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 static int
  38 is_ascii_string2(VALUE str)
  39 {
  40     return is_ascii_string(str);
  41 }
  42
  43 RBIMPL_ATTR_FORMAT(RBIMPL_PRINTF_FORMAT, 6, 0)
  44 static VALUE
  45 syntax_error_append(VALUE exc, VALUE file, int line, int column,
  46                     parser_encoding *enc, const char *fmt, va_list args)
  47 {
  48     return rb_syntax_error_append(exc, file, line, column, enc, fmt, args);
  49 }
  50
  51 static int
  52 local_defined(ID id, const void *p)
  53 {
  54     return rb_local_defined(id, (const rb_iseq_t *)p);
  55 }
  56
  57 static int
  58 dvar_defined(ID id, const void *p)
  59 {
  60     return rb_dvar_defined(id, (const rb_iseq_t *)p);
  61 }
  62
  63 static int
  64 is_usascii_enc(parser_encoding *enc)
  65 {
  66     return rb_is_usascii_enc(enc);
  67 }
  68
  69 static int
  70 is_local_id2(ID id)
  71 {
  72     return is_local_id(id);
  73 }
  74
  75 static int
  76 is_attrset_id2(ID id)
  77 {
  78     return is_attrset_id(id);
  79 }
  80
  81 static int
  82 is_notop_id2(ID id)
  83 {
  84     return is_notop_id(id);
  85 }
  86
  87 static VALUE
  88 enc_str_new(const char *ptr, long len, parser_encoding *enc)
  89 {
  90     return rb_enc_str_new(ptr, len, enc);
  91 }
  92
  93 static int
  94 enc_isalnum(OnigCodePoint c, parser_encoding *enc)
  95 {
  96     return rb_enc_isalnum(c, enc);
  97 }
  98
  99 static int
 100 enc_precise_mbclen(const char *p, const char *e, parser_encoding *enc)
 101 {
 102     return rb_enc_precise_mbclen(p, e, enc);
 103 }
 104
 105 static int
 106 mbclen_charfound_p(int len)
 107 {
 108     return MBCLEN_CHARFOUND_P(len);
 109 }
 110
 111 static int
 112 mbclen_charfound_len(int len)
 113 {
 114     return MBCLEN_CHARFOUND_LEN(len);
 115 }
 116
 117 static const char *
 118 enc_name(parser_encoding *enc)
 119 {
 120     return rb_enc_name(enc);
 121 }
 122
 123 static char *
 124 enc_prev_char(const char *s, const char *p, const char *e, parser_encoding *enc)
 125 {
 126     return rb_enc_prev_char(s, p, e, enc);
 127 }
 128
 129 static parser_encoding *
 130 enc_get(VALUE obj)
 131 {
 132     return rb_enc_get(obj);
 133 }
 134
 135 static int
 136 enc_asciicompat(parser_encoding *enc)
 137 {
 138     return rb_enc_asciicompat(enc);
 139 }
 140
 141 static parser_encoding *
 142 utf8_encoding(void)
 143 {
 144     return rb_utf8_encoding();
 145 }
 146
 147 static VALUE
 148 enc_associate(VALUE obj, parser_encoding *enc)
 149 {
 150     return rb_enc_associate(obj, enc);
 151 }
 152
 153 static parser_encoding *
 154 ascii8bit_encoding(void)
 155 {
 156     return rb_ascii8bit_encoding();
 157 }
 158
 159 static int
 160 enc_codelen(int c, parser_encoding *enc)
 161 {
 162     return rb_enc_codelen(c, enc);
 163 }
 164
 165 static int
 166 enc_mbcput(unsigned int c, void *buf, parser_encoding *enc)
 167 {
 168     return rb_enc_mbcput(c, buf, enc);
 169 }
 170
 171 static parser_encoding *
 172 enc_from_index(int idx)
 173 {
 174     return rb_enc_from_index(idx);
 175 }
 176
 177 static int
 178 enc_isspace(OnigCodePoint c, parser_encoding *enc)
 179 {
 180     return rb_enc_isspace(c, enc);
 181 }
 182
 183 static ID
 184 intern3(const char *name, long len, parser_encoding *enc)
 185 {
 186     return rb_intern3(name, len, enc);
 187 }
 188
 189 static parser_encoding *
 190 usascii_encoding(void)
 191 {
 192     return rb_usascii_encoding();
 193 }
 194
 195 static int
 196 enc_symname_type(const char *name, long len, parser_encoding *enc, unsigned int allowed_attrset)
 197 {
 198     return rb_enc_symname_type(name, len, enc, allowed_attrset);
 199 }
 200
 201 typedef struct {
 202     struct parser_params *parser;
 203     rb_encoding *enc;
 204     NODE *succ_block;
 205     const rb_code_location_t *loc;
 206 } reg_named_capture_assign_t;
 207
 208 static int
 209 reg_named_capture_assign_iter(const OnigUChar *name, const OnigUChar *name_end,
 210           int back_num, int *back_refs, OnigRegex regex, void *arg0)
 211 {
 212     reg_named_capture_assign_t *arg = (reg_named_capture_assign_t*)arg0;
 213     struct parser_params* p = arg->parser;
 214     rb_encoding *enc = arg->enc;
 215     const rb_code_location_t *loc = arg->loc;
 216     long len = name_end - name;
 217     const char *s = (const char *)name;
 218
 219     return rb_reg_named_capture_assign_iter_impl(p, s, len, enc, &arg->succ_block, loc);
 220 }
 221
 222 static NODE *
 223 reg_named_capture_assign(struct parser_params* p, VALUE regexp, const rb_code_location_t *loc)
 224 {
 225     reg_named_capture_assign_t arg;
 226
 227     arg.parser = p;
 228     arg.enc = rb_enc_get(regexp);
 229     arg.succ_block = 0;
 230     arg.loc = loc;
 231     onig_foreach_name(RREGEXP_PTR(regexp), reg_named_capture_assign_iter, &arg);
 232
 233     if (!arg.succ_block) return 0;
 234     return RNODE_BLOCK(arg.succ_block)->nd_next;
 235 }
 236
 237 static int
 238 rtest(VALUE obj)
 239 {
 240     return (int)RB_TEST(obj);
 241 }
 242
 243 static int
 244 nil_p(VALUE obj)
 245 {
 246     return (int)NIL_P(obj);
 247 }
 248
 249 static VALUE
 250 syntax_error_new(void)
 251 {
 252     return rb_class_new_instance(0, 0, rb_eSyntaxError);
 253 }
 254
 255 static void *
 256 memmove2(void *dest, const void *src, size_t t, size_t n)
 257 {
 258     return memmove(dest, src, rbimpl_size_mul_or_raise(t, n));
 259 }
 260
 261 static void *
 262 nonempty_memcpy(void *dest, const void *src, size_t t, size_t n)
 263 {
 264     return ruby_nonempty_memcpy(dest, src, rbimpl_size_mul_or_raise(t, n));
 265 }
 266
 267 static VALUE
 268 ruby_verbose2(void)
 269 {
 270     return ruby_verbose;
 271 }
 272
 273 static int *
 274 rb_errno_ptr2(void)
 275 {
 276     return rb_errno_ptr();
 277 }
 278
 279 static void *
 280 zalloc(size_t elemsiz)
 281 {
 282     return ruby_xcalloc(1, elemsiz);
 283 }
 284
 285 static void
 286 gc_guard(VALUE obj)
 287 {
 288     RB_GC_GUARD(obj);
 289 }
 290
 291 static VALUE
 292 arg_error(void)
 293 {
 294     return rb_eArgError;
 295 }
 296
 297 static VALUE
 298 static_id2sym(ID id)
 299 {
 300     return (((VALUE)(id)<<RUBY_SPECIAL_SHIFT)|SYMBOL_FLAG);
 301 }
 302
 303 static long
 304 str_coderange_scan_restartable(const char *s, const char *e, parser_encoding *enc, int *cr)
 305 {
 306     return rb_str_coderange_scan_restartable(s, e, enc, cr);
 307 }
 308
 309 static int
 310 enc_mbminlen(parser_encoding *enc)
 311 {
 312     return rb_enc_mbminlen(enc);
 313 }
 314
 315 static bool
 316 enc_isascii(OnigCodePoint c, parser_encoding *enc)
 317 {
 318     return rb_enc_isascii(c, enc);
 319 }
 320
 321 static OnigCodePoint
 322 enc_mbc_to_codepoint(const char *p, const char *e, parser_encoding *enc)
 323 {
 324     const OnigUChar *up = RBIMPL_CAST((const OnigUChar *)p);
 325     const OnigUChar *ue = RBIMPL_CAST((const OnigUChar *)e);
 326
 327     return ONIGENC_MBC_TO_CODE((rb_encoding *)enc, up, ue);
 328 }
 329
 330 extern VALUE rb_eArgError;
 331
 332 static const rb_parser_config_t rb_global_parser_config = {
 333     .malloc = ruby_xmalloc,
 334     .calloc = ruby_xcalloc,
 335     .realloc = ruby_xrealloc,
 336     .free = ruby_xfree,
 337     .alloc_n = ruby_xmalloc2,
 338     .alloc = ruby_xmalloc,
 339     .realloc_n = ruby_xrealloc2,
 340     .zalloc = zalloc,
 341     .rb_memmove = memmove2,
 342     .nonempty_memcpy = nonempty_memcpy,
 343     .xmalloc_mul_add = rb_xmalloc_mul_add,
 344
 345     .compile_callback = rb_suppress_tracing,
 346     .reg_named_capture_assign = reg_named_capture_assign,
 347
 348     .attr_get = rb_attr_get,
 349
 350     .ary_new = rb_ary_new,
 351     .ary_push = rb_ary_push,
 352     .ary_new_from_args = rb_ary_new_from_args,
 353     .ary_unshift = rb_ary_unshift,
 354
 355     .make_temporary_id = rb_make_temporary_id,
 356     .is_local_id = is_local_id2,
 357     .is_attrset_id = is_attrset_id2,
 358     .is_global_name_punct = is_global_name_punct,
 359     .id_type = id_type,
 360     .id_attrset = rb_id_attrset,
 361     .intern = rb_intern,
 362     .intern2 = rb_intern2,
 363     .intern3 = intern3,
 364     .intern_str = rb_intern_str,
 365     .is_notop_id = is_notop_id2,
 366     .enc_symname_type = enc_symname_type,
 367     .id2name = rb_id2name,
 368     .id2str = rb_id2str,
 369     .id2sym = rb_id2sym,
 370     .sym2id = rb_sym2id,
 371
 372     .str_catf = rb_str_catf,
 373     .str_cat_cstr = rb_str_cat_cstr,
 374     .str_modify = rb_str_modify,
 375     .str_set_len = rb_str_set_len,
 376     .str_cat = rb_str_cat,
 377     .str_resize = rb_str_resize,
 378     .str_new = rb_str_new,
 379     .str_new_cstr = rb_str_new_cstr,
 380     .str_to_interned_str = rb_str_to_interned_str,
 381     .is_ascii_string = is_ascii_string2,
 382     .enc_str_new = enc_str_new,
 383     .str_vcatf = rb_str_vcatf,
 384     .rb_sprintf = rb_sprintf,
 385     .rstring_ptr = RSTRING_PTR,
 386     .rstring_end = RSTRING_END,
 387     .rstring_len = RSTRING_LEN,
 388     .obj_as_string = rb_obj_as_string,
 389
 390     .int2num = rb_int2num_inline,
 391
 392     .stderr_tty_p = rb_stderr_tty_p,
 393     .write_error_str = rb_write_error_str,
 394     .io_write = rb_io_write,
 395     .io_flush = rb_io_flush,
 396     .io_puts = rb_io_puts,
 397
 398     .debug_output_stdout = rb_ractor_stdout,
 399     .debug_output_stderr = rb_ractor_stderr,
 400
 401     .is_usascii_enc = is_usascii_enc,
 402     .enc_isalnum = enc_isalnum,
 403     .enc_precise_mbclen = enc_precise_mbclen,
 404     .mbclen_charfound_p = mbclen_charfound_p,
 405     .mbclen_charfound_len = mbclen_charfound_len,
 406     .enc_name = enc_name,
 407     .enc_prev_char = enc_prev_char,
 408     .enc_get = enc_get,
 409     .enc_asciicompat = enc_asciicompat,
 410     .utf8_encoding = utf8_encoding,
 411     .enc_associate = enc_associate,
 412     .ascii8bit_encoding = ascii8bit_encoding,
 413     .enc_codelen = enc_codelen,
 414     .enc_mbcput = enc_mbcput,
 415     .enc_find_index = rb_enc_find_index,
 416     .enc_from_index = enc_from_index,
 417     .enc_isspace = enc_isspace,
 418     .enc_coderange_7bit = ENC_CODERANGE_7BIT,
 419     .enc_coderange_unknown = ENC_CODERANGE_UNKNOWN,
 420     .usascii_encoding = usascii_encoding,
 421     .enc_coderange_broken = ENC_CODERANGE_BROKEN,
 422     .enc_mbminlen = enc_mbminlen,
 423     .enc_isascii = enc_isascii,
 424     .enc_mbc_to_codepoint = enc_mbc_to_codepoint,
 425
 426     .local_defined = local_defined,
 427     .dvar_defined = dvar_defined,
 428
 429     .syntax_error_append = syntax_error_append,
 430     .raise = rb_raise,
 431     .syntax_error_new = syntax_error_new,
 432
 433     .errinfo = rb_errinfo,
 434     .set_errinfo = rb_set_errinfo,
 435     .exc_raise = rb_exc_raise,
 436     .make_exception = rb_make_exception,
 437
 438     .sized_xfree = ruby_sized_xfree,
 439     .sized_realloc_n = ruby_sized_realloc_n,
 440     .gc_guard = gc_guard,
 441     .gc_mark = rb_gc_mark,
 442
 443     .reg_compile = rb_reg_compile,
 444     .reg_check_preprocess = rb_reg_check_preprocess,
 445     .memcicmp = rb_memcicmp,
 446
 447     .compile_warn = rb_compile_warn,
 448     .compile_warning = rb_compile_warning,
 449     .bug = rb_bug,
 450     .fatal = rb_fatal,
 451     .verbose = ruby_verbose2,
 452     .errno_ptr = rb_errno_ptr2,
 453
 454     .make_backtrace = rb_make_backtrace,
 455
 456     .scan_hex = ruby_scan_hex,
 457     .scan_oct = ruby_scan_oct,
 458     .scan_digits = ruby_scan_digits,
 459     .strtod = ruby_strtod,
 460
 461     .rtest = rtest,
 462     .nil_p = nil_p,
 463     .qnil = Qnil,
 464     .qfalse = Qfalse,
 465     .eArgError = arg_error,
 466     .long2int = rb_long2int,
 467
 468     /* For Ripper */
 469     .static_id2sym = static_id2sym,
 470     .str_coderange_scan_restartable = str_coderange_scan_restartable,
 471 };
 472 #endif
 473
 474 enum lex_type {
 475     lex_type_str,
 476     lex_type_io,
 477     lex_type_array,
 478     lex_type_generic,
 479 };
 480
 481 struct ruby_parser {
 482     rb_parser_t *parser_params;
 483     enum lex_type type;
 484     union {
 485         struct lex_pointer_string lex_str;
 486         struct {
 487             VALUE file;
 488         } lex_io;
 489         struct {
 490             VALUE ary;
 491         } lex_array;
 492     } data;
 493 };
 494
 495 static void
 496 parser_mark(void *ptr)
 497 {
 498     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 499     rb_ruby_parser_mark(parser->parser_params);
 500
 501     switch (parser->type) {
 502       case lex_type_str:
 503         rb_gc_mark(parser->data.lex_str.str);
 504         break;
 505       case lex_type_io:
 506         rb_gc_mark(parser->data.lex_io.file);
 507         break;
 508       case lex_type_array:
 509         rb_gc_mark(parser->data.lex_array.ary);
 510         break;
 511       case lex_type_generic:
 512         /* noop. Caller of rb_parser_compile_generic should mark the objects. */
 513         break;
 514     }
 515 }
 516
 517 static void
 518 parser_free(void *ptr)
 519 {
 520     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 521     rb_ruby_parser_free(parser->parser_params);
 522     xfree(parser);
 523 }
 524
 525 static size_t
 526 parser_memsize(const void *ptr)
 527 {
 528     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 529     return rb_ruby_parser_memsize(parser->parser_params);
 530 }
 531
 532 static const rb_data_type_t ruby_parser_data_type = {
 533     "parser",
 534     {
 535         parser_mark,
 536         parser_free,
 537         parser_memsize,
 538     },
 539     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 540 };
 541
 542 #ifdef UNIVERSAL_PARSER
 543 const rb_parser_config_t *
 544 rb_ruby_parser_config(void)
 545 {
 546     return &rb_global_parser_config;
 547 }
 548
 549 rb_parser_t *
 550 rb_parser_params_new(void)
 551 {
 552     return rb_ruby_parser_new(&rb_global_parser_config);
 553 }
 554 #else
 555 rb_parser_t *
 556 rb_parser_params_new(void)
 557 {
 558     return rb_ruby_parser_new();
 559 }
 560 #endif /* UNIVERSAL_PARSER */
 561
 562 VALUE
 563 rb_parser_new(void)
 564 {
 565     struct ruby_parser *parser;
 566     rb_parser_t *parser_params;
 567
 568     /*
 569      * Create parser_params ahead of vparser because
 570      * rb_ruby_parser_new can run GC so if create vparser
 571      * first, parser_mark tries to mark not initialized parser_params.
 572      */
 573     parser_params = rb_parser_params_new();
 574     VALUE vparser = TypedData_Make_Struct(0, struct ruby_parser,
 575                                          &ruby_parser_data_type, parser);
 576     parser->parser_params = parser_params;
 577
 578     return vparser;
 579 }
 580
 581 void
 582 rb_parser_set_options(VALUE vparser, int print, int loop, int chomp, int split)
 583 {
 584     struct ruby_parser *parser;
 585
 586     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 587     rb_ruby_parser_set_options(parser->parser_params, print, loop, chomp, split);
 588 }
 589
 590 VALUE
 591 rb_parser_set_context(VALUE vparser, const struct rb_iseq_struct *base, int main)
 592 {
 593     struct ruby_parser *parser;
 594
 595     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 596     rb_ruby_parser_set_context(parser->parser_params, base, main);
 597     return vparser;
 598 }
 599
 600 void
 601 rb_parser_set_script_lines(VALUE vparser)
 602 {
 603     struct ruby_parser *parser;
 604
 605     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 606     rb_ruby_parser_set_script_lines(parser->parser_params);
 607 }
 608
 609 void
 610 rb_parser_error_tolerant(VALUE vparser)
 611 {
 612     struct ruby_parser *parser;
 613
 614     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 615     rb_ruby_parser_error_tolerant(parser->parser_params);
 616 }
 617
 618 void
 619 rb_parser_keep_tokens(VALUE vparser)
 620 {
 621     struct ruby_parser *parser;
 622
 623     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 624     rb_ruby_parser_keep_tokens(parser->parser_params);
 625 }
 626
 627 rb_parser_string_t *
 628 rb_parser_lex_get_str(struct parser_params *p, struct lex_pointer_string *ptr_str)
 629 {
 630     char *beg, *end, *start;
 631     long len;
 632     VALUE s = ptr_str->str;
 633
 634     beg = RSTRING_PTR(s);
 635     len = RSTRING_LEN(s);
 636     start = beg;
 637     if (ptr_str->ptr) {
 638         if (len == ptr_str->ptr) return 0;
 639         beg += ptr_str->ptr;
 640         len -= ptr_str->ptr;
 641     }
 642     end = memchr(beg, '\n', len);
 643     if (end) len = ++end - beg;
 644     ptr_str->ptr += len;
 645     return rb_str_to_parser_string(p, rb_str_subseq(s, beg - start, len));
 646 }
 647
 648 static rb_parser_string_t *
 649 lex_get_str(struct parser_params *p, rb_parser_input_data input, int line_count)
 650 {
 651     return rb_parser_lex_get_str(p, (struct lex_pointer_string *)input);
 652 }
 653
 654 static void parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines);
 655
 656 static rb_ast_t*
 657 parser_compile(rb_parser_t *p, rb_parser_lex_gets_func *gets, VALUE fname, rb_parser_input_data input, int line)
 658 {
 659     rb_ast_t *ast = rb_parser_compile(p, gets, fname, input, line);
 660     parser_aset_script_lines_for(fname, ast->body.script_lines);
 661     return ast;
 662 }
 663
 664 static rb_ast_t*
 665 parser_compile_string0(struct ruby_parser *parser, VALUE fname, VALUE s, int line)
 666 {
 667     VALUE str = rb_str_new_frozen(s);
 668
 669     parser->type = lex_type_str;
 670     parser->data.lex_str.str = str;
 671     parser->data.lex_str.ptr = 0;
 672
 673     return parser_compile(parser->parser_params, lex_get_str, fname, (rb_parser_input_data)&parser->data, line);
 674 }
 675
 676 static rb_encoding *
 677 must_be_ascii_compatible(VALUE s)
 678 {
 679     rb_encoding *enc = rb_enc_get(s);
 680     if (!rb_enc_asciicompat(enc)) {
 681         rb_raise(rb_eArgError, "invalid source encoding");
 682     }
 683     return enc;
 684 }
 685
 686 static rb_ast_t*
 687 parser_compile_string_path(struct ruby_parser *parser, VALUE f, VALUE s, int line)
 688 {
 689     must_be_ascii_compatible(s);
 690     return parser_compile_string0(parser, f, s, line);
 691 }
 692
 693 static rb_ast_t*
 694 parser_compile_string(struct ruby_parser *parser, const char *f, VALUE s, int line)
 695 {
 696     return parser_compile_string_path(parser, rb_filesystem_str_new_cstr(f), s, line);
 697 }
 698
 699 VALUE rb_io_gets_internal(VALUE io);
 700
 701 static rb_parser_string_t *
 702 lex_io_gets(struct parser_params *p, rb_parser_input_data input, int line_count)
 703 {
 704     VALUE io = (VALUE)input;
 705     VALUE line = rb_io_gets_internal(io);
 706     if (NIL_P(line)) return 0;
 707     return rb_str_to_parser_string(p, line);
 708 }
 709
 710 static rb_parser_string_t *
 711 lex_gets_array(struct parser_params *p, rb_parser_input_data data, int index)
 712 {
 713     VALUE array = (VALUE)data;
 714     VALUE str = rb_ary_entry(array, index);
 715     if (!NIL_P(str)) {
 716         StringValue(str);
 717         if (!rb_enc_asciicompat(rb_enc_get(str))) {
 718             rb_raise(rb_eArgError, "invalid source encoding");
 719         }
 720         return rb_str_to_parser_string(p, str);
 721     }
 722     else {
 723         return 0;
 724     }
 725 }
 726
 727 static rb_ast_t*
 728 parser_compile_file_path(struct ruby_parser *parser, VALUE fname, VALUE file, int start)
 729 {
 730     parser->type = lex_type_io;
 731     parser->data.lex_io.file = file;
 732
 733     return parser_compile(parser->parser_params, lex_io_gets, fname, (rb_parser_input_data)file, start);
 734 }
 735
 736 static rb_ast_t*
 737 parser_compile_array(struct ruby_parser *parser, VALUE fname, VALUE array, int start)
 738 {
 739     parser->type = lex_type_array;
 740     parser->data.lex_array.ary = array;
 741
 742     return parser_compile(parser->parser_params, lex_gets_array, fname, (rb_parser_input_data)array, start);
 743 }
 744
 745 static rb_ast_t*
 746 parser_compile_generic(struct ruby_parser *parser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 747 {
 748     parser->type = lex_type_generic;
 749
 750     return parser_compile(parser->parser_params, lex_gets, fname, (rb_parser_input_data)input, start);
 751 }
 752
 753 static void
 754 ast_free(void *ptr)
 755 {
 756     rb_ast_t *ast = (rb_ast_t *)ptr;
 757     rb_ast_free(ast);
 758 }
 759
 760 static const rb_data_type_t ast_data_type = {
 761     "AST",
 762     {
 763         NULL,
 764         ast_free,
 765         NULL, // No dsize() because this object does not appear in ObjectSpace.
 766     },
 767     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 768 };
 769
 770 static VALUE
 771 ast_alloc(void)
 772 {
 773     return TypedData_Wrap_Struct(0, &ast_data_type, NULL);
 774 }
 775
 776 VALUE
 777 rb_parser_compile_file_path(VALUE vparser, VALUE fname, VALUE file, int start)
 778 {
 779     struct ruby_parser *parser;
 780     VALUE ast_value = ast_alloc();
 781
 782     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 783     DATA_PTR(ast_value) = parser_compile_file_path(parser, fname, file, start);
 784     RB_GC_GUARD(vparser);
 785
 786     return ast_value;
 787 }
 788
 789 VALUE
 790 rb_parser_compile_array(VALUE vparser, VALUE fname, VALUE array, int start)
 791 {
 792     struct ruby_parser *parser;
 793     VALUE ast_value = ast_alloc();
 794
 795     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 796     DATA_PTR(ast_value) = parser_compile_array(parser, fname, array, start);
 797     RB_GC_GUARD(vparser);
 798
 799     return ast_value;
 800 }
 801
 802 VALUE
 803 rb_parser_compile_generic(VALUE vparser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 804 {
 805     struct ruby_parser *parser;
 806     VALUE ast_value = ast_alloc();
 807
 808     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 809     DATA_PTR(ast_value) = parser_compile_generic(parser, lex_gets, fname, input, start);
 810     RB_GC_GUARD(vparser);
 811
 812     return ast_value;
 813 }
 814
 815 VALUE
 816 rb_parser_compile_string(VALUE vparser, const char *f, VALUE s, int line)
 817 {
 818     struct ruby_parser *parser;
 819     VALUE ast_value = ast_alloc();
 820
 821     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 822     DATA_PTR(ast_value) = parser_compile_string(parser, f, s, line);
 823     RB_GC_GUARD(vparser);
 824
 825     return ast_value;
 826 }
 827
 828 VALUE
 829 rb_parser_compile_string_path(VALUE vparser, VALUE f, VALUE s, int line)
 830 {
 831     struct ruby_parser *parser;
 832     VALUE ast_value = ast_alloc();
 833
 834     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 835     DATA_PTR(ast_value) = parser_compile_string_path(parser, f, s, line);
 836     RB_GC_GUARD(vparser);
 837
 838     return ast_value;
 839 }
 840
 841 VALUE
 842 rb_parser_encoding(VALUE vparser)
 843 {
 844     struct ruby_parser *parser;
 845
 846     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 847     return rb_enc_from_encoding(rb_ruby_parser_encoding(parser->parser_params));
 848 }
 849
 850 VALUE
 851 rb_parser_end_seen_p(VALUE vparser)
 852 {
 853     struct ruby_parser *parser;
 854
 855     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 856     return RBOOL(rb_ruby_parser_end_seen_p(parser->parser_params));
 857 }
 858
 859 VALUE
 860 rb_parser_set_yydebug(VALUE vparser, VALUE flag)
 861 {
 862     struct ruby_parser *parser;
 863
 864     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 865     rb_ruby_parser_set_yydebug(parser->parser_params, RTEST(flag));
 866     return flag;
 867 }
 868
 869 void
 870 rb_set_script_lines_for(VALUE vparser, VALUE path)
 871 {
 872     struct ruby_parser *parser;
 873     VALUE hash;
 874     ID script_lines;
 875     CONST_ID(script_lines, "SCRIPT_LINES__");
 876     if (!rb_const_defined_at(rb_cObject, script_lines)) return;
 877     hash = rb_const_get_at(rb_cObject, script_lines);
 878     if (RB_TYPE_P(hash, T_HASH)) {
 879         rb_hash_aset(hash, path, Qtrue);
 880         TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 881         rb_ruby_parser_set_script_lines(parser->parser_params);
 882     }
 883 }
 884
 885 VALUE
 886 rb_parser_build_script_lines_from(rb_parser_ary_t *lines)
 887 {
 888     int i;
 889     if (!lines) return Qnil;
 890     if (lines->data_type != PARSER_ARY_DATA_SCRIPT_LINE) {
 891         rb_bug("unexpected rb_parser_ary_data_type (%d) for script lines", lines->data_type);
 892     }
 893     VALUE script_lines = rb_ary_new_capa(lines->len);
 894     for (i = 0; i < lines->len; i++) {
 895         rb_parser_string_t *str = (rb_parser_string_t *)lines->data[i];
 896         rb_ary_push(script_lines, rb_enc_str_new(str->ptr, str->len, str->enc));
 897     }
 898     return script_lines;
 899 }
 900
 901 VALUE
 902 rb_str_new_parser_string(rb_parser_string_t *str)
 903 {
 904     VALUE string = rb_enc_literal_str(str->ptr, str->len, str->enc);
 905     rb_enc_str_coderange(string);
 906     return string;
 907 }
 908
 909 VALUE
 910 rb_str_new_mutable_parser_string(rb_parser_string_t *str)
 911 {
 912     return rb_enc_str_new(str->ptr, str->len, str->enc);
 913 }
 914
 915 static VALUE
 916 negative_numeric(VALUE val)
 917 {
 918     if (FIXNUM_P(val)) {
 919         return LONG2FIX(-FIX2LONG(val));
 920     }
 921     if (SPECIAL_CONST_P(val)) {
 922 #if USE_FLONUM
 923         if (FLONUM_P(val)) {
 924             return DBL2NUM(-RFLOAT_VALUE(val));
 925         }
 926 #endif
 927         goto unknown;
 928     }
 929     switch (BUILTIN_TYPE(val)) {
 930       case T_BIGNUM:
 931         BIGNUM_NEGATE(val);
 932         val = rb_big_norm(val);
 933         break;
 934       case T_RATIONAL:
 935         RATIONAL_SET_NUM(val, negative_numeric(RRATIONAL(val)->num));
 936         break;
 937       case T_COMPLEX:
 938         RCOMPLEX_SET_REAL(val, negative_numeric(RCOMPLEX(val)->real));
 939         RCOMPLEX_SET_IMAG(val, negative_numeric(RCOMPLEX(val)->imag));
 940         break;
 941       case T_FLOAT:
 942         val = DBL2NUM(-RFLOAT_VALUE(val));
 943         break;
 944       unknown:
 945       default:
 946         rb_bug("unknown literal type (%s) passed to negative_numeric",
 947                rb_builtin_class_name(val));
 948         break;
 949     }
 950     return val;
 951 }
 952
 953 static VALUE
 954 integer_value(const char *val, int base)
 955 {
 956     return rb_cstr_to_inum(val, base, FALSE);
 957 }
 958
 959 static VALUE
 960 rational_value(const char *node_val, int base, int seen_point)
 961 {
 962     VALUE lit;
 963     char* val = strdup(node_val);
 964     if (seen_point > 0) {
 965         int len = (int)(strlen(val));
 966         char *point = &val[seen_point];
 967         size_t fraclen = len-seen_point-1;
 968         memmove(point, point+1, fraclen+1);
 969
 970         lit = rb_rational_new(integer_value(val, base), rb_int_positive_pow(10, fraclen));
 971     }
 972     else {
 973         lit = rb_rational_raw1(integer_value(val, base));
 974     }
 975
 976     free(val);
 977
 978     return lit;
 979 }
 980
 981 VALUE
 982 rb_node_integer_literal_val(const NODE *n)
 983 {
 984     const rb_node_integer_t *node = RNODE_INTEGER(n);
 985     VALUE val = integer_value(node->val, node->base);
 986     if (node->minus) {
 987         val = negative_numeric(val);
 988     }
 989     return val;
 990 }
 991
 992 VALUE
 993 rb_node_float_literal_val(const NODE *n)
 994 {
 995     const rb_node_float_t *node = RNODE_FLOAT(n);
 996     double d = strtod(node->val, 0);
 997     if (node->minus) {
 998         d = -d;
 999     }
1000     VALUE val = DBL2NUM(d);
1001     return val;
1002 }
1003
1004 VALUE
1005 rb_node_rational_literal_val(const NODE *n)
1006 {
1007     VALUE lit;
1008     const rb_node_rational_t *node = RNODE_RATIONAL(n);
1009
1010     lit = rational_value(node->val, node->base, node->seen_point);
1011
1012     if (node->minus) {
1013         lit = negative_numeric(lit);
1014     }
1015
1016     return lit;
1017 }
1018
1019 VALUE
1020 rb_node_imaginary_literal_val(const NODE *n)
1021 {
1022     VALUE lit;
1023     const rb_node_imaginary_t *node = RNODE_IMAGINARY(n);
1024
1025     enum rb_numeric_type type = node->type;
1026
1027     switch (type) {
1028       case integer_literal:
1029         lit = integer_value(node->val, node->base);
1030         break;
1031       case float_literal:{
1032         double d = strtod(node->val, 0);
1033         lit = DBL2NUM(d);
1034         break;
1035       }
1036       case rational_literal:
1037         lit = rational_value(node->val, node->base, node->seen_point);
1038         break;
1039       default:
1040         rb_bug("unreachable");
1041     }
1042
1043     lit = rb_complex_raw(INT2FIX(0), lit);
1044
1045     if (node->minus) {
1046         lit = negative_numeric(lit);
1047     }
1048     return lit;
1049 }
1050
1051 VALUE
1052 rb_node_str_string_val(const NODE *node)
1053 {
1054     rb_parser_string_t *str = RNODE_STR(node)->string;
1055     return rb_str_new_parser_string(str);
1056 }
1057
1058 VALUE
1059 rb_node_sym_string_val(const NODE *node)
1060 {
1061     rb_parser_string_t *str = RNODE_SYM(node)->string;
1062     return ID2SYM(rb_intern3(str->ptr, str->len, str->enc));
1063 }
1064
1065 VALUE
1066 rb_node_dstr_string_val(const NODE *node)
1067 {
1068     rb_parser_string_t *str = RNODE_DSTR(node)->string;
1069     return str ? rb_str_new_parser_string(str) : Qnil;
1070 }
1071
1072 VALUE
1073 rb_node_dregx_string_val(const NODE *node)
1074 {
1075     rb_parser_string_t *str = RNODE_DREGX(node)->string;
1076     return rb_str_new_parser_string(str);
1077 }
1078
1079 VALUE
1080 rb_node_regx_string_val(const NODE *node)
1081 {
1082     rb_node_regx_t *node_reg = RNODE_REGX(node);
1083     rb_parser_string_t *string = node_reg->string;
1084     VALUE str = rb_enc_str_new(string->ptr, string->len, string->enc);
1085
1086     return rb_reg_compile(str, node_reg->options, NULL, 0);
1087 }
1088
1089 VALUE
1090 rb_node_line_lineno_val(const NODE *node)
1091 {
1092     return INT2FIX(node->nd_loc.beg_pos.lineno);
1093 }
1094
1095 VALUE
1096 rb_node_file_path_val(const NODE *node)
1097 {
1098     return rb_str_new_parser_string(RNODE_FILE(node)->path);
1099 }
1100
1101 VALUE
1102 rb_node_encoding_val(const NODE *node)
1103 {
1104     return rb_enc_from_encoding(RNODE_ENCODING(node)->enc);
1105 }
1106
1107 static void
1108 parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines)
1109 {
1110     VALUE hash, script_lines;
1111     ID script_lines_id;
1112     if (NIL_P(path) || !lines) return;
1113     CONST_ID(script_lines_id, "SCRIPT_LINES__");
1114     if (!rb_const_defined_at(rb_cObject, script_lines_id)) return;
1115     hash = rb_const_get_at(rb_cObject, script_lines_id);
1116     if (!RB_TYPE_P(hash, T_HASH)) return;
1117     if (rb_hash_lookup(hash, path) == Qnil) return;
1118     script_lines = rb_parser_build_script_lines_from(lines);
1119     rb_hash_aset(hash, path, script_lines);
1120 }
1121
1122 VALUE
1123 rb_ruby_ast_new(const NODE *const root)
1124 {
1125     rb_ast_t *ast;
1126     VALUE ast_value = TypedData_Make_Struct(0, rb_ast_t, &ast_data_type, ast);
1127 #ifdef UNIVERSAL_PARSER
1128     ast->config = &rb_global_parser_config;
1129 #endif
1130     ast->body = (rb_ast_body_t){
1131         .root = root,
1132         .frozen_string_literal = -1,
1133         .coverage_enabled = -1,
1134         .script_lines = NULL,
1135         .line_count = 0,
1136     };
1137     return ast_value;
1138 }
1139
1140 rb_ast_t *
1141 rb_ruby_ast_data_get(VALUE ast_value)
1142 {
1143     rb_ast_t *ast;
1144     if (NIL_P(ast_value)) return NULL;
1145     TypedData_Get_Struct(ast_value, rb_ast_t, &ast_data_type, ast);
1146     return ast;
1147 }