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_resize = rb_str_resize,
 375     .str_new = rb_str_new,
 376     .str_new_cstr = rb_str_new_cstr,
 377     .str_to_interned_str = rb_str_to_interned_str,
 378     .is_ascii_string = is_ascii_string2,
 379     .enc_str_new = enc_str_new,
 380     .str_vcatf = rb_str_vcatf,
 381     .rb_sprintf = rb_sprintf,
 382     .rstring_ptr = RSTRING_PTR,
 383     .rstring_end = RSTRING_END,
 384     .rstring_len = RSTRING_LEN,
 385     .obj_as_string = rb_obj_as_string,
 386
 387     .int2num = rb_int2num_inline,
 388
 389     .stderr_tty_p = rb_stderr_tty_p,
 390     .write_error_str = rb_write_error_str,
 391     .io_write = rb_io_write,
 392     .io_flush = rb_io_flush,
 393     .io_puts = rb_io_puts,
 394
 395     .debug_output_stdout = rb_ractor_stdout,
 396     .debug_output_stderr = rb_ractor_stderr,
 397
 398     .is_usascii_enc = is_usascii_enc,
 399     .enc_isalnum = enc_isalnum,
 400     .enc_precise_mbclen = enc_precise_mbclen,
 401     .mbclen_charfound_p = mbclen_charfound_p,
 402     .mbclen_charfound_len = mbclen_charfound_len,
 403     .enc_name = enc_name,
 404     .enc_prev_char = enc_prev_char,
 405     .enc_get = enc_get,
 406     .enc_asciicompat = enc_asciicompat,
 407     .utf8_encoding = utf8_encoding,
 408     .enc_associate = enc_associate,
 409     .ascii8bit_encoding = ascii8bit_encoding,
 410     .enc_codelen = enc_codelen,
 411     .enc_mbcput = enc_mbcput,
 412     .enc_find_index = rb_enc_find_index,
 413     .enc_from_index = enc_from_index,
 414     .enc_isspace = enc_isspace,
 415     .enc_coderange_7bit = ENC_CODERANGE_7BIT,
 416     .enc_coderange_unknown = ENC_CODERANGE_UNKNOWN,
 417     .usascii_encoding = usascii_encoding,
 418     .enc_mbminlen = enc_mbminlen,
 419     .enc_isascii = enc_isascii,
 420     .enc_mbc_to_codepoint = enc_mbc_to_codepoint,
 421
 422     .local_defined = local_defined,
 423     .dvar_defined = dvar_defined,
 424
 425     .syntax_error_append = syntax_error_append,
 426     .raise = rb_raise,
 427     .syntax_error_new = syntax_error_new,
 428
 429     .errinfo = rb_errinfo,
 430     .set_errinfo = rb_set_errinfo,
 431     .exc_raise = rb_exc_raise,
 432     .make_exception = rb_make_exception,
 433
 434     .sized_xfree = ruby_sized_xfree,
 435     .sized_realloc_n = ruby_sized_realloc_n,
 436     .gc_guard = gc_guard,
 437     .gc_mark = rb_gc_mark,
 438
 439     .reg_compile = rb_reg_compile,
 440     .reg_check_preprocess = rb_reg_check_preprocess,
 441     .memcicmp = rb_memcicmp,
 442
 443     .compile_warn = rb_compile_warn,
 444     .compile_warning = rb_compile_warning,
 445     .bug = rb_bug,
 446     .fatal = rb_fatal,
 447     .verbose = ruby_verbose2,
 448     .errno_ptr = rb_errno_ptr2,
 449
 450     .make_backtrace = rb_make_backtrace,
 451
 452     .scan_hex = ruby_scan_hex,
 453     .scan_oct = ruby_scan_oct,
 454     .scan_digits = ruby_scan_digits,
 455     .strtod = ruby_strtod,
 456
 457     .rtest = rtest,
 458     .nil_p = nil_p,
 459     .qnil = Qnil,
 460     .qfalse = Qfalse,
 461     .eArgError = arg_error,
 462     .long2int = rb_long2int,
 463
 464     /* For Ripper */
 465     .static_id2sym = static_id2sym,
 466     .str_coderange_scan_restartable = str_coderange_scan_restartable,
 467 };
 468 #endif
 469
 470 enum lex_type {
 471     lex_type_str,
 472     lex_type_io,
 473     lex_type_array,
 474     lex_type_generic,
 475 };
 476
 477 struct ruby_parser {
 478     rb_parser_t *parser_params;
 479     enum lex_type type;
 480     union {
 481         struct lex_pointer_string lex_str;
 482         struct {
 483             VALUE file;
 484         } lex_io;
 485         struct {
 486             VALUE ary;
 487         } lex_array;
 488     } data;
 489 };
 490
 491 static void
 492 parser_mark(void *ptr)
 493 {
 494     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 495     rb_ruby_parser_mark(parser->parser_params);
 496
 497     switch (parser->type) {
 498       case lex_type_str:
 499         rb_gc_mark(parser->data.lex_str.str);
 500         break;
 501       case lex_type_io:
 502         rb_gc_mark(parser->data.lex_io.file);
 503         break;
 504       case lex_type_array:
 505         rb_gc_mark(parser->data.lex_array.ary);
 506         break;
 507       case lex_type_generic:
 508         /* noop. Caller of rb_parser_compile_generic should mark the objects. */
 509         break;
 510     }
 511 }
 512
 513 static void
 514 parser_free(void *ptr)
 515 {
 516     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 517     rb_ruby_parser_free(parser->parser_params);
 518     xfree(parser);
 519 }
 520
 521 static size_t
 522 parser_memsize(const void *ptr)
 523 {
 524     struct ruby_parser *parser = (struct ruby_parser*)ptr;
 525     return rb_ruby_parser_memsize(parser->parser_params);
 526 }
 527
 528 static const rb_data_type_t ruby_parser_data_type = {
 529     "parser",
 530     {
 531         parser_mark,
 532         parser_free,
 533         parser_memsize,
 534     },
 535     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 536 };
 537
 538 #ifdef UNIVERSAL_PARSER
 539 const rb_parser_config_t *
 540 rb_ruby_parser_config(void)
 541 {
 542     return &rb_global_parser_config;
 543 }
 544
 545 rb_parser_t *
 546 rb_parser_params_new(void)
 547 {
 548     return rb_ruby_parser_new(&rb_global_parser_config);
 549 }
 550 #else
 551 rb_parser_t *
 552 rb_parser_params_new(void)
 553 {
 554     return rb_ruby_parser_new();
 555 }
 556 #endif /* UNIVERSAL_PARSER */
 557
 558 VALUE
 559 rb_parser_new(void)
 560 {
 561     struct ruby_parser *parser;
 562     rb_parser_t *parser_params;
 563
 564     /*
 565      * Create parser_params ahead of vparser because
 566      * rb_ruby_parser_new can run GC so if create vparser
 567      * first, parser_mark tries to mark not initialized parser_params.
 568      */
 569     parser_params = rb_parser_params_new();
 570     VALUE vparser = TypedData_Make_Struct(0, struct ruby_parser,
 571                                          &ruby_parser_data_type, parser);
 572     parser->parser_params = parser_params;
 573
 574     return vparser;
 575 }
 576
 577 void
 578 rb_parser_set_options(VALUE vparser, int print, int loop, int chomp, int split)
 579 {
 580     struct ruby_parser *parser;
 581
 582     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 583     rb_ruby_parser_set_options(parser->parser_params, print, loop, chomp, split);
 584 }
 585
 586 VALUE
 587 rb_parser_set_context(VALUE vparser, const struct rb_iseq_struct *base, int main)
 588 {
 589     struct ruby_parser *parser;
 590
 591     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 592     rb_ruby_parser_set_context(parser->parser_params, base, main);
 593     return vparser;
 594 }
 595
 596 void
 597 rb_parser_set_script_lines(VALUE vparser)
 598 {
 599     struct ruby_parser *parser;
 600
 601     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 602     rb_ruby_parser_set_script_lines(parser->parser_params);
 603 }
 604
 605 void
 606 rb_parser_error_tolerant(VALUE vparser)
 607 {
 608     struct ruby_parser *parser;
 609
 610     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 611     rb_ruby_parser_error_tolerant(parser->parser_params);
 612 }
 613
 614 void
 615 rb_parser_keep_tokens(VALUE vparser)
 616 {
 617     struct ruby_parser *parser;
 618
 619     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 620     rb_ruby_parser_keep_tokens(parser->parser_params);
 621 }
 622
 623 rb_parser_string_t *
 624 rb_parser_lex_get_str(struct parser_params *p, struct lex_pointer_string *ptr_str)
 625 {
 626     char *beg, *end, *start;
 627     long len;
 628     VALUE s = ptr_str->str;
 629
 630     beg = RSTRING_PTR(s);
 631     len = RSTRING_LEN(s);
 632     start = beg;
 633     if (ptr_str->ptr) {
 634         if (len == ptr_str->ptr) return 0;
 635         beg += ptr_str->ptr;
 636         len -= ptr_str->ptr;
 637     }
 638     end = memchr(beg, '\n', len);
 639     if (end) len = ++end - beg;
 640     ptr_str->ptr += len;
 641     return rb_str_to_parser_string(p, rb_str_subseq(s, beg - start, len));
 642 }
 643
 644 static rb_parser_string_t *
 645 lex_get_str(struct parser_params *p, rb_parser_input_data input, int line_count)
 646 {
 647     return rb_parser_lex_get_str(p, (struct lex_pointer_string *)input);
 648 }
 649
 650 static void parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines);
 651
 652 static rb_ast_t*
 653 parser_compile(rb_parser_t *p, rb_parser_lex_gets_func *gets, VALUE fname, rb_parser_input_data input, int line)
 654 {
 655     rb_ast_t *ast = rb_parser_compile(p, gets, fname, input, line);
 656     parser_aset_script_lines_for(fname, ast->body.script_lines);
 657     return ast;
 658 }
 659
 660 static rb_ast_t*
 661 parser_compile_string0(struct ruby_parser *parser, VALUE fname, VALUE s, int line)
 662 {
 663     VALUE str = rb_str_new_frozen(s);
 664
 665     parser->type = lex_type_str;
 666     parser->data.lex_str.str = str;
 667     parser->data.lex_str.ptr = 0;
 668
 669     return parser_compile(parser->parser_params, lex_get_str, fname, (rb_parser_input_data)&parser->data, line);
 670 }
 671
 672 static rb_encoding *
 673 must_be_ascii_compatible(VALUE s)
 674 {
 675     rb_encoding *enc = rb_enc_get(s);
 676     if (!rb_enc_asciicompat(enc)) {
 677         rb_raise(rb_eArgError, "invalid source encoding");
 678     }
 679     return enc;
 680 }
 681
 682 static rb_ast_t*
 683 parser_compile_string_path(struct ruby_parser *parser, VALUE f, VALUE s, int line)
 684 {
 685     must_be_ascii_compatible(s);
 686     return parser_compile_string0(parser, f, s, line);
 687 }
 688
 689 static rb_ast_t*
 690 parser_compile_string(struct ruby_parser *parser, const char *f, VALUE s, int line)
 691 {
 692     return parser_compile_string_path(parser, rb_filesystem_str_new_cstr(f), s, line);
 693 }
 694
 695 VALUE rb_io_gets_internal(VALUE io);
 696
 697 static rb_parser_string_t *
 698 lex_io_gets(struct parser_params *p, rb_parser_input_data input, int line_count)
 699 {
 700     VALUE io = (VALUE)input;
 701     VALUE line = rb_io_gets_internal(io);
 702     if (NIL_P(line)) return 0;
 703     return rb_str_to_parser_string(p, line);
 704 }
 705
 706 static rb_parser_string_t *
 707 lex_gets_array(struct parser_params *p, rb_parser_input_data data, int index)
 708 {
 709     VALUE array = (VALUE)data;
 710     VALUE str = rb_ary_entry(array, index);
 711     if (!NIL_P(str)) {
 712         StringValue(str);
 713         if (!rb_enc_asciicompat(rb_enc_get(str))) {
 714             rb_raise(rb_eArgError, "invalid source encoding");
 715         }
 716         return rb_str_to_parser_string(p, str);
 717     }
 718     else {
 719         return 0;
 720     }
 721 }
 722
 723 static rb_ast_t*
 724 parser_compile_file_path(struct ruby_parser *parser, VALUE fname, VALUE file, int start)
 725 {
 726     parser->type = lex_type_io;
 727     parser->data.lex_io.file = file;
 728
 729     return parser_compile(parser->parser_params, lex_io_gets, fname, (rb_parser_input_data)file, start);
 730 }
 731
 732 static rb_ast_t*
 733 parser_compile_array(struct ruby_parser *parser, VALUE fname, VALUE array, int start)
 734 {
 735     parser->type = lex_type_array;
 736     parser->data.lex_array.ary = array;
 737
 738     return parser_compile(parser->parser_params, lex_gets_array, fname, (rb_parser_input_data)array, start);
 739 }
 740
 741 static rb_ast_t*
 742 parser_compile_generic(struct ruby_parser *parser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 743 {
 744     parser->type = lex_type_generic;
 745
 746     return parser_compile(parser->parser_params, lex_gets, fname, (rb_parser_input_data)input, start);
 747 }
 748
 749 static void
 750 ast_free(void *ptr)
 751 {
 752     rb_ast_t *ast = (rb_ast_t *)ptr;
 753     rb_ast_free(ast);
 754 }
 755
 756 static const rb_data_type_t ast_data_type = {
 757     "AST",
 758     {
 759         NULL,
 760         ast_free,
 761         NULL, // No dsize() because this object does not appear in ObjectSpace.
 762     },
 763     0, 0, RUBY_TYPED_FREE_IMMEDIATELY
 764 };
 765
 766 static VALUE
 767 ast_alloc(void)
 768 {
 769     return TypedData_Wrap_Struct(0, &ast_data_type, NULL);
 770 }
 771
 772 VALUE
 773 rb_parser_compile_file_path(VALUE vparser, VALUE fname, VALUE file, int start)
 774 {
 775     struct ruby_parser *parser;
 776     VALUE ast_value = ast_alloc();
 777
 778     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 779     DATA_PTR(ast_value) = parser_compile_file_path(parser, fname, file, start);
 780     RB_GC_GUARD(vparser);
 781
 782     return ast_value;
 783 }
 784
 785 VALUE
 786 rb_parser_compile_array(VALUE vparser, VALUE fname, VALUE array, int start)
 787 {
 788     struct ruby_parser *parser;
 789     VALUE ast_value = ast_alloc();
 790
 791     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 792     DATA_PTR(ast_value) = parser_compile_array(parser, fname, array, start);
 793     RB_GC_GUARD(vparser);
 794
 795     return ast_value;
 796 }
 797
 798 VALUE
 799 rb_parser_compile_generic(VALUE vparser, rb_parser_lex_gets_func *lex_gets, VALUE fname, VALUE input, int start)
 800 {
 801     struct ruby_parser *parser;
 802     VALUE ast_value = ast_alloc();
 803
 804     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 805     DATA_PTR(ast_value) = parser_compile_generic(parser, lex_gets, fname, input, start);
 806     RB_GC_GUARD(vparser);
 807
 808     return ast_value;
 809 }
 810
 811 VALUE
 812 rb_parser_compile_string(VALUE vparser, const char *f, VALUE s, int line)
 813 {
 814     struct ruby_parser *parser;
 815     VALUE ast_value = ast_alloc();
 816
 817     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 818     DATA_PTR(ast_value) = parser_compile_string(parser, f, s, line);
 819     RB_GC_GUARD(vparser);
 820
 821     return ast_value;
 822 }
 823
 824 VALUE
 825 rb_parser_compile_string_path(VALUE vparser, VALUE f, VALUE s, int line)
 826 {
 827     struct ruby_parser *parser;
 828     VALUE ast_value = ast_alloc();
 829
 830     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 831     DATA_PTR(ast_value) = parser_compile_string_path(parser, f, s, line);
 832     RB_GC_GUARD(vparser);
 833
 834     return ast_value;
 835 }
 836
 837 VALUE
 838 rb_parser_encoding(VALUE vparser)
 839 {
 840     struct ruby_parser *parser;
 841
 842     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 843     return rb_enc_from_encoding(rb_ruby_parser_encoding(parser->parser_params));
 844 }
 845
 846 VALUE
 847 rb_parser_end_seen_p(VALUE vparser)
 848 {
 849     struct ruby_parser *parser;
 850
 851     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 852     return RBOOL(rb_ruby_parser_end_seen_p(parser->parser_params));
 853 }
 854
 855 VALUE
 856 rb_parser_set_yydebug(VALUE vparser, VALUE flag)
 857 {
 858     struct ruby_parser *parser;
 859
 860     TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 861     rb_ruby_parser_set_yydebug(parser->parser_params, RTEST(flag));
 862     return flag;
 863 }
 864
 865 void
 866 rb_set_script_lines_for(VALUE vparser, VALUE path)
 867 {
 868     struct ruby_parser *parser;
 869     VALUE hash;
 870     ID script_lines;
 871     CONST_ID(script_lines, "SCRIPT_LINES__");
 872     if (!rb_const_defined_at(rb_cObject, script_lines)) return;
 873     hash = rb_const_get_at(rb_cObject, script_lines);
 874     if (RB_TYPE_P(hash, T_HASH)) {
 875         rb_hash_aset(hash, path, Qtrue);
 876         TypedData_Get_Struct(vparser, struct ruby_parser, &ruby_parser_data_type, parser);
 877         rb_ruby_parser_set_script_lines(parser->parser_params);
 878     }
 879 }
 880
 881 VALUE
 882 rb_parser_build_script_lines_from(rb_parser_ary_t *lines)
 883 {
 884     int i;
 885     if (!lines) return Qnil;
 886     if (lines->data_type != PARSER_ARY_DATA_SCRIPT_LINE) {
 887         rb_bug("unexpected rb_parser_ary_data_type (%d) for script lines", lines->data_type);
 888     }
 889     VALUE script_lines = rb_ary_new_capa(lines->len);
 890     for (i = 0; i < lines->len; i++) {
 891         rb_parser_string_t *str = (rb_parser_string_t *)lines->data[i];
 892         rb_ary_push(script_lines, rb_enc_str_new(str->ptr, str->len, str->enc));
 893     }
 894     return script_lines;
 895 }
 896
 897 VALUE
 898 rb_str_new_parser_string(rb_parser_string_t *str)
 899 {
 900     VALUE string = rb_enc_literal_str(str->ptr, str->len, str->enc);
 901     rb_enc_str_coderange(string);
 902     return string;
 903 }
 904
 905 VALUE
 906 rb_str_new_mutable_parser_string(rb_parser_string_t *str)
 907 {
 908     return rb_enc_str_new(str->ptr, str->len, str->enc);
 909 }
 910
 911 static VALUE
 912 negative_numeric(VALUE val)
 913 {
 914     if (FIXNUM_P(val)) {
 915         return LONG2FIX(-FIX2LONG(val));
 916     }
 917     if (SPECIAL_CONST_P(val)) {
 918 #if USE_FLONUM
 919         if (FLONUM_P(val)) {
 920             return DBL2NUM(-RFLOAT_VALUE(val));
 921         }
 922 #endif
 923         goto unknown;
 924     }
 925     switch (BUILTIN_TYPE(val)) {
 926       case T_BIGNUM:
 927         BIGNUM_NEGATE(val);
 928         val = rb_big_norm(val);
 929         break;
 930       case T_RATIONAL:
 931         RATIONAL_SET_NUM(val, negative_numeric(RRATIONAL(val)->num));
 932         break;
 933       case T_COMPLEX:
 934         RCOMPLEX_SET_REAL(val, negative_numeric(RCOMPLEX(val)->real));
 935         RCOMPLEX_SET_IMAG(val, negative_numeric(RCOMPLEX(val)->imag));
 936         break;
 937       case T_FLOAT:
 938         val = DBL2NUM(-RFLOAT_VALUE(val));
 939         break;
 940       unknown:
 941       default:
 942         rb_bug("unknown literal type (%s) passed to negative_numeric",
 943                rb_builtin_class_name(val));
 944         break;
 945     }
 946     return val;
 947 }
 948
 949 static VALUE
 950 integer_value(const char *val, int base)
 951 {
 952     return rb_cstr_to_inum(val, base, FALSE);
 953 }
 954
 955 static VALUE
 956 rational_value(const char *node_val, int base, int seen_point)
 957 {
 958     VALUE lit;
 959     char* val = strdup(node_val);
 960     if (seen_point > 0) {
 961         int len = (int)(strlen(val));
 962         char *point = &val[seen_point];
 963         size_t fraclen = len-seen_point-1;
 964         memmove(point, point+1, fraclen+1);
 965
 966         lit = rb_rational_new(integer_value(val, base), rb_int_positive_pow(10, fraclen));
 967     }
 968     else {
 969         lit = rb_rational_raw1(integer_value(val, base));
 970     }
 971
 972     free(val);
 973
 974     return lit;
 975 }
 976
 977 VALUE
 978 rb_node_integer_literal_val(const NODE *n)
 979 {
 980     const rb_node_integer_t *node = RNODE_INTEGER(n);
 981     VALUE val = integer_value(node->val, node->base);
 982     if (node->minus) {
 983         val = negative_numeric(val);
 984     }
 985     return val;
 986 }
 987
 988 VALUE
 989 rb_node_float_literal_val(const NODE *n)
 990 {
 991     const rb_node_float_t *node = RNODE_FLOAT(n);
 992     double d = strtod(node->val, 0);
 993     if (node->minus) {
 994         d = -d;
 995     }
 996     VALUE val = DBL2NUM(d);
 997     return val;
 998 }
 999
1000 VALUE
1001 rb_node_rational_literal_val(const NODE *n)
1002 {
1003     VALUE lit;
1004     const rb_node_rational_t *node = RNODE_RATIONAL(n);
1005
1006     lit = rational_value(node->val, node->base, node->seen_point);
1007
1008     if (node->minus) {
1009         lit = negative_numeric(lit);
1010     }
1011
1012     return lit;
1013 }
1014
1015 VALUE
1016 rb_node_imaginary_literal_val(const NODE *n)
1017 {
1018     VALUE lit;
1019     const rb_node_imaginary_t *node = RNODE_IMAGINARY(n);
1020
1021     enum rb_numeric_type type = node->type;
1022
1023     switch (type) {
1024       case integer_literal:
1025         lit = integer_value(node->val, node->base);
1026         break;
1027       case float_literal:{
1028         double d = strtod(node->val, 0);
1029         lit = DBL2NUM(d);
1030         break;
1031       }
1032       case rational_literal:
1033         lit = rational_value(node->val, node->base, node->seen_point);
1034         break;
1035       default:
1036         rb_bug("unreachable");
1037     }
1038
1039     lit = rb_complex_raw(INT2FIX(0), lit);
1040
1041     if (node->minus) {
1042         lit = negative_numeric(lit);
1043     }
1044     return lit;
1045 }
1046
1047 VALUE
1048 rb_node_str_string_val(const NODE *node)
1049 {
1050     rb_parser_string_t *str = RNODE_STR(node)->string;
1051     return rb_str_new_parser_string(str);
1052 }
1053
1054 VALUE
1055 rb_node_sym_string_val(const NODE *node)
1056 {
1057     rb_parser_string_t *str = RNODE_SYM(node)->string;
1058     return ID2SYM(rb_intern3(str->ptr, str->len, str->enc));
1059 }
1060
1061 VALUE
1062 rb_node_dstr_string_val(const NODE *node)
1063 {
1064     rb_parser_string_t *str = RNODE_DSTR(node)->string;
1065     return str ? rb_str_new_parser_string(str) : Qnil;
1066 }
1067
1068 VALUE
1069 rb_node_dregx_string_val(const NODE *node)
1070 {
1071     rb_parser_string_t *str = RNODE_DREGX(node)->string;
1072     return rb_str_new_parser_string(str);
1073 }
1074
1075 VALUE
1076 rb_node_regx_string_val(const NODE *node)
1077 {
1078     rb_node_regx_t *node_reg = RNODE_REGX(node);
1079     rb_parser_string_t *string = node_reg->string;
1080     VALUE str = rb_enc_str_new(string->ptr, string->len, string->enc);
1081
1082     return rb_reg_compile(str, node_reg->options, NULL, 0);
1083 }
1084
1085 VALUE
1086 rb_node_line_lineno_val(const NODE *node)
1087 {
1088     return INT2FIX(node->nd_loc.beg_pos.lineno);
1089 }
1090
1091 VALUE
1092 rb_node_file_path_val(const NODE *node)
1093 {
1094     return rb_str_new_parser_string(RNODE_FILE(node)->path);
1095 }
1096
1097 VALUE
1098 rb_node_encoding_val(const NODE *node)
1099 {
1100     return rb_enc_from_encoding(RNODE_ENCODING(node)->enc);
1101 }
1102
1103 static void
1104 parser_aset_script_lines_for(VALUE path, rb_parser_ary_t *lines)
1105 {
1106     VALUE hash, script_lines;
1107     ID script_lines_id;
1108     if (NIL_P(path) || !lines) return;
1109     CONST_ID(script_lines_id, "SCRIPT_LINES__");
1110     if (!rb_const_defined_at(rb_cObject, script_lines_id)) return;
1111     hash = rb_const_get_at(rb_cObject, script_lines_id);
1112     if (!RB_TYPE_P(hash, T_HASH)) return;
1113     if (rb_hash_lookup(hash, path) == Qnil) return;
1114     script_lines = rb_parser_build_script_lines_from(lines);
1115     rb_hash_aset(hash, path, script_lines);
1116 }
1117
1118 VALUE
1119 rb_ruby_ast_new(const NODE *const root)
1120 {
1121     rb_ast_t *ast;
1122     VALUE ast_value = TypedData_Make_Struct(0, rb_ast_t, &ast_data_type, ast);
1123 #ifdef UNIVERSAL_PARSER
1124     ast->config = &rb_global_parser_config;
1125 #endif
1126     ast->body = (rb_ast_body_t){
1127         .root = root,
1128         .frozen_string_literal = -1,
1129         .coverage_enabled = -1,
1130         .script_lines = NULL,
1131         .line_count = 0,
1132     };
1133     return ast_value;
1134 }
1135
1136 rb_ast_t *
1137 rb_ruby_ast_data_get(VALUE ast_value)
1138 {
1139     rb_ast_t *ast;
1140     if (NIL_P(ast_value)) return NULL;
1141     TypedData_Get_Struct(ast_value, rb_ast_t, &ast_data_type, ast);
1142     return ast;
1143 }