch_scan_event(parser, tIGNORED_NL); 06919 } 06920 fallthru = FALSE; 06921 #endif 06922 goto retry; 06923 } 06924 while ((c = nextc())) { 06925 switch (c) { 06926 case ' ': case '\t': case '\f': case '\r': 06927 case '\13': /* '\v' */ 06928 space_seen = 1; 06929 break; 06930 case '.': { 06931 if ((c = nextc()) != '.') { 06932 pushback(c); 06933 pushback('.'); 06934 goto retry; 06935 } 06936 } 06937 default: 06938 --ruby_sourceline; 06939 lex_nextline = lex_lastline; 06940 case -1: /* EOF no decrement*/ 06941 lex_goto_eol(parser); 06942 #ifdef RIPPER 06943 if (c != -1) { 06944 parser->tokp = lex_p; 06945 } 06946 #endif 06947 goto normal_newline; 06948 } 06949 } 06950 normal_newline: 06951 command_start = TRUE; 06952 lex_state = EXPR_BEG; 06953 return '\n'; 06954 06955 case '*': 06956 if ((c = nextc()) == '*') { 06957 if ((c = nextc()) == '=') { 06958 set_yylval_id(tPOW); 06959 lex_state = EXPR_BEG; 06960 return tOP_ASGN; 06961 } 06962 pushback(c); 06963 if (IS_SPCARG(c)) { 06964 rb_warning0("`**' interpreted as argument prefix"); 06965 c = tDSTAR; 06966 } 06967 else if (IS_BEG()) { 06968 c = tDSTAR; 06969 } 06970 else { 06971 warn_balanced("**", "argument prefix"); 06972 c = tPOW; 06973 } 06974 } 06975 else { 06976 if (c == '=') { 06977 set_yylval_id('*'); 06978 lex_state = EXPR_BEG; 06979 return tOP_ASGN; 06980 } 06981 pushback(c); 06982 if (IS_SPCARG(c)) { 06983 rb_warning0("`*' interpreted as argument prefix"); 06984 c = tSTAR; 06985 } 06986 else if (IS_BEG()) { 06987 c = tSTAR; 06988 } 06989 else { 06990 warn_balanced("*", "argument prefix"); 06991 c = '*'; 06992 } 06993 } 06994 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 06995 return c; 06996 06997 case '!': 06998 c = nextc(); 06999 if (IS_AFTER_OPERATOR()) { 07000 lex_state = EXPR_ARG; 07001 if (c == '@') { 07002 return '!'; 07003 } 07004 } 07005 else { 07006 lex_state = EXPR_BEG; 07007 } 07008 if (c == '=') { 07009 return tNEQ; 07010 } 07011 if (c == '~') { 07012 return tNMATCH; 07013 } 07014 pushback(c); 07015 return '!'; 07016 07017 case '=': 07018 if (was_bol()) { 07019 /* skip embedded rd document */ 07020 if (strncmp(lex_p, "begin", 5) == 0 && ISSPACE(lex_p[5])) { 07021 #ifdef RIPPER 07022 int first_p = TRUE; 07023 07024 lex_goto_eol(parser); 07025 ripper_dispatch_scan_event(parser, tEMBDOC_BEG); 07026 #endif 07027 for (;;) { 07028 lex_goto_eol(parser); 07029 #ifdef RIPPER 07030 if (!first_p) { 07031 ripper_dispatch_scan_event(parser, tEMBDOC); 07032 } 07033 first_p = FALSE; 07034 #endif 07035 c = nextc(); 07036 if (c == -1) { 07037 compile_error(PARSER_ARG "embedded document meets end of file"); 07038 return 0; 07039 } 07040 if (c != '=') continue; 07041 if (strncmp(lex_p, "end", 3) == 0 && 07042 (lex_p + 3 == lex_pend || ISSPACE(lex_p[3]))) { 07043 break; 07044 } 07045 } 07046 lex_goto_eol(parser); 07047 #ifdef RIPPER 07048 ripper_dispatch_scan_event(parser, tEMBDOC_END); 07049 #endif 07050 goto retry; 07051 } 07052 } 07053 07054 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07055 if ((c = nextc()) == '=') { 07056 if ((c = nextc()) == '=') { 07057 return tEQQ; 07058 } 07059 pushback(c); 07060 return tEQ; 07061 } 07062 if (c == '~') { 07063 return tMATCH; 07064 } 07065 else if (c == '>') { 07066 return tASSOC; 07067 } 07068 pushback(c); 07069 return '='; 07070 07071 case '<': 07072 last_state = lex_state; 07073 c = nextc(); 07074 if (c == '<' && 07075 !IS_lex_state(EXPR_DOT | EXPR_CLASS) && 07076 !IS_END() && 07077 (!IS_ARG() || space_seen)) { 07078 int token = heredoc_identifier(); 07079 if (token) return token; 07080 } 07081 if (IS_AFTER_OPERATOR()) { 07082 lex_state = EXPR_ARG; 07083 } 07084 else { 07085 if (IS_lex_state(EXPR_CLASS)) 07086 command_start = TRUE; 07087 lex_state = EXPR_BEG; 07088 } 07089 if (c == '=') { 07090 if ((c = nextc()) == '>') { 07091 return tCMP; 07092 } 07093 pushback(c); 07094 return tLEQ; 07095 } 07096 if (c == '<') { 07097 if ((c = nextc()) == '=') { 07098 set_yylval_id(tLSHFT); 07099 lex_state = EXPR_BEG; 07100 return tOP_ASGN; 07101 } 07102 pushback(c); 07103 warn_balanced("<<", "here document"); 07104 return tLSHFT; 07105 } 07106 pushback(c); 07107 return '<'; 07108 07109 case '>': 07110 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07111 if ((c = nextc()) == '=') { 07112 return tGEQ; 07113 } 07114 if (c == '>') { 07115 if ((c = nextc()) == '=') { 07116 set_yylval_id(tRSHFT); 07117 lex_state = EXPR_BEG; 07118 return tOP_ASGN; 07119 } 07120 pushback(c); 07121 return tRSHFT; 07122 } 07123 pushback(c); 07124 return '>'; 07125 07126 case '"': 07127 lex_strterm = NEW_STRTERM(str_dquote, '"', 0); 07128 return tSTRING_BEG; 07129 07130 case '`': 07131 if (IS_lex_state(EXPR_FNAME)) { 07132 lex_state = EXPR_ENDFN; 07133 return c; 07134 } 07135 if (IS_lex_state(EXPR_DOT)) { 07136 if (cmd_state) 07137 lex_state = EXPR_CMDARG; 07138 else 07139 lex_state = EXPR_ARG; 07140 return c; 07141 } 07142 lex_strterm = NEW_STRTERM(str_xquote, '`', 0); 07143 return tXSTRING_BEG; 07144 07145 case '\'': 07146 lex_strterm = NEW_STRTERM(str_squote, '\'', 0); 07147 return tSTRING_BEG; 07148 07149 case '?': 07150 if (IS_END()) { 07151 lex_state = EXPR_VALUE; 07152 return '?'; 07153 } 07154 c = nextc(); 07155 if (c == -1) { 07156 compile_error(PARSER_ARG "incomplete character syntax"); 07157 return 0; 07158 } 07159 if (rb_enc_isspace(c, current_enc)) { 07160 if (!IS_ARG()) { 07161 int c2 = 0; 07162 switch (c) { 07163 case ' ': 07164 c2 = 's'; 07165 break; 07166 case '\n': 07167 c2 = 'n'; 07168 break; 07169 case '\t': 07170 c2 = 't'; 07171 break; 07172 case '\v': 07173 c2 = 'v'; 07174 break; 07175 case '\r': 07176 c2 = 'r'; 07177 break; 07178 case '\f': 07179 c2 = 'f'; 07180 break; 07181 } 07182 if (c2) { 07183 rb_warnI("invalid character syntax; use ?\\%c", c2); 07184 } 07185 } 07186 ternary: 07187 pushback(c); 07188 lex_state = EXPR_VALUE; 07189 return '?'; 07190 } 07191 newtok(); 07192 enc = current_enc; 07193 if (!parser_isascii()) { 07194 if (tokadd_mbchar(c) == -1) return 0; 07195 } 07196 else if ((rb_enc_isalnum(c, current_enc) || c == '_') && 07197 lex_p < lex_pend && is_identchar(lex_p, lex_pend, current_enc)) { 07198 goto ternary; 07199 } 07200 else if (c == '\\') { 07201 if (peek('u')) { 07202 nextc(); 07203 c = parser_tokadd_utf8(parser, &enc, 0, 0, 0); 07204 if (0x80 <= c) { 07205 tokaddmbc(c, enc); 07206 } 07207 else { 07208 tokadd(c); 07209 } 07210 } 07211 else if (!lex_eol_p() && !(c = *lex_p, ISASCII(c))) { 07212 nextc(); 07213 if (tokadd_mbchar(c) == -1) return 0; 07214 } 07215 else { 07216 c = read_escape(0, &enc); 07217 tokadd(c); 07218 } 07219 } 07220 else { 07221 tokadd(c); 07222 } 07223 tokfix(); 07224 set_yylval_str(STR_NEW3(tok(), toklen(), enc, 0)); 07225 lex_state = EXPR_END; 07226 return tCHAR; 07227 07228 case '&': 07229 if ((c = nextc()) == '&') { 07230 lex_state = EXPR_BEG; 07231 if ((c = nextc()) == '=') { 07232 set_yylval_id(tANDOP); 07233 lex_state = EXPR_BEG; 07234 return tOP_ASGN; 07235 } 07236 pushback(c); 07237 return tANDOP; 07238 } 07239 else if (c == '=') { 07240 set_yylval_id('&'); 07241 lex_state = EXPR_BEG; 07242 return tOP_ASGN; 07243 } 07244 pushback(c); 07245 if (IS_SPCARG(c)) { 07246 rb_warning0("`&' interpreted as argument prefix"); 07247 c = tAMPER; 07248 } 07249 else if (IS_BEG()) { 07250 c = tAMPER; 07251 } 07252 else { 07253 warn_balanced("&", "argument prefix"); 07254 c = '&'; 07255 } 07256 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07257 return c; 07258 07259 case '|': 07260 if ((c = nextc()) == '|') { 07261 lex_state = EXPR_BEG; 07262 if ((c = nextc()) == '=') { 07263 set_yylval_id(tOROP); 07264 lex_state = EXPR_BEG; 07265 return tOP_ASGN; 07266 } 07267 pushback(c); 07268 return tOROP; 07269 } 07270 if (c == '=') { 07271 set_yylval_id('|'); 07272 lex_state = EXPR_BEG; 07273 return tOP_ASGN; 07274 } 07275 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07276 pushback(c); 07277 return '|'; 07278 07279 case '+': 07280 c = nextc(); 07281 if (IS_AFTER_OPERATOR()) { 07282 lex_state = EXPR_ARG; 07283 if (c == '@') { 07284 return tUPLUS; 07285 } 07286 pushback(c); 07287 return '+'; 07288 } 07289 if (c == '=') { 07290 set_yylval_id('+'); 07291 lex_state = EXPR_BEG; 07292 return tOP_ASGN; 07293 } 07294 if (IS_BEG() || (IS_SPCARG(c) && arg_ambiguous())) { 07295 lex_state = EXPR_BEG; 07296 pushback(c); 07297 if (c != -1 && ISDIGIT(c)) { 07298 c = '+'; 07299 goto start_num; 07300 } 07301 return tUPLUS; 07302 } 07303 lex_state = EXPR_BEG; 07304 pushback(c); 07305 warn_balanced("+", "unary operator"); 07306 return '+'; 07307 07308 case '-': 07309 c = nextc(); 07310 if (IS_AFTER_OPERATOR()) { 07311 lex_state = EXPR_ARG; 07312 if (c == '@') { 07313 return tUMINUS; 07314 } 07315 pushback(c); 07316 return '-'; 07317 } 07318 if (c == '=') { 07319 set_yylval_id('-'); 07320 lex_state = EXPR_BEG; 07321 return tOP_ASGN; 07322 } 07323 if (c == '>') { 07324 lex_state = EXPR_ENDFN; 07325 return tLAMBDA; 07326 } 07327 if (IS_BEG() || (IS_SPCARG(c) && arg_ambiguous())) { 07328 lex_state = EXPR_BEG; 07329 pushback(c); 07330 if (c != -1 && ISDIGIT(c)) { 07331 return tUMINUS_NUM; 07332 } 07333 return tUMINUS; 07334 } 07335 lex_state = EXPR_BEG; 07336 pushback(c); 07337 warn_balanced("-", "unary operator"); 07338 return '-'; 07339 07340 case '.': 07341 lex_state = EXPR_BEG; 07342 if ((c = nextc()) == '.') { 07343 if ((c = nextc()) == '.') { 07344 return tDOT3; 07345 } 07346 pushback(c); 07347 return tDOT2; 07348 } 07349 pushback(c); 07350 if (c != -1 && ISDIGIT(c)) { 07351 yyerror("no .<digit> floating literal anymore; put 0 before dot"); 07352 } 07353 lex_state = EXPR_DOT; 07354 return '.'; 07355 07356 start_num: 07357 case '0': case '1': case '2': case '3': case '4': 07358 case '5': case '6': case '7': case '8': case '9': 07359 { 07360 int is_float, seen_point, seen_e, nondigit; 07361 07362 is_float = seen_point = seen_e = nondigit = 0; 07363 lex_state = EXPR_END; 07364 newtok(); 07365 if (c == '-' || c == '+') { 07366 tokadd(c); 07367 c = nextc(); 07368 } 07369 if (c == '0') { 07370 #define no_digits() do {yyerror("numeric literal without digits"); return 0;} while (0) 07371 int start = toklen(); 07372 c = nextc(); 07373 if (c == 'x' || c == 'X') { 07374 /* hexadecimal */ 07375 c = nextc(); 07376 if (c != -1 && ISXDIGIT(c)) { 07377 do { 07378 if (c == '_') { 07379 if (nondigit) break; 07380 nondigit = c; 07381 continue; 07382 } 07383 if (!ISXDIGIT(c)) break; 07384 nondigit = 0; 07385 tokadd(c); 07386 } while ((c = nextc()) != -1); 07387 } 07388 pushback(c); 07389 tokfix(); 07390 if (toklen() == start) { 07391 no_digits(); 07392 } 07393 else if (nondigit) goto trailing_uc; 07394 set_yylval_literal(rb_cstr_to_inum(tok(), 16, FALSE)); 07395 return tINTEGER; 07396 } 07397 if (c == 'b' || c == 'B') { 07398 /* binary */ 07399 c = nextc(); 07400 if (c == '0' || c == '1') { 07401 do { 07402 if (c == '_') { 07403 if (nondigit) break; 07404 nondigit = c; 07405 continue; 07406 } 07407 if (c != '0' && c != '1') break; 07408 nondigit = 0; 07409 tokadd(c); 07410 } while ((c = nextc()) != -1); 07411 } 07412 pushback(c); 07413 tokfix(); 07414 if (toklen() == start) { 07415 no_digits(); 07416 } 07417 else if (nondigit) goto trailing_uc; 07418 set_yylval_literal(rb_cstr_to_inum(tok(), 2, FALSE)); 07419 return tINTEGER; 07420 } 07421 if (c == 'd' || c == 'D') { 07422 /* decimal */ 07423 c = nextc(); 07424 if (c != -1 && ISDIGIT(c)) { 07425 do { 07426 if (c == '_') { 07427 if (nondigit) break; 07428 nondigit = c; 07429 continue; 07430 } 07431 if (!ISDIGIT(c)) break; 07432 nondigit = 0; 07433 tokadd(c); 07434 } while ((c = nextc()) != -1); 07435 } 07436 pushback(c); 07437 tokfix(); 07438 if (toklen() == start) { 07439 no_digits(); 07440 } 07441 else if (nondigit) goto trailing_uc; 07442 set_yylval_literal(rb_cstr_to_inum(tok(), 10, FALSE)); 07443 return tINTEGER; 07444 } 07445 if (c == '_') { 07446 /* 0_0 */ 07447 goto octal_number; 07448 } 07449 if (c == 'o' || c == 'O') { 07450 /* prefixed octal */ 07451 c = nextc(); 07452 if (c == -1 || c == '_' || !ISDIGIT(c)) { 07453 no_digits(); 07454 } 07455 } 07456 if (c >= '0' && c <= '7') { 07457 /* octal */ 07458 octal_number: 07459 do { 07460 if (c == '_') { 07461 if (nondigit) break; 07462 nondigit = c; 07463 continue; 07464 } 07465 if (c < '0' || c > '9') break; 07466 if (c > '7') goto invalid_octal; 07467 nondigit = 0; 07468 tokadd(c); 07469 } while ((c = nextc()) != -1); 07470 if (toklen() > start) { 07471 pushback(c); 07472 tokfix(); 07473 if (nondigit) goto trailing_uc; 07474 set_yylval_literal(rb_cstr_to_inum(tok(), 8, FALSE)); 07475 return tINTEGER; 07476 } 07477 if (nondigit) { 07478 pushback(c); 07479 goto trailing_uc; 07480 } 07481 } 07482 if (c > '7' && c <= '9') { 07483 invalid_octal: 07484 yyerror("Invalid octal digit"); 07485 } 07486 else if (c == '.' || c == 'e' || c == 'E') { 07487 tokadd('0'); 07488 } 07489 else { 07490 pushback(c); 07491 set_yylval_literal(INT2FIX(0)); 07492 return tINTEGER; 07493 } 07494 } 07495 07496 for (;;) { 07497 switch (c) { 07498 case '0': case '1': case '2': case '3': case '4': 07499 case '5': case '6': case '7': case '8': case '9': 07500 nondigit = 0; 07501 tokadd(c); 07502 break; 07503 07504 case '.': 07505 if (nondigit) goto trailing_uc; 07506 if (seen_point || seen_e) { 07507 goto decode_num; 07508 } 07509 else { 07510 int c0 = nextc(); 07511 if (c0 == -1 || !ISDIGIT(c0)) { 07512 pushback(c0); 07513 goto decode_num; 07514 } 07515 c = c0; 07516 } 07517 tokadd('.'); 07518 tokadd(c); 07519 is_float++; 07520 seen_point++; 07521 nondigit = 0; 07522 break; 07523 07524 case 'e': 07525 case 'E': 07526 if (nondigit) { 07527 pushback(c); 07528 c = nondigit; 07529 goto decode_num; 07530 } 07531 if (seen_e) { 07532 goto decode_num; 07533 } 07534 tokadd(c); 07535 seen_e++; 07536 is_float++; 07537 nondigit = c; 07538 c = nextc(); 07539 if (c != '-' && c != '+') continue; 07540 tokadd(c); 07541 nondigit = c; 07542 break; 07543 07544 case '_': /* `_' in number just ignored */ 07545 if (nondigit) goto decode_num; 07546 nondigit = c; 07547 break; 07548 07549 default: 07550 goto decode_num; 07551 } 07552 c = nextc(); 07553 } 07554 07555 decode_num: 07556 pushback(c); 07557 if (nondigit) { 07558 char tmp[30]; 07559 trailing_uc: 07560 snprintf(tmp, sizeof(tmp), "trailing `%c' in number", nondigit); 07561 yyerror(tmp); 07562 } 07563 tokfix(); 07564 if (is_float) { 07565 double d = strtod(tok(), 0); 07566 if (errno == ERANGE) { 07567 rb_warningS("Float %s out of range", tok()); 07568 errno = 0; 07569 } 07570 set_yylval_literal(DBL2NUM(d)); 07571 return tFLOAT; 07572 } 07573 set_yylval_literal(rb_cstr_to_inum(tok(), 10, FALSE)); 07574 return tINTEGER; 07575 } 07576 07577 case ')': 07578 case ']': 07579 paren_nest--; 07580 case '}': 07581 COND_LEXPOP(); 07582 CMDARG_LEXPOP(); 07583 if (c == ')') 07584 lex_state = EXPR_ENDFN; 07585 else 07586 lex_state = EXPR_ENDARG; 07587 if (c == '}') { 07588 if (!brace_nest--) c = tSTRING_DEND; 07589 } 07590 return c; 07591 07592 case ':': 07593 c = nextc(); 07594 if (c == ':') { 07595 if (IS_BEG() || IS_lex_state(EXPR_CLASS) || IS_SPCARG(-1)) { 07596 lex_state = EXPR_BEG; 07597 return tCOLON3; 07598 } 07599 lex_state = EXPR_DOT; 07600 return tCOLON2; 07601 } 07602 if (IS_END() || ISSPACE(c)) { 07603 pushback(c); 07604 warn_balanced(":", "symbol literal"); 07605 lex_state = EXPR_BEG; 07606 return ':'; 07607 } 07608 switch (c) { 07609 case '\'': 07610 lex_strterm = NEW_STRTERM(str_ssym, c, 0); 07611 break; 07612 case '"': 07613 lex_strterm = NEW_STRTERM(str_dsym, c, 0); 07614 break; 07615 default: 07616 pushback(c); 07617 break; 07618 } 07619 lex_state = EXPR_FNAME; 07620 return tSYMBEG; 07621 07622 case '/': 07623 if (IS_lex_state(EXPR_BEG_ANY)) { 07624 lex_strterm = NEW_STRTERM(str_regexp, '/', 0); 07625 return tREGEXP_BEG; 07626 } 07627 if ((c = nextc()) == '=') { 07628 set_yylval_id('/'); 07629 lex_state = EXPR_BEG; 07630 return tOP_ASGN; 07631 } 07632 pushback(c); 07633 if (IS_SPCARG(c)) { 07634 (void)arg_ambiguous(); 07635 lex_strterm = NEW_STRTERM(str_regexp, '/', 0); 07636 return tREGEXP_BEG; 07637 } 07638 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07639 warn_balanced("/", "regexp literal"); 07640 return '/'; 07641 07642 case '^': 07643 if ((c = nextc()) == '=') { 07644 set_yylval_id('^'); 07645 lex_state = EXPR_BEG; 07646 return tOP_ASGN; 07647 } 07648 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07649 pushback(c); 07650 return '^'; 07651 07652 case ';': 07653 lex_state = EXPR_BEG; 07654 command_start = TRUE; 07655 return ';'; 07656 07657 case ',': 07658 lex_state = EXPR_BEG; 07659 return ','; 07660 07661 case '~': 07662 if (IS_AFTER_OPERATOR()) { 07663 if ((c = nextc()) != '@') { 07664 pushback(c); 07665 } 07666 lex_state = EXPR_ARG; 07667 } 07668 else { 07669 lex_state = EXPR_BEG; 07670 } 07671 return '~'; 07672 07673 case '(': 07674 if (IS_BEG()) { 07675 c = tLPAREN; 07676 } 07677 else if (IS_SPCARG(-1)) { 07678 c = tLPAREN_ARG; 07679 } 07680 paren_nest++; 07681 COND_PUSH(0); 07682 CMDARG_PUSH(0); 07683 lex_state = EXPR_BEG; 07684 return c; 07685 07686 case '[': 07687 paren_nest++; 07688 if (IS_AFTER_OPERATOR()) { 07689 lex_state = EXPR_ARG; 07690 if ((c = nextc()) == ']') { 07691 if ((c = nextc()) == '=') { 07692 return tASET; 07693 } 07694 pushback(c); 07695 return tAREF; 07696 } 07697 pushback(c); 07698 return '['; 07699 } 07700 else if (IS_BEG()) { 07701 c = tLBRACK; 07702 } 07703 else if (IS_ARG() && space_seen) { 07704 c = tLBRACK; 07705 } 07706 lex_state = EXPR_BEG; 07707 COND_PUSH(0); 07708 CMDARG_PUSH(0); 07709 return c; 07710 07711 case '{': 07712 ++brace_nest; 07713 if (lpar_beg && lpar_beg == paren_nest) { 07714 lex_state = EXPR_BEG; 07715 lpar_beg = 0; 07716 --paren_nest; 07717 COND_PUSH(0); 07718 CMDARG_PUSH(0); 07719 return tLAMBEG; 07720 } 07721 if (IS_ARG() || IS_lex_state(EXPR_END | EXPR_ENDFN)) 07722 c = '{'; /* block (primary) */ 07723 else if (IS_lex_state(EXPR_ENDARG)) 07724 c = tLBRACE_ARG; /* block (expr) */ 07725 else 07726 c = tLBRACE; /* hash */ 07727 COND_PUSH(0); 07728 CMDARG_PUSH(0); 07729 lex_state = EXPR_BEG; 07730 if (c != tLBRACE) command_start = TRUE; 07731 return c; 07732 07733 case '\\': 07734 c = nextc(); 07735 if (c == '\n') { 07736 space_seen = 1; 07737 #ifdef RIPPER 07738 ripper_dispatch_scan_event(parser, tSP); 07739 #endif 07740 goto retry; /* skip \\n */ 07741 } 07742 pushback(c); 07743 return '\\'; 07744 07745 case '%': 07746 if (IS_lex_state(EXPR_BEG_ANY)) { 07747 int term; 07748 int paren; 07749 07750 c = nextc(); 07751 quotation: 07752 if (c == -1 || !ISALNUM(c)) { 07753 term = c; 07754 c = 'Q'; 07755 } 07756 else { 07757 term = nextc(); 07758 if (rb_enc_isalnum(term, current_enc) || !parser_isascii()) { 07759 yyerror("unknown type of %string"); 07760 return 0; 07761 } 07762 } 07763 if (c == -1 || term == -1) { 07764 compile_error(PARSER_ARG "unterminated quoted string meets end of file"); 07765 return 0; 07766 } 07767 paren = term; 07768 if (term == '(') term = ')'; 07769 else if (term == '[') term = ']'; 07770 else if (term == '{') term = '}'; 07771 else if (term == '<') term = '>'; 07772 else paren = 0; 07773 07774 switch (c) { 07775 case 'Q': 07776 lex_strterm = NEW_STRTERM(str_dquote, term, paren); 07777 return tSTRING_BEG; 07778 07779 case 'q': 07780 lex_strterm = NEW_STRTERM(str_squote, term, paren); 07781 return tSTRING_BEG; 07782 07783 case 'W': 07784 lex_strterm = NEW_STRTERM(str_dword, term, paren); 07785 do {c = nextc();} while (ISSPACE(c)); 07786 pushback(c); 07787 return tWORDS_BEG; 07788 07789 case 'w': 07790 lex_strterm = NEW_STRTERM(str_sword, term, paren); 07791 do {c = nextc();} while (ISSPACE(c)); 07792 pushback(c); 07793 return tQWORDS_BEG; 07794 07795 case 'I': 07796 lex_strterm = NEW_STRTERM(str_dword, term, paren); 07797 do {c = nextc();} while (ISSPACE(c)); 07798 pushback(c); 07799 return tSYMBOLS_BEG; 07800 07801 case 'i': 07802 lex_strterm = NEW_STRTERM(str_sword, term, paren); 07803 do {c = nextc();} while (ISSPACE(c)); 07804 pushback(c); 07805 return tQSYMBOLS_BEG; 07806 07807 case 'x': 07808 lex_strterm = NEW_STRTERM(str_xquote, term, paren); 07809 return tXSTRING_BEG; 07810 07811 case 'r': 07812 lex_strterm = NEW_STRTERM(str_regexp, term, paren); 07813 return tREGEXP_BEG; 07814 07815 case 's': 07816 lex_strterm = NEW_STRTERM(str_ssym, term, paren); 07817 lex_state = EXPR_FNAME; 07818 return tSYMBEG; 07819 07820 default: 07821 yyerror("unknown type of %string"); 07822 return 0; 07823 } 07824 } 07825 if ((c = nextc()) == '=') { 07826 set_yylval_id('%'); 07827 lex_state = EXPR_BEG; 07828 return tOP_ASGN; 07829 } 07830 if (IS_SPCARG(c)) { 07831 goto quotation; 07832 } 07833 lex_state = IS_AFTER_OPERATOR() ? EXPR_ARG : EXPR_BEG; 07834 pushback(c); 07835 warn_balanced("%%", "string literal"); 07836 return '%'; 07837 07838 case '$': 07839 lex_state = EXPR_END; 07840 newtok(); 07841 c = nextc(); 07842 switch (c) { 07843 case '_': /* $_: last read line string */ 07844 c = nextc(); 07845 if (parser_is_identchar()) { 07846 tokadd('$'); 07847 tokadd('_'); 07848 break; 07849 } 07850 pushback(c); 07851 c = '_'; 07852 /* fall through */ 07853 case '~': /* $~: match-data */ 07854 case '*': /* $*: argv */ 07855 case '$': /* $$: pid */ 07856 case '?': /* $?: last status */ 07857 case '!': /* $!: error string */ 07858 case '@': /* $@: error position */ 07859 case '/': /* $/: input record separator */ 07860 case '\\': /* $\: output record separator */ 07861 case ';': /* $;: field separator */ 07862 case ',': /* $,: output field separator */ 07863 case '.': /* $.: last read line number */ 07864 case '=': /* $=: ignorecase */ 07865 case ':': /* $:: load path */ 07866 case '<': /* $<: reading filename */ 07867 case '>': /* $>: default output handle */ 07868 case '\"': /* $": already loaded files */ 07869 tokadd('$'); 07870 tokadd(c); 07871 tokfix(); 07872 set_yylval_name(rb_intern(tok())); 07873 return tGVAR; 07874 07875 case '-': 07876 tokadd('$'); 07877 tokadd(c); 07878 c = nextc(); 07879 if (parser_is_identchar()) { 07880 if (tokadd_mbchar(c) == -1) return 0; 07881 } 07882 else { 07883 pushback(c); 07884 } 07885 gvar: 07886 tokfix(); 07887 set_yylval_name(rb_intern(tok())); 07888 return tGVAR; 07889 07890 case '&': /* $&: last match */ 07891 case '`': /* $`: string before last match */ 07892 case '\'': /* $': string after last match */ 07893 case '+': /* $+: string matches last paren. */ 07894 if (IS_lex_state_for(last_state, EXPR_FNAME)) { 07895 tokadd('$'); 07896 tokadd(c); 07897 goto gvar; 07898 } 07899 set_yylval_node(NEW_BACK_REF(c)); 07900 return tBACK_REF; 07901 07902 case '1': case '2': case '3': 07903 case '4': case '5': case '6': 07904 case '7': case '8': case '9': 07905 tokadd('$'); 07906 do { 07907 tokadd(c); 07908 c = nextc(); 07909 } while (c != -1 && ISDIGIT(c)); 07910 pushback(c); 07911 if (IS_lex_state_for(last_state, EXPR_FNAME)) goto gvar; 07912 tokfix(); 07913 set_yylval_node(NEW_NTH_REF(atoi(tok()+1))); 07914 return tNTH_REF; 07915 07916 default: 07917 if (!parser_is_identchar()) { 07918 pushback(c); 07919 compile_error(PARSER_ARG "`$%c' is not allowed as a global variable name", c); 07920 return 0; 07921 } 07922 case '0': 07923 tokadd('$'); 07924 } 07925 break; 07926 07927 case '@': 07928 c = nextc(); 07929 newtok(); 07930 tokadd('@'); 07931 if (c == '@') { 07932 tokadd('@'); 07933 c = nextc(); 07934 } 07935 if (c != -1 && (ISDIGIT(c) || !parser_is_identchar())) { 07936 pushback(c); 07937 if (tokidx == 1) { 07938 compile_error(PARSER_ARG "`@%c' is not allowed as an instance variable name", c); 07939 } 07940 else { 07941 compile_error(PARSER_ARG "`@@%c' is not allowed as a class variable name", c); 07942 } 07943 return 0; 07944 } 07945 break; 07946 07947 case '_': 07948 if (was_bol() && whole_match_p("__END__", 7, 0)) { 07949 ruby__end__seen = 1; 07950 parser->eofp = Qtrue; 07951 #ifndef RIPPER 07952 return -1; 07953 #else 07954 lex_goto_eol(parser); 07955 ripper_dispatch_scan_event(parser, k__END__); 07956 return 0; 07957 #endif 07958 } 07959 newtok(); 07960 break; 07961 07962 default: 07963 if (!parser_is_identchar()) { 07964 rb_compile_error(PARSER_ARG "Invalid char `\\x%02X' in expression", c); 07965 goto retry; 07966 } 07967 07968 newtok(); 07969 break; 07970 } 07971 07972 mb = ENC_CODERANGE_7BIT; 07973 do { 07974 if (!ISASCII(c)) mb = ENC_CODERANGE_UNKNOWN; 07975 if (tokadd_mbchar(c) == -1) return 0; 07976 c = nextc(); 07977 } while (parser_is_identchar()); 07978 switch (tok()[0]) { 07979 case '@': case '$': 07980 pushback(c); 07981 break; 07982 default: 07983 if ((c == '!' || c == '?') && !peek('=')) { 07984 tokadd(c); 07985 } 07986 else { 07987 pushback(c); 07988 } 07989 } 07990 tokfix(); 07991 07992 { 07993 int result = 0; 07994 07995 last_state = lex_state; 07996 switch (tok()[0]) { 07997 case '$': 07998 lex_state = EXPR_END; 07999 result = tGVAR; 08000 break; 08001 case '@': 08002 lex_state = EXPR_END; 08003 if (tok()[1] == '@') 08004 result = tCVAR; 08005 else 08006 result = tIVAR; 08007 break; 08008 08009 default: 08010 if (toklast() == '!' || toklast() == '?') { 08011 result = tFID; 08012 } 08013 else { 08014 if (IS_lex_state(EXPR_FNAME)) { 08015 if ((c = nextc()) == '=' && !peek('~') && !peek('>') && 08016 (!peek('=') || (peek_n('>', 1)))) { 08017 result = tIDENTIFIER; 08018 tokadd(c); 08019 tokfix(); 08020 } 08021 else { 08022 pushback(c); 08023 } 08024 } 08025 if (result == 0 && ISUPPER(tok()[0])) { 08026 result = tCONSTANT; 08027 } 08028 else { 08029 result = tIDENTIFIER; 08030 } 08031 } 08032 08033 if (IS_LABEL_POSSIBLE()) { 08034 if (IS_LABEL_SUFFIX(0)) { 08035 lex_state = EXPR_BEG; 08036 nextc(); 08037 set_yylval_name(TOK_INTERN(!ENC_SINGLE(mb))); 08038 return tLABEL; 08039 } 08040 } 08041 if (mb == ENC_CODERANGE_7BIT && !IS_lex_state(EXPR_DOT)) { 08042 const struct kwtable *kw; 08043 08044 /* See if it is a reserved word. */ 08045 kw = rb_reserved_word(tok(), toklen()); 08046 if (kw) { 08047 enum lex_state_e state = lex_state; 08048 lex_state = kw->state; 08049 if (state == EXPR_FNAME) { 08050 set_yylval_name(rb_intern(kw->name)); 08051 return kw->id[0]; 08052 } 08053 if (lex_state == EXPR_BEG) { 08054 command_start = TRUE; 08055 } 08056 if (kw->id[0] == keyword_do) { 08057 if (lpar_beg && lpar_beg == paren_nest) { 08058 lpar_beg = 0; 08059 --paren_nest; 08060 return keyword_do_LAMBDA; 08061 } 08062 if (COND_P()) return keyword_do_cond; 08063 if (CMDARG_P() && state != EXPR_CMDARG) 08064 return keyword_do_block; 08065 if (state & (EXPR_BEG | EXPR_ENDARG)) 08066 return keyword_do_block; 08067 return keyword_do; 08068 } 08069 if (state & (EXPR_BEG | EXPR_VALUE)) 08070 return kw->id[0]; 08071 else { 08072 if (kw->id[0] != kw->id[1]) 08073 lex_state = EXPR_BEG; 08074 return kw->id[1]; 08075 } 08076 } 08077 } 08078 08079 if (IS_lex_state(EXPR_BEG_ANY | EXPR_ARG_ANY | EXPR_DOT)) { 08080 if (cmd_state) { 08081 lex_state = EXPR_CMDARG; 08082 } 08083 else { 08084 lex_state = EXPR_ARG; 08085 } 08086 } 08087 else if (lex_state == EXPR_FNAME) { 08088 lex_state = EXPR_ENDFN; 08089 } 08090 else { 08091 lex_state = EXPR_END; 08092 } 08093 } 08094 { 08095 ID ident = TOK_INTERN(!ENC_SINGLE(mb)); 08096 08097 set_yylval_name(ident); 08098 if (!IS_lex_state_for(last_state, EXPR_DOT|EXPR_FNAME) && 08099 is_local_id(ident) && lvar_defined(ident)) { 08100 lex_state = EXPR_END; 08101 } 08102 } 08103 return result; 08104 } 08105 } 08106 08107 #if YYPURE 08108 static int 08109 yylex(void *lval, void *p) 08110 #else 08111 yylex(void *p) 08112 #endif 08113 { 08114 struct parser_params *parser = (struct parser_params*)p; 08115 int t; 08116 08117 #if YYPURE 08118 parser->parser_yylval = lval; 08119 parser->parser_yylval->val = Qundef; 08120 #endif 08121 t = parser_yylex(parser); 08122 #ifdef RIPPER 08123 if (!NIL_P(parser->delayed)) { 08124 ripper_dispatch_delayed_token(parser, t); 08125 return t; 08126 } 08127 if (t != 0) 08128 ripper_dispatch_scan_event(parser, t); 08129 #endif 08130 08131 return t; 08132 } 08133 08134 #ifndef RIPPER 08135 static NODE* 08136 node_newnode(struct parser_params *parser, enum node_type type, VALUE a0, VALUE a1, VALUE a2) 08137 { 08138 NODE *n = (rb_node_newnode)(type, a0, a1, a2); 08139 nd_set_line(n, ruby_sourceline); 08140 return n; 08141 } 08142 08143 static enum node_type 08144 nodetype(NODE *node) /* for debug */ 08145 { 08146 return (enum node_type)nd_type(node); 08147 } 08148 08149 static int 08150 nodeline(NODE *node) 08151 { 08152 return nd_line(node); 08153 } 08154 08155 static NODE* 08156 newline_node(NODE *node) 08157 { 08158 if (node) { 08159 node = remove_begin(node); 08160 node->flags |= NODE_FL_NEWLINE; 08161 } 08162 return node; 08163 } 08164 08165 static void 08166 fixpos(NODE *node, NODE *orig) 08167 { 08168 if (!node) return; 08169 if (!orig) return; 08170 if (orig == (NODE*)1) return; 08171 nd_set_line(node, nd_line(orig)); 08172 } 08173 08174 static void 08175 parser_warning(struct parser_params *parser, NODE *node, const char *mesg) 08176 { 08177 rb_compile_warning(ruby_sourcefile, nd_line(node), "%s", mesg); 08178 } 08179 #define parser_warning(node, mesg) parser_warning(parser, (node), (mesg)) 08180 08181 static void 08182 parser_warn(struct parser_params *parser, NODE *node, const char *mesg) 08183 { 08184 rb_compile_warn(ruby_sourcefile, nd_line(node), "%s", mesg); 08185 } 08186 #define parser_warn(node, mesg) parser_warn(parser, (node), (mesg)) 08187 08188 static NODE* 08189 block_append_gen(struct parser_params *parser, NODE *head, NODE *tail) 08190 { 08191 NODE *end, *h = head, *nd; 08192 08193 if (tail == 0) return head; 08194 08195 if (h == 0) return tail; 08196 switch (nd_type(h)) { 08197 case NODE_LIT: 08198 case NODE_STR: 08199 case NODE_SELF: 08200 case NODE_TRUE: 08201 case NODE_FALSE: 08202 case NODE_NIL: 08203 parser_warning(h, "unused literal ignored"); 08204 return tail; 08205 default: 08206 h = end = NEW_BLOCK(head); 08207 end->nd_end = end; 08208 fixpos(end, head); 08209 head = end; 08210 break; 08211 case NODE_BLOCK: 08212 end = h->nd_end; 08213 break; 08214 } 08215 08216 nd = end->nd_head; 08217 switch (nd_type(nd)) { 08218 case NODE_RETURN: 08219 case NODE_BREAK: 08220 case NODE_NEXT: 08221 case NODE_REDO: 08222 case NODE_RETRY: 08223 if (RTEST(ruby_verbose)) { 08224 parser_warning(tail, "statement not reached"); 08225 } 08226 break; 08227 08228 default: 08229 break; 08230 } 08231 08232 if (nd_type(tail) != NODE_BLOCK) { 08233 tail = NEW_BLOCK(tail); 08234 tail->nd_end = tail; 08235 } 08236 end->nd_next = tail; 08237 h->nd_end = tail->nd_end; 08238 return head; 08239 } 08240 08241 /* append item to the list */ 08242 static NODE* 08243 list_append_gen(struct parser_params *parser, NODE *list, NODE *item) 08244 { 08245 NODE *last; 08246 08247 if (list == 0) return NEW_LIST(item); 08248 if (list->nd_next) { 08249 last = list->nd_next->nd_end; 08250 } 08251 else { 08252 last = list; 08253 } 08254 08255 list->nd_alen += 1; 08256 last->nd_next = NEW_LIST(item); 08257 list->nd_next->nd_end = last->nd_next; 08258 return list; 08259 } 08260 08261 /* concat two lists */ 08262 static NODE* 08263 list_concat_gen(struct parser_params *parser, NODE *head, NODE *tail) 08264 { 08265 NODE *last; 08266 08267 if (head->nd_next) { 08268 last = head->nd_next->nd_end; 08269 } 08270 else { 08271 last = head; 08272 } 08273 08274 head->nd_alen += tail->nd_alen; 08275 last->nd_next = tail; 08276 if (tail->nd_next) { 08277 head->nd_next->nd_end = tail->nd_next->nd_end; 08278 } 08279 else { 08280 head->nd_next->nd_end = tail; 08281 } 08282 08283 return head; 08284 } 08285 08286 static int 08287 literal_concat0(struct parser_params *parser, VALUE head, VALUE tail) 08288 { 08289 if (NIL_P(tail)) return 1; 08290 if (!rb_enc_compatible(head, tail)) { 08291 compile_error(PARSER_ARG "string literal encodings differ (%s / %s)", 08292 rb_enc_name(rb_enc_get(head)), 08293 rb_enc_name(rb_enc_get(tail))); 08294 rb_str_resize(head, 0); 08295 rb_str_resize(tail, 0); 08296 return 0; 08297 } 08298 rb_str_buf_append(head, tail); 08299 return 1; 08300 } 08301 08302 /* concat two string literals */ 08303 static NODE * 08304 literal_concat_gen(struct parser_params *parser, NODE *head, NODE *tail) 08305 { 08306 enum node_type htype; 08307 NODE *headlast; 08308 VALUE lit; 08309 08310 if (!head) return tail; 08311 if (!tail) return head; 08312 08313 htype = nd_type(head); 08314 if (htype == NODE_EVSTR) { 08315 NODE *node = NEW_DSTR(Qnil); 08316 head = list_append(node, head); 08317 htype = NODE_DSTR; 08318 } 08319 switch (nd_type(tail)) { 08320 case NODE_STR: 08321 if (htype == NODE_DSTR && (headlast = head->nd_next->nd_end->nd_head) && 08322 nd_type(headlast) == NODE_STR) { 08323 htype = NODE_STR; 08324 lit = headlast->nd_lit; 08325 } 08326 else { 08327 lit = head->nd_lit; 08328 } 08329 if (htype == NODE_STR) { 08330 if (!literal_concat0(parser, lit, tail->nd_lit)) { 08331 error: 08332 rb_gc_force_recycle((VALUE)head); 08333 rb_gc_force_recycle((VALUE)tail); 08334 return 0; 08335 } 08336 rb_gc_force_recycle((VALUE)tail); 08337 } 08338 else { 08339 list_append(head, tail); 08340 } 08341 break; 08342 08343 case NODE_DSTR: 08344 if (htype == NODE_STR) { 08345 if (!literal_concat0(parser, head->nd_lit, tail->nd_lit)) 08346 goto error; 08347 tail->nd_lit = head->nd_lit; 08348 rb_gc_force_recycle((VALUE)head); 08349 head = tail; 08350 } 08351 else if (NIL_P(tail->nd_lit)) { 08352 append: 08353 head->nd_alen += tail->nd_alen - 1; 08354 head->nd_next->nd_end->nd_next = tail->nd_next; 08355 head->nd_next->nd_end = tail->nd_next->nd_end; 08356 rb_gc_force_recycle((VALUE)tail); 08357 } 08358 else if (htype == NODE_DSTR && (headlast = head->nd_next->nd_end->nd_head) && 08359 nd_type(headlast) == NODE_STR) { 08360 lit = headlast->nd_lit; 08361 if (!literal_concat0(parser, lit, tail->nd_lit)) 08362 goto error; 08363 tail->nd_lit = Qnil; 08364 goto append; 08365 } 08366 else { 08367 nd_set_type(tail, NODE_ARRAY); 08368 tail->nd_head = NEW_STR(tail->nd_lit); 08369 list_concat(head, tail); 08370 } 08371 break; 08372 08373 case NODE_EVSTR: 08374 if (htype == NODE_STR) { 08375 nd_set_type(head, NODE_DSTR); 08376 head->nd_alen = 1; 08377 } 08378 list_append(head, tail); 08379 break; 08380 } 08381 return head; 08382 } 08383 08384 static NODE * 08385 evstr2dstr_gen(struct parser_params *parser, NODE *node) 08386 { 08387 if (nd_type(node) == NODE_EVSTR) { 08388 node = list_append(NEW_DSTR(Qnil), node); 08389 } 08390 return node; 08391 } 08392 08393 static NODE * 08394 new_evstr_gen(struct parser_params *parser, NODE *node) 08395 { 08396 NODE *head = node; 08397 08398 if (node) { 08399 switch (nd_type(node)) { 08400 case NODE_STR: case NODE_DSTR: case NODE_EVSTR: 08401 return node; 08402 } 08403 } 08404 return NEW_EVSTR(head); 08405 } 08406 08407 static NODE * 08408 call_bin_op_gen(struct parser_params *parser, NODE *recv, ID id, NODE *arg1) 08409 { 08410 value_expr(recv); 08411 value_expr(arg1); 08412 return NEW_CALL(recv, id, NEW_LIST(arg1)); 08413 } 08414 08415 static NODE * 08416 call_uni_op_gen(struct parser_params *parser, NODE *recv, ID id) 08417 { 08418 value_expr(recv); 08419 return NEW_CALL(recv, id, 0); 08420 } 08421 08422 static NODE* 08423 match_op_gen(struct parser_params *parser, NODE *node1, NODE *node2) 08424 { 08425 value_expr(node1); 08426 value_expr(node2); 08427 if (node1) { 08428 switch (nd_type(node1)) { 08429 case NODE_DREGX: 08430 case NODE_DREGX_ONCE: 08431 return NEW_MATCH2(node1, node2); 08432 08433 case NODE_LIT: 08434 if (RB_TYPE_P(node1->nd_lit, T_REGEXP)) { 08435 return NEW_MATCH2(node1, node2); 08436 } 08437 } 08438 } 08439 08440 if (node2) { 08441 switch (nd_type(node2)) { 08442 case NODE_DREGX: 08443 case NODE_DREGX_ONCE: 08444 return NEW_MATCH3(node2, node1); 08445 08446 case NODE_LIT: 08447 if (RB_TYPE_P(node2->nd_lit, T_REGEXP)) { 08448 return NEW_MATCH3(node2, node1); 08449 } 08450 } 08451 } 08452 08453 return NEW_CALL(node1, tMATCH, NEW_LIST(node2)); 08454 } 08455 08456 static NODE* 08457 gettable_gen(struct parser_params *parser, ID id) 08458 { 08459 switch (id) { 08460 case keyword_self: 08461 return NEW_SELF(); 08462 case keyword_nil: 08463 return NEW_NIL(); 08464 case keyword_true: 08465 return NEW_TRUE(); 08466 case keyword_false: 08467 return NEW_FALSE(); 08468 case keyword__FILE__: 08469 return NEW_STR(rb_external_str_new_with_enc(ruby_sourcefile, strlen(ruby_sourcefile), 08470 rb_filesystem_encoding())); 08471 case keyword__LINE__: 08472 return NEW_LIT(INT2FIX(tokline)); 08473 case keyword__ENCODING__: 08474 return NEW_LIT(rb_enc_from_encoding(current_enc)); 08475 } 08476 switch (id_type(id)) { 08477 case ID_LOCAL: 08478 if (dyna_in_block() && dvar_defined(id)) return NEW_DVAR(id); 08479 if (local_id(id)) return NEW_LVAR(id); 08480 /* method call without arguments */ 08481 return NEW_VCALL(id); 08482 case ID_GLOBAL: 08483 return NEW_GVAR(id); 08484 case ID_INSTANCE: 08485 return NEW_IVAR(id); 08486 case ID_CONST: 08487 return NEW_CONST(id); 08488 case ID_CLASS: 08489 return NEW_CVAR(id); 08490 } 08491 compile_error(PARSER_ARG "identifier %s is not valid to get", rb_id2name(id)); 08492 return 0; 08493 } 08494 #else /* !RIPPER */ 08495 static int 08496 id_is_var_gen(struct parser_params *parser, ID id) 08497 { 08498 if (is_notop_id(id)) { 08499 switch (id & ID_SCOPE_MASK) { 08500 case ID_GLOBAL: case ID_INSTANCE: case ID_CONST: case ID_CLASS: 08501 return 1; 08502 case ID_LOCAL: 08503 if (dyna_in_block() && dvar_defined(id)) return 1; 08504 if (local_id(id)) return 1; 08505 /* method call without arguments */ 08506 return 0; 08507 } 08508 } 08509 compile_error(PARSER_ARG "identifier %s is not valid to get", rb_id2name(id)); 08510 return 0; 08511 } 08512 #endif /* !RIPPER */ 08513 08514 #if PARSER_DEBUG 08515 static const char * 08516 lex_state_name(enum lex_state_e state) 08517 { 08518 static const char names[][12] = { 08519 "EXPR_BEG", "EXPR_END", "EXPR_ENDARG", "EXPR_ENDFN", "EXPR_ARG", 08520 "EXPR_CMDARG", "EXPR_MID", "EXPR_FNAME", "EXPR_DOT", "EXPR_CLASS", 08521 "EXPR_VALUE", 08522 }; 08523 08524 if ((unsigned)state & ~(~0u << EXPR_MAX_STATE)) 08525 return names[ffs(state)]; 08526 return NULL; 08527 } 08528 #endif 08529 08530 #ifdef RIPPER 08531 static VALUE 08532 assignable_gen(struct parser_params *parser, VALUE lhs) 08533 #else 08534 static NODE* 08535 assignable_gen(struct parser_params *parser, ID id, NODE *val) 08536 #endif 08537 { 08538 #ifdef RIPPER 08539 ID id = get_id(lhs); 08540 # define assignable_result(x) get_value(lhs) 08541 # define parser_yyerror(parser, x) dispatch1(assign_error, lhs) 08542 #else 08543 # define assignable_result(x) (x) 08544 #endif 08545 if (!id) return assignable_result(0); 08546 switch (id) { 08547 case keyword_self: 08548 yyerror("Can't change the value of self"); 08549 goto error; 08550 case keyword_nil: 08551 yyerror("Can't assign to nil"); 08552 goto error; 08553 case keyword_true: 08554 yyerror("Can't assign to true"); 08555 goto error; 08556 case keyword_false: 08557 yyerror("Can't assign to false"); 08558 goto error; 08559 case keyword__FILE__: 08560 yyerror("Can't assign to __FILE__"); 08561 goto error; 08562 case keyword__LINE__: 08563 yyerror("Can't assign to __LINE__"); 08564 goto error; 08565 case keyword__ENCODING__: 08566 yyerror("Can't assign to __ENCODING__"); 08567 goto error; 08568 } 08569 switch (id_type(id)) { 08570 case ID_LOCAL: 08571 if (dyna_in_block()) { 08572 if (dvar_curr(id)) { 08573 return assignable_result(NEW_DASGN_CURR(id, val)); 08574 } 08575 else if (dvar_defined(id)) { 08576 return assignable_result(NEW_DASGN(id, val)); 08577 } 08578 else if (local_id(id)) { 08579 return assignable_result(NEW_LASGN(id, val)); 08580 } 08581 else { 08582 dyna_var(id); 08583 return assignable_result(NEW_DASGN_CURR(id, val)); 08584 } 08585 } 08586 else { 08587 if (!local_id(id)) { 08588 local_var(id); 08589 } 08590 return assignable_result(NEW_LASGN(id, val)); 08591 } 08592 break; 08593 case ID_GLOBAL: 08594 return assignable_result(NEW_GASGN(id, val)); 08595 case ID_INSTANCE: 08596 return assignable_result(NEW_IASGN(id, val)); 08597 case ID_CONST: 08598 if (!in_def && !in_single) 08599 return assignable_result(NEW_CDECL(id, val, 0)); 08600 yyerror("dynamic constant assignment"); 08601 break; 08602 case ID_CLASS: 08603 return assignable_result(NEW_CVASGN(id, val)); 08604 default: 08605 compile_error(PARSER_ARG "identifier %s is not valid to set", rb_id2name(id)); 08606 } 08607 error: 08608 return assignable_result(0); 08609 #undef assignable_result 08610 #undef parser_yyerror 08611 } 08612 08613 static int 08614 is_private_local_id(ID name) 08615 { 08616 VALUE s; 08617 if (name == idUScore) return 1; 08618 if (!is_local_id(name)) return 0; 08619 s = rb_id2str(name); 08620 if (!s) return 0; 08621 return RSTRING_PTR(s)[0] == '_'; 08622 } 08623 08624 #define LVAR_USED ((ID)1 << (sizeof(ID) * CHAR_BIT - 1)) 08625 08626 static ID 08627 shadowing_lvar_gen(struct parser_params *parser, ID name) 08628 { 08629 if (is_private_local_id(name)) return name; 08630 if (dyna_in_block()) { 08631 if (dvar_curr(name)) { 08632 yyerror("duplicated argument name"); 08633 } 08634 else if (dvar_defined_get(name) || local_id(name)) { 08635 rb_warningS("shadowing outer local variable - %s", rb_id2name(name)); 08636 vtable_add(lvtbl->vars, name); 08637 if (lvtbl->used) { 08638 vtable_add(lvtbl->used, (ID)ruby_sourceline | LVAR_USED); 08639 } 08640 } 08641 } 08642 else { 08643 if (local_id(name)) { 08644 yyerror("duplicated argument name"); 08645 } 08646 } 08647 return name; 08648 } 08649 08650 static void 08651 new_bv_gen(struct parser_params *parser, ID name) 08652 { 08653 if (!name) return; 08654 if (!is_local_id(name)) { 08655 compile_error(PARSER_ARG "invalid local variable - %s", 08656 rb_id2name(name)); 08657 return; 08658 } 08659 shadowing_lvar(name); 08660 dyna_var(name); 08661 } 08662 08663 #ifndef RIPPER 08664 static NODE * 08665 aryset_gen(struct parser_params *parser, NODE *recv, NODE *idx) 08666 { 08667 if (recv && nd_type(recv) == NODE_SELF) 08668 recv = (NODE *)1; 08669 return NEW_ATTRASGN(recv, tASET, idx); 08670 } 08671 08672 static void 08673 block_dup_check_gen(struct parser_params *parser, NODE *node1, NODE *node2) 08674 { 08675 if (node2 && node1 && nd_type(node1) == NODE_BLOCK_PASS) { 08676 compile_error(PARSER_ARG "both block arg and actual block given"); 08677 } 08678 } 08679 08680 ID 08681 rb_id_attrset(ID id) 08682 { 08683 id &= ~ID_SCOPE_MASK; 08684 id |= ID_ATTRSET; 08685 return id; 08686 } 08687 08688 static NODE * 08689 attrset_gen(struct parser_params *parser, NODE *recv, ID id) 08690 { 08691 if (recv && nd_type(recv) == NODE_SELF) 08692 recv = (NODE *)1; 08693 return NEW_ATTRASGN(recv, rb_id_attrset(id), 0); 08694 } 08695 08696 static void 08697 rb_backref_error_gen(struct parser_params *parser, NODE *node) 08698 { 08699 switch (nd_type(node)) { 08700 case NODE_NTH_REF: 08701 compile_error(PARSER_ARG "Can't set variable $%ld", node->nd_nth); 08702 break; 08703 case NODE_BACK_REF: 08704 compile_error(PARSER_ARG "Can't set variable $%c", (int)node->nd_nth); 08705 break; 08706 } 08707 } 08708 08709 static NODE * 08710 arg_concat_gen(struct parser_params *parser, NODE *node1, NODE *node2) 08711 { 08712 if (!node2) return node1; 08713 switch (nd_type(node1)) { 08714 case NODE_BLOCK_PASS: 08715 if (node1->nd_head) 08716 node1->nd_head = arg_concat(node1->nd_head, node2); 08717 else 08718 node1->nd_head = NEW_LIST(node2); 08719 return node1; 08720 case NODE_ARGSPUSH: 08721 if (nd_type(node2) != NODE_ARRAY) break; 08722 node1->nd_body = list_concat(NEW_LIST(node1->nd_body), node2); 08723 nd_set_type(node1, NODE_ARGSCAT); 08724 return node1; 08725 case NODE_ARGSCAT: 08726 if (nd_type(node2) != NODE_ARRAY || 08727 nd_type(node1->nd_body) != NODE_ARRAY) break; 08728 node1->nd_body = list_concat(node1->nd_body, node2); 08729 return node1; 08730 } 08731 return NEW_ARGSCAT(node1, node2); 08732 } 08733 08734 static NODE * 08735 arg_append_gen(struct parser_params *parser, NODE *node1, NODE *node2) 08736 { 08737 if (!node1) return NEW_LIST(node2); 08738 switch (nd_type(node1)) { 08739 case NODE_ARRAY: 08740 return list_append(node1, node2); 08741 case NODE_BLOCK_PASS: 08742 node1->nd_head = arg_append(node1->nd_head, node2); 08743 return node1; 08744 case NODE_ARGSPUSH: 08745 node1->nd_body = list_append(NEW_LIST(node1->nd_body), node2); 08746 nd_set_type(node1, NODE_ARGSCAT); 08747 return node1; 08748 } 08749 return NEW_ARGSPUSH(node1, node2); 08750 } 08751 08752 static NODE * 08753 splat_array(NODE* node) 08754 { 08755 if (nd_type(node) == NODE_SPLAT) node = node->nd_head; 08756 if (nd_type(node) == NODE_ARRAY) return node; 08757 return 0; 08758 } 08759 08760 static NODE * 08761 node_assign_gen(struct parser_params *parser, NODE *lhs, NODE *rhs) 08762 { 08763 if (!lhs) return 0; 08764 08765 switch (nd_type(lhs)) { 08766 case NODE_GASGN: 08767 case NODE_IASGN: 08768 case NODE_IASGN2: 08769 case NODE_LASGN: 08770 case NODE_DASGN: 08771 case NODE_DASGN_CURR: 08772 case NODE_MASGN: 08773 case NODE_CDECL: 08774 case NODE_CVASGN: 08775 lhs->nd_value = rhs; 08776 break; 08777 08778 case NODE_ATTRASGN: 08779 case NODE_CALL: 08780 lhs->nd_args = arg_append(lhs->nd_args, rhs); 08781 break; 08782 08783 default: 08784 /* should not happen */ 08785 break; 08786 } 08787 08788 return lhs; 08789 } 08790 08791 static int 08792 value_expr_gen(struct parser_params *parser, NODE *node) 08793 { 08794 int cond = 0; 08795 08796 if (!node) { 08797 rb_warning0("empty expression"); 08798 } 08799 while (node) { 08800 switch (nd_type(node)) { 08801 case NODE_DEFN: 08802 case NODE_DEFS: 08803 parser_warning(node, "void value expression"); 08804 return FALSE; 08805 08806 case NODE_RETURN: 08807 case NODE_BREAK: 08808 case NODE_NEXT: 08809 case NODE_REDO: 08810 case NODE_RETRY: 08811 if (!cond) yyerror("void value expression"); 08812 /* or "control never reach"? */ 08813 return FALSE; 08814 08815 case NODE_BLOCK: 08816 while (node->nd_next) { 08817 node = node->nd_next; 08818 } 08819 node = node->nd_head; 08820 break; 08821 08822 case NODE_BEGIN: 08823 node = node->nd_body; 08824 break; 08825 08826 case NODE_IF: 08827 if (!node->nd_body) { 08828 node = node->nd_else; 08829 break; 08830 } 08831 else if (!node->nd_else) { 08832 node = node->nd_body; 08833 break; 08834 } 08835 if (!value_expr(node->nd_body)) return FALSE; 08836 node = node->nd_else; 08837 break; 08838 08839 case NODE_AND: 08840 case NODE_OR: 08841 cond = 1; 08842 node = node->nd_2nd; 08843 break; 08844 08845 default: 08846 return TRUE; 08847 } 08848 } 08849 08850 return TRUE; 08851 } 08852 08853 static void 08854 void_expr_gen(struct parser_params *parser, NODE *node) 08855 { 08856 const char *useless = 0; 08857 08858 if (!RTEST(ruby_verbose)) return; 08859 08860 if (!node) return; 08861 switch (nd_type(node)) { 08862 case NODE_CALL: 08863 switch (node->nd_mid) { 08864 case '+': 08865 case '-': 08866 case '*': 08867 case '/': 08868 case '%': 08869 case tPOW: 08870 case tUPLUS: 08871 case tUMINUS: 08872 case '|': 08873 case '^': 08874 case '&': 08875 case tCMP: 08876 case '>': 08877 case tGEQ: 08878 case '<': 08879 case tLEQ: 08880 case tEQ: 08881 case tNEQ: 08882 useless = rb_id2name(node->nd_mid); 08883 break; 08884 } 08885 break; 08886 08887 case NODE_LVAR: 08888 case NODE_DVAR: 08889 case NODE_GVAR: 08890 case NODE_IVAR: 08891 case NODE_CVAR: 08892 case NODE_NTH_REF: 08893 case NODE_BACK_REF: 08894 useless = "a variable"; 08895 break; 08896 case NODE_CONST: 08897 useless = "a constant"; 08898 break; 08899 case NODE_LIT: 08900 case NODE_STR: 08901 case NODE_DSTR: 08902 case NODE_DREGX: 08903 case NODE_DREGX_ONCE: 08904 useless = "a literal"; 08905 break; 08906 case NODE_COLON2: 08907 case NODE_COLON3: 08908 useless = "::"; 08909 break; 08910 case NODE_DOT2: 08911 useless = ".."; 08912 break; 08913 case NODE_DOT3: 08914 useless = "..."; 08915 break; 08916 case NODE_SELF: 08917 useless = "self"; 08918 break; 08919 case NODE_NIL: 08920 useless = "nil"; 08921 break; 08922 case NODE_TRUE: 08923 useless = "true"; 08924 break; 08925 case NODE_FALSE: 08926 useless = "false"; 08927 break; 08928 case NODE_DEFINED: 08929 useless = "defined?"; 08930 break; 08931 } 08932 08933 if (useless) { 08934 int line = ruby_sourceline; 08935 08936 ruby_sourceline = nd_line(node); 08937 rb_warnS("possibly useless use of %s in void context", useless); 08938 ruby_sourceline = line; 08939 } 08940 } 08941 08942 static void 08943 void_stmts_gen(struct parser_params *parser, NODE *node) 08944 { 08945 if (!RTEST(ruby_verbose)) return; 08946 if (!node) return; 08947 if (nd_type(node) != NODE_BLOCK) return; 08948 08949 for (;;) { 08950 if (!node->nd_next) return; 08951 void_expr0(node->nd_head); 08952 node = node->nd_next; 08953 } 08954 } 08955 08956 static NODE * 08957 remove_begin(NODE *node) 08958 { 08959 NODE **n = &node, *n1 = node; 08960 while (n1 && nd_type(n1) == NODE_BEGIN && n1->nd_body) { 08961 *n = n1 = n1->nd_body; 08962 } 08963 return node; 08964 } 08965 08966 static void 08967 reduce_nodes_gen(struct parser_params *parser, NODE **body) 08968 { 08969 NODE *node = *body; 08970 08971 if (!node) { 08972 *body = NEW_NIL(); 08973 return; 08974 } 08975 #define subnodes(n1, n2) \ 08976 ((!node->n1) ? (node->n2 ? (body = &node->n2, 1) : 0) : \ 08977 (!node->n2) ? (body = &node->n1, 1) : \ 08978 (reduce_nodes(&node->n1), body = &node->n2, 1)) 08979 08980 while (node) { 08981 int newline = (int)(node->flags & NODE_FL_NEWLINE); 08982 switch (nd_type(node)) { 08983 end: 08984 case NODE_NIL: 08985 *body = 0; 08986 return; 08987 case NODE_RETURN: 08988 *body = node = node->nd_stts; 08989 if (newline && node) node->flags |= NODE_FL_NEWLINE; 08990 continue; 08991 case NODE_BEGIN: 08992 *body = node = node->nd_body; 08993 if (newline && node) node->flags |= NODE_FL_NEWLINE; 08994 continue; 08995 case NODE_BLOCK: 08996 body = &node->nd_end->nd_head; 08997 break; 08998 case NODE_IF: 08999 if (subnodes(nd_body, nd_else)) break; 09000 return; 09001 case NODE_CASE: 09002 body = &node->nd_body; 09003 break; 09004 case NODE_WHEN: 09005 if (!subnodes(nd_body, nd_next)) goto end; 09006 break; 09007 case NODE_ENSURE: 09008 if (!subnodes(nd_head, nd_resq)) goto end; 09009 break; 09010 case NODE_RESCUE: 09011 if (node->nd_else) { 09012 body = &node->nd_resq; 09013 break; 09014 } 09015 if (!subnodes(nd_head, nd_resq)) goto end; 09016 break; 09017 default: 09018 return; 09019 } 09020 node = *body; 09021 if (newline && node) node->flags |= NODE_FL_NEWLINE; 09022 } 09023 09024 #undef subnodes 09025 } 09026 09027 static int 09028 is_static_content(NODE *node) 09029 { 09030 if (!node) return 1; 09031 switch (nd_type(node)) { 09032 case NODE_HASH: 09033 if (!(node = node->nd_head)) break; 09034 case NODE_ARRAY: 09035 do { 09036 if (!is_static_content(node->nd_head)) return 0; 09037 } while ((node = node->nd_next) != 0); 09038 case NODE_LIT: 09039 case NODE_STR: 09040 case NODE_NIL: 09041 case NODE_TRUE: 09042 case NODE_FALSE: 09043 case NODE_ZARRAY: 09044 break; 09045 default: 09046 return 0; 09047 } 09048 return 1; 09049 } 09050 09051 static int 09052 assign_in_cond(struct parser_params *parser, NODE *node) 09053 { 09054 switch (nd_type(node)) { 09055 case NODE_MASGN: 09056 yyerror("multiple assignment in conditional"); 09057 return 1; 09058 09059 case NODE_LASGN: 09060 case NODE_DASGN: 09061 case NODE_DASGN_CURR: 09062 case NODE_GASGN: 09063 case NODE_IASGN: 09064 break; 09065 09066 default: 09067 return 0; 09068 } 09069 09070 if (!node->nd_value) return 1; 09071 if (is_static_content(node->nd_value)) { 09072 /* reports always */ 09073 parser_warn(node->nd_value, "found = in conditional, should be =="); 09074 } 09075 return 1; 09076 } 09077 09078 static void 09079 warn_unless_e_option(struct parser_params *parser, NODE *node, const char *str) 09080 { 09081 if (!e_option_supplied(parser)) parser_warn(node, str); 09082 } 09083 09084 static void 09085 warning_unless_e_option(struct parser_params *parser, NODE *node, const char *str) 09086 { 09087 if (!e_option_supplied(parser)) parser_warning(node, str); 09088 } 09089 09090 static void 09091 fixup_nodes(NODE **rootnode) 09092 { 09093 NODE *node, *next, *head; 09094 09095 for (node = *rootnode; node; node = next) { 09096 enum node_type type; 09097 VALUE val; 09098 09099 next = node->nd_next; 09100 head = node->nd_head; 09101 rb_gc_force_recycle((VALUE)node); 09102 *rootnode = next; 09103 switch (type = nd_type(head)) { 09104 case NODE_DOT2: 09105 case NODE_DOT3: 09106 val = rb_range_new(head->nd_beg->nd_lit, head->nd_end->nd_lit, 09107 type == NODE_DOT3); 09108 rb_gc_force_recycle((VALUE)head->nd_beg); 09109 rb_gc_force_recycle((VALUE)head->nd_end); 09110 nd_set_type(head, NODE_LIT); 09111 head->nd_lit = val; 09112 break; 09113 default: 09114 break; 09115 } 09116 } 09117 } 09118 09119 static NODE *cond0(struct parser_params*,NODE*); 09120 09121 static NODE* 09122 range_op(struct parser_params *parser, NODE *node) 09123 { 09124 enum node_type type; 09125 09126 if (node == 0) return 0; 09127 09128 type = nd_type(node); 09129 value_expr(node); 09130 if (type == NODE_LIT && FIXNUM_P(node->nd_lit)) { 09131 warn_unless_e_option(parser, node, "integer literal in conditional range"); 09132 return NEW_CALL(node, tEQ, NEW_LIST(NEW_GVAR(rb_intern("$.")))); 09133 } 09134 return cond0(parser, node); 09135 } 09136 09137 static int 09138 literal_node(NODE *node) 09139 { 09140 if (!node) return 1; /* same as NODE_NIL */ 09141 switch (nd_type(node)) { 09142 case NODE_LIT: 09143 case NODE_STR: 09144 case NODE_DSTR: 09145 case NODE_EVSTR: 09146 case NODE_DREGX: 09147 case NODE_DREGX_ONCE: 09148 case NODE_DSYM: 09149 return 2; 09150 case NODE_TRUE: 09151 case NODE_FALSE: 09152 case NODE_NIL: 09153 return 1; 09154 } 09155 return 0; 09156 } 09157 09158 static NODE* 09159 cond0(struct parser_params *parser, NODE *node) 09160 { 09161 if (node == 0) return 0; 09162 assign_in_cond(parser, node); 09163 09164 switch (nd_type(node)) { 09165 case NODE_DSTR: 09166 case NODE_EVSTR: 09167 case NODE_STR: 09168 rb_warn0("string literal in condition"); 09169 break; 09170 09171 case NODE_DREGX: 09172 case NODE_DREGX_ONCE: 09173 warning_unless_e_option(parser, node, "regex literal in condition"); 09174 return NEW_MATCH2(node, NEW_GVAR(rb_intern("$_"))); 09175 09176 case NODE_AND: 09177 case NODE_OR: 09178 node->nd_1st = cond0(parser, node->nd_1st); 09179 node->nd_2nd = cond0(parser, node->nd_2nd); 09180 break; 09181 09182 case NODE_DOT2: 09183 case NODE_DOT3: 09184 node->nd_beg = range_op(parser, node->nd_beg); 09185 node->nd_end = range_op(parser, node->nd_end); 09186 if (nd_type(node) == NODE_DOT2) nd_set_type(node,NODE_FLIP2); 09187 else if (nd_type(node) == NODE_DOT3) nd_set_type(node, NODE_FLIP3); 09188 if (!e_option_supplied(parser)) { 09189 int b = literal_node(node->nd_beg); 09190 int e = literal_node(node->nd_end); 09191 if ((b == 1 && e == 1) || (b + e >= 2 && RTEST(ruby_verbose))) { 09192 parser_warn(node, "range literal in condition"); 09193 } 09194 } 09195 break; 09196 09197 case NODE_DSYM: 09198 parser_warning(node, "literal in condition"); 09199 break; 09200 09201 case NODE_LIT: 09202 if (RB_TYPE_P(node->nd_lit, T_REGEXP)) { 09203 warn_unless_e_option(parser, node, "regex literal in condition"); 09204 nd_set_type(node, NODE_MATCH); 09205 } 09206 else { 09207 parser_warning(node, "literal in condition"); 09208 } 09209 default: 09210 break; 09211 } 09212 return node; 09213 } 09214 09215 static NODE* 09216 cond_gen(struct parser_params *parser, NODE *node) 09217 { 09218 if (node == 0) return 0; 09219 return cond0(parser, node); 09220 } 09221 09222 static NODE* 09223 logop_gen(struct parser_params *parser, enum node_type type, NODE *left, NODE *right) 09224 { 09225 value_expr(left); 09226 if (left && (enum node_type)nd_type(left) == type) { 09227 NODE *node = left, *second; 09228 while ((second = node->nd_2nd) != 0 && (enum node_type)nd_type(second) == type) { 09229 node = second; 09230 } 09231 node->nd_2nd = NEW_NODE(type, second, right, 0); 09232 return left; 09233 } 09234 return NEW_NODE(type, left, right, 0); 09235 } 09236 09237 static void 09238 no_blockarg(struct parser_params *parser, NODE *node) 09239 { 09240 if (node && nd_type(node) == NODE_BLOCK_PASS) { 09241 compile_error(PARSER_ARG "block argument should not be given"); 09242 } 09243 } 09244 09245 static NODE * 09246 ret_args_gen(struct parser_params *parser, NODE *node) 09247 { 09248 if (node) { 09249 no_blockarg(parser, node); 09250 if (nd_type(node) == NODE_ARRAY) { 09251 if (node->nd_next == 0) { 09252 node = node->nd_head; 09253 } 09254 else { 09255 nd_set_type(node, NODE_VALUES); 09256 } 09257 } 09258 } 09259 return node; 09260 } 09261 09262 static NODE * 09263 new_yield_gen(struct parser_params *parser, NODE *node) 09264 { 09265 if (node) no_blockarg(parser, node); 09266 09267 return NEW_YIELD(node); 09268 } 09269 09270 static NODE* 09271 negate_lit(NODE *node) 09272 { 09273 switch (TYPE(node->nd_lit)) { 09274 case T_FIXNUM: 09275 node->nd_lit = LONG2FIX(-FIX2LONG(node->nd_lit)); 09276 break; 09277 case T_BIGNUM: 09278 node->nd_lit = rb_funcall(node->nd_lit,tUMINUS,0,0); 09279 break; 09280 case T_FLOAT: 09281 #if USE_FLONUM 09282 if (FLONUM_P(node->nd_lit)) { 09283 node->nd_lit = DBL2NUM(-RFLOAT_VALUE(node->nd_lit)); 09284 } 09285 else { 09286 RFLOAT(node->nd_lit)->float_value = -RFLOAT_VALUE(node->nd_lit); 09287 } 09288 #else 09289 RFLOAT(node->nd_lit)->float_value = -RFLOAT_VALUE(node->nd_lit); 09290 #endif 09291 break; 09292 default: 09293 break; 09294 } 09295 return node; 09296 } 09297 09298 static NODE * 09299 arg_blk_pass(NODE *node1, NODE *node2) 09300 { 09301 if (node2) { 09302 node2->nd_head = node1; 09303 return node2; 09304 } 09305 return node1; 09306 } 09307 09308 09309 static NODE* 09310 new_args_gen(struct parser_params *parser, NODE *m, NODE *o, ID r, NODE *p, NODE *tail) 09311 { 09312 int saved_line = ruby_sourceline; 09313 struct rb_args_info *args = tail->nd_ainfo; 09314 09315 args->pre_args_num = m ? rb_long2int(m->nd_plen) : 0; 09316 args->pre_init = m ? m->nd_next : 0; 09317 09318 args->post_args_num = p ? rb_long2int(p->nd_plen) : 0; 09319 args->post_init = p ? p->nd_next : 0; 09320 args->first_post_arg = p ? p->nd_pid : 0; 09321 09322 args->rest_arg = r; 09323 09324 args->opt_args = o; 09325 09326 ruby_sourceline = saved_line; 09327 09328 return tail; 09329 } 09330 09331 static NODE* 09332 new_args_tail_gen(struct parser_params *parser, NODE *k, ID kr, ID b) 09333 { 09334 int saved_line = ruby_sourceline; 09335 struct rb_args_info *args; 09336 NODE *kw_rest_arg = 0; 09337 NODE *node; 09338 09339 args = ALLOC(struct rb_args_info); 09340 MEMZERO(args, struct rb_args_info, 1); 09341 node = NEW_NODE(NODE_ARGS, 0, 0, args); 09342 09343 args->block_arg = b; 09344 args->kw_args = k; 09345 if (k && !kr) kr = internal_id(); 09346 if (kr) { 09347 arg_var(kr); 09348 kw_rest_arg = NEW_DVAR(kr); 09349 } 09350 args->kw_rest_arg = kw_rest_arg; 09351 09352 ruby_sourceline = saved_line; 09353 return node; 09354 } 09355 09356 static NODE* 09357 dsym_node_gen(struct parser_params *parser, NODE *node) 09358 { 09359 VALUE lit; 09360 09361 if (!node) { 09362 return NEW_LIT(ID2SYM(idNULL)); 09363 } 09364 09365 switch (nd_type(node)) { 09366 case NODE_DSTR: 09367 nd_set_type(node, NODE_DSYM); 09368 break; 09369 case NODE_STR: 09370 lit = node->nd_lit; 09371 node->nd_lit = ID2SYM(rb_intern_str(lit)); 09372 nd_set_type(node, NODE_LIT); 09373 break; 09374 default: 09375 node = NEW_NODE(NODE_DSYM, Qnil, 1, NEW_LIST(node)); 09376 break; 09377 } 09378 return node; 09379 } 09380 #endif /* !RIPPER */ 09381 09382 #ifndef RIPPER 09383 static NODE * 09384 new_op_assign_gen(struct parser_params *parser, NODE *lhs, ID op, NODE *rhs) 09385 { 09386 NODE *asgn; 09387 09388 if (lhs) { 09389 ID vid = lhs->nd_vid; 09390 if (op == tOROP) { 09391 lhs->nd_value = rhs; 09392 asgn = NEW_OP_ASGN_OR(gettable(vid), lhs); 09393 if (is_asgn_or_id(vid)) { 09394 asgn->nd_aid = vid; 09395 } 09396 } 09397 else if (op == tANDOP) { 09398 lhs->nd_value = rhs; 09399 asgn = NEW_OP_ASGN_AND(gettable(vid), lhs); 09400 } 09401 else { 09402 asgn = lhs; 09403 asgn->nd_value = NEW_CALL(gettable(vid), op, NEW_LIST(rhs)); 09404 } 09405 } 09406 else { 09407 asgn = NEW_BEGIN(0); 09408 } 09409 return asgn; 09410 } 09411 09412 static NODE * 09413 new_attr_op_assign_gen(struct parser_params *parser, NODE *lhs, ID attr, ID op, NODE *rhs) 09414 { 09415 NODE *asgn; 09416 09417 if (op == tOROP) { 09418 op = 0; 09419 } 09420 else if (op == tANDOP) { 09421 op = 1; 09422 } 09423 asgn = NEW_OP_ASGN2(lhs, attr, op, rhs); 09424 fixpos(asgn, lhs); 09425 return asgn; 09426 } 09427 09428 static NODE * 09429 new_const_op_assign_gen(struct parser_params *parser, NODE *lhs, ID op, NODE *rhs) 09430 { 09431 NODE *asgn; 09432 09433 if (op == tOROP) { 09434 op = 0; 09435 } 09436 else if (op == tANDOP) { 09437 op = 1; 09438 } 09439 if (lhs) { 09440 asgn = NEW_OP_CDECL(lhs, op, rhs); 09441 } 09442 else { 09443 asgn = NEW_BEGIN(0); 09444 } 09445 fixpos(asgn, lhs); 09446 return asgn; 09447 } 09448 #else 09449 static VALUE 09450 new_op_assign_gen(struct parser_params *parser, VALUE lhs, VALUE op, VALUE rhs) 09451 { 09452 return dispatch3(opassign, lhs, op, rhs); 09453 } 09454 09455 static VALUE 09456 new_attr_op_assign_gen(struct parser_params *parser, VALUE lhs, VALUE type, VALUE attr, VALUE op, VALUE rhs) 09457 { 09458 VALUE recv = dispatch3(field, lhs, type, attr); 09459 return dispatch3(opassign, recv, op, rhs); 09460 } 09461 #endif 09462 09463 static void 09464 warn_unused_var(struct parser_params *parser, struct local_vars *local) 09465 { 09466 int i, cnt; 09467 ID *v, *u; 09468 09469 if (!local->used) return; 09470 v = local->vars->tbl; 09471 u = local->used->tbl; 09472 cnt = local->used->pos; 09473 if (cnt != local->vars->pos) { 09474 rb_bug("local->used->pos != local->vars->pos"); 09475 } 09476 for (i = 0; i < cnt; ++i) { 09477 if (!v[i] || (u[i] & LVAR_USED)) continue; 09478 if (is_private_local_id(v[i])) continue; 09479 rb_warn4S(ruby_sourcefile, (int)u[i], "assigned but unused variable - %s", rb_id2name(v[i])); 09480 } 09481 } 09482 09483 static void 09484 local_push_gen(struct parser_params *parser, int inherit_dvars) 09485 { 09486 struct local_vars *local; 09487 09488 local = ALLOC(struct local_vars); 09489 local->prev = lvtbl; 09490 local->args = vtable_alloc(0); 09491 local->vars = vtable_alloc(inherit_dvars ? DVARS_INHERIT : DVARS_TOPSCOPE); 09492 local->used = !(inherit_dvars && 09493 (ifndef_ripper(compile_for_eval || e_option_supplied(parser))+0)) && 09494 RTEST(ruby_verbose) ? vtable_alloc(0) : 0; 09495 lvtbl = local; 09496 } 09497 09498 static void 09499 local_pop_gen(struct parser_params *parser) 09500 { 09501 struct local_vars *local = lvtbl->prev; 09502 if (lvtbl->used) { 09503 warn_unused_var(parser, lvtbl); 09504 vtable_free(lvtbl->used); 09505 } 09506 vtable_free(lvtbl->args); 09507 vtable_free(lvtbl->vars); 09508 xfree(lvtbl); 09509 lvtbl = local; 09510 } 09511 09512 #ifndef RIPPER 09513 static ID* 09514 vtable_tblcpy(ID *buf, const struct vtable *src) 09515 { 09516 int i, cnt = vtable_size(src); 09517 09518 if (cnt > 0) { 09519 buf[0] = cnt; 09520 for (i = 0; i < cnt; i++) { 09521 buf[i] = src->tbl[i]; 09522 } 09523 return buf; 09524 } 09525 return 0; 09526 } 09527 09528 static ID* 09529 local_tbl_gen(struct parser_params *parser) 09530 { 09531 int cnt = vtable_size(lvtbl->args) + vtable_size(lvtbl->vars); 09532 ID *buf; 09533 09534 if (cnt <= 0) return 0; 09535 buf = ALLOC_N(ID, cnt + 1); 09536 vtable_tblcpy(buf+1, lvtbl->args); 09537 vtable_tblcpy(buf+vtable_size(lvtbl->args)+1, lvtbl->vars); 09538 buf[0] = cnt; 09539 return buf; 09540 } 09541 #endif 09542 09543 static int 09544 arg_var_gen(struct parser_params *parser, ID id) 09545 { 09546 vtable_add(lvtbl->args, id); 09547 return vtable_size(lvtbl->args) - 1; 09548 } 09549 09550 static int 09551 local_var_gen(struct parser_params *parser, ID id) 09552 { 09553 vtable_add(lvtbl->vars, id); 09554 if (lvtbl->used) { 09555 vtable_add(lvtbl->used, (ID)ruby_sourceline); 09556 } 09557 return vtable_size(lvtbl->vars) - 1; 09558 } 09559 09560 static int 09561 local_id_gen(struct parser_params *parser, ID id) 09562 { 09563 struct vtable *vars, *args, *used; 09564 09565 vars = lvtbl->vars; 09566 args = lvtbl->args; 09567 used = lvtbl->used; 09568 09569 while (vars && POINTER_P(vars->prev)) { 09570 vars = vars->prev; 09571 args = args->prev; 09572 if (used) used = used->prev; 09573 } 09574 09575 if (vars && vars->prev == DVARS_INHERIT) { 09576 return rb_local_defined(id); 09577 } 09578 else if (vtable_included(args, id)) { 09579 return 1; 09580 } 09581 else { 09582 int i = vtable_included(vars, id); 09583 if (i && used) used->tbl[i-1] |= LVAR_USED; 09584 return i != 0; 09585 } 09586 } 09587 09588 static const struct vtable * 09589 dyna_push_gen(struct parser_params *parser) 09590 { 09591 lvtbl->args = vtable_alloc(lvtbl->args); 09592 lvtbl->vars = vtable_alloc(lvtbl->vars); 09593 if (lvtbl->used) { 09594 lvtbl->used = vtable_alloc(lvtbl->used); 09595 } 09596 return lvtbl->args; 09597 } 09598 09599 static void 09600 dyna_pop_1(struct parser_params *parser) 09601 { 09602 struct vtable *tmp; 09603 09604 if ((tmp = lvtbl->used) != 0) { 09605 warn_unused_var(parser, lvtbl); 09606 lvtbl->used = lvtbl->used->prev; 09607 vtable_free(tmp); 09608 } 09609 tmp = lvtbl->args; 09610 lvtbl->args = lvtbl->args->prev; 09611 vtable_free(tmp); 09612 tmp = lvtbl->vars; 09613 lvtbl->vars = lvtbl->vars->prev; 09614 vtable_free(tmp); 09615 } 09616 09617 static void 09618 dyna_pop_gen(struct parser_params *parser, const struct vtable *lvargs) 09619 { 09620 while (lvtbl->args != lvargs) { 09621 dyna_pop_1(parser); 09622 if (!lvtbl->args) { 09623 struct local_vars *local = lvtbl->prev; 09624 xfree(lvtbl); 09625 lvtbl = local; 09626 } 09627 } 09628 dyna_pop_1(parser); 09629 } 09630 09631 static int 09632 dyna_in_block_gen(struct parser_params *parser) 09633 { 09634 return POINTER_P(lvtbl->vars) && lvtbl->vars->prev != DVARS_TOPSCOPE; 09635 } 09636 09637 static int 09638 dvar_defined_gen(struct parser_params *parser, ID id, int get) 09639 { 09640 struct vtable *vars, *args, *used; 09641 int i; 09642 09643 args = lvtbl->args; 09644 vars = lvtbl->vars; 09645 used = lvtbl->used; 09646 09647 while (POINTER_P(vars)) { 09648 if (vtable_included(args, id)) { 09649 return 1; 09650 } 09651 if ((i = vtable_included(vars, id)) != 0) { 09652 if (used) used->tbl[i-1] |= LVAR_USED; 09653 return 1; 09654 } 09655 args = args->prev; 09656 vars = vars->prev; 09657 if (get) used = 0; 09658 if (used) used = used->prev; 09659 } 09660 09661 if (vars == DVARS_INHERIT) { 09662 return rb_dvar_defined(id); 09663 } 09664 09665 return 0; 09666 } 09667 09668 static int 09669 dvar_curr_gen(struct parser_params *parser, ID id) 09670 { 09671 return (vtable_included(lvtbl->args, id) || 09672 vtable_included(lvtbl->vars, id)); 09673 } 09674 09675 #ifndef RIPPER 09676 static void 09677 reg_fragment_setenc_gen(struct parser_params* parser, VALUE str, int options) 09678 { 09679 int c = RE_OPTION_ENCODING_IDX(options); 09680 09681 if (c) { 09682 int opt, idx; 09683 rb_char_to_option_kcode(c, &opt, &idx); 09684 if (idx != ENCODING_GET(str) && 09685 rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) { 09686 goto error; 09687 } 09688 ENCODING_SET(str, idx); 09689 } 09690 else if (RE_OPTION_ENCODING_NONE(options)) { 09691 if (!ENCODING_IS_ASCII8BIT(str) && 09692 rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) { 09693 c = 'n'; 09694 goto error; 09695 } 09696 rb_enc_associate(str, rb_ascii8bit_encoding()); 09697 } 09698 else if (current_enc == rb_usascii_encoding()) { 09699 if (rb_enc_str_coderange(str) != ENC_CODERANGE_7BIT) { 09700 /* raise in re.c */ 09701 rb_enc_associate(str, rb_usascii_encoding()); 09702 } 09703 else { 09704 rb_enc_associate(str, rb_ascii8bit_encoding()); 09705 } 09706 } 09707 return; 09708 09709 error: 09710 compile_error(PARSER_ARG 09711 "regexp encoding option '%c' differs from source encoding '%s'", 09712 c, rb_enc_name(rb_enc_get(str))); 09713 } 09714 09715 static int 09716 reg_fragment_check_gen(struct parser_params* parser, VALUE str, int options) 09717 { 09718 VALUE err; 09719 reg_fragment_setenc(str, options); 09720 err = rb_reg_check_preprocess(str); 09721 if (err != Qnil) { 09722 err = rb_obj_as_string(err); 09723 compile_error(PARSER_ARG "%s", RSTRING_PTR(err)); 09724 RB_GC_GUARD(err); 09725 return 0; 09726 } 09727 return 1; 09728 } 09729 09730 typedef struct { 09731 struct parser_params* parser; 09732 rb_encoding *enc; 09733 NODE *succ_block; 09734 NODE *fail_block; 09735 int num; 09736 } reg_named_capture_assign_t; 09737 09738 static int 09739 reg_named_capture_assign_iter(const OnigUChar *name, const OnigUChar *name_end, 09740 int back_num, int *back_refs, OnigRegex regex, void *arg0) 09741 { 09742 reg_named_capture_assign_t *arg = (reg_named_capture_assign_t*)arg0; 09743 struct parser_params* parser = arg->parser; 09744 rb_encoding *enc = arg->enc; 09745 long len = name_end - name; 09746 const char *s = (const char *)name; 09747 ID var; 09748 09749 arg->num++; 09750 09751 if (arg->succ_block == 0) { 09752 arg->succ_block = NEW_BEGIN(0); 09753 arg->fail_block = NEW_BEGIN(0); 09754 } 09755 09756 if (!len || (*name != '_' && ISASCII(*name) && !rb_enc_islower(*name, enc)) || 09757 (len < MAX_WORD_LENGTH && rb_reserved_word(s, (int)len)) || 09758 !rb_enc_symname2_p(s, len, enc)) { 09759 return ST_CONTINUE; 09760 } 09761 var = rb_intern3(s, len, enc); 09762 if (dvar_defined(var) || local_id(var)) { 09763 rb_warningS("named capture conflicts a local variable - %s", 09764 rb_id2name(var)); 09765 } 09766 arg->succ_block = block_append(arg->succ_block, 09767 newline_node(node_assign(assignable(var,0), 09768 NEW_CALL( 09769 gettable(rb_intern("$~")), 09770 idAREF, 09771 NEW_LIST(NEW_LIT(ID2SYM(var)))) 09772 ))); 09773 arg->fail_block = block_append(arg->fail_block, 09774 newline_node(node_assign(assignable(var,0), NEW_LIT(Qnil)))); 09775 return ST_CONTINUE; 09776 } 09777 09778 static NODE * 09779 reg_named_capture_assign_gen(struct parser_params* parser, VALUE regexp, NODE *match) 09780 { 09781 reg_named_capture_assign_t arg; 09782 09783 arg.parser = parser; 09784 arg.enc = rb_enc_get(regexp); 09785 arg.succ_block = 0; 09786 arg.fail_block = 0; 09787 arg.num = 0; 09788 onig_foreach_name(RREGEXP(regexp)->ptr, reg_named_capture_assign_iter, (void*)&arg); 09789 09790 if (arg.num == 0) 09791 return match; 09792 09793 return 09794 block_append( 09795 newline_node(match), 09796 NEW_IF(gettable(rb_intern("$~")), 09797 block_append( 09798 newline_node(arg.succ_block), 09799 newline_node( 09800 NEW_CALL( 09801 gettable(rb_intern("$~")), 09802 rb_intern("begin"), 09803 NEW_LIST(NEW_LIT(INT2FIX(0)))))), 09804 block_append( 09805 newline_node(arg.fail_block), 09806 newline_node( 09807 NEW_LIT(Qnil))))); 09808 } 09809 09810 static VALUE 09811 reg_compile_gen(struct parser_params* parser, VALUE str, int options) 09812 { 09813 VALUE re; 09814 VALUE err; 09815 09816 reg_fragment_setenc(str, options); 09817 err = rb_errinfo(); 09818 re = rb_reg_compile(str, options & RE_OPTION_MASK, ruby_sourcefile, ruby_sourceline); 09819 if (NIL_P(re)) { 09820 ID mesg = rb_intern("mesg"); 09821 VALUE m = rb_attr_get(rb_errinfo(), mesg); 09822 rb_set_errinfo(err); 09823 if (!NIL_P(err)) { 09824 rb_str_append(rb_str_cat(rb_attr_get(err, mesg), "\n", 1), m); 09825 } 09826 else { 09827 compile_error(PARSER_ARG "%s", RSTRING_PTR(m)); 09828 } 09829 return Qnil; 09830 } 09831 return re; 09832 } 09833 09834 void 09835 rb_gc_mark_parser(void) 09836 { 09837 } 09838 09839 NODE* 09840 rb_parser_append_print(VALUE vparser, NODE *node) 09841 { 09842 NODE *prelude = 0; 09843 NODE *scope = node; 09844 struct parser_params *parser; 09845 09846 if (!node) return node; 09847 09848 TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, parser); 09849 09850 node = node->nd_body; 09851 09852 if (nd_type(node) == NODE_PRELUDE) { 09853 prelude = node; 09854 node = node->nd_body; 09855 } 09856 09857 node = block_append(node, 09858 NEW_FCALL(rb_intern("print"), 09859 NEW_ARRAY(NEW_GVAR(rb_intern("$_"))))); 09860 if (prelude) { 09861 prelude->nd_body = node; 09862 scope->nd_body = prelude; 09863 } 09864 else { 09865 scope->nd_body = node; 09866 } 09867 09868 return scope; 09869 } 09870 09871 NODE * 09872 rb_parser_while_loop(VALUE vparser, NODE *node, int chop, int split) 09873 { 09874 NODE *prelude = 0; 09875 NODE *scope = node; 09876 struct parser_params *parser; 09877 09878 if (!node) return node; 09879 09880 TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, parser); 09881 09882 node = node->nd_body; 09883 09884 if (nd_type(node) == NODE_PRELUDE) { 09885 prelude = node; 09886 node = node->nd_body; 09887 } 09888 if (split) { 09889 node = block_append(NEW_GASGN(rb_intern("$F"), 09890 NEW_CALL(NEW_GVAR(rb_intern("$_")), 09891 rb_intern("split"), 0)), 09892 node); 09893 } 09894 if (chop) { 09895 node = block_append(NEW_CALL(NEW_GVAR(rb_intern("$_")), 09896 rb_intern("chop!"), 0), node); 09897 } 09898 09899 node = NEW_OPT_N(node); 09900 09901 if (prelude) { 09902 prelude->nd_body = node; 09903 scope->nd_body = prelude; 09904 } 09905 else { 09906 scope->nd_body = node; 09907 } 09908 09909 return scope; 09910 } 09911 09912 static const struct { 09913 ID token; 09914 const char *name; 09915 } op_tbl[] = { 09916 {tDOT2, ".."}, 09917 {tDOT3, "..."}, 09918 {tPOW, "**"}, 09919 {tDSTAR, "**"}, 09920 {tUPLUS, "+@"}, 09921 {tUMINUS, "-@"}, 09922 {tCMP, "<=>"}, 09923 {tGEQ, ">="}, 09924 {tLEQ, "<="}, 09925 {tEQ, "=="}, 09926 {tEQQ, "==="}, 09927 {tNEQ, "!="}, 09928 {tMATCH, "=~"}, 09929 {tNMATCH, "!~"}, 09930 {tAREF, "[]"}, 09931 {tASET, "[]="}, 09932 {tLSHFT, "<<"}, 09933 {tRSHFT, ">>"}, 09934 {tCOLON2, "::"}, 09935 }; 09936 09937 #define op_tbl_count numberof(op_tbl) 09938 09939 #ifndef ENABLE_SELECTOR_NAMESPACE 09940 #define ENABLE_SELECTOR_NAMESPACE 0 09941 #endif 09942 09943 static struct symbols { 09944 ID last_id; 09945 st_table *sym_id; 09946 st_table *id_str; 09947 #if ENABLE_SELECTOR_NAMESPACE 09948 st_table *ivar2_id; 09949 st_table *id_ivar2; 09950 #endif 09951 VALUE op_sym[tLAST_OP_ID]; 09952 } global_symbols = {tLAST_TOKEN}; 09953 09954 static const struct st_hash_type symhash = { 09955 rb_str_hash_cmp, 09956 rb_str_hash, 09957 }; 09958 09959 #if ENABLE_SELECTOR_NAMESPACE 09960 struct ivar2_key { 09961 ID id; 09962 VALUE klass; 09963 }; 09964 09965 static int 09966 ivar2_cmp(struct ivar2_key *key1, struct ivar2_key *key2) 09967 { 09968 if (key1->id == key2->id && key1->klass == key2->klass) { 09969 return 0; 09970 } 09971 return 1; 09972 } 09973 09974 static int 09975 ivar2_hash(struct ivar2_key *key) 09976 { 09977 return (key->id << 8) ^ (key->klass >> 2); 09978 } 09979 09980 static const struct st_hash_type ivar2_hash_type = { 09981 ivar2_cmp, 09982 ivar2_hash, 09983 }; 09984 #endif 09985 09986 void 09987 Init_sym(void) 09988 { 09989 global_symbols.sym_id = st_init_table_with_size(&symhash, 1000); 09990 global_symbols.id_str = st_init_numtable_with_size(1000); 09991 #if ENABLE_SELECTOR_NAMESPACE 09992 global_symbols.ivar2_id = st_init_table_with_size(&ivar2_hash_type, 1000); 09993 global_symbols.id_ivar2 = st_init_numtable_with_size(1000); 09994 #endif 09995 09996 (void)nodetype; 09997 (void)nodeline; 09998 #if PARSER_DEBUG 09999 (void)lex_state_name(-1); 10000 #endif 10001 10002 Init_id(); 10003 } 10004 10005 void 10006 rb_gc_mark_symbols(void) 10007 { 10008 rb_mark_tbl(global_symbols.id_str); 10009 rb_gc_mark_locations(global_symbols.op_sym, 10010 global_symbols.op_sym + numberof(global_symbols.op_sym)); 10011 } 10012 #endif /* !RIPPER */ 10013 10014 static ID 10015 internal_id_gen(struct parser_params *parser) 10016 { 10017 ID id = (ID)vtable_size(lvtbl->args) + (ID)vtable_size(lvtbl->vars); 10018 id += ((tLAST_TOKEN - ID_INTERNAL) >> ID_SCOPE_SHIFT) + 1; 10019 return ID_INTERNAL | (id << ID_SCOPE_SHIFT); 10020 } 10021 10022 #ifndef RIPPER 10023 static int 10024 is_special_global_name(const char *m, const char *e, rb_encoding *enc) 10025 { 10026 int mb = 0; 10027 10028 if (m >= e) return 0; 10029 if (is_global_name_punct(*m)) { 10030 ++m; 10031 } 10032 else if (*m == '-') { 10033 ++m; 10034 if (m < e && is_identchar(m, e, enc)) { 10035 if (!ISASCII(*m)) mb = 1; 10036 m += rb_enc_mbclen(m, e, enc); 10037 } 10038 } 10039 else { 10040 if (!rb_enc_isdigit(*m, enc)) return 0; 10041 do { 10042 if (!ISASCII(*m)) mb = 1; 10043 ++m; 10044 } while (m < e && rb_enc_isdigit(*m, enc)); 10045 } 10046 return m == e ? mb + 1 : 0; 10047 } 10048 10049 int 10050 rb_symname_p(const char *name) 10051 { 10052 return rb_enc_symname_p(name, rb_ascii8bit_encoding()); 10053 } 10054 10055 int 10056 rb_enc_symname_p(const char *name, rb_encoding *enc) 10057 { 10058 return rb_enc_symname2_p(name, strlen(name), enc); 10059 } 10060 10061 static int 10062 rb_enc_symname_type(const char *name, long len, rb_encoding *enc) 10063 { 10064 const char *m = name; 10065 const char *e = m + len; 10066 int type = ID_JUNK; 10067 10068 if (!m || len <= 0) return -1; 10069 switch (*m) { 10070 case '\0': 10071 return -1; 10072 10073 case '$': 10074 type = ID_GLOBAL; 10075 if (is_special_global_name(++m, e, enc)) return type; 10076 goto id; 10077 10078 case '@': 10079 type = ID_INSTANCE; 10080 if (*++m == '@') { 10081 ++m; 10082 type = ID_CLASS; 10083 } 10084 goto id; 10085 10086 case '<': 10087 switch (*++m) { 10088 case '<': ++m; break; 10089 case '=': if (*++m == '>') ++m; break; 10090 default: break; 10091 } 10092 break; 10093 10094 case '>': 10095 switch (*++m) { 10096 case '>': case '=': ++m; break; 10097 } 10098 break; 10099 10100 case '=': 10101 switch (*++m) { 10102 case '~': ++m; break; 10103 case '=': if (*++m == '=') ++m; break; 10104 default: return -1; 10105 } 10106 break; 10107 10108 case '*': 10109 if (*++m == '*') ++m; 10110 break; 10111 10112 case '+': case '-': 10113 if (*++m == '@') ++m; 10114 break; 10115 10116 case '|': case '^': case '&': case '/': case '%': case '~': case '`': 10117 ++m; 10118 break; 10119 10120 case '[': 10121 if (*++m != ']') return -1; 10122 if (*++m == '=') ++m; 10123 break; 10124 10125 case '!': 10126 if (len == 1) return ID_JUNK; 10127 switch (*++m) { 10128 case '=': case '~': ++m; break; 10129 default: return -1; 10130 } 10131 break; 10132 10133 default: 10134 type = rb_enc_isupper(*m, enc) ? ID_CONST : ID_LOCAL; 10135 id: 10136 if (m >= e || (*m != '_' && !rb_enc_isalpha(*m, enc) && ISASCII(*m))) 10137 return -1; 10138 while (m < e && is_identchar(m, e, enc)) m += rb_enc_mbclen(m, e, enc); 10139 switch (*m) { 10140 case '!': case '?': 10141 if (type == ID_GLOBAL || type == ID_CLASS || type == ID_INSTANCE) return -1; 10142 type = ID_JUNK; 10143 ++m; 10144 break; 10145 case '=': 10146 if (type != ID_CONST && type != ID_LOCAL) return -1; 10147 type = ID_ATTRSET; 10148 ++m; 10149 break; 10150 } 10151 break; 10152 } 10153 return m == e ? type : -1; 10154 } 10155 10156 int 10157 rb_enc_symname2_p(const char *name, long len, rb_encoding *enc) 10158 { 10159 return rb_enc_symname_type(name, len, enc) != -1; 10160 } 10161 10162 static int 10163 rb_str_symname_type(VALUE name) 10164 { 10165 const char *ptr = StringValuePtr(name); 10166 long len = RSTRING_LEN(name); 10167 int type = rb_enc_symname_type(ptr, len, rb_enc_get(name)); 10168 RB_GC_GUARD(name); 10169 return type; 10170 } 10171 10172 static ID 10173 register_symid(ID id, const char *name, long len, rb_encoding *enc) 10174 { 10175 VALUE str = rb_enc_str_new(name, len, enc); 10176 return register_symid_str(id, str); 10177 } 10178 10179 static ID 10180 register_symid_str(ID id, VALUE str) 10181 { 10182 OBJ_FREEZE(str); 10183 st_add_direct(global_symbols.sym_id, (st_data_t)str, id); 10184 st_add_direct(global_symbols.id_str, id, (st_data_t)str); 10185 return id; 10186 } 10187 10188 static int 10189 sym_check_asciionly(VALUE str) 10190 { 10191 if (!rb_enc_asciicompat(rb_enc_get(str))) return FALSE; 10192 switch (rb_enc_str_coderange(str)) { 10193 case ENC_CODERANGE_BROKEN: 10194 rb_raise(rb_eEncodingError, "invalid encoding symbol"); 10195 case ENC_CODERANGE_7BIT: 10196 return TRUE; 10197 } 10198 return FALSE; 10199 } 10200 10201 /* 10202 * _str_ itself will be registered at the global symbol table. _str_ 10203 * can be modified before the registration, since the encoding will be 10204 * set to ASCII-8BIT if it is a special global name. 10205 */ 10206 static ID intern_str(VALUE str); 10207 10208 ID 10209 rb_intern3(const char *name, long len, rb_encoding *enc) 10210 { 10211 VALUE str; 10212 st_data_t data; 10213 struct RString fake_str; 10214 fake_str.basic.flags = T_STRING|RSTRING_NOEMBED; 10215 fake_str.basic.klass = rb_cString; 10216 fake_str.as.heap.len = len; 10217 fake_str.as.heap.ptr = (char *)name; 10218 fake_str.as.heap.aux.capa = len; 10219 str = (VALUE)&fake_str; 10220 rb_enc_associate(str, enc); 10221 OBJ_FREEZE(str); 10222 10223 if (st_lookup(global_symbols.sym_id, str, &data)) 10224 return (ID)data; 10225 10226 str = rb_enc_str_new(name, len, enc); /* make true string */ 10227 return intern_str(str); 10228 } 10229 10230 static ID 10231 intern_str(VALUE str) 10232 { 10233 const char *name, *m, *e; 10234 long len, last; 10235 rb_encoding *enc, *symenc; 10236 unsigned char c; 10237 ID id; 10238 int mb; 10239 10240 RSTRING_GETMEM(str, name, len); 10241 m = name; 10242 e = m + len; 10243 enc = rb_enc_get(str); 10244 symenc = enc; 10245 10246 if (rb_cString && !rb_enc_asciicompat(enc)) { 10247 id = ID_JUNK; 10248 goto new_id; 10249 } 10250 last = len-1; 10251 id = 0; 10252 switch (*m) { 10253 case '$': 10254 id |= ID_GLOBAL; 10255 if ((mb = is_special_global_name(++m, e, enc)) != 0) { 10256 if (!--mb) symenc = rb_usascii_encoding(); 10257 goto new_id; 10258 } 10259 break; 10260 case '@': 10261 if (m[1] == '@') { 10262 m++; 10263 id |= ID_CLASS; 10264 } 10265 else { 10266 id |= ID_INSTANCE; 10267 } 10268 m++; 10269 break; 10270 default: 10271 c = m[0]; 10272 if (c != '_' && rb_enc_isascii(c, enc) && rb_enc_ispunct(c, enc)) { 10273 /* operators */ 10274 int i; 10275 10276 if (len == 1) { 10277 id = c; 10278 goto id_register; 10279 } 10280 for (i = 0; i < op_tbl_count; i++) { 10281 if (*op_tbl[i].name == *m && 10282 strcmp(op_tbl[i].name, m) == 0) { 10283 id = op_tbl[i].token; 10284 goto id_register; 10285 } 10286 } 10287 } 10288 10289 if (m[last] == '=') { 10290 /* attribute assignment */ 10291 id = rb_intern3(name, last, enc); 10292 if (id > tLAST_OP_ID && !is_attrset_id(id)) { 10293 enc = rb_enc_get(rb_id2str(id)); 10294 id = rb_id_attrset(id); 10295 goto id_register; 10296 } 10297 id = ID_ATTRSET; 10298 } 10299 else if (rb_enc_isupper(m[0], enc)) { 10300 id = ID_CONST; 10301 } 10302 else { 10303 id = ID_LOCAL; 10304 } 10305 break; 10306 } 10307 if (!rb_enc_isdigit(*m, enc)) { 10308 while (m <= name + last && is_identchar(m, e, enc)) { 10309 if (ISASCII(*m)) { 10310 m++; 10311 } 10312 else { 10313 m += rb_enc_mbclen(m, e, enc); 10314 } 10315 } 10316 } 10317 if (m - name < len) id = ID_JUNK; 10318 if (sym_check_asciionly(str)) symenc = rb_usascii_encoding(); 10319 new_id: 10320 if (symenc != enc) rb_enc_associate(str, symenc); 10321 if (global_symbols.last_id >= ~(ID)0 >> (ID_SCOPE_SHIFT+RUBY_SPECIAL_SHIFT)) { 10322 if (len > 20) { 10323 rb_raise(rb_eRuntimeError, "symbol table overflow (symbol %.20s...)", 10324 name); 10325 } 10326 else { 10327 rb_raise(rb_eRuntimeError, "symbol table overflow (symbol %.*s)", 10328 (int)len, name); 10329 } 10330 } 10331 id |= ++global_symbols.last_id << ID_SCOPE_SHIFT; 10332 id_register: 10333 return register_symid_str(id, str); 10334 } 10335 10336 ID 10337 rb_intern2(const char *name, long len) 10338 { 10339 return rb_intern3(name, len, rb_usascii_encoding()); 10340 } 10341 10342 #undef rb_intern 10343 ID 10344 rb_intern(const char *name) 10345 { 10346 return rb_intern2(name, strlen(name)); 10347 } 10348 10349 ID 10350 rb_intern_str(VALUE str) 10351 { 10352 st_data_t id; 10353 10354 if (st_lookup(global_symbols.sym_id, str, &id)) 10355 return (ID)id; 10356 return intern_str(rb_str_dup(str)); 10357 } 10358 10359 VALUE 10360 rb_id2str(ID id) 10361 { 10362 st_data_t data; 10363 10364 if (id < tLAST_TOKEN) { 10365 int i = 0; 10366 10367 if (id < INT_MAX && rb_ispunct((int)id)) { 10368 VALUE str = global_symbols.op_sym[i = (int)id]; 10369 if (!str) { 10370 char name[2]; 10371 name[0] = (char)id; 10372 name[1] = 0; 10373 str = rb_usascii_str_new(name, 1); 10374 OBJ_FREEZE(str); 10375 global_symbols.op_sym[i] = str; 10376 } 10377 return str; 10378 } 10379 for (i = 0; i < op_tbl_count; i++) { 10380 if (op_tbl[i].token == id) { 10381 VALUE str = global_symbols.op_sym[i]; 10382 if (!str) { 10383 str = rb_usascii_str_new2(op_tbl[i].name); 10384 OBJ_FREEZE(str); 10385 global_symbols.op_sym[i] = str; 10386 } 10387 return str; 10388 } 10389 } 10390 } 10391 10392 if (st_lookup(global_symbols.id_str, id, &data)) { 10393 VALUE str = (VALUE)data; 10394 if (RBASIC(str)->klass == 0) 10395 RBASIC(str)->klass = rb_cString; 10396 return str; 10397 } 10398 10399 if (is_attrset_id(id)) { 10400 ID id2 = (id & ~ID_SCOPE_MASK) | ID_LOCAL; 10401 VALUE str; 10402 10403 while (!(str = rb_id2str(id2))) { 10404 if (!is_local_id(id2)) return 0; 10405 id2 = (id & ~ID_SCOPE_MASK) | ID_CONST; 10406 } 10407 str = rb_str_dup(str); 10408 rb_str_cat(str, "=", 1); 10409 rb_intern_str(str); 10410 if (st_lookup(global_symbols.id_str, id, &data)) { 10411 VALUE str = (VALUE)data; 10412 if (RBASIC(str)->klass == 0) 10413 RBASIC(str)->klass = rb_cString; 10414 return str; 10415 } 10416 } 10417 return 0; 10418 } 10419 10420 const char * 10421 rb_id2name(ID id) 10422 { 10423 VALUE str = rb_id2str(id); 10424 10425 if (!str) return 0; 10426 return RSTRING_PTR(str); 10427 } 10428 10429 static int 10430 symbols_i(VALUE sym, ID value, VALUE ary) 10431 { 10432 rb_ary_push(ary, ID2SYM(value)); 10433 return ST_CONTINUE; 10434 } 10435 10436 /* 10437 * call-seq: 10438 * Symbol.all_symbols => array 10439 * 10440 * Returns an array of all the symbols currently in Ruby's symbol 10441 * table. 10442 * 10443 * Symbol.all_symbols.size #=> 903 10444 * Symbol.all_symbols[1,20] #=> [:floor, :ARGV, :Binding, :symlink, 10445 * :chown, :EOFError, :$;, :String, 10446 * :LOCK_SH, :"setuid?", :$<, 10447 * :default_proc, :compact, :extend, 10448 * :Tms, :getwd, :$=, :ThreadGroup, 10449 * :wait2, :$>] 10450 */ 10451 10452 VALUE 10453 rb_sym_all_symbols(void) 10454 { 10455 VALUE ary = rb_ary_new2(global_symbols.sym_id->num_entries); 10456 10457 st_foreach(global_symbols.sym_id, symbols_i, ary); 10458 return ary; 10459 } 10460 10461 int 10462 rb_is_const_id(ID id) 10463 { 10464 return is_const_id(id); 10465 } 10466 10467 int 10468 rb_is_class_id(ID id) 10469 { 10470 return is_class_id(id); 10471 } 10472 10473 int 10474 rb_is_global_id(ID id) 10475 { 10476 return is_global_id(id); 10477 } 10478 10479 int 10480 rb_is_instance_id(ID id) 10481 { 10482 return is_instance_id(id); 10483 } 10484 10485 int 10486 rb_is_attrset_id(ID id) 10487 { 10488 return is_attrset_id(id); 10489 } 10490 10491 int 10492 rb_is_local_id(ID id) 10493 { 10494 return is_local_id(id); 10495 } 10496 10497 int 10498 rb_is_junk_id(ID id) 10499 { 10500 return is_junk_id(id); 10501 } 10502 10514 ID 10515 rb_check_id(volatile VALUE *namep) 10516 { 10517 st_data_t id; 10518 VALUE tmp; 10519 VALUE name = *namep; 10520 10521 if (SYMBOL_P(name)) { 10522 return SYM2ID(name); 10523 } 10524 else if (!RB_TYPE_P(name, T_STRING)) { 10525 tmp = rb_check_string_type(name); 10526 if (NIL_P(tmp)) { 10527 tmp = rb_inspect(name); 10528 rb_raise(rb_eTypeError, "%s is not a symbol", 10529 RSTRING_PTR(tmp)); 10530 } 10531 name = tmp; 10532 *namep = name; 10533 } 10534 10535 sym_check_asciionly(name); 10536 10537 if (st_lookup(global_symbols.sym_id, (st_data_t)name, &id)) 10538 return (ID)id; 10539 10540 if (rb_is_attrset_name(name)) { 10541 struct RString fake_str; 10542 const VALUE localname = (VALUE)&fake_str; 10543 /* make local name by chopping '=' */ 10544 fake_str.basic.flags = T_STRING|RSTRING_NOEMBED; 10545 fake_str.basic.klass = rb_cString; 10546 fake_str.as.heap.len = RSTRING_LEN(name) - 1; 10547 fake_str.as.heap.ptr = RSTRING_PTR(name); 10548 fake_str.as.heap.aux.capa = fake_str.as.heap.len; 10549 rb_enc_copy(localname, name); 10550 OBJ_FREEZE(localname); 10551 10552 if (st_lookup(global_symbols.sym_id, (st_data_t)localname, &id)) { 10553 return rb_id_attrset((ID)id); 10554 } 10555 RB_GC_GUARD(name); 10556 } 10557 10558 return (ID)0; 10559 } 10560 10561 ID 10562 rb_check_id_cstr(const char *ptr, long len, rb_encoding *enc) 10563 { 10564 st_data_t id; 10565 struct RString fake_str; 10566 const VALUE name = (VALUE)&fake_str; 10567 fake_str.basic.flags = T_STRING|RSTRING_NOEMBED; 10568 fake_str.basic.klass = rb_cString; 10569 fake_str.as.heap.len = len; 10570 fake_str.as.heap.ptr = (char *)ptr; 10571 fake_str.as.heap.aux.capa = len; 10572 rb_enc_associate(name, enc); 10573 10574 sym_check_asciionly(name); 10575 10576 if (st_lookup(global_symbols.sym_id, (st_data_t)name, &id)) 10577 return (ID)id; 10578 10579 if (rb_is_attrset_name(name)) { 10580 fake_str.as.heap.len = len - 1; 10581 if (st_lookup(global_symbols.sym_id, (st_data_t)name, &id)) { 10582 return rb_id_attrset((ID)id); 10583 } 10584 } 10585 10586 return (ID)0; 10587 } 10588 10589 int 10590 rb_is_const_name(VALUE name) 10591 { 10592 return rb_str_symname_type(name) == ID_CONST; 10593 } 10594 10595 int 10596 rb_is_class_name(VALUE name) 10597 { 10598 return rb_str_symname_type(name) == ID_CLASS; 10599 } 10600 10601 int 10602 rb_is_global_name(VALUE name) 10603 { 10604 return rb_str_symname_type(name) == ID_GLOBAL; 10605 } 10606 10607 int 10608 rb_is_instance_name(VALUE name) 10609 { 10610 return rb_str_symname_type(name) == ID_INSTANCE; 10611 } 10612 10613 int 10614 rb_is_attrset_name(VALUE name) 10615 { 10616 return rb_str_symname_type(name) == ID_ATTRSET; 10617 } 10618 10619 int 10620 rb_is_local_name(VALUE name) 10621 { 10622 return rb_str_symname_type(name) == ID_LOCAL; 10623 } 10624 10625 int 10626 rb_is_method_name(VALUE name) 10627 { 10628 switch (rb_str_symname_type(name)) { 10629 case ID_LOCAL: case ID_ATTRSET: case ID_JUNK: 10630 return TRUE; 10631 } 10632 return FALSE; 10633 } 10634 10635 int 10636 rb_is_junk_name(VALUE name) 10637 { 10638 return rb_str_symname_type(name) == -1; 10639 } 10640 10641 #endif /* !RIPPER */ 10642 10643 static void 10644 parser_initialize(struct parser_params *parser) 10645 { 10646 parser->eofp = Qfalse; 10647 10648 parser->parser_lex_strterm = 0; 10649 parser->parser_cond_stack = 0; 10650 parser->parser_cmdarg_stack = 0; 10651 parser->parser_class_nest = 0; 10652 parser->parser_paren_nest = 0; 10653 parser->parser_lpar_beg = 0; 10654 parser->parser_brace_nest = 0; 10655 parser->parser_in_single = 0; 10656 parser->parser_in_def = 0; 10657 parser->parser_in_defined = 0; 10658 parser->parser_compile_for_eval = 0; 10659 parser->parser_cur_mid = 0; 10660 parser->parser_tokenbuf = NULL; 10661 parser->parser_tokidx = 0; 10662 parser->parser_toksiz = 0; 10663 parser->parser_heredoc_end = 0; 10664 parser->parser_command_start = TRUE; 10665 parser->parser_deferred_nodes = 0; 10666 parser->parser_lex_pbeg = 0; 10667 parser->parser_lex_p = 0; 10668 parser->parser_lex_pend = 0; 10669 parser->parser_lvtbl = 0; 10670 parser->parser_ruby__end__seen = 0; 10671 parser->parser_ruby_sourcefile = 0; 10672 #ifndef RIPPER 10673 parser->is_ripper = 0; 10674 parser->parser_eval_tree_begin = 0; 10675 parser->parser_eval_tree = 0; 10676 #else 10677 parser->is_ripper = 1; 10678 parser->parser_ruby_sourcefile_string = Qnil; 10679 parser->delayed = Qnil; 10680 10681 parser->result = Qnil; 10682 parser->parsing_thread = Qnil; 10683 parser->toplevel_p = TRUE; 10684 #endif 10685 #ifdef YYMALLOC 10686 parser->heap = NULL; 10687 #endif 10688 parser->enc = rb_utf8_encoding(); 10689 } 10690 10691 #ifdef RIPPER 10692 #define parser_mark ripper_parser_mark 10693 #define parser_free ripper_parser_free 10694 #endif 10695 10696 static void 10697 parser_mark(void *ptr) 10698 { 10699 struct parser_params *p = (struct parser_params*)ptr; 10700 10701 rb_gc_mark((VALUE)p->parser_lex_strterm); 10702 rb_gc_mark((VALUE)p->parser_deferred_nodes); 10703 rb_gc_mark(p->parser_lex_input); 10704 rb_gc_mark(p->parser_lex_lastline); 10705 rb_gc_mark(p->parser_lex_nextline); 10706 #ifndef RIPPER 10707 rb_gc_mark((VALUE)p->parser_eval_tree_begin) ; 10708 rb_gc_mark((VALUE)p->parser_eval_tree) ; 10709 rb_gc_mark(p->debug_lines); 10710 #else 10711 rb_gc_mark(p->parser_ruby_sourcefile_string); 10712 rb_gc_mark(p->delayed); 10713 rb_gc_mark(p->value); 10714 rb_gc_mark(p->result); 10715 rb_gc_mark(p->parsing_thread); 10716 #endif 10717 #ifdef YYMALLOC 10718 rb_gc_mark((VALUE)p->heap); 10719 #endif 10720 } 10721 10722 static void 10723 parser_free(void *ptr) 10724 { 10725 struct parser_params *p = (struct parser_params*)ptr; 10726 struct local_vars *local, *prev; 10727 10728 if (p->parser_tokenbuf) { 10729 xfree(p->parser_tokenbuf); 10730 } 10731 for (local = p->parser_lvtbl; local; local = prev) { 10732 if (local->vars) xfree(local->vars); 10733 prev = local->prev; 10734 xfree(local); 10735 } 10736 #ifndef RIPPER 10737 xfree(p->parser_ruby_sourcefile); 10738 #endif 10739 xfree(p); 10740 } 10741 10742 static size_t 10743 parser_memsize(const void *ptr) 10744 { 10745 struct parser_params *p = (struct parser_params*)ptr; 10746 struct local_vars *local; 10747 size_t size = sizeof(*p); 10748 10749 if (!ptr) return 0; 10750 size += p->parser_toksiz; 10751 for (local = p->parser_lvtbl; local; local = local->prev) { 10752 size += sizeof(*local); 10753 if (local->vars) size += local->vars->capa * sizeof(ID); 10754 } 10755 #ifndef RIPPER 10756 if (p->parser_ruby_sourcefile) { 10757 size += strlen(p->parser_ruby_sourcefile) + 1; 10758 } 10759 #endif 10760 return size; 10761 } 10762 10763 static 10764 #ifndef RIPPER 10765 const 10766 #endif 10767 rb_data_type_t parser_data_type = { 10768 "parser", 10769 { 10770 parser_mark, 10771 parser_free, 10772 parser_memsize, 10773 }, 10774 }; 10775 10776 #ifndef RIPPER 10777 #undef rb_reserved_word 10778 10779 const struct kwtable * 10780 rb_reserved_word(const char *str, unsigned int len) 10781 { 10782 return reserved_word(str, len); 10783 } 10784 10785 static struct parser_params * 10786 parser_new(void) 10787 { 10788 struct parser_params *p; 10789 10790 p = ALLOC_N(struct parser_params, 1); 10791 MEMZERO(p, struct parser_params, 1); 10792 parser_initialize(p); 10793 return p; 10794 } 10795 10796 VALUE 10797 rb_parser_new(void) 10798 { 10799 struct parser_params *p = parser_new(); 10800 10801 return TypedData_Wrap_Struct(0, &parser_data_type, p); 10802 } 10803 10804 /* 10805 * call-seq: 10806 * ripper#end_seen? -> Boolean 10807 * 10808 * Return true if parsed source ended by +\_\_END\_\_+. 10809 */ 10810 VALUE 10811 rb_parser_end_seen_p(VALUE vparser) 10812 { 10813 struct parser_params *parser; 10814 10815 TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, parser); 10816 return ruby__end__seen ? Qtrue : Qfalse; 10817 } 10818 10819 /* 10820 * call-seq: 10821 * ripper#encoding -> encoding 10822 * 10823 * Return encoding of the source. 10824 */ 10825 VALUE 10826 rb_parser_encoding(VALUE vparser) 10827 { 10828 struct parser_params *parser; 10829 10830 TypedData_Get_Struct(vparser, struct parser_params, &parser_data_type, parser); 10831 return rb_enc_from_encoding(current_enc); 10832 } 10833 10834 /* 10835 * call-seq: 10836 * ripper.yydebug -> true or false 10837 * 10838 * Get yydebug. 10839 */ 10840 VALUE 10841 rb_parser_get_yydebug(VALUE self) 10842 { 10843 struct parser_params *parser; 10844 10845 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 10846 return yydebug ? Qtrue : Qfalse; 10847 } 10848 10849 /* 10850 * call-seq: 10851 * ripper.yydebug = flag 10852 * 10853 * Set yydebug. 10854 */ 10855 VALUE 10856 rb_parser_set_yydebug(VALUE self, VALUE flag) 10857 { 10858 struct parser_params *parser; 10859 10860 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 10861 yydebug = RTEST(flag); 10862 return flag; 10863 } 10864 10865 #ifdef YYMALLOC 10866 #define HEAPCNT(n, size) ((n) * (size) / sizeof(YYSTYPE)) 10867 #define NEWHEAP() rb_node_newnode(NODE_ALLOCA, 0, (VALUE)parser->heap, 0) 10868 #define ADD2HEAP(n, c, p) ((parser->heap = (n))->u1.node = (p), \ 10869 (n)->u3.cnt = (c), (p)) 10870 10871 void * 10872 rb_parser_malloc(struct parser_params *parser, size_t size) 10873 { 10874 size_t cnt = HEAPCNT(1, size); 10875 NODE *n = NEWHEAP(); 10876 void *ptr = xmalloc(size); 10877 10878 return ADD2HEAP(n, cnt, ptr); 10879 } 10880 10881 void * 10882 rb_parser_calloc(struct parser_params *parser, size_t nelem, size_t size) 10883 { 10884 size_t cnt = HEAPCNT(nelem, size); 10885 NODE *n = NEWHEAP(); 10886 void *ptr = xcalloc(nelem, size); 10887 10888 return ADD2HEAP(n, cnt, ptr); 10889 } 10890 10891 void * 10892 rb_parser_realloc(struct parser_params *parser, void *ptr, size_t size) 10893 { 10894 NODE *n; 10895 size_t cnt = HEAPCNT(1, size); 10896 10897 if (ptr && (n = parser->heap) != NULL) { 10898 do { 10899 if (n->u1.node == ptr) { 10900 n->u1.node = ptr = xrealloc(ptr, size); 10901 if (n->u3.cnt) n->u3.cnt = cnt; 10902 return ptr; 10903 } 10904 } while ((n = n->u2.node) != NULL); 10905 } 10906 n = NEWHEAP(); 10907 ptr = xrealloc(ptr, size); 10908 return ADD2HEAP(n, cnt, ptr); 10909 } 10910 10911 void 10912 rb_parser_free(struct parser_params *parser, void *ptr) 10913 { 10914 NODE **prev = &parser->heap, *n; 10915 10916 while ((n = *prev) != NULL) { 10917 if (n->u1.node == ptr) { 10918 *prev = n->u2.node; 10919 rb_gc_force_recycle((VALUE)n); 10920 break; 10921 } 10922 prev = &n->u2.node; 10923 } 10924 xfree(ptr); 10925 } 10926 #endif 10927 #endif 10928 10929 #ifdef RIPPER 10930 #ifdef RIPPER_DEBUG 10931 extern int rb_is_pointer_to_heap(VALUE); 10932 10933 /* :nodoc: */ 10934 static VALUE 10935 ripper_validate_object(VALUE self, VALUE x) 10936 { 10937 if (x == Qfalse) return x; 10938 if (x == Qtrue) return x; 10939 if (x == Qnil) return x; 10940 if (x == Qundef) 10941 rb_raise(rb_eArgError, "Qundef given"); 10942 if (FIXNUM_P(x)) return x; 10943 if (SYMBOL_P(x)) return x; 10944 if (!rb_is_pointer_to_heap(x)) 10945 rb_raise(rb_eArgError, "invalid pointer: %p", x); 10946 switch (TYPE(x)) { 10947 case T_STRING: 10948 case T_OBJECT: 10949 case T_ARRAY: 10950 case T_BIGNUM: 10951 case T_FLOAT: 10952 return x; 10953 case T_NODE: 10954 if (nd_type(x) != NODE_LASGN) { 10955 rb_raise(rb_eArgError, "NODE given: %p", x); 10956 } 10957 return ((NODE *)x)->nd_rval; 10958 default: 10959 rb_raise(rb_eArgError, "wrong type of ruby object: %p (%s)", 10960 x, rb_obj_classname(x)); 10961 } 10962 return x; 10963 } 10964 #endif 10965 10966 #define validate(x) ((x) = get_value(x)) 10967 10968 static VALUE 10969 ripper_dispatch0(struct parser_params *parser, ID mid) 10970 { 10971 return rb_funcall(parser->value, mid, 0); 10972 } 10973 10974 static VALUE 10975 ripper_dispatch1(struct parser_params *parser, ID mid, VALUE a) 10976 { 10977 validate(a); 10978 return rb_funcall(parser->value, mid, 1, a); 10979 } 10980 10981 static VALUE 10982 ripper_dispatch2(struct parser_params *parser, ID mid, VALUE a, VALUE b) 10983 { 10984 validate(a); 10985 validate(b); 10986 return rb_funcall(parser->value, mid, 2, a, b); 10987 } 10988 10989 static VALUE 10990 ripper_dispatch3(struct parser_params *parser, ID mid, VALUE a, VALUE b, VALUE c) 10991 { 10992 validate(a); 10993 validate(b); 10994 validate(c); 10995 return rb_funcall(parser->value, mid, 3, a, b, c); 10996 } 10997 10998 static VALUE 10999 ripper_dispatch4(struct parser_params *parser, ID mid, VALUE a, VALUE b, VALUE c, VALUE d) 11000 { 11001 validate(a); 11002 validate(b); 11003 validate(c); 11004 validate(d); 11005 return rb_funcall(parser->value, mid, 4, a, b, c, d); 11006 } 11007 11008 static VALUE 11009 ripper_dispatch5(struct parser_params *parser, ID mid, VALUE a, VALUE b, VALUE c, VALUE d, VALUE e) 11010 { 11011 validate(a); 11012 validate(b); 11013 validate(c); 11014 validate(d); 11015 validate(e); 11016 return rb_funcall(parser->value, mid, 5, a, b, c, d, e); 11017 } 11018 11019 static VALUE 11020 ripper_dispatch7(struct parser_params *parser, ID mid, VALUE a, VALUE b, VALUE c, VALUE d, VALUE e, VALUE f, VALUE g) 11021 { 11022 validate(a); 11023 validate(b); 11024 validate(c); 11025 validate(d); 11026 validate(e); 11027 validate(f); 11028 validate(g); 11029 return rb_funcall(parser->value, mid, 7, a, b, c, d, e, f, g); 11030 } 11031 11032 static const struct kw_assoc { 11033 ID id; 11034 const char *name; 11035 } keyword_to_name[] = { 11036 {keyword_class, "class"}, 11037 {keyword_module, "module"}, 11038 {keyword_def, "def"}, 11039 {keyword_undef, "undef"}, 11040 {keyword_begin, "begin"}, 11041 {keyword_rescue, "rescue"}, 11042 {keyword_ensure, "ensure"}, 11043 {keyword_end, "end"}, 11044 {keyword_if, "if"}, 11045 {keyword_unless, "unless"}, 11046 {keyword_then, "then"}, 11047 {keyword_elsif, "elsif"}, 11048 {keyword_else, "else"}, 11049 {keyword_case, "case"}, 11050 {keyword_when, "when"}, 11051 {keyword_while, "while"}, 11052 {keyword_until, "until"}, 11053 {keyword_for, "for"}, 11054 {keyword_break, "break"}, 11055 {keyword_next, "next"}, 11056 {keyword_redo, "redo"}, 11057 {keyword_retry, "retry"}, 11058 {keyword_in, "in"}, 11059 {keyword_do, "do"}, 11060 {keyword_do_cond, "do"}, 11061 {keyword_do_block, "do"}, 11062 {keyword_return, "return"}, 11063 {keyword_yield, "yield"}, 11064 {keyword_super, "super"}, 11065 {keyword_self, "self"}, 11066 {keyword_nil, "nil"}, 11067 {keyword_true, "true"}, 11068 {keyword_false, "false"}, 11069 {keyword_and, "and"}, 11070 {keyword_or, "or"}, 11071 {keyword_not, "not"}, 11072 {modifier_if, "if"}, 11073 {modifier_unless, "unless"}, 11074 {modifier_while, "while"}, 11075 {modifier_until, "until"}, 11076 {modifier_rescue, "rescue"}, 11077 {keyword_alias, "alias"}, 11078 {keyword_defined, "defined?"}, 11079 {keyword_BEGIN, "BEGIN"}, 11080 {keyword_END, "END"}, 11081 {keyword__LINE__, "__LINE__"}, 11082 {keyword__FILE__, "__FILE__"}, 11083 {keyword__ENCODING__, "__ENCODING__"}, 11084 {0, NULL} 11085 }; 11086 11087 static const char* 11088 keyword_id_to_str(ID id) 11089 { 11090 const struct kw_assoc *a; 11091 11092 for (a = keyword_to_name; a->id; a++) { 11093 if (a->id == id) 11094 return a->name; 11095 } 11096 return NULL; 11097 } 11098 11099 #undef ripper_id2sym 11100 static VALUE 11101 ripper_id2sym(ID id) 11102 { 11103 const char *name; 11104 char buf[8]; 11105 11106 if (id <= 256) { 11107 buf[0] = (char)id; 11108 buf[1] = '\0'; 11109 return ID2SYM(rb_intern2(buf, 1)); 11110 } 11111 if ((name = keyword_id_to_str(id))) { 11112 return ID2SYM(rb_intern(name)); 11113 } 11114 switch (id) { 11115 case tOROP: 11116 name = "||"; 11117 break; 11118 case tANDOP: 11119 name = "&&"; 11120 break; 11121 default: 11122 name = rb_id2name(id); 11123 if (!name) { 11124 rb_bug("cannot convert ID to string: %ld", (unsigned long)id); 11125 } 11126 return ID2SYM(id); 11127 } 11128 return ID2SYM(rb_intern(name)); 11129 } 11130 11131 static ID 11132 ripper_get_id(VALUE v) 11133 { 11134 NODE *nd; 11135 if (!RB_TYPE_P(v, T_NODE)) return 0; 11136 nd = (NODE *)v; 11137 if (nd_type(nd) != NODE_LASGN) return 0; 11138 return nd->nd_vid; 11139 } 11140 11141 static VALUE 11142 ripper_get_value(VALUE v) 11143 { 11144 NODE *nd; 11145 if (v == Qundef) return Qnil; 11146 if (!RB_TYPE_P(v, T_NODE)) return v; 11147 nd = (NODE *)v; 11148 if (nd_type(nd) != NODE_LASGN) return Qnil; 11149 return nd->nd_rval; 11150 } 11151 11152 static void 11153 ripper_compile_error(struct parser_params *parser, const char *fmt, ...) 11154 { 11155 VALUE str; 11156 va_list args; 11157 11158 va_start(args, fmt); 11159 str = rb_vsprintf(fmt, args); 11160 va_end(args); 11161 rb_funcall(parser->value, rb_intern("compile_error"), 1, str); 11162 } 11163 11164 static void 11165 ripper_warn0(struct parser_params *parser, const char *fmt) 11166 { 11167 rb_funcall(parser->value, rb_intern("warn"), 1, STR_NEW2(fmt)); 11168 } 11169 11170 static void 11171 ripper_warnI(struct parser_params *parser, const char *fmt, int a) 11172 { 11173 rb_funcall(parser->value, rb_intern("warn"), 2, 11174 STR_NEW2(fmt), INT2NUM(a)); 11175 } 11176 11177 static void 11178 ripper_warnS(struct parser_params *parser, const char *fmt, const char *str) 11179 { 11180 rb_funcall(parser->value, rb_intern("warn"), 2, 11181 STR_NEW2(fmt), STR_NEW2(str)); 11182 } 11183 11184 static void 11185 ripper_warning0(struct parser_params *parser, const char *fmt) 11186 { 11187 rb_funcall(parser->value, rb_intern("warning"), 1, STR_NEW2(fmt)); 11188 } 11189 11190 static void 11191 ripper_warningS(struct parser_params *parser, const char *fmt, const char *str) 11192 { 11193 rb_funcall(parser->value, rb_intern("warning"), 2, 11194 STR_NEW2(fmt), STR_NEW2(str)); 11195 } 11196 11197 static VALUE 11198 ripper_lex_get_generic(struct parser_params *parser, VALUE src) 11199 { 11200 return rb_io_gets(src); 11201 } 11202 11203 static VALUE 11204 ripper_s_allocate(VALUE klass) 11205 { 11206 struct parser_params *p; 11207 VALUE self; 11208 11209 p = ALLOC_N(struct parser_params, 1); 11210 MEMZERO(p, struct parser_params, 1); 11211 self = TypedData_Wrap_Struct(klass, &parser_data_type, p); 11212 p->value = self; 11213 return self; 11214 } 11215 11216 #define ripper_initialized_p(r) ((r)->parser_lex_input != 0) 11217 11218 /* 11219 * call-seq: 11220 * Ripper.new(src, filename="(ripper)", lineno=1) -> ripper 11221 * 11222 * Create a new Ripper object. 11223 * _src_ must be a String, an IO, or an Object which has #gets method. 11224 * 11225 * This method does not starts parsing. 11226 * See also Ripper#parse and Ripper.parse. 11227 */ 11228 static VALUE 11229 ripper_initialize(int argc, VALUE *argv, VALUE self) 11230 { 11231 struct parser_params *parser; 11232 VALUE src, fname, lineno; 11233 11234 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 11235 rb_scan_args(argc, argv, "12", &src, &fname, &lineno); 11236 if (RB_TYPE_P(src, T_FILE)) { 11237 parser->parser_lex_gets = ripper_lex_get_generic; 11238 } 11239 else { 11240 StringValue(src); 11241 parser->parser_lex_gets = lex_get_str; 11242 } 11243 parser->parser_lex_input = src; 11244 parser->eofp = Qfalse; 11245 if (NIL_P(fname)) { 11246 fname = STR_NEW2("(ripper)"); 11247 } 11248 else { 11249 StringValue(fname); 11250 } 11251 parser_initialize(parser); 11252 11253 parser->parser_ruby_sourcefile_string = fname; 11254 parser->parser_ruby_sourcefile = RSTRING_PTR(fname); 11255 parser->parser_ruby_sourceline = NIL_P(lineno) ? 0 : NUM2INT(lineno) - 1; 11256 11257 return Qnil; 11258 } 11259 11260 struct ripper_args { 11261 struct parser_params *parser; 11262 int argc; 11263 VALUE *argv; 11264 }; 11265 11266 static VALUE 11267 ripper_parse0(VALUE parser_v) 11268 { 11269 struct parser_params *parser; 11270 11271 TypedData_Get_Struct(parser_v, struct parser_params, &parser_data_type, parser); 11272 parser_prepare(parser); 11273 ripper_yyparse((void*)parser); 11274 return parser->result; 11275 } 11276 11277 static VALUE 11278 ripper_ensure(VALUE parser_v) 11279 { 11280 struct parser_params *parser; 11281 11282 TypedData_Get_Struct(parser_v, struct parser_params, &parser_data_type, parser); 11283 parser->parsing_thread = Qnil; 11284 return Qnil; 11285 } 11286 11287 /* 11288 * call-seq: 11289 * ripper#parse 11290 * 11291 * Start parsing and returns the value of the root action. 11292 */ 11293 static VALUE 11294 ripper_parse(VALUE self) 11295 { 11296 struct parser_params *parser; 11297 11298 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 11299 if (!ripper_initialized_p(parser)) { 11300 rb_raise(rb_eArgError, "method called for uninitialized object"); 11301 } 11302 if (!NIL_P(parser->parsing_thread)) { 11303 if (parser->parsing_thread == rb_thread_current()) 11304 rb_raise(rb_eArgError, "Ripper#parse is not reentrant"); 11305 else 11306 rb_raise(rb_eArgError, "Ripper#parse is not multithread-safe"); 11307 } 11308 parser->parsing_thread = rb_thread_current(); 11309 rb_ensure(ripper_parse0, self, ripper_ensure, self); 11310 11311 return parser->result; 11312 } 11313 11314 /* 11315 * call-seq: 11316 * ripper#column -> Integer 11317 * 11318 * Return column number of current parsing line. 11319 * This number starts from 0. 11320 */ 11321 static VALUE 11322 ripper_column(VALUE self) 11323 { 11324 struct parser_params *parser; 11325 long col; 11326 11327 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 11328 if (!ripper_initialized_p(parser)) { 11329 rb_raise(rb_eArgError, "method called for uninitialized object"); 11330 } 11331 if (NIL_P(parser->parsing_thread)) return Qnil; 11332 col = parser->tokp - parser->parser_lex_pbeg; 11333 return LONG2NUM(col); 11334 } 11335 11336 /* 11337 * call-seq: 11338 * ripper#filename -> String 11339 * 11340 * Return current parsing filename. 11341 */ 11342 static VALUE 11343 ripper_filename(VALUE self) 11344 { 11345 struct parser_params *parser; 11346 11347 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 11348 if (!ripper_initialized_p(parser)) { 11349 rb_raise(rb_eArgError, "method called for uninitialized object"); 11350 } 11351 return parser->parser_ruby_sourcefile_string; 11352 } 11353 11354 /* 11355 * call-seq: 11356 * ripper#lineno -> Integer 11357 * 11358 * Return line number of current parsing line. 11359 * This number starts from 1. 11360 */ 11361 static VALUE 11362 ripper_lineno(VALUE self) 11363 { 11364 struct parser_params *parser; 11365 11366 TypedData_Get_Struct(self, struct parser_params, &parser_data_type, parser); 11367 if (!ripper_initialized_p(parser)) { 11368 rb_raise(rb_eArgError, "method called for uninitialized object"); 11369 } 11370 if (NIL_P(parser->parsing_thread)) return Qnil; 11371 return INT2NUM(parser->parser_ruby_sourceline); 11372 } 11373 11374 #ifdef RIPPER_DEBUG 11375 /* :nodoc: */ 11376 static VALUE 11377 ripper_assert_Qundef(VALUE self, VALUE obj, VALUE msg) 11378 { 11379 StringValue(msg); 11380 if (obj == Qundef) { 11381 rb_raise(rb_eArgError, "%s", RSTRING_PTR(msg)); 11382 } 11383 return Qnil; 11384 } 11385 11386 /* :nodoc: */ 11387 static VALUE 11388 ripper_value(VALUE self, VALUE obj) 11389 { 11390 return ULONG2NUM(obj); 11391 } 11392 #endif 11393 11394 11395 void 11396 Init_ripper(void) 11397 { 11398 parser_data_type.parent = RTYPEDDATA_TYPE(rb_parser_new()); 11399 11400 ripper_init_eventids1(); 11401 ripper_init_eventids2(); 11402 /* ensure existing in symbol table */ 11403 (void)rb_intern("||"); 11404 (void)rb_intern("&&"); 11405 11406 InitVM(ripper); 11407 } 11408 11409 void 11410 InitVM_ripper(void) 11411 { 11412 VALUE Ripper; 11413 11414 Ripper = rb_define_class("Ripper", rb_cObject); 11415 rb_define_const(Ripper, "Version", rb_usascii_str_new2(RIPPER_VERSION)); 11416 rb_define_alloc_func(Ripper, ripper_s_allocate); 11417 rb_define_method(Ripper, "initialize", ripper_initialize, -1); 11418 rb_define_method(Ripper, "parse", ripper_parse, 0); 11419 rb_define_method(Ripper, "column", ripper_column, 0); 11420 rb_define_method(Ripper, "filename", ripper_filename, 0); 11421 rb_define_method(Ripper, "lineno", ripper_lineno, 0); 11422 rb_define_method(Ripper, "end_seen?", rb_parser_end_seen_p, 0); 11423 rb_define_method(Ripper, "encoding", rb_parser_encoding, 0); 11424 rb_define_method(Ripper, "yydebug", rb_parser_get_yydebug, 0); 11425 rb_define_method(Ripper, "yydebug=", rb_parser_set_yydebug, 1); 11426 #ifdef RIPPER_DEBUG 11427 rb_define_method(rb_mKernel, "assert_Qundef", ripper_assert_Qundef, 2); 11428 rb_define_method(rb_mKernel, "rawVALUE", ripper_value, 1); 11429 rb_define_method(rb_mKernel, "validate_object", ripper_validate_object, 1); 11430 #endif 11431 11432 ripper_init_eventids1_table(Ripper); 11433 ripper_init_eventids2_table(Ripper); 11434 11435 # if 0 11436 /* Hack to let RDoc document SCRIPT_LINES__ */ 11437 11438 /* 11439 * When a Hash is assigned to +SCRIPT_LINES__+ the contents of files loaded 11440 * after the assignment will be added as an Array of lines with the file 11441 * name as the key. 11442 */ 11443 rb_define_global_const("SCRIPT_LINES__", Qnil); 11444 #endif 11445 11446 } 11447 #endif /* RIPPER */ 11448
1.7.6.1