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_cat = rb_str_cat,
 375     .str_resize = rb_str_resize,
 376     .str_new = rb_str_new,
 377     .str_new_cstr = rb_str_new_cstr,
 378     .str_to_interned_str = rb_str_to_interned_str,
 379     .is_ascii_string = is_ascii_string2,
 380     .enc_str_new = enc_str_new,
 381     .str_vcatf = rb_str_vcatf,
 382     .rb_sprintf = rb_sprintf,
 383     .rstring_ptr = RSTRING_PTR,
 384     .rstring_end = RSTRING_END,
 385     .rstring_len = RSTRING_LEN,
 386     .obj_as_string = rb_obj_as_string,
 387
 388     .int2num = rb_int2num_inline,
 389
 390     .stderr_tty_p = rb_stderr_tty_p,
 391     .write_error_str = rb_write_error_str,
 392     .io_write = rb_io_write,
 393     .io_flush = rb_io_flush,
 394     .io_puts = rb_io_puts,
 395
 396     .debug_output_stdout = rb_ractor_stdout,
 397     .debug_output_stderr = rb_ractor_stderr,
 398
 399     .is_usascii_enc = is_usascii_enc,
 400     .enc_isalnum = enc_isalnum,
 401     .enc_precise_mbclen = enc_precise_mbclen,
 402     .mbclen_charfound_p = mbclen_charfound_p,
 403     .mbclen_charfound_len = mbclen_charfound_len,
 404     .enc_name = enc_name,
 405     .enc_prev_char = enc_prev_char,
 406     .enc_get = enc_get,
 407     .enc_asciicompat = enc_asciicompat,
 408     .utf8_encoding = utf8_encoding,
 409     .enc_associate = enc_associate,
 410     .ascii8bit_encoding = ascii8bit_encoding,
 411     .enc_codelen = enc_codelen,
 412     .enc_mbcput = enc_mbcput,
 413     .enc_find_index = rb_enc_find_index,
 414     .enc_from_index = enc_from_index,
 415     .enc_isspace = enc_isspace,
 416     .enc_coderange_7bit = ENC_CODERANGE_7BIT,
 417     .enc_coderange_unknown = ENC_CODERANGE_UNKNOWN,
 418     .usascii_encoding = usascii_encoding,
 419     .enc_mbminlen = enc_mbminlen,
 420     .enc_isascii = enc_isascii,
 421     .enc_mbc_to_codepoint = enc_mbc_to_codepoint,
 422
 423     .local_defined = local_defined,
 424     .dvar_defined = dvar_defined,
 425
 426     .syntax_error_append = syntax_error_append,
 427     .raise = rb_raise,
 428     .syntax_error_new = syntax_error_new,
 429
 430     .errinfo = rb_errinfo,
 431     .set_errinfo = rb_set_errinfo,
 432     .exc_raise = rb_exc_raise,
 433     .make_exception = rb_make_exception,
 434
 435     .sized_xfree = ruby_sized_xfree,
 436     .sized_realloc_n = ruby_sized_realloc_n,
 437     .gc_guard = gc_guard,
 438     .gc_mark = rb_gc_mark,
 439
 440     .reg_compile = rb_reg_compile,
 441     .reg_check_preprocess = rb_reg_check_preprocess,
 442     .memcicmp = rb_memcicmp,
 443
 444     .compile_warn = rb_compile_warn,
 445     .compile_warning = rb_compile_warning,
 446     .bug = rb_bug,
 447     .fatal = rb_fatal,
 448     .verbose = ruby_verbose2,
 449     .errno_ptr = rb_errno_ptr2,
 450
 451     .make_backtrace = rb_make_backtrace,
 452
 453     .scan_hex = ruby_scan_hex,
 454     .scan_oct = ruby_scan_oct,
 455     .scan_digits = ruby_scan_digits,
 456     .strtod = ruby_strtod,
 457
 458     .rtest = rtest,
 459     .nil_p = nil_p,
 460     .qnil = Qnil,
 461     .qfalse = Qfalse,
 462     .eArgError = arg_error,
 463     .long2int = rb_long2int,
 464
 465     /* For Ripper */
 466     .static_id2sym = static_id2sym,
 467     .str_coderange_scan_restartable = str_coderange_scan_restartable,
 468 };
 469 #endif
 470
 471 enum lex_type {
 472     lex_type_str,
 473     lex_type_io,
 474     lex_type_array,
 475     lex_type_generic,
 476 };
 477
 478 struct ruby_parser {
 479     rb_parser_t *parser_params;
 480     enum lex_type type;
 481     union {
 482         struct lex_pointer_string lex_str;
 483         struct {
 484             VALUE file;
 485         } lex_io;
 486         struct {
 487             VALUE ary;
 488         } lex_array;
 489     } data;
 490 };
 491
 492 static void
 493 parser_mark(void *ptr)
 494 {
 495     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 496     rb_ruby_parser_mark(parser->parser_params);
 497
 498     switch (parser->type) {
 499       case lex_type_str:
 500         rb_gc_mark(parser->data.lex_str.str);
 501         break;
 502       case lex_type_io:
 503         rb_gc_mark(parser->data.lex_io.file);
 504         break;
 505       case lex_type_array:
 506         rb_gc_mark(parser->data.lex_array.ary);
 507         break;
 508       case lex_type_generic:
 509         /* noop. Caller of rb_parser_compile_generic should mark the objects. */
 510         break;
 511     }
 512 }
 513
 514 static void
 515 parser_free(void *ptr)
 516 {
 517     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 518     rb_ruby_parser_free(parser->parser_params);
 519     xfree(parser);
 520 }
 521
 522 static size_t
 523 parser_memsize(const void *ptr)
 524 {
 525     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 526     return rb_ruby_parser_memsize(parser->parser_params);
 527 }
 528
 529 static const rb_data_type_t ruby_parser_data_type = {
 530     "parser",
 531     {
 532         parser_mark,
 533         parser_free,
 534         parser_memsize,
 535     },
 536     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 537 };
 538
 539 #ifdef UNIVERSAL_PARSER
 540 const rb_parser_config_t *
 541 rb_ruby_parser_config(void)
 542 {
 543     return &rb_global_parser_config;
 544 }
 545
 546 rb_parser_t *
 547 rb_parser_params_new(void)
 548 {
 549     return rb_ruby_parser_new(&rb_global_parser_config);
 550 }
 551 #else
 552 rb_parser_t *
 553 rb_parser_params_new(void)
 554 {
 555     return rb_ruby_parser_new();
 556 }
 557 #endif /* UNIVERSAL_PARSER */
 558
 559 VALUE
 560 rb_parser_new(void)
 561 {
 562     struct ruby_parser *parser;
 563     rb_parser_t *parser_params;
 564
 565     /*
 566      * Create parser_params ahead of vparser because
 567      * rb_ruby_parser_new can run GC so if create vparser
 568      * first, parser_mark tries to mark not initialized parser_params.
 569      */
 570     parser_params = rb_parser_params_new();
 571     VALUE vparser = TypedData_Make_Struct(0, struct ruby_parser,
 572                                          &ruby_parser_data_type, parser);
 573     parser->parser_params = parser_params;
 574
 575     return vparser;
 576 }
 577
 578 void
 579 rb_parser_set_options(VALUE vparser, int print, int loop, int chomp, int split)
 580 {
 581     struct ruby_parser *parser;
 582
 583     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 584     rb_ruby_parser_set_options(parser->parser_params, print, loop, chomp, split);
 585 }
 586
 587 VALUE
 588 rb_parser_set_context(VALUE vparser, const struct rb_iseq_struct *base, int main)
 589 {
 590     struct ruby_parser *parser;
 591
 592     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 593     rb_ruby_parser_set_context(parser->parser_params, base, main);
 594     return vparser;
 595 }
 596
 597 void
 598 rb_parser_set_script_lines(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_set_script_lines(parser->parser_params);
 604 }
 605
 606 void
 607 rb_parser_error_tolerant(VALUE vparser)
 608 {
 609     struct ruby_parser *parser;
 610
 611     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 612     rb_ruby_parser_error_tolerant(parser->parser_params);
 613 }
 614
 615 void
 616 rb_parser_keep_tokens(VALUE vparser)
 617 {
 618     struct ruby_parser *parser;
 619
 620     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 621     rb_ruby_parser_keep_tokens(parser->parser_params);
 622 }
 623
 624 rb_parser_string_t *
 625 rb_parser_lex_get_str(struct parser_params *p, struct lex_pointer_string *ptr_str)
 626 {
 627     char *beg, *end, *start;
 628     long len;
 629     VALUE s = ptr_str->str;
 630
 631     beg = RSTRING_PTR(s);
 632     len = RSTRING_LEN(s);
 633     start = beg;
 634     if (ptr_str->ptr) {
 635         if (len == ptr_str->ptr) return 0;
 636         beg += ptr_str->ptr;
 637         len -= ptr_str->ptr;
 638     }
 639     end = memchr(beg, '\n', len);
 640     if (end) len = ++end - beg;
 641     ptr_str->ptr += len;
 642     return rb_str_to_parser_string(p, rb_str_subseq(s, beg - start, len));
 643 }
 644
 645 static rb_parser_string_t *
 646 lex_get_str(struct parser_params *p, rb_parser_input_data input, int line_count)
 647 {
 648     return rb_parser_lex_get_str(p, (struct lex_pointer_string *)input);
 649 }
 650
 651 static void parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines);
 652
 653 static rb_ast_t*
 654 parser_compile(rb_parser_t *p, rb_parser_lex_gets_func *gets, VALUE fname, rb_parser_input_data input, int line)
 655 {
 656     rb_ast_t *ast = rb_parser_compile(p, gets, fname, input, line);
 657     parser_aset_script_lines_for(fname, ast->body.script_lines);
 658     return ast;
 659 }
 660
 661 static rb_ast_t*
 662 parser_compile_string0(struct ruby_parser *parser, VALUE fname, VALUE s, int line)
 663 {
 664     VALUE str = rb_str_new_frozen(s);
 665
 666     parser->type = lex_type_str;
 667     parser->data.lex_str.str = str;
 668     parser->data.lex_str.ptr = 0;
 669
 670     return parser_compile(parser->parser_params, lex_get_str, fname, (rb_parser_input_data)&parser->data, line);
 671 }
 672
 673 static rb_encoding *
 674 must_be_ascii_compatible(VALUE s)
 675 {
 676     rb_encoding *enc = rb_enc_get(s);
 677     if (!rb_enc_asciicompat(enc)) {
 678         rb_raise(rb_eArgError, "invalid source encoding");
 679     }
 680     return enc;
 681 }
 682
 683 static rb_ast_t*
 684 parser_compile_string_path(struct ruby_parser *parser, VALUE f, VALUE s, int line)
 685 {
 686     must_be_ascii_compatible(s);
 687     return parser_compile_string0(parser, f, s, line);
 688 }
 689
 690 static rb_ast_t*
 691 parser_compile_string(struct ruby_parser *parser, const char *f, VALUE s, int line)
 692 {
 693     return parser_compile_string_path(parser, rb_filesystem_str_new_cstr(f), s, line);
 694 }
 695
 696 VALUE rb_io_gets_internal(VALUE io);
 697
 698 static rb_parser_string_t *
 699 lex_io_gets(struct parser_params *p, rb_parser_input_data input, int line_count)
 700 {
 701     VALUE io = (VALUE)input;
 702     VALUE line = rb_io_gets_internal(io);
 703     if (NIL_P(line)) return 0;
 704     return rb_str_to_parser_string(p, line);
 705 }
 706
 707 static rb_parser_string_t *
 708 lex_gets_array(struct parser_params *p, rb_parser_input_data data, int index)
 709 {
 710     VALUE array = (VALUE)data;
 711     VALUE str = rb_ary_entry(array, index);
 712     if (!NIL_P(str)) {
 713         StringValue(str);
 714         if (!rb_enc_asciicompat(rb_enc_get(str))) {
 715             rb_raise(rb_eArgError, "invalid source encoding");
 716         }
 717         return rb_str_to_parser_string(p, str);
 718     }
 719     else {
 720         return 0;
 721     }
 722 }
 723
 724 static rb_ast_t*
 725 parser_compile_file_path(struct ruby_parser *parser, VALUE fname, VALUE file, int start)
 726 {
 727     parser->type = lex_type_io;
 728     parser->data.lex_io.file = file;
 729
 730     return parser_compile(parser->parser_params, lex_io_gets, fname, (rb_parser_input_data)file, start);
 731 }
 732
 733 static rb_ast_t*
 734 parser_compile_array(struct ruby_parser *parser, VALUE fname, VALUE array, int start)
 735 {
 736     parser->type = lex_type_array;
 737     parser->data.lex_array.ary = array;
 738
 739     return parser_compile(parser->parser_params, lex_gets_array, fname, (rb_parser_input_data)array, start);
 740 }
 741
 742 static rb_ast_t*
 743 parser_compile_generic(struct ruby_parser *parser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 744 {
 745     parser->type = lex_type_generic;
 746
 747     return parser_compile(parser->parser_params, lex_gets, fname, (rb_parser_input_data)input, start);
 748 }
 749
 750 static void
 751 ast_free(void *ptr)
 752 {
 753     rb_ast_t *ast = (rb_ast_t *)ptr;
 754     rb_ast_free(ast);
 755 }
 756
 757 static const rb_data_type_t ast_data_type = {
 758     "AST",
 759     {
 760         NULL,
 761         ast_free,
 762         NULL, // No dsize() because this object does not appear in ObjectSpace.
 763     },
 764     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 765 };
 766
 767 static VALUE
 768 ast_alloc(void)
 769 {
 770     return TypedData_Wrap_Struct(0, &ast_data_type, NULL);
 771 }
 772
 773 VALUE
 774 rb_parser_compile_file_path(VALUE vparser, VALUE fname, VALUE file, int start)
 775 {
 776     struct ruby_parser *parser;
 777     VALUE ast_value = ast_alloc();
 778
 779     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 780     DATA_PTR(ast_value) = parser_compile_file_path(parser, fname, file, start);
 781     RB_GC_GUARD(vparser);
 782
 783     return ast_value;
 784 }
 785
 786 VALUE
 787 rb_parser_compile_array(VALUE vparser, VALUE fname, VALUE array, int start)
 788 {
 789     struct ruby_parser *parser;
 790     VALUE ast_value = ast_alloc();
 791
 792     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 793     DATA_PTR(ast_value) = parser_compile_array(parser, fname, array, start);
 794     RB_GC_GUARD(vparser);
 795
 796     return ast_value;
 797 }
 798
 799 VALUE
 800 rb_parser_compile_generic(VALUE vparser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 801 {
 802     struct ruby_parser *parser;
 803     VALUE ast_value = ast_alloc();
 804
 805     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 806     DATA_PTR(ast_value) = parser_compile_generic(parser, lex_gets, fname, input, start);
 807     RB_GC_GUARD(vparser);
 808
 809     return ast_value;
 810 }
 811
 812 VALUE
 813 rb_parser_compile_string(VALUE vparser, const char *f, VALUE s, int line)
 814 {
 815     struct ruby_parser *parser;
 816     VALUE ast_value = ast_alloc();
 817
 818     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 819     DATA_PTR(ast_value) = parser_compile_string(parser, f, s, line);
 820     RB_GC_GUARD(vparser);
 821
 822     return ast_value;
 823 }
 824
 825 VALUE
 826 rb_parser_compile_string_path(VALUE vparser, VALUE f, VALUE s, int line)
 827 {
 828     struct ruby_parser *parser;
 829     VALUE ast_value = ast_alloc();
 830
 831     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 832     DATA_PTR(ast_value) = parser_compile_string_path(parser, f, s, line);
 833     RB_GC_GUARD(vparser);
 834
 835     return ast_value;
 836 }
 837
 838 VALUE
 839 rb_parser_encoding(VALUE vparser)
 840 {
 841     struct ruby_parser *parser;
 842
 843     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 844     return rb_enc_from_encoding(rb_ruby_parser_encoding(parser->parser_params));
 845 }
 846
 847 VALUE
 848 rb_parser_end_seen_p(VALUE vparser)
 849 {
 850     struct ruby_parser *parser;
 851
 852     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 853     return RBOOL(rb_ruby_parser_end_seen_p(parser->parser_params));
 854 }
 855
 856 VALUE
 857 rb_parser_set_yydebug(VALUE vparser, VALUE flag)
 858 {
 859     struct ruby_parser *parser;
 860
 861     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 862     rb_ruby_parser_set_yydebug(parser->parser_params, RTEST(flag));
 863     return flag;
 864 }
 865
 866 void
 867 rb_set_script_lines_for(VALUE vparser, VALUE path)
 868 {
 869     struct ruby_parser *parser;
 870     VALUE hash;
 871     ID script_lines;
 872     CONST_ID(script_lines, "SCRIPT_LINES__");
 873     if (!rb_const_defined_at(rb_cObject, script_lines)) return;
 874     hash = rb_const_get_at(rb_cObject, script_lines);
 875     if (RB_TYPE_P(hash, T_HASH)) {
 876         rb_hash_aset(hash, path, Qtrue);
 877         TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 878         rb_ruby_parser_set_script_lines(parser->parser_params);
 879     }
 880 }
 881
 882 VALUE
 883 rb_parser_build_script_lines_from(rb_parser_ary_t *lines)
 884 {
 885     int i;
 886     if (!lines) return Qnil;
 887     if (lines->data_type != PARSER_ARY_DATA_SCRIPT_LINE) {
 888         rb_bug("unexpected rb_parser_ary_data_type (%d) for script lines", lines->data_type);
 889     }
 890     VALUE script_lines = rb_ary_new_capa(lines->len);
 891     for (i = 0; i < lines->len; i++) {
 892         rb_parser_string_t *str = (rb_parser_string_t *)lines->data[i];
 893         rb_ary_push(script_lines, rb_enc_str_new(str->ptr, str->len, str->enc));
 894     }
 895     return script_lines;
 896 }
 897
 898 VALUE
 899 rb_str_new_parser_string(rb_parser_string_t *str)
 900 {
 901     VALUE string = rb_enc_literal_str(str->ptr, str->len, str->enc);
 902     rb_enc_str_coderange(string);
 903     return string;
 904 }
 905
 906 VALUE
 907 rb_str_new_mutable_parser_string(rb_parser_string_t *str)
 908 {
 909     return rb_enc_str_new(str->ptr, str->len, str->enc);
 910 }
 911
 912 static VALUE
 913 negative_numeric(VALUE val)
 914 {
 915     if (FIXNUM_P(val)) {
 916         return LONG2FIX(-FIX2LONG(val));
 917     }
 918     if (SPECIAL_CONST_P(val)) {
 919 #if USE_FLONUM
 920         if (FLONUM_P(val)) {
 921             return DBL2NUM(-RFLOAT_VALUE(val));
 922         }
 923 #endif
 924         goto unknown;
 925     }
 926     switch (BUILTIN_TYPE(val)) {
 927       case T_BIGNUM:
 928         BIGNUM_NEGATE(val);
 929         val = rb_big_norm(val);
 930         break;
 931       case T_RATIONAL:
 932         RATIONAL_SET_NUM(val, negative_numeric(RRATIONAL(val)->num));
 933         break;
 934       case T_COMPLEX:
 935         RCOMPLEX_SET_REAL(val, negative_numeric(RCOMPLEX(val)->real));
 936         RCOMPLEX_SET_IMAG(val, negative_numeric(RCOMPLEX(val)->imag));
 937         break;
 938       case T_FLOAT:
 939         val = DBL2NUM(-RFLOAT_VALUE(val));
 940         break;
 941       unknown:
 942       default:
 943         rb_bug("unknown literal type (%s) passed to negative_numeric",
 944                rb_builtin_class_name(val));
 945         break;
 946     }
 947     return val;
 948 }
 949
 950 static VALUE
 951 integer_value(const char *val, int base)
 952 {
 953     return rb_cstr_to_inum(val, base, FALSE);
 954 }
 955
 956 static VALUE
 957 rational_value(const char *node_val, int base, int seen_point)
 958 {
 959     VALUE lit;
 960     char* val = strdup(node_val);
 961     if (seen_point > 0) {
 962         int len = (int)(strlen(val));
 963         char *point = &val[seen_point];
 964         size_t fraclen = len-seen_point-1;
 965         memmove(point, point+1, fraclen+1);
 966
 967         lit = rb_rational_new(integer_value(val, base), rb_int_positive_pow(10, fraclen));
 968     }
 969     else {
 970         lit = rb_rational_raw1(integer_value(val, base));
 971     }
 972
 973     free(val);
 974
 975     return lit;
 976 }
 977
 978 VALUE
 979 rb_node_integer_literal_val(const NODE *n)
 980 {
 981     const rb_node_integer_t *node = RNODE_INTEGER(n);
 982     VALUE val = integer_value(node->val, node->base);
 983     if (node->minus) {
 984         val = negative_numeric(val);
 985     }
 986     return val;
 987 }
 988
 989 VALUE
 990 rb_node_float_literal_val(const NODE *n)
 991 {
 992     const rb_node_float_t *node = RNODE_FLOAT(n);
 993     double d = strtod(node->val, 0);
 994     if (node->minus) {
 995         d = -d;
 996     }
 997     VALUE val = DBL2NUM(d);
 998     return val;
 999 }
1000
1001 VALUE
1002 rb_node_rational_literal_val(const NODE *n)
1003 {
1004     VALUE lit;
1005     const rb_node_rational_t *node = RNODE_RATIONAL(n);
1006
1007     lit = rational_value(node->val, node->base, node->seen_point);
1008
1009     if (node->minus) {
1010         lit = negative_numeric(lit);
1011     }
1012
1013     return lit;
1014 }
1015
1016 VALUE
1017 rb_node_imaginary_literal_val(const NODE *n)
1018 {
1019     VALUE lit;
1020     const rb_node_imaginary_t *node = RNODE_IMAGINARY(n);
1021
1022     enum rb_numeric_type type = node->type;
1023
1024     switch (type) {
1025       case integer_literal:
1026         lit = integer_value(node->val, node->base);
1027         break;
1028       case float_literal:{
1029         double d = strtod(node->val, 0);
1030         lit = DBL2NUM(d);
1031         break;
1032       }
1033       case rational_literal:
1034         lit = rational_value(node->val, node->base, node->seen_point);
1035         break;
1036       default:
1037         rb_bug("unreachable");
1038     }
1039
1040     lit = rb_complex_raw(INT2FIX(0), lit);
1041
1042     if (node->minus) {
1043         lit = negative_numeric(lit);
1044     }
1045     return lit;
1046 }
1047
1048 VALUE
1049 rb_node_str_string_val(const NODE *node)
1050 {
1051     rb_parser_string_t *str = RNODE_STR(node)->string;
1052     return rb_str_new_parser_string(str);
1053 }
1054
1055 VALUE
1056 rb_node_sym_string_val(const NODE *node)
1057 {
1058     rb_parser_string_t *str = RNODE_SYM(node)->string;
1059     return ID2SYM(rb_intern3(str->ptr, str->len, str->enc));
1060 }
1061
1062 VALUE
1063 rb_node_dstr_string_val(const NODE *node)
1064 {
1065     rb_parser_string_t *str = RNODE_DSTR(node)->string;
1066     return str ? rb_str_new_parser_string(str) : Qnil;
1067 }
1068
1069 VALUE
1070 rb_node_dregx_string_val(const NODE *node)
1071 {
1072     rb_parser_string_t *str = RNODE_DREGX(node)->string;
1073     return rb_str_new_parser_string(str);
1074 }
1075
1076 VALUE
1077 rb_node_regx_string_val(const NODE *node)
1078 {
1079     rb_node_regx_t *node_reg = RNODE_REGX(node);
1080     rb_parser_string_t *string = node_reg->string;
1081     VALUE str = rb_enc_str_new(string->ptr, string->len, string->enc);
1082
1083     return rb_reg_compile(str, node_reg->options, NULL, 0);
1084 }
1085
1086 VALUE
1087 rb_node_line_lineno_val(const NODE *node)
1088 {
1089     return INT2FIX(node->nd_loc.beg_pos.lineno);
1090 }
1091
1092 VALUE
1093 rb_node_file_path_val(const NODE *node)
1094 {
1095     return rb_str_new_parser_string(RNODE_FILE(node)->path);
1096 }
1097
1098 VALUE
1099 rb_node_encoding_val(const NODE *node)
1100 {
1101     return rb_enc_from_encoding(RNODE_ENCODING(node)->enc);
1102 }
1103
1104 static void
1105 parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines)
1106 {
1107     VALUE hash, script_lines;
1108     ID script_lines_id;
1109     if (NIL_P(path) || !lines) return;
1110     CONST_ID(script_lines_id, "SCRIPT_LINES__");
1111     if (!rb_const_defined_at(rb_cObject, script_lines_id)) return;
1112     hash = rb_const_get_at(rb_cObject, script_lines_id);
1113     if (!RB_TYPE_P(hash, T_HASH)) return;
1114     if (rb_hash_lookup(hash, path) == Qnil) return;
1115     script_lines = rb_parser_build_script_lines_from(lines);
1116     rb_hash_aset(hash, path, script_lines);
1117 }
1118
1119 VALUE
1120 rb_ruby_ast_new(const NODE *const root)
1121 {
1122     rb_ast_t *ast;
1123     VALUE ast_value = TypedData_Make_Struct(0, rb_ast_t, &ast_data_type, ast);
1124 #ifdef UNIVERSAL_PARSER
1125     ast->config = &rb_global_parser_config;
1126 #endif
1127     ast->body = (rb_ast_body_t){
1128         .root = root,
1129         .frozen_string_literal = -1,
1130         .coverage_enabled = -1,
1131         .script_lines = NULL,
1132         .line_count = 0,
1133     };
1134     return ast_value;
1135 }
1136
1137 rb_ast_t *
1138 rb_ruby_ast_data_get(VALUE ast_value)
1139 {
1140     rb_ast_t *ast;
1141     if (NIL_P(ast_value)) return NULL;
1142     TypedData_Get_Struct(ast_value, rb_ast_t, &ast_data_type, ast);
1143     return ast;
1144 }