ruby_parser.c

   1 /* This is a wrapper for parse.y */
   2
   3 #include "internal/parse.h"
   4 #include "internal/re.h"
   5 #include "internal/ruby_parser.h"
   6
   7 #include "node.h"
   8 #include "rubyparser.h"
   9 #include "internal/error.h"
  10
  11 #ifdef UNIVERSAL_PARSER
  12
  13 #include "internal.h"
  14 #include "internal/array.h"
  15 #include "internal/bignum.h"
  16 #include "internal/compile.h"
  17 #include "internal/complex.h"
  18 #include "internal/encoding.h"
  19 #include "internal/gc.h"
  20 #include "internal/hash.h"
  21 #include "internal/io.h"
  22 #include "internal/rational.h"
  23 #include "internal/re.h"
  24 #include "internal/string.h"
  25 #include "internal/symbol.h"
  26 #include "internal/thread.h"
  27
  28 #include "ruby/ractor.h"
  29 #include "ruby/ruby.h"
  30 #include "ruby/util.h"
  31 #include "internal.h"
  32 #include "vm_core.h"
  33 #include "symbol.h"
  34
  35 #define parser_encoding const void
  36
  37 RBIMPL_ATTR_FORMAT(RBIMPL_PRINTF_FORMAT, 6, 0)
  38 static VALUE
  39 syntax_error_append(VALUE exc, VALUE file, int line, int column,
  40                     parser_encoding *enc, const char *fmt, va_list args)
  41 {
  42     return rb_syntax_error_append(exc, file, line, column, enc, fmt, args);
  43 }
  44
  45 static int
  46 local_defined(ID id, const void *p)
  47 {
  48     return rb_local_defined(id, (const rb_iseq_t *)p);
  49 }
  50
  51 static int
  52 dvar_defined(ID id, const void *p)
  53 {
  54     return rb_dvar_defined(id, (const rb_iseq_t *)p);
  55 }
  56
  57 static int
  58 is_usascii_enc(parser_encoding *enc)
  59 {
  60     return rb_is_usascii_enc(enc);
  61 }
  62
  63 static int
  64 is_local_id2(ID id)
  65 {
  66     return is_local_id(id);
  67 }
  68
  69 static int
  70 is_attrset_id2(ID id)
  71 {
  72     return is_attrset_id(id);
  73 }
  74
  75 static int
  76 is_notop_id2(ID id)
  77 {
  78     return is_notop_id(id);
  79 }
  80
  81 static VALUE
  82 enc_str_new(const char *ptr, long len, parser_encoding *enc)
  83 {
  84     return rb_enc_str_new(ptr, len, enc);
  85 }
  86
  87 static int
  88 enc_isalnum(OnigCodePoint c, parser_encoding *enc)
  89 {
  90     return rb_enc_isalnum(c, enc);
  91 }
  92
  93 static int
  94 enc_precise_mbclen(const char *p, const char *e, parser_encoding *enc)
  95 {
  96     return rb_enc_precise_mbclen(p, e, enc);
  97 }
  98
  99 static int
 100 mbclen_charfound_p(int len)
 101 {
 102     return MBCLEN_CHARFOUND_P(len);
 103 }
 104
 105 static int
 106 mbclen_charfound_len(int len)
 107 {
 108     return MBCLEN_CHARFOUND_LEN(len);
 109 }
 110
 111 static const char *
 112 enc_name(parser_encoding *enc)
 113 {
 114     return rb_enc_name(enc);
 115 }
 116
 117 static char *
 118 enc_prev_char(const char *s, const char *p, const char *e, parser_encoding *enc)
 119 {
 120     return rb_enc_prev_char(s, p, e, enc);
 121 }
 122
 123 static parser_encoding *
 124 enc_get(VALUE obj)
 125 {
 126     return rb_enc_get(obj);
 127 }
 128
 129 static int
 130 enc_asciicompat(parser_encoding *enc)
 131 {
 132     return rb_enc_asciicompat(enc);
 133 }
 134
 135 static parser_encoding *
 136 utf8_encoding(void)
 137 {
 138     return rb_utf8_encoding();
 139 }
 140
 141 static VALUE
 142 enc_associate(VALUE obj, parser_encoding *enc)
 143 {
 144     return rb_enc_associate(obj, enc);
 145 }
 146
 147 static parser_encoding *
 148 ascii8bit_encoding(void)
 149 {
 150     return rb_ascii8bit_encoding();
 151 }
 152
 153 static int
 154 enc_codelen(int c, parser_encoding *enc)
 155 {
 156     return rb_enc_codelen(c, enc);
 157 }
 158
 159 static int
 160 enc_mbcput(unsigned int c, void *buf, parser_encoding *enc)
 161 {
 162     return rb_enc_mbcput(c, buf, enc);
 163 }
 164
 165 static parser_encoding *
 166 enc_from_index(int idx)
 167 {
 168     return rb_enc_from_index(idx);
 169 }
 170
 171 static int
 172 enc_isspace(OnigCodePoint c, parser_encoding *enc)
 173 {
 174     return rb_enc_isspace(c, enc);
 175 }
 176
 177 static ID
 178 intern3(const char *name, long len, parser_encoding *enc)
 179 {
 180     return rb_intern3(name, len, enc);
 181 }
 182
 183 static int
 184 enc_symname_type(const char *name, long len, parser_encoding *enc, unsigned int allowed_attrset)
 185 {
 186     return rb_enc_symname_type(name, len, enc, allowed_attrset);
 187 }
 188
 189 typedef struct {
 190     struct parser_params *parser;
 191     rb_encoding *enc;
 192     NODE *succ_block;
 193     const rb_code_location_t *loc;
 194     rb_parser_assignable_func assignable;
 195 } reg_named_capture_assign_t;
 196
 197 static int
 198 reg_named_capture_assign_iter(const OnigUChar *name, const OnigUChar *name_end,
 199           int back_num, int *back_refs, OnigRegex regex, void *arg0)
 200 {
 201     reg_named_capture_assign_t *arg = (reg_named_capture_assign_t*)arg0;
 202     struct parser_params* p = arg->parser;
 203     rb_encoding *enc = arg->enc;
 204     const rb_code_location_t *loc = arg->loc;
 205     long len = name_end - name;
 206     const char *s = (const char *)name;
 207
 208     return rb_reg_named_capture_assign_iter_impl(p, s, len, enc, &arg->succ_block, loc, arg->assignable);
 209 }
 210
 211 static NODE *
 212 reg_named_capture_assign(struct parser_params* p, VALUE regexp, const rb_code_location_t *loc,
 213                          rb_parser_assignable_func assignable)
 214 {
 215     reg_named_capture_assign_t arg;
 216
 217     arg.parser = p;
 218     arg.enc = rb_enc_get(regexp);
 219     arg.succ_block = 0;
 220     arg.loc = loc;
 221     arg.assignable = assignable;
 222     onig_foreach_name(RREGEXP_PTR(regexp), reg_named_capture_assign_iter, &arg);
 223
 224     if (!arg.succ_block) return 0;
 225     return RNODE_BLOCK(arg.succ_block)->nd_next;
 226 }
 227
 228 static int
 229 rtest(VALUE obj)
 230 {
 231     return (int)RB_TEST(obj);
 232 }
 233
 234 static int
 235 nil_p(VALUE obj)
 236 {
 237     return (int)NIL_P(obj);
 238 }
 239
 240 static VALUE
 241 syntax_error_new(void)
 242 {
 243     return rb_class_new_instance(0, 0, rb_eSyntaxError);
 244 }
 245
 246 static void *
 247 memmove2(void *dest, const void *src, size_t t, size_t n)
 248 {
 249     return memmove(dest, src, rbimpl_size_mul_or_raise(t, n));
 250 }
 251
 252 static void *
 253 nonempty_memcpy(void *dest, const void *src, size_t t, size_t n)
 254 {
 255     return ruby_nonempty_memcpy(dest, src, rbimpl_size_mul_or_raise(t, n));
 256 }
 257
 258 static VALUE
 259 ruby_verbose2(void)
 260 {
 261     return ruby_verbose;
 262 }
 263
 264 static int *
 265 rb_errno_ptr2(void)
 266 {
 267     return rb_errno_ptr();
 268 }
 269
 270 static void *
 271 zalloc(size_t elemsiz)
 272 {
 273     return ruby_xcalloc(1, elemsiz);
 274 }
 275
 276 static void
 277 gc_guard(VALUE obj)
 278 {
 279     RB_GC_GUARD(obj);
 280 }
 281
 282 static VALUE
 283 arg_error(void)
 284 {
 285     return rb_eArgError;
 286 }
 287
 288 static VALUE
 289 static_id2sym(ID id)
 290 {
 291     return (((VALUE)(id)<<RUBY_SPECIAL_SHIFT)|SYMBOL_FLAG);
 292 }
 293
 294 static long
 295 str_coderange_scan_restartable(const char *s, const char *e, parser_encoding *enc, int *cr)
 296 {
 297     return rb_str_coderange_scan_restartable(s, e, enc, cr);
 298 }
 299
 300 static int
 301 enc_mbminlen(parser_encoding *enc)
 302 {
 303     return rb_enc_mbminlen(enc);
 304 }
 305
 306 static bool
 307 enc_isascii(OnigCodePoint c, parser_encoding *enc)
 308 {
 309     return rb_enc_isascii(c, enc);
 310 }
 311
 312 static OnigCodePoint
 313 enc_mbc_to_codepoint(const char *p, const char *e, parser_encoding *enc)
 314 {
 315     const OnigUChar *up = RBIMPL_CAST((const OnigUChar *)p);
 316     const OnigUChar *ue = RBIMPL_CAST((const OnigUChar *)e);
 317
 318     return ONIGENC_MBC_TO_CODE((rb_encoding *)enc, up, ue);
 319 }
 320
 321 extern VALUE rb_eArgError;
 322
 323 static const rb_parser_config_t rb_global_parser_config = {
 324     .malloc = ruby_xmalloc,
 325     .calloc = ruby_xcalloc,
 326     .realloc = ruby_xrealloc,
 327     .free = ruby_xfree,
 328     .alloc_n = ruby_xmalloc2,
 329     .alloc = ruby_xmalloc,
 330     .realloc_n = ruby_xrealloc2,
 331     .zalloc = zalloc,
 332     .rb_memmove = memmove2,
 333     .nonempty_memcpy = nonempty_memcpy,
 334     .xmalloc_mul_add = rb_xmalloc_mul_add,
 335
 336     .compile_callback = rb_suppress_tracing,
 337     .reg_named_capture_assign = reg_named_capture_assign,
 338
 339     .attr_get = rb_attr_get,
 340
 341     .ary_new_from_args = rb_ary_new_from_args,
 342     .ary_unshift = rb_ary_unshift,
 343
 344     .make_temporary_id = rb_make_temporary_id,
 345     .is_local_id = is_local_id2,
 346     .is_attrset_id = is_attrset_id2,
 347     .is_global_name_punct = is_global_name_punct,
 348     .id_type = id_type,
 349     .id_attrset = rb_id_attrset,
 350     .intern = rb_intern,
 351     .intern2 = rb_intern2,
 352     .intern3 = intern3,
 353     .intern_str = rb_intern_str,
 354     .is_notop_id = is_notop_id2,
 355     .enc_symname_type = enc_symname_type,
 356     .id2name = rb_id2name,
 357     .id2str = rb_id2str,
 358     .id2sym = rb_id2sym,
 359
 360     .str_catf = rb_str_catf,
 361     .str_cat_cstr = rb_str_cat_cstr,
 362     .str_resize = rb_str_resize,
 363     .str_new = rb_str_new,
 364     .str_new_cstr = rb_str_new_cstr,
 365     .str_to_interned_str = rb_str_to_interned_str,
 366     .enc_str_new = enc_str_new,
 367     .str_vcatf = rb_str_vcatf,
 368     .rb_sprintf = rb_sprintf,
 369     .rstring_ptr = RSTRING_PTR,
 370     .rstring_len = RSTRING_LEN,
 371
 372     .int2num = rb_int2num_inline,
 373
 374     .stderr_tty_p = rb_stderr_tty_p,
 375     .write_error_str = rb_write_error_str,
 376     .io_write = rb_io_write,
 377     .io_flush = rb_io_flush,
 378     .io_puts = rb_io_puts,
 379
 380     .debug_output_stdout = rb_ractor_stdout,
 381     .debug_output_stderr = rb_ractor_stderr,
 382
 383     .is_usascii_enc = is_usascii_enc,
 384     .enc_isalnum = enc_isalnum,
 385     .enc_precise_mbclen = enc_precise_mbclen,
 386     .mbclen_charfound_p = mbclen_charfound_p,
 387     .mbclen_charfound_len = mbclen_charfound_len,
 388     .enc_name = enc_name,
 389     .enc_prev_char = enc_prev_char,
 390     .enc_get = enc_get,
 391     .enc_asciicompat = enc_asciicompat,
 392     .utf8_encoding = utf8_encoding,
 393     .enc_associate = enc_associate,
 394     .ascii8bit_encoding = ascii8bit_encoding,
 395     .enc_codelen = enc_codelen,
 396     .enc_mbcput = enc_mbcput,
 397     .enc_find_index = rb_enc_find_index,
 398     .enc_from_index = enc_from_index,
 399     .enc_isspace = enc_isspace,
 400     .enc_coderange_7bit = ENC_CODERANGE_7BIT,
 401     .enc_coderange_unknown = ENC_CODERANGE_UNKNOWN,
 402     .enc_mbminlen = enc_mbminlen,
 403     .enc_isascii = enc_isascii,
 404     .enc_mbc_to_codepoint = enc_mbc_to_codepoint,
 405
 406     .local_defined = local_defined,
 407     .dvar_defined = dvar_defined,
 408
 409     .syntax_error_append = syntax_error_append,
 410     .raise = rb_raise,
 411     .syntax_error_new = syntax_error_new,
 412
 413     .errinfo = rb_errinfo,
 414     .set_errinfo = rb_set_errinfo,
 415     .make_exception = rb_make_exception,
 416
 417     .sized_xfree = ruby_sized_xfree,
 418     .sized_realloc_n = ruby_sized_realloc_n,
 419     .gc_guard = gc_guard,
 420     .gc_mark = rb_gc_mark,
 421
 422     .reg_compile = rb_reg_compile,
 423     .reg_check_preprocess = rb_reg_check_preprocess,
 424     .memcicmp = rb_memcicmp,
 425
 426     .compile_warn = rb_compile_warn,
 427     .compile_warning = rb_compile_warning,
 428     .bug = rb_bug,
 429     .fatal = rb_fatal,
 430     .verbose = ruby_verbose2,
 431     .errno_ptr = rb_errno_ptr2,
 432
 433     .make_backtrace = rb_make_backtrace,
 434
 435     .scan_hex = ruby_scan_hex,
 436     .scan_oct = ruby_scan_oct,
 437     .scan_digits = ruby_scan_digits,
 438     .strtod = ruby_strtod,
 439
 440     .rtest = rtest,
 441     .nil_p = nil_p,
 442     .qnil = Qnil,
 443     .qfalse = Qfalse,
 444     .eArgError = arg_error,
 445     .long2int = rb_long2int,
 446
 447     /* For Ripper */
 448     .static_id2sym = static_id2sym,
 449     .str_coderange_scan_restartable = str_coderange_scan_restartable,
 450 };
 451 #endif
 452
 453 enum lex_type {
 454     lex_type_str,
 455     lex_type_io,
 456     lex_type_array,
 457     lex_type_generic,
 458 };
 459
 460 struct ruby_parser {
 461     rb_parser_t *parser_params;
 462     enum lex_type type;
 463     union {
 464         struct lex_pointer_string lex_str;
 465         struct {
 466             VALUE file;
 467         } lex_io;
 468         struct {
 469             VALUE ary;
 470         } lex_array;
 471     } data;
 472 };
 473
 474 static void
 475 parser_mark(void *ptr)
 476 {
 477     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 478     rb_ruby_parser_mark(parser->parser_params);
 479
 480     switch (parser->type) {
 481       case lex_type_str:
 482         rb_gc_mark(parser->data.lex_str.str);
 483         break;
 484       case lex_type_io:
 485         rb_gc_mark(parser->data.lex_io.file);
 486         break;
 487       case lex_type_array:
 488         rb_gc_mark(parser->data.lex_array.ary);
 489         break;
 490       case lex_type_generic:
 491         /* noop. Caller of rb_parser_compile_generic should mark the objects. */
 492         break;
 493     }
 494 }
 495
 496 static void
 497 parser_free(void *ptr)
 498 {
 499     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 500     rb_ruby_parser_free(parser->parser_params);
 501     xfree(parser);
 502 }
 503
 504 static size_t
 505 parser_memsize(const void *ptr)
 506 {
 507     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 508     return rb_ruby_parser_memsize(parser->parser_params);
 509 }
 510
 511 static const rb_data_type_t ruby_parser_data_type = {
 512     "parser",
 513     {
 514         parser_mark,
 515         parser_free,
 516         parser_memsize,
 517     },
 518     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 519 };
 520
 521 #ifdef UNIVERSAL_PARSER
 522 const rb_parser_config_t *
 523 rb_ruby_parser_config(void)
 524 {
 525     return &rb_global_parser_config;
 526 }
 527
 528 rb_parser_t *
 529 rb_parser_params_new(void)
 530 {
 531     return rb_ruby_parser_new(&rb_global_parser_config);
 532 }
 533 #else
 534 rb_parser_t *
 535 rb_parser_params_new(void)
 536 {
 537     return rb_ruby_parser_new();
 538 }
 539 #endif /* UNIVERSAL_PARSER */
 540
 541 VALUE
 542 rb_parser_new(void)
 543 {
 544     struct ruby_parser *parser;
 545     rb_parser_t *parser_params;
 546
 547     /*
 548      * Create parser_params ahead of vparser because
 549      * rb_ruby_parser_new can run GC so if create vparser
 550      * first, parser_mark tries to mark not initialized parser_params.
 551      */
 552     parser_params = rb_parser_params_new();
 553     VALUE vparser = TypedData_Make_Struct(0, struct ruby_parser,
 554                                          &ruby_parser_data_type, parser);
 555     parser->parser_params = parser_params;
 556
 557     return vparser;
 558 }
 559
 560 void
 561 rb_parser_set_options(VALUE vparser, int print, int loop, int chomp, int split)
 562 {
 563     struct ruby_parser *parser;
 564
 565     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 566     rb_ruby_parser_set_options(parser->parser_params, print, loop, chomp, split);
 567 }
 568
 569 VALUE
 570 rb_parser_set_context(VALUE vparser, const struct rb_iseq_struct *base, int main)
 571 {
 572     struct ruby_parser *parser;
 573
 574     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 575     rb_ruby_parser_set_context(parser->parser_params, base, main);
 576     return vparser;
 577 }
 578
 579 void
 580 rb_parser_set_script_lines(VALUE vparser)
 581 {
 582     struct ruby_parser *parser;
 583
 584     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 585     rb_ruby_parser_set_script_lines(parser->parser_params);
 586 }
 587
 588 void
 589 rb_parser_error_tolerant(VALUE vparser)
 590 {
 591     struct ruby_parser *parser;
 592
 593     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 594     rb_ruby_parser_error_tolerant(parser->parser_params);
 595 }
 596
 597 void
 598 rb_parser_keep_tokens(VALUE vparser)
 599 {
 600     struct ruby_parser *parser;
 601
 602     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 603     rb_ruby_parser_keep_tokens(parser->parser_params);
 604 }
 605
 606 rb_parser_string_t *
 607 rb_parser_lex_get_str(struct parser_params *p, struct lex_pointer_string *ptr_str)
 608 {
 609     char *beg, *end, *start;
 610     long len;
 611     VALUE s = ptr_str->str;
 612
 613     beg = RSTRING_PTR(s);
 614     len = RSTRING_LEN(s);
 615     start = beg;
 616     if (ptr_str->ptr) {
 617         if (len == ptr_str->ptr) return 0;
 618         beg += ptr_str->ptr;
 619         len -= ptr_str->ptr;
 620     }
 621     end = memchr(beg, '\n', len);
 622     if (end) len = ++end - beg;
 623     ptr_str->ptr += len;
 624     return rb_str_to_parser_string(p, rb_str_subseq(s, beg - start, len));
 625 }
 626
 627 static rb_parser_string_t *
 628 lex_get_str(struct parser_params *p, rb_parser_input_data input, int line_count)
 629 {
 630     return rb_parser_lex_get_str(p, (struct lex_pointer_string *)input);
 631 }
 632
 633 static void parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines);
 634
 635 static rb_ast_t*
 636 parser_compile(rb_parser_t *p, rb_parser_lex_gets_func *gets, VALUE fname, rb_parser_input_data input, int line)
 637 {
 638     rb_ast_t *ast = rb_parser_compile(p, gets, fname, input, line);
 639     parser_aset_script_lines_for(fname, ast->body.script_lines);
 640     return ast;
 641 }
 642
 643 static rb_ast_t*
 644 parser_compile_string0(struct ruby_parser *parser, VALUE fname, VALUE s, int line)
 645 {
 646     VALUE str = rb_str_new_frozen(s);
 647
 648     parser->type = lex_type_str;
 649     parser->data.lex_str.str = str;
 650     parser->data.lex_str.ptr = 0;
 651
 652     return parser_compile(parser->parser_params, lex_get_str, fname, (rb_parser_input_data)&parser->data, line);
 653 }
 654
 655 static rb_encoding *
 656 must_be_ascii_compatible(VALUE s)
 657 {
 658     rb_encoding *enc = rb_enc_get(s);
 659     if (!rb_enc_asciicompat(enc)) {
 660         rb_raise(rb_eArgError, "invalid source encoding");
 661     }
 662     return enc;
 663 }
 664
 665 static rb_ast_t*
 666 parser_compile_string_path(struct ruby_parser *parser, VALUE f, VALUE s, int line)
 667 {
 668     must_be_ascii_compatible(s);
 669     return parser_compile_string0(parser, f, s, line);
 670 }
 671
 672 static rb_ast_t*
 673 parser_compile_string(struct ruby_parser *parser, const char *f, VALUE s, int line)
 674 {
 675     return parser_compile_string_path(parser, rb_filesystem_str_new_cstr(f), s, line);
 676 }
 677
 678 VALUE rb_io_gets_internal(VALUE io);
 679
 680 static rb_parser_string_t *
 681 lex_io_gets(struct parser_params *p, rb_parser_input_data input, int line_count)
 682 {
 683     VALUE io = (VALUE)input;
 684     VALUE line = rb_io_gets_internal(io);
 685     if (NIL_P(line)) return 0;
 686     return rb_str_to_parser_string(p, line);
 687 }
 688
 689 static rb_parser_string_t *
 690 lex_gets_array(struct parser_params *p, rb_parser_input_data data, int index)
 691 {
 692     VALUE array = (VALUE)data;
 693     VALUE str = rb_ary_entry(array, index);
 694     if (!NIL_P(str)) {
 695         StringValue(str);
 696         if (!rb_enc_asciicompat(rb_enc_get(str))) {
 697             rb_raise(rb_eArgError, "invalid source encoding");
 698         }
 699         return rb_str_to_parser_string(p, str);
 700     }
 701     else {
 702         return 0;
 703     }
 704 }
 705
 706 static rb_ast_t*
 707 parser_compile_file_path(struct ruby_parser *parser, VALUE fname, VALUE file, int start)
 708 {
 709     parser->type = lex_type_io;
 710     parser->data.lex_io.file = file;
 711
 712     return parser_compile(parser->parser_params, lex_io_gets, fname, (rb_parser_input_data)file, start);
 713 }
 714
 715 static rb_ast_t*
 716 parser_compile_array(struct ruby_parser *parser, VALUE fname, VALUE array, int start)
 717 {
 718     parser->type = lex_type_array;
 719     parser->data.lex_array.ary = array;
 720
 721     return parser_compile(parser->parser_params, lex_gets_array, fname, (rb_parser_input_data)array, start);
 722 }
 723
 724 static rb_ast_t*
 725 parser_compile_generic(struct ruby_parser *parser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 726 {
 727     parser->type = lex_type_generic;
 728
 729     return parser_compile(parser->parser_params, lex_gets, fname, (rb_parser_input_data)input, start);
 730 }
 731
 732 static void
 733 ast_free(void *ptr)
 734 {
 735     rb_ast_t *ast = (rb_ast_t *)ptr;
 736     rb_ast_free(ast);
 737 }
 738
 739 static const rb_data_type_t ast_data_type = {
 740     "AST",
 741     {
 742         NULL,
 743         ast_free,
 744         NULL, // No dsize() because this object does not appear in ObjectSpace.
 745     },
 746     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 747 };
 748
 749 static VALUE
 750 ast_alloc(void)
 751 {
 752     return TypedData_Wrap_Struct(0, &ast_data_type, NULL);
 753 }
 754
 755 VALUE
 756 rb_parser_compile_file_path(VALUE vparser, VALUE fname, VALUE file, int start)
 757 {
 758     struct ruby_parser *parser;
 759     VALUE ast_value = ast_alloc();
 760
 761     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 762     DATA_PTR(ast_value) = parser_compile_file_path(parser, fname, file, start);
 763     RB_GC_GUARD(vparser);
 764
 765     return ast_value;
 766 }
 767
 768 VALUE
 769 rb_parser_compile_array(VALUE vparser, VALUE fname, VALUE array, int start)
 770 {
 771     struct ruby_parser *parser;
 772     VALUE ast_value = ast_alloc();
 773
 774     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 775     DATA_PTR(ast_value) = parser_compile_array(parser, fname, array, start);
 776     RB_GC_GUARD(vparser);
 777
 778     return ast_value;
 779 }
 780
 781 VALUE
 782 rb_parser_compile_generic(VALUE vparser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 783 {
 784     struct ruby_parser *parser;
 785     VALUE ast_value = ast_alloc();
 786
 787     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 788     DATA_PTR(ast_value) = parser_compile_generic(parser, lex_gets, fname, input, start);
 789     RB_GC_GUARD(vparser);
 790
 791     return ast_value;
 792 }
 793
 794 VALUE
 795 rb_parser_compile_string(VALUE vparser, const char *f, VALUE s, int line)
 796 {
 797     struct ruby_parser *parser;
 798     VALUE ast_value = ast_alloc();
 799
 800     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 801     DATA_PTR(ast_value) = parser_compile_string(parser, f, s, line);
 802     RB_GC_GUARD(vparser);
 803
 804     return ast_value;
 805 }
 806
 807 VALUE
 808 rb_parser_compile_string_path(VALUE vparser, VALUE f, VALUE s, int line)
 809 {
 810     struct ruby_parser *parser;
 811     VALUE ast_value = ast_alloc();
 812
 813     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 814     DATA_PTR(ast_value) = parser_compile_string_path(parser, f, s, line);
 815     RB_GC_GUARD(vparser);
 816
 817     return ast_value;
 818 }
 819
 820 VALUE
 821 rb_parser_encoding(VALUE vparser)
 822 {
 823     struct ruby_parser *parser;
 824
 825     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 826     return rb_enc_from_encoding(rb_ruby_parser_encoding(parser->parser_params));
 827 }
 828
 829 VALUE
 830 rb_parser_end_seen_p(VALUE vparser)
 831 {
 832     struct ruby_parser *parser;
 833
 834     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 835     return RBOOL(rb_ruby_parser_end_seen_p(parser->parser_params));
 836 }
 837
 838 VALUE
 839 rb_parser_set_yydebug(VALUE vparser, VALUE flag)
 840 {
 841     struct ruby_parser *parser;
 842
 843     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 844     rb_ruby_parser_set_yydebug(parser->parser_params, RTEST(flag));
 845     return flag;
 846 }
 847
 848 void
 849 rb_set_script_lines_for(VALUE vparser, VALUE path)
 850 {
 851     struct ruby_parser *parser;
 852     VALUE hash;
 853     ID script_lines;
 854     CONST_ID(script_lines, "SCRIPT_LINES__");
 855     if (!rb_const_defined_at(rb_cObject, script_lines)) return;
 856     hash = rb_const_get_at(rb_cObject, script_lines);
 857     if (RB_TYPE_P(hash, T_HASH)) {
 858         rb_hash_aset(hash, path, Qtrue);
 859         TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 860         rb_ruby_parser_set_script_lines(parser->parser_params);
 861     }
 862 }
 863
 864 VALUE
 865 rb_parser_build_script_lines_from(rb_parser_ary_t *lines)
 866 {
 867     int i;
 868     if (!lines) return Qnil;
 869     if (lines->data_type != PARSER_ARY_DATA_SCRIPT_LINE) {
 870         rb_bug("unexpected rb_parser_ary_data_type (%d) for script lines", lines->data_type);
 871     }
 872     VALUE script_lines = rb_ary_new_capa(lines->len);
 873     for (i = 0; i < lines->len; i++) {
 874         rb_parser_string_t *str = (rb_parser_string_t *)lines->data[i];
 875         rb_ary_push(script_lines, rb_enc_str_new(str->ptr, str->len, str->enc));
 876     }
 877     return script_lines;
 878 }
 879
 880 VALUE
 881 rb_str_new_parser_string(rb_parser_string_t *str)
 882 {
 883     VALUE string = rb_enc_literal_str(str->ptr, str->len, str->enc);
 884     rb_enc_str_coderange(string);
 885     return string;
 886 }
 887
 888 VALUE
 889 rb_str_new_mutable_parser_string(rb_parser_string_t *str)
 890 {
 891     return rb_enc_str_new(str->ptr, str->len, str->enc);
 892 }
 893
 894 static VALUE
 895 negative_numeric(VALUE val)
 896 {
 897     if (FIXNUM_P(val)) {
 898         return LONG2FIX(-FIX2LONG(val));
 899     }
 900     if (SPECIAL_CONST_P(val)) {
 901 #if USE_FLONUM
 902         if (FLONUM_P(val)) {
 903             return DBL2NUM(-RFLOAT_VALUE(val));
 904         }
 905 #endif
 906         goto unknown;
 907     }
 908     switch (BUILTIN_TYPE(val)) {
 909       case T_BIGNUM:
 910         BIGNUM_NEGATE(val);
 911         val = rb_big_norm(val);
 912         break;
 913       case T_RATIONAL:
 914         RATIONAL_SET_NUM(val, negative_numeric(RRATIONAL(val)->num));
 915         break;
 916       case T_COMPLEX:
 917         RCOMPLEX_SET_REAL(val, negative_numeric(RCOMPLEX(val)->real));
 918         RCOMPLEX_SET_IMAG(val, negative_numeric(RCOMPLEX(val)->imag));
 919         break;
 920       case T_FLOAT:
 921         val = DBL2NUM(-RFLOAT_VALUE(val));
 922         break;
 923       unknown:
 924       default:
 925         rb_bug("unknown literal type (%s) passed to negative_numeric",
 926                rb_builtin_class_name(val));
 927         break;
 928     }
 929     return val;
 930 }
 931
 932 static VALUE
 933 integer_value(const char *val, int base)
 934 {
 935     return rb_cstr_to_inum(val, base, FALSE);
 936 }
 937
 938 static VALUE
 939 rational_value(const char *node_val, int base, int seen_point)
 940 {
 941     VALUE lit;
 942     char* val = strdup(node_val);
 943     if (seen_point > 0) {
 944         int len = (int)(strlen(val));
 945         char *point = &val[seen_point];
 946         size_t fraclen = len-seen_point-1;
 947         memmove(point, point+1, fraclen+1);
 948
 949         lit = rb_rational_new(integer_value(val, base), rb_int_positive_pow(10, fraclen));
 950     }
 951     else {
 952         lit = rb_rational_raw1(integer_value(val, base));
 953     }
 954
 955     free(val);
 956
 957     return lit;
 958 }
 959
 960 VALUE
 961 rb_node_integer_literal_val(const NODE *n)
 962 {
 963     const rb_node_integer_t *node = RNODE_INTEGER(n);
 964     VALUE val = integer_value(node->val, node->base);
 965     if (node->minus) {
 966         val = negative_numeric(val);
 967     }
 968     return val;
 969 }
 970
 971 VALUE
 972 rb_node_float_literal_val(const NODE *n)
 973 {
 974     const rb_node_float_t *node = RNODE_FLOAT(n);
 975     double d = strtod(node->val, 0);
 976     if (node->minus) {
 977         d = -d;
 978     }
 979     VALUE val = DBL2NUM(d);
 980     return val;
 981 }
 982
 983 VALUE
 984 rb_node_rational_literal_val(const NODE *n)
 985 {
 986     VALUE lit;
 987     const rb_node_rational_t *node = RNODE_RATIONAL(n);
 988
 989     lit = rational_value(node->val, node->base, node->seen_point);
 990
 991     if (node->minus) {
 992         lit = negative_numeric(lit);
 993     }
 994
 995     return lit;
 996 }
 997
 998 VALUE
 999 rb_node_imaginary_literal_val(const NODE *n)
1000 {
1001     VALUE lit;
1002     const rb_node_imaginary_t *node = RNODE_IMAGINARY(n);
1003
1004     enum rb_numeric_type type = node->type;
1005
1006     switch (type) {
1007       case integer_literal:
1008         lit = integer_value(node->val, node->base);
1009         break;
1010       case float_literal:{
1011         double d = strtod(node->val, 0);
1012         lit = DBL2NUM(d);
1013         break;
1014       }
1015       case rational_literal:
1016         lit = rational_value(node->val, node->base, node->seen_point);
1017         break;
1018       default:
1019         rb_bug("unreachable");
1020     }
1021
1022     lit = rb_complex_raw(INT2FIX(0), lit);
1023
1024     if (node->minus) {
1025         lit = negative_numeric(lit);
1026     }
1027     return lit;
1028 }
1029
1030 VALUE
1031 rb_node_str_string_val(const NODE *node)
1032 {
1033     rb_parser_string_t *str = RNODE_STR(node)->string;
1034     return rb_str_new_parser_string(str);
1035 }
1036
1037 VALUE
1038 rb_node_sym_string_val(const NODE *node)
1039 {
1040     rb_parser_string_t *str = RNODE_SYM(node)->string;
1041     return ID2SYM(rb_intern3(str->ptr, str->len, str->enc));
1042 }
1043
1044 VALUE
1045 rb_node_dstr_string_val(const NODE *node)
1046 {
1047     rb_parser_string_t *str = RNODE_DSTR(node)->string;
1048     return str ? rb_str_new_parser_string(str) : Qnil;
1049 }
1050
1051 VALUE
1052 rb_node_dregx_string_val(const NODE *node)
1053 {
1054     rb_parser_string_t *str = RNODE_DREGX(node)->string;
1055     return rb_str_new_parser_string(str);
1056 }
1057
1058 VALUE
1059 rb_node_regx_string_val(const NODE *node)
1060 {
1061     rb_node_regx_t *node_reg = RNODE_REGX(node);
1062     rb_parser_string_t *string = node_reg->string;
1063     VALUE str = rb_enc_str_new(string->ptr, string->len, string->enc);
1064
1065     return rb_reg_compile(str, node_reg->options, NULL, 0);
1066 }
1067
1068 VALUE
1069 rb_node_line_lineno_val(const NODE *node)
1070 {
1071     return INT2FIX(node->nd_loc.beg_pos.lineno);
1072 }
1073
1074 VALUE
1075 rb_node_file_path_val(const NODE *node)
1076 {
1077     return rb_str_new_parser_string(RNODE_FILE(node)->path);
1078 }
1079
1080 VALUE
1081 rb_node_encoding_val(const NODE *node)
1082 {
1083     return rb_enc_from_encoding(RNODE_ENCODING(node)->enc);
1084 }
1085
1086 static void
1087 parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines)
1088 {
1089     VALUE hash, script_lines;
1090     ID script_lines_id;
1091     if (NIL_P(path) || !lines) return;
1092     CONST_ID(script_lines_id, "SCRIPT_LINES__");
1093     if (!rb_const_defined_at(rb_cObject, script_lines_id)) return;
1094     hash = rb_const_get_at(rb_cObject, script_lines_id);
1095     if (!RB_TYPE_P(hash, T_HASH)) return;
1096     if (rb_hash_lookup(hash, path) == Qnil) return;
1097     script_lines = rb_parser_build_script_lines_from(lines);
1098     rb_hash_aset(hash, path, script_lines);
1099 }
1100
1101 VALUE
1102 rb_ruby_ast_new(const NODE *const root)
1103 {
1104     rb_ast_t *ast;
1105     VALUE ast_value = TypedData_Make_Struct(0, rb_ast_t, &ast_data_type, ast);
1106 #ifdef UNIVERSAL_PARSER
1107     ast->config = &rb_global_parser_config;
1108 #endif
1109     ast->body = (rb_ast_body_t){
1110         .root = root,
1111         .frozen_string_literal = -1,
1112         .coverage_enabled = -1,
1113         .script_lines = NULL,
1114         .line_count = 0,
1115     };
1116     return ast_value;
1117 }
1118
1119 rb_ast_t *
1120 rb_ruby_ast_data_get(VALUE ast_value)
1121 {
1122     rb_ast_t *ast;
1123     if (NIL_P(ast_value)) return NULL;
1124     TypedData_Get_Struct(ast_value, rb_ast_t, &ast_data_type, ast);
1125     return ast;
1126 }