Ruby 3.5.0dev (2025-04-27 revision 3ec7bfff2e7ac4f6f69d26676edcfd2e73ea3b05)
string.h
1#ifndef INTERNAL_STRING_H /*-*-C-*-vi:se ft=c:*/
2#define INTERNAL_STRING_H
11#include "ruby/internal/config.h"
12#include <stddef.h> /* for size_t */
13#include "internal/compilers.h" /* for __has_builtin */
14#include "ruby/internal/stdbool.h" /* for bool */
15#include "ruby/encoding.h" /* for rb_encoding */
16#include "ruby/ruby.h" /* for VALUE */
17
18#define STR_SHARED FL_USER0 /* = ELTS_SHARED */
19#define STR_NOEMBED FL_USER1
20#define STR_CHILLED (FL_USER2 | FL_USER3)
21#define STR_CHILLED_LITERAL FL_USER2
22#define STR_CHILLED_SYMBOL_TO_S FL_USER3
23
24enum ruby_rstring_private_flags {
25 RSTRING_CHILLED = STR_CHILLED,
26};
27
28#ifdef rb_fstring_cstr
29# undef rb_fstring_cstr
30#endif
31
32/* string.c */
33VALUE rb_fstring(VALUE);
34VALUE rb_fstring_cstr(const char *str);
35VALUE rb_fstring_enc_new(const char *ptr, long len, rb_encoding *enc);
36int rb_str_buf_cat_escaped_char(VALUE result, unsigned int c, int unicode_p);
37int rb_str_symname_p(VALUE);
38VALUE rb_str_quote_unprintable(VALUE);
39char *rb_str_fill_terminator(VALUE str, const int termlen);
40void rb_str_change_terminator_length(VALUE str, const int oldtermlen, const int termlen);
41VALUE rb_str_locktmp_ensure(VALUE str, VALUE (*func)(VALUE), VALUE arg);
42VALUE rb_str_chomp_string(VALUE str, VALUE chomp);
43VALUE rb_external_str_with_enc(VALUE str, rb_encoding *eenc);
44VALUE rb_str_cat_conv_enc_opts(VALUE newstr, long ofs, const char *ptr, long len,
45 rb_encoding *from, int ecflags, VALUE ecopts);
46VALUE rb_enc_str_scrub(rb_encoding *enc, VALUE str, VALUE repl);
47VALUE rb_str_escape(VALUE str);
48size_t rb_str_memsize(VALUE);
49char *rb_str_to_cstr(VALUE str);
50const char *ruby_escaped_char(int c);
51void rb_str_make_independent(VALUE str);
52int rb_enc_str_coderange_scan(VALUE str, rb_encoding *enc);
53int rb_ascii8bit_appendable_encoding_index(rb_encoding *enc, unsigned int code);
54VALUE rb_str_include(VALUE str, VALUE arg);
55VALUE rb_str_byte_substr(VALUE str, VALUE beg, VALUE len);
56VALUE rb_str_substr_two_fixnums(VALUE str, VALUE beg, VALUE len, int empty);
57VALUE rb_str_tmp_frozen_no_embed_acquire(VALUE str);
58void rb_str_make_embedded(VALUE);
59VALUE rb_str_upto_each(VALUE, VALUE, int, int (*each)(VALUE, VALUE), VALUE);
60size_t rb_str_size_as_embedded(VALUE);
61bool rb_str_reembeddable_p(VALUE);
62VALUE rb_str_upto_endless_each(VALUE, int (*each)(VALUE, VALUE), VALUE);
63VALUE rb_str_with_debug_created_info(VALUE, VALUE, int);
64VALUE rb_str_frozen_bare_string(VALUE);
65
66/* error.c */
67void rb_warn_unchilled_literal(VALUE str);
68void rb_warn_unchilled_symbol_to_s(VALUE str);
69
70static inline bool STR_EMBED_P(VALUE str);
71static inline bool STR_SHARED_P(VALUE str);
72static inline VALUE QUOTE(VALUE v);
73static inline VALUE QUOTE_ID(ID v);
74static inline bool is_ascii_string(VALUE str);
75static inline bool is_broken_string(VALUE str);
76static inline VALUE rb_str_eql_internal(const VALUE str1, const VALUE str2);
77
78RUBY_SYMBOL_EXPORT_BEGIN
79/* string.c (export) */
80VALUE rb_str_tmp_frozen_acquire(VALUE str);
81void rb_str_tmp_frozen_release(VALUE str, VALUE tmp);
82VALUE rb_setup_fake_str(struct RString *fake_str, const char *name, long len, rb_encoding *enc);
83RUBY_SYMBOL_EXPORT_END
84
85VALUE rb_fstring_new(const char *ptr, long len);
86void rb_gc_free_fstring(VALUE obj);
87bool rb_obj_is_fstring_table(VALUE obj);
88void Init_fstring_table();
89VALUE rb_obj_as_string_result(VALUE str, VALUE obj);
90VALUE rb_str_opt_plus(VALUE x, VALUE y);
91VALUE rb_str_concat_literals(size_t num, const VALUE *strary);
92VALUE rb_str_eql(VALUE str1, VALUE str2);
93VALUE rb_id_quote_unprintable(ID);
94VALUE rb_sym_proc_call(ID mid, int argc, const VALUE *argv, int kw_splat, VALUE passed_proc);
95VALUE rb_enc_literal_str(const char *ptr, long len, rb_encoding *enc);
96
98VALUE rb_ec_str_resurrect(struct rb_execution_context_struct *ec, VALUE str, bool chilled);
99
100#define rb_fstring_lit(str) rb_fstring_new((str), rb_strlen_lit(str))
101#define rb_fstring_literal(str) rb_fstring_lit(str)
102#define rb_fstring_enc_lit(str, enc) rb_fstring_enc_new((str), rb_strlen_lit(str), (enc))
103#define rb_fstring_enc_literal(str, enc) rb_fstring_enc_lit(str, enc)
104
105static inline VALUE
106QUOTE(VALUE v)
107{
108 return rb_str_quote_unprintable(v);
109}
110
111static inline VALUE
112QUOTE_ID(ID i)
113{
114 return rb_id_quote_unprintable(i);
115}
116
117static inline bool
118STR_EMBED_P(VALUE str)
119{
120 return ! FL_TEST_RAW(str, STR_NOEMBED);
121}
122
123static inline bool
124STR_SHARED_P(VALUE str)
125{
126 return FL_ALL_RAW(str, STR_NOEMBED | STR_SHARED);
127}
128
129static inline bool
130CHILLED_STRING_P(VALUE obj)
131{
132 return RB_TYPE_P(obj, T_STRING) && FL_TEST_RAW(obj, STR_CHILLED);
133}
134
135static inline void
136CHILLED_STRING_MUTATED(VALUE str)
137{
138 VALUE chilled_reason = RB_FL_TEST_RAW(str, STR_CHILLED);
139 FL_UNSET_RAW(str, STR_CHILLED);
140 switch (chilled_reason) {
141 case STR_CHILLED_SYMBOL_TO_S:
142 rb_warn_unchilled_symbol_to_s(str);
143 break;
144 case STR_CHILLED_LITERAL:
145 rb_warn_unchilled_literal(str);
146 break;
147 default:
148 rb_bug("RString was chilled for multiple reasons");
149 }
150}
151
152static inline bool
153is_ascii_string(VALUE str)
154{
156}
157
158static inline bool
159is_broken_string(VALUE str)
160{
162}
163
164static inline bool
165at_char_boundary(const char *s, const char *p, const char *e, rb_encoding *enc)
166{
167 return rb_enc_left_char_head(s, p, e, enc) == p;
168}
169
170static inline bool
171at_char_right_boundary(const char *s, const char *p, const char *e, rb_encoding *enc)
172{
173 RUBY_ASSERT(s <= p);
174 RUBY_ASSERT(p <= e);
175
176 return rb_enc_right_char_head(s, p, e, enc) == p;
177}
178
179/* expect tail call optimization */
180// YJIT needs this function to never allocate and never raise
181static inline VALUE
182rb_str_eql_internal(const VALUE str1, const VALUE str2)
183{
184 const long len = RSTRING_LEN(str1);
185 const char *ptr1, *ptr2;
186
187 if (len != RSTRING_LEN(str2)) return Qfalse;
188 if (!rb_str_comparable(str1, str2)) return Qfalse;
189 if ((ptr1 = RSTRING_PTR(str1)) == (ptr2 = RSTRING_PTR(str2)))
190 return Qtrue;
191 if (memcmp(ptr1, ptr2, len) == 0)
192 return Qtrue;
193 return Qfalse;
194}
195
196#if __has_builtin(__builtin_constant_p)
197# define rb_fstring_cstr(str) \
198 (__builtin_constant_p(str) ? \
199 rb_fstring_new((str), (long)strlen(str)) : \
200 (rb_fstring_cstr)(str))
201#endif
202#endif /* INTERNAL_STRING_H */
#define RUBY_ASSERT(...)
Asserts that the given expression is truthy if and only if RUBY_DEBUG is truthy.
Definition assert.h:219
static VALUE RB_FL_TEST_RAW(VALUE obj, VALUE flags)
This is an implementation detail of RB_FL_TEST().
Definition fl_type.h:469
#define ENC_CODERANGE_7BIT
Old name of RUBY_ENC_CODERANGE_7BIT.
Definition coderange.h:180
#define FL_UNSET_RAW
Old name of RB_FL_UNSET_RAW.
Definition fl_type.h:134
#define T_STRING
Old name of RUBY_T_STRING.
Definition value_type.h:78
#define FL_TEST_RAW
Old name of RB_FL_TEST_RAW.
Definition fl_type.h:132
#define Qtrue
Old name of RUBY_Qtrue.
#define Qfalse
Old name of RUBY_Qfalse.
#define ENC_CODERANGE_BROKEN
Old name of RUBY_ENC_CODERANGE_BROKEN.
Definition coderange.h:182
#define FL_ALL_RAW
Old name of RB_FL_ALL_RAW.
Definition fl_type.h:124
Encoding relates APIs.
static char * rb_enc_left_char_head(const char *s, const char *p, const char *e, rb_encoding *enc)
Queries the left boundary of a character.
Definition encoding.h:683
static char * rb_enc_right_char_head(const char *s, const char *p, const char *e, rb_encoding *enc)
Queries the right boundary of a character.
Definition encoding.h:704
int rb_enc_str_coderange(VALUE str)
Scans the passed string to collect its code range.
Definition string.c:1270
int rb_str_comparable(VALUE str1, VALUE str2)
Checks if two strings are comparable each other or not.
Definition string.c:4468
int len
Length of the buffer.
Definition io.h:8
C99 shim for <stdbool.h>
Ruby's String.
Definition rstring.h:196
uintptr_t ID
Type that represents a Ruby identifier such as a variable name.
Definition value.h:52
uintptr_t VALUE
Type that represents a Ruby object.
Definition value.h:40
static bool RB_TYPE_P(VALUE obj, enum ruby_value_type t)
Queries if the given object is of given type.
Definition value_type.h:376