1 /* This is a wrapper for parse.y */
3 #include "internal/parse.h"
4 #include "internal/re.h"
5 #include "internal/ruby_parser.h"
8 #include "rubyparser.h"
9 #include "internal/error.h"
11 #ifdef UNIVERSAL_PARSER
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"
28 #include "ruby/ractor.h"
29 #include "ruby/ruby.h"
30 #include "ruby/util.h"
35 #define parser_encoding const void
38 is_ascii_string2(VALUE str
)
40 return is_ascii_string(str
);
43 RBIMPL_ATTR_FORMAT(RBIMPL_PRINTF_FORMAT
, 6, 0)
45 syntax_error_append(VALUE exc
, VALUE file
, int line
, int column
,
46 parser_encoding
*enc
, const char *fmt
, va_list args
)
48 return rb_syntax_error_append(exc
, file
, line
, column
, enc
, fmt
, args
);
52 local_defined(ID id
, const void *p
)
54 return rb_local_defined(id
, (const rb_iseq_t
*)p
);
58 dvar_defined(ID id
, const void *p
)
60 return rb_dvar_defined(id
, (const rb_iseq_t
*)p
);
64 is_usascii_enc(parser_encoding
*enc
)
66 return rb_is_usascii_enc(enc
);
72 return is_local_id(id
);
78 return is_attrset_id(id
);
84 return is_notop_id(id
);
88 enc_str_new(const char *ptr
, long len
, parser_encoding
*enc
)
90 return rb_enc_str_new(ptr
, len
, enc
);
94 enc_isalnum(OnigCodePoint c
, parser_encoding
*enc
)
96 return rb_enc_isalnum(c
, enc
);
100 enc_precise_mbclen(const char *p
, const char *e
, parser_encoding
*enc
)
102 return rb_enc_precise_mbclen(p
, e
, enc
);
106 mbclen_charfound_p(int len
)
108 return MBCLEN_CHARFOUND_P(len
);
112 mbclen_charfound_len(int len
)
114 return MBCLEN_CHARFOUND_LEN(len
);
118 enc_name(parser_encoding
*enc
)
120 return rb_enc_name(enc
);
124 enc_prev_char(const char *s
, const char *p
, const char *e
, parser_encoding
*enc
)
126 return rb_enc_prev_char(s
, p
, e
, enc
);
129 static parser_encoding
*
132 return rb_enc_get(obj
);
136 enc_asciicompat(parser_encoding
*enc
)
138 return rb_enc_asciicompat(enc
);
141 static parser_encoding
*
144 return rb_utf8_encoding();
148 enc_associate(VALUE obj
, parser_encoding
*enc
)
150 return rb_enc_associate(obj
, enc
);
153 static parser_encoding
*
154 ascii8bit_encoding(void)
156 return rb_ascii8bit_encoding();
160 enc_codelen(int c
, parser_encoding
*enc
)
162 return rb_enc_codelen(c
, enc
);
166 enc_mbcput(unsigned int c
, void *buf
, parser_encoding
*enc
)
168 return rb_enc_mbcput(c
, buf
, enc
);
172 enc_mbclen(const char *p
, const char *e
, parser_encoding
*enc
)
174 return rb_enc_mbclen(p
, e
, enc
);
177 static parser_encoding
*
178 enc_from_index(int idx
)
180 return rb_enc_from_index(idx
);
184 enc_isspace(OnigCodePoint c
, parser_encoding
*enc
)
186 return rb_enc_isspace(c
, enc
);
190 intern3(const char *name
, long len
, parser_encoding
*enc
)
192 return rb_intern3(name
, len
, enc
);
195 static parser_encoding
*
196 usascii_encoding(void)
198 return rb_usascii_encoding();
202 enc_symname_type(const char *name
, long len
, parser_encoding
*enc
, unsigned int allowed_attrset
)
204 return rb_enc_symname_type(name
, len
, enc
, allowed_attrset
);
208 struct parser_params
*parser
;
211 const rb_code_location_t
*loc
;
212 } reg_named_capture_assign_t
;
215 reg_named_capture_assign_iter(const OnigUChar
*name
, const OnigUChar
*name_end
,
216 int back_num
, int *back_refs
, OnigRegex regex
, void *arg0
)
218 reg_named_capture_assign_t
*arg
= (reg_named_capture_assign_t
*)arg0
;
219 struct parser_params
* p
= arg
->parser
;
220 rb_encoding
*enc
= arg
->enc
;
221 const rb_code_location_t
*loc
= arg
->loc
;
222 long len
= name_end
- name
;
223 const char *s
= (const char *)name
;
225 return rb_reg_named_capture_assign_iter_impl(p
, s
, len
, enc
, &arg
->succ_block
, loc
);
229 reg_named_capture_assign(struct parser_params
* p
, VALUE regexp
, const rb_code_location_t
*loc
)
231 reg_named_capture_assign_t arg
;
234 arg
.enc
= rb_enc_get(regexp
);
237 onig_foreach_name(RREGEXP_PTR(regexp
), reg_named_capture_assign_iter
, &arg
);
239 if (!arg
.succ_block
) return 0;
240 return RNODE_BLOCK(arg
.succ_block
)->nd_next
;
246 return (int)RB_TEST(obj
);
252 return (int)NIL_P(obj
);
256 syntax_error_new(void)
258 return rb_class_new_instance(0, 0, rb_eSyntaxError
);
262 memmove2(void *dest
, const void *src
, size_t t
, size_t n
)
264 return memmove(dest
, src
, rbimpl_size_mul_or_raise(t
, n
));
268 nonempty_memcpy(void *dest
, const void *src
, size_t t
, size_t n
)
270 return ruby_nonempty_memcpy(dest
, src
, rbimpl_size_mul_or_raise(t
, n
));
282 return rb_errno_ptr();
286 zalloc(size_t elemsiz
)
288 return ruby_xcalloc(1, elemsiz
);
306 return (((VALUE
)(id
)<<RUBY_SPECIAL_SHIFT
)|SYMBOL_FLAG
);
310 str_coderange_scan_restartable(const char *s
, const char *e
, parser_encoding
*enc
, int *cr
)
312 return rb_str_coderange_scan_restartable(s
, e
, enc
, cr
);
316 enc_mbminlen(parser_encoding
*enc
)
318 return rb_enc_mbminlen(enc
);
322 enc_isascii(OnigCodePoint c
, parser_encoding
*enc
)
324 return rb_enc_isascii(c
, enc
);
328 enc_mbc_to_codepoint(const char *p
, const char *e
, parser_encoding
*enc
)
330 const OnigUChar
*up
= RBIMPL_CAST((const OnigUChar
*)p
);
331 const OnigUChar
*ue
= RBIMPL_CAST((const OnigUChar
*)e
);
333 return ONIGENC_MBC_TO_CODE((rb_encoding
*)enc
, up
, ue
);
336 extern VALUE rb_eArgError
;
338 static const rb_parser_config_t rb_global_parser_config
= {
339 .malloc
= ruby_xmalloc
,
340 .calloc
= ruby_xcalloc
,
341 .realloc
= ruby_xrealloc
,
343 .alloc_n
= ruby_xmalloc2
,
344 .alloc
= ruby_xmalloc
,
345 .realloc_n
= ruby_xrealloc2
,
347 .rb_memmove
= memmove2
,
348 .nonempty_memcpy
= nonempty_memcpy
,
349 .xmalloc_mul_add
= rb_xmalloc_mul_add
,
351 .compile_callback
= rb_suppress_tracing
,
352 .reg_named_capture_assign
= reg_named_capture_assign
,
354 .attr_get
= rb_attr_get
,
356 .ary_new
= rb_ary_new
,
357 .ary_push
= rb_ary_push
,
358 .ary_new_from_args
= rb_ary_new_from_args
,
359 .ary_unshift
= rb_ary_unshift
,
361 .make_temporary_id
= rb_make_temporary_id
,
362 .is_local_id
= is_local_id2
,
363 .is_attrset_id
= is_attrset_id2
,
364 .is_global_name_punct
= is_global_name_punct
,
366 .id_attrset
= rb_id_attrset
,
368 .intern2
= rb_intern2
,
370 .intern_str
= rb_intern_str
,
371 .is_notop_id
= is_notop_id2
,
372 .enc_symname_type
= enc_symname_type
,
373 .id2name
= rb_id2name
,
378 .str_catf
= rb_str_catf
,
379 .str_cat_cstr
= rb_str_cat_cstr
,
380 .str_modify
= rb_str_modify
,
381 .str_set_len
= rb_str_set_len
,
382 .str_cat
= rb_str_cat
,
383 .str_resize
= rb_str_resize
,
384 .str_new
= rb_str_new
,
385 .str_new_cstr
= rb_str_new_cstr
,
386 .str_to_interned_str
= rb_str_to_interned_str
,
387 .is_ascii_string
= is_ascii_string2
,
388 .enc_str_new
= enc_str_new
,
389 .str_vcatf
= rb_str_vcatf
,
390 .string_value_cstr
= rb_string_value_cstr
,
391 .rb_sprintf
= rb_sprintf
,
392 .rstring_ptr
= RSTRING_PTR
,
393 .rstring_end
= RSTRING_END
,
394 .rstring_len
= RSTRING_LEN
,
395 .obj_as_string
= rb_obj_as_string
,
397 .int2num
= rb_int2num_inline
,
399 .stderr_tty_p
= rb_stderr_tty_p
,
400 .write_error_str
= rb_write_error_str
,
401 .io_write
= rb_io_write
,
402 .io_flush
= rb_io_flush
,
403 .io_puts
= rb_io_puts
,
405 .debug_output_stdout
= rb_ractor_stdout
,
406 .debug_output_stderr
= rb_ractor_stderr
,
408 .is_usascii_enc
= is_usascii_enc
,
409 .enc_isalnum
= enc_isalnum
,
410 .enc_precise_mbclen
= enc_precise_mbclen
,
411 .mbclen_charfound_p
= mbclen_charfound_p
,
412 .mbclen_charfound_len
= mbclen_charfound_len
,
413 .enc_name
= enc_name
,
414 .enc_prev_char
= enc_prev_char
,
416 .enc_asciicompat
= enc_asciicompat
,
417 .utf8_encoding
= utf8_encoding
,
418 .enc_associate
= enc_associate
,
419 .ascii8bit_encoding
= ascii8bit_encoding
,
420 .enc_codelen
= enc_codelen
,
421 .enc_mbcput
= enc_mbcput
,
422 .enc_mbclen
= enc_mbclen
,
423 .enc_find_index
= rb_enc_find_index
,
424 .enc_from_index
= enc_from_index
,
425 .enc_isspace
= enc_isspace
,
426 .enc_coderange_7bit
= ENC_CODERANGE_7BIT
,
427 .enc_coderange_unknown
= ENC_CODERANGE_UNKNOWN
,
428 .usascii_encoding
= usascii_encoding
,
429 .enc_coderange_broken
= ENC_CODERANGE_BROKEN
,
430 .enc_mbminlen
= enc_mbminlen
,
431 .enc_isascii
= enc_isascii
,
432 .enc_mbc_to_codepoint
= enc_mbc_to_codepoint
,
434 .local_defined
= local_defined
,
435 .dvar_defined
= dvar_defined
,
437 .syntax_error_append
= syntax_error_append
,
439 .syntax_error_new
= syntax_error_new
,
441 .errinfo
= rb_errinfo
,
442 .set_errinfo
= rb_set_errinfo
,
443 .exc_raise
= rb_exc_raise
,
444 .make_exception
= rb_make_exception
,
446 .sized_xfree
= ruby_sized_xfree
,
447 .sized_realloc_n
= ruby_sized_realloc_n
,
448 .gc_guard
= gc_guard
,
449 .gc_mark
= rb_gc_mark
,
451 .reg_compile
= rb_reg_compile
,
452 .reg_check_preprocess
= rb_reg_check_preprocess
,
453 .memcicmp
= rb_memcicmp
,
455 .compile_warn
= rb_compile_warn
,
456 .compile_warning
= rb_compile_warning
,
459 .verbose
= ruby_verbose2
,
460 .errno_ptr
= rb_errno_ptr2
,
462 .make_backtrace
= rb_make_backtrace
,
464 .scan_hex
= ruby_scan_hex
,
465 .scan_oct
= ruby_scan_oct
,
466 .scan_digits
= ruby_scan_digits
,
467 .strtod
= ruby_strtod
,
473 .eArgError
= arg_error
,
474 .long2int
= rb_long2int
,
477 .static_id2sym
= static_id2sym
,
478 .str_coderange_scan_restartable
= str_coderange_scan_restartable
,
490 rb_parser_t
*parser_params
;
493 struct lex_pointer_string lex_str
;
504 parser_mark(void *ptr
)
506 struct ruby_parser
*parser
= (struct ruby_parser
*)ptr
;
507 rb_ruby_parser_mark(parser
->parser_params
);
509 switch (parser
->type
) {
511 rb_gc_mark(parser
->data
.lex_str
.str
);
514 rb_gc_mark(parser
->data
.lex_io
.file
);
517 rb_gc_mark(parser
->data
.lex_array
.ary
);
519 case lex_type_generic
:
520 /* noop. Caller of rb_parser_compile_generic should mark the objects. */
526 parser_free(void *ptr
)
528 struct ruby_parser
*parser
= (struct ruby_parser
*)ptr
;
529 rb_ruby_parser_free(parser
->parser_params
);
534 parser_memsize(const void *ptr
)
536 struct ruby_parser
*parser
= (struct ruby_parser
*)ptr
;
537 return rb_ruby_parser_memsize(parser
->parser_params
);
540 static const rb_data_type_t ruby_parser_data_type
= {
547 0, 0, RUBY_TYPED_FREE_IMMEDIATELY
550 #ifdef UNIVERSAL_PARSER
551 const rb_parser_config_t
*
552 rb_ruby_parser_config(void)
554 return &rb_global_parser_config
;
558 rb_parser_params_new(void)
560 return rb_ruby_parser_new(&rb_global_parser_config
);
564 rb_parser_params_new(void)
566 return rb_ruby_parser_new();
568 #endif /* UNIVERSAL_PARSER */
573 struct ruby_parser
*parser
;
574 rb_parser_t
*parser_params
;
577 * Create parser_params ahead of vparser because
578 * rb_ruby_parser_new can run GC so if create vparser
579 * first, parser_mark tries to mark not initialized parser_params.
581 parser_params
= rb_parser_params_new();
582 VALUE vparser
= TypedData_Make_Struct(0, struct ruby_parser
,
583 &ruby_parser_data_type
, parser
);
584 parser
->parser_params
= parser_params
;
590 rb_parser_set_options(VALUE vparser
, int print
, int loop
, int chomp
, int split
)
592 struct ruby_parser
*parser
;
594 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
595 rb_ruby_parser_set_options(parser
->parser_params
, print
, loop
, chomp
, split
);
599 rb_parser_set_context(VALUE vparser
, const struct rb_iseq_struct
*base
, int main
)
601 struct ruby_parser
*parser
;
603 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
604 rb_ruby_parser_set_context(parser
->parser_params
, base
, main
);
609 rb_parser_set_script_lines(VALUE vparser
)
611 struct ruby_parser
*parser
;
613 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
614 rb_ruby_parser_set_script_lines(parser
->parser_params
);
618 rb_parser_error_tolerant(VALUE vparser
)
620 struct ruby_parser
*parser
;
622 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
623 rb_ruby_parser_error_tolerant(parser
->parser_params
);
627 rb_parser_keep_tokens(VALUE vparser
)
629 struct ruby_parser
*parser
;
631 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
632 rb_ruby_parser_keep_tokens(parser
->parser_params
);
636 rb_parser_lex_get_str(struct parser_params
*p
, struct lex_pointer_string
*ptr_str
)
638 char *beg
, *end
, *start
;
640 VALUE s
= ptr_str
->str
;
642 beg
= RSTRING_PTR(s
);
643 len
= RSTRING_LEN(s
);
646 if (len
== ptr_str
->ptr
) return 0;
650 end
= memchr(beg
, '\n', len
);
651 if (end
) len
= ++end
- beg
;
653 return rb_str_to_parser_string(p
, rb_str_subseq(s
, beg
- start
, len
));
656 static rb_parser_string_t
*
657 lex_get_str(struct parser_params
*p
, rb_parser_input_data input
, int line_count
)
659 return rb_parser_lex_get_str(p
, (struct lex_pointer_string
*)input
);
662 static void parser_aset_script_lines_for(VALUE path
, rb_parser_ary_t
*lines
);
665 parser_compile(rb_parser_t
*p
, rb_parser_lex_gets_func
*gets
, VALUE fname
, rb_parser_input_data input
, int line
)
670 rb_encoding
*enc
= 0;
673 StringValueCStr(fname
);
674 ptr
= RSTRING_PTR(fname
);
675 len
= RSTRING_LEN(fname
);
676 enc
= rb_enc_get(fname
);
679 ast
= rb_parser_compile(p
, gets
, ptr
, len
, enc
, input
, line
);
680 parser_aset_script_lines_for(fname
, ast
->body
.script_lines
);
685 parser_compile_string0(struct ruby_parser
*parser
, VALUE fname
, VALUE s
, int line
)
687 VALUE str
= rb_str_new_frozen(s
);
689 parser
->type
= lex_type_str
;
690 parser
->data
.lex_str
.str
= str
;
691 parser
->data
.lex_str
.ptr
= 0;
693 return parser_compile(parser
->parser_params
, lex_get_str
, fname
, (rb_parser_input_data
)&parser
->data
, line
);
697 must_be_ascii_compatible(VALUE s
)
699 rb_encoding
*enc
= rb_enc_get(s
);
700 if (!rb_enc_asciicompat(enc
)) {
701 rb_raise(rb_eArgError
, "invalid source encoding");
707 parser_compile_string_path(struct ruby_parser
*parser
, VALUE f
, VALUE s
, int line
)
709 must_be_ascii_compatible(s
);
710 return parser_compile_string0(parser
, f
, s
, line
);
714 parser_compile_string(struct ruby_parser
*parser
, const char *f
, VALUE s
, int line
)
716 return parser_compile_string_path(parser
, rb_filesystem_str_new_cstr(f
), s
, line
);
719 VALUE
rb_io_gets_internal(VALUE io
);
721 static rb_parser_string_t
*
722 lex_io_gets(struct parser_params
*p
, rb_parser_input_data input
, int line_count
)
724 VALUE io
= (VALUE
)input
;
725 VALUE line
= rb_io_gets_internal(io
);
726 if (NIL_P(line
)) return 0;
727 return rb_str_to_parser_string(p
, line
);
730 static rb_parser_string_t
*
731 lex_gets_array(struct parser_params
*p
, rb_parser_input_data data
, int index
)
733 VALUE array
= (VALUE
)data
;
734 VALUE str
= rb_ary_entry(array
, index
);
737 if (!rb_enc_asciicompat(rb_enc_get(str
))) {
738 rb_raise(rb_eArgError
, "invalid source encoding");
740 return rb_str_to_parser_string(p
, str
);
748 parser_compile_file_path(struct ruby_parser
*parser
, VALUE fname
, VALUE file
, int start
)
750 parser
->type
= lex_type_io
;
751 parser
->data
.lex_io
.file
= file
;
753 return parser_compile(parser
->parser_params
, lex_io_gets
, fname
, (rb_parser_input_data
)file
, start
);
757 parser_compile_array(struct ruby_parser
*parser
, VALUE fname
, VALUE array
, int start
)
759 parser
->type
= lex_type_array
;
760 parser
->data
.lex_array
.ary
= array
;
762 return parser_compile(parser
->parser_params
, lex_gets_array
, fname
, (rb_parser_input_data
)array
, start
);
766 parser_compile_generic(struct ruby_parser
*parser
, rb_parser_lex_gets_func
*lex_gets
, VALUE fname
, VALUE input
, int start
)
768 parser
->type
= lex_type_generic
;
770 return parser_compile(parser
->parser_params
, lex_gets
, fname
, (rb_parser_input_data
)input
, start
);
776 rb_ast_t
*ast
= (rb_ast_t
*)ptr
;
780 static const rb_data_type_t ast_data_type
= {
785 NULL
, // No dsize() because this object does not appear in ObjectSpace.
787 0, 0, RUBY_TYPED_FREE_IMMEDIATELY
793 return TypedData_Wrap_Struct(0, &ast_data_type
, NULL
);
797 rb_parser_compile_file_path(VALUE vparser
, VALUE fname
, VALUE file
, int start
)
799 struct ruby_parser
*parser
;
800 VALUE ast_value
= ast_alloc();
802 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
803 DATA_PTR(ast_value
) = parser_compile_file_path(parser
, fname
, file
, start
);
804 RB_GC_GUARD(vparser
);
810 rb_parser_compile_array(VALUE vparser
, VALUE fname
, VALUE array
, int start
)
812 struct ruby_parser
*parser
;
813 VALUE ast_value
= ast_alloc();
815 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
816 DATA_PTR(ast_value
) = parser_compile_array(parser
, fname
, array
, start
);
817 RB_GC_GUARD(vparser
);
823 rb_parser_compile_generic(VALUE vparser
, rb_parser_lex_gets_func
*lex_gets
, VALUE fname
, VALUE input
, int start
)
825 struct ruby_parser
*parser
;
826 VALUE ast_value
= ast_alloc();
828 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
829 DATA_PTR(ast_value
) = parser_compile_generic(parser
, lex_gets
, fname
, input
, start
);
830 RB_GC_GUARD(vparser
);
836 rb_parser_compile_string(VALUE vparser
, const char *f
, VALUE s
, int line
)
838 struct ruby_parser
*parser
;
839 VALUE ast_value
= ast_alloc();
841 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
842 DATA_PTR(ast_value
) = parser_compile_string(parser
, f
, s
, line
);
843 RB_GC_GUARD(vparser
);
849 rb_parser_compile_string_path(VALUE vparser
, VALUE f
, VALUE s
, int line
)
851 struct ruby_parser
*parser
;
852 VALUE ast_value
= ast_alloc();
854 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
855 DATA_PTR(ast_value
) = parser_compile_string_path(parser
, f
, s
, line
);
856 RB_GC_GUARD(vparser
);
862 rb_parser_encoding(VALUE vparser
)
864 struct ruby_parser
*parser
;
866 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
867 return rb_enc_from_encoding(rb_ruby_parser_encoding(parser
->parser_params
));
871 rb_parser_end_seen_p(VALUE vparser
)
873 struct ruby_parser
*parser
;
875 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
876 return RBOOL(rb_ruby_parser_end_seen_p(parser
->parser_params
));
880 rb_parser_set_yydebug(VALUE vparser
, VALUE flag
)
882 struct ruby_parser
*parser
;
884 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
885 rb_ruby_parser_set_yydebug(parser
->parser_params
, RTEST(flag
));
890 rb_set_script_lines_for(VALUE vparser
, VALUE path
)
892 struct ruby_parser
*parser
;
895 CONST_ID(script_lines
, "SCRIPT_LINES__");
896 if (!rb_const_defined_at(rb_cObject
, script_lines
)) return;
897 hash
= rb_const_get_at(rb_cObject
, script_lines
);
898 if (RB_TYPE_P(hash
, T_HASH
)) {
899 rb_hash_aset(hash
, path
, Qtrue
);
900 TypedData_Get_Struct(vparser
, struct ruby_parser
, &ruby_parser_data_type
, parser
);
901 rb_ruby_parser_set_script_lines(parser
->parser_params
);
906 rb_parser_build_script_lines_from(rb_parser_ary_t
*lines
)
909 if (!lines
) return Qnil
;
910 if (lines
->data_type
!= PARSER_ARY_DATA_SCRIPT_LINE
) {
911 rb_bug("unexpected rb_parser_ary_data_type (%d) for script lines", lines
->data_type
);
913 VALUE script_lines
= rb_ary_new_capa(lines
->len
);
914 for (i
= 0; i
< lines
->len
; i
++) {
915 rb_parser_string_t
*str
= (rb_parser_string_t
*)lines
->data
[i
];
916 rb_ary_push(script_lines
, rb_enc_str_new(str
->ptr
, str
->len
, str
->enc
));
922 rb_str_new_parser_string(rb_parser_string_t
*str
)
924 VALUE string
= rb_enc_interned_str(str
->ptr
, str
->len
, str
->enc
);
925 rb_enc_str_coderange(string
);
930 rb_str_new_mutable_parser_string(rb_parser_string_t
*str
)
932 return rb_enc_str_new(str
->ptr
, str
->len
, str
->enc
);
936 negative_numeric(VALUE val
)
939 return LONG2FIX(-FIX2LONG(val
));
941 if (SPECIAL_CONST_P(val
)) {
944 return DBL2NUM(-RFLOAT_VALUE(val
));
949 switch (BUILTIN_TYPE(val
)) {
952 val
= rb_big_norm(val
);
955 RATIONAL_SET_NUM(val
, negative_numeric(RRATIONAL(val
)->num
));
958 RCOMPLEX_SET_REAL(val
, negative_numeric(RCOMPLEX(val
)->real
));
959 RCOMPLEX_SET_IMAG(val
, negative_numeric(RCOMPLEX(val
)->imag
));
962 val
= DBL2NUM(-RFLOAT_VALUE(val
));
966 rb_bug("unknown literal type (%s) passed to negative_numeric",
967 rb_builtin_class_name(val
));
974 integer_value(const char *val
, int base
)
976 return rb_cstr_to_inum(val
, base
, FALSE
);
980 rational_value(const char *node_val
, int base
, int seen_point
)
983 char* val
= strdup(node_val
);
984 if (seen_point
> 0) {
985 int len
= (int)(strlen(val
));
986 char *point
= &val
[seen_point
];
987 size_t fraclen
= len
-seen_point
-1;
988 memmove(point
, point
+1, fraclen
+1);
990 lit
= rb_rational_new(integer_value(val
, base
), rb_int_positive_pow(10, fraclen
));
993 lit
= rb_rational_raw1(integer_value(val
, base
));
1002 rb_node_integer_literal_val(const NODE
*n
)
1004 const rb_node_integer_t
*node
= RNODE_INTEGER(n
);
1005 VALUE val
= integer_value(node
->val
, node
->base
);
1007 val
= negative_numeric(val
);
1013 rb_node_float_literal_val(const NODE
*n
)
1015 const rb_node_float_t
*node
= RNODE_FLOAT(n
);
1016 double d
= strtod(node
->val
, 0);
1020 VALUE val
= DBL2NUM(d
);
1025 rb_node_rational_literal_val(const NODE
*n
)
1028 const rb_node_rational_t
*node
= RNODE_RATIONAL(n
);
1030 lit
= rational_value(node
->val
, node
->base
, node
->seen_point
);
1033 lit
= negative_numeric(lit
);
1040 rb_node_imaginary_literal_val(const NODE
*n
)
1043 const rb_node_imaginary_t
*node
= RNODE_IMAGINARY(n
);
1045 enum rb_numeric_type type
= node
->type
;
1048 case integer_literal
:
1049 lit
= integer_value(node
->val
, node
->base
);
1051 case float_literal
:{
1052 double d
= strtod(node
->val
, 0);
1056 case rational_literal
:
1057 lit
= rational_value(node
->val
, node
->base
, node
->seen_point
);
1060 rb_bug("unreachable");
1063 lit
= rb_complex_raw(INT2FIX(0), lit
);
1066 lit
= negative_numeric(lit
);
1072 rb_node_str_string_val(const NODE
*node
)
1074 rb_parser_string_t
*str
= RNODE_STR(node
)->string
;
1075 return rb_str_new_parser_string(str
);
1079 rb_node_sym_string_val(const NODE
*node
)
1081 rb_parser_string_t
*str
= RNODE_SYM(node
)->string
;
1082 return ID2SYM(rb_intern3(str
->ptr
, str
->len
, str
->enc
));
1086 rb_node_dstr_string_val(const NODE
*node
)
1088 rb_parser_string_t
*str
= RNODE_DSTR(node
)->string
;
1089 return str
? rb_str_new_parser_string(str
) : Qnil
;
1093 rb_node_dregx_string_val(const NODE
*node
)
1095 rb_parser_string_t
*str
= RNODE_DREGX(node
)->string
;
1096 return rb_str_new_parser_string(str
);
1100 rb_node_regx_string_val(const NODE
*node
)
1102 rb_node_regx_t
*node_reg
= RNODE_REGX(node
);
1103 rb_parser_string_t
*string
= node_reg
->string
;
1104 VALUE str
= rb_enc_str_new(string
->ptr
, string
->len
, string
->enc
);
1106 return rb_reg_compile(str
, node_reg
->options
, NULL
, 0);
1110 rb_node_line_lineno_val(const NODE
*node
)
1112 return INT2FIX(node
->nd_loc
.beg_pos
.lineno
);
1116 rb_node_file_path_val(const NODE
*node
)
1118 return rb_str_new_parser_string(RNODE_FILE(node
)->path
);
1122 rb_node_encoding_val(const NODE
*node
)
1124 return rb_enc_from_encoding(RNODE_ENCODING(node
)->enc
);
1128 parser_aset_script_lines_for(VALUE path
, rb_parser_ary_t
*lines
)
1130 VALUE hash
, script_lines
;
1132 if (NIL_P(path
) || !lines
) return;
1133 CONST_ID(script_lines_id
, "SCRIPT_LINES__");
1134 if (!rb_const_defined_at(rb_cObject
, script_lines_id
)) return;
1135 hash
= rb_const_get_at(rb_cObject
, script_lines_id
);
1136 if (!RB_TYPE_P(hash
, T_HASH
)) return;
1137 if (rb_hash_lookup(hash
, path
) == Qnil
) return;
1138 script_lines
= rb_parser_build_script_lines_from(lines
);
1139 rb_hash_aset(hash
, path
, script_lines
);
1143 rb_ruby_ast_new(const NODE
*const root
)
1146 VALUE ast_value
= TypedData_Make_Struct(0, rb_ast_t
, &ast_data_type
, ast
);
1147 #ifdef UNIVERSAL_PARSER
1148 ast
->config
= &rb_global_parser_config
;
1150 ast
->body
= (rb_ast_body_t
){
1152 .frozen_string_literal
= -1,
1153 .coverage_enabled
= -1,
1154 .script_lines
= NULL
,
1161 rb_ruby_ast_data_get(VALUE ast_value
)
1164 if (NIL_P(ast_value
)) return NULL
;
1165 TypedData_Get_Struct(ast_value
, rb_ast_t
, &ast_data_type
, ast
);