1 #ifndef RUBY_INTERNAL_ENCODING_ENCODING_H
2 #define RUBY_INTERNAL_ENCODING_ENCODING_H
24 #include "ruby/oniguruma.h"
51 enum ruby_encoding_consts {
54 RUBY_ENCODING_INLINE_MAX = 127,
60 RUBY_ENCODING_MASK = (RUBY_ENCODING_INLINE_MAX<<RUBY_ENCODING_SHIFT
64 RUBY_ENCODING_MAXNAMELEN = 42
67 #define ENCODING_INLINE_MAX RUBY_ENCODING_INLINE_MAX
68 #define ENCODING_SHIFT RUBY_ENCODING_SHIFT
69 #define ENCODING_MASK RUBY_ENCODING_MASK
85 f <<= RUBY_ENCODING_SHIFT;
103 return RBIMPL_CAST((
int)ret);
106 #define ENCODING_SET_INLINED(obj,i) RB_ENCODING_SET_INLINED(obj,i)
107 #define ENCODING_SET(obj,i) RB_ENCODING_SET(obj,i)
108 #define ENCODING_GET_INLINED(obj) RB_ENCODING_GET_INLINED(obj)
109 #define ENCODING_GET(obj) RB_ENCODING_GET(obj)
110 #define ENCODING_IS_ASCII8BIT(obj) RB_ENCODING_IS_ASCII8BIT(obj)
111 #define ENCODING_MAXNAMELEN RUBY_ENCODING_MAXNAMELEN
199 if (encindex == RUBY_ENCODING_INLINE_MAX) {
416 static
inline const
char *
434 return enc->min_enc_len;
449 return enc->max_enc_len;
516 #define MBCLEN_CHARFOUND_P(ret) ONIGENC_MBCLEN_CHARFOUND_P(ret)
517 #define MBCLEN_CHARFOUND_LEN(ret) ONIGENC_MBCLEN_CHARFOUND_LEN(ret)
518 #define MBCLEN_INVALID_P(ret) ONIGENC_MBCLEN_INVALID_P(ret)
519 #define MBCLEN_NEEDMORE_P(ret) ONIGENC_MBCLEN_NEEDMORE_P(ret)
520 #define MBCLEN_NEEDMORE_LEN(ret) ONIGENC_MBCLEN_NEEDMORE_LEN(ret)
570 static inline unsigned int
590 static inline OnigCodePoint
593 const OnigUChar *up = RBIMPL_CAST((
const OnigUChar *)p);
594 const OnigUChar *ue = RBIMPL_CAST((
const OnigUChar *)e);
596 return ONIGENC_MBC_TO_CODE(enc, up, ue);
621 OnigCodePoint uc = RBIMPL_CAST((OnigCodePoint)c);
623 return ONIGENC_CODE_TO_MBCLEN(enc, uc);
645 OnigCodePoint uc = RBIMPL_CAST((OnigCodePoint)c);
646 OnigUChar *ubuf = RBIMPL_CAST((OnigUChar *)buf);
648 return ONIGENC_CODE_TO_MBC(enc, uc, ubuf);
664 const OnigUChar *us = RBIMPL_CAST((
const OnigUChar *)s);
665 const OnigUChar *up = RBIMPL_CAST((
const OnigUChar *)p);
666 const OnigUChar *ue = RBIMPL_CAST((
const OnigUChar *)e);
667 OnigUChar *ur = onigenc_get_prev_char_head(enc, us, up, ue);
669 return RBIMPL_CAST((
char *)ur);
685 const OnigUChar *us = RBIMPL_CAST((
const OnigUChar *)s);
686 const OnigUChar *up = RBIMPL_CAST((
const OnigUChar *)p);
687 const OnigUChar *ue = RBIMPL_CAST((
const OnigUChar *)e);
688 OnigUChar *ur = onigenc_get_left_adjust_char_head(enc, us, up, ue);
690 return RBIMPL_CAST((
char *)ur);
706 const OnigUChar *us = RBIMPL_CAST((
const OnigUChar *)s);
707 const OnigUChar *up = RBIMPL_CAST((
const OnigUChar *)p);
708 const OnigUChar *ue = RBIMPL_CAST((
const OnigUChar *)e);
709 OnigUChar *ur = onigenc_get_right_adjust_char_head(enc, us, up, ue);
711 return RBIMPL_CAST((
char *)ur);
728 const OnigUChar *us = RBIMPL_CAST((
const OnigUChar *)s);
729 const OnigUChar *up = RBIMPL_CAST((
const OnigUChar *)p);
730 const OnigUChar *ue = RBIMPL_CAST((
const OnigUChar *)e);
731 const OnigUChar *ur = onigenc_step_back(enc, us, up, ue, n);
733 return RBIMPL_CAST((
char *)ur);
902 #ifndef rb_ascii8bit_encindex
932 #ifndef rb_utf8_encindex
943 #ifndef rb_usascii_encindex
1024 #define RB_ENCODING_GET RB_ENCODING_GET
1025 #define RB_ENCODING_GET_INLINED RB_ENCODING_GET_INLINED
1026 #define RB_ENCODING_IS_ASCII8BIT RB_ENCODING_IS_ASCII8BIT
1027 #define RB_ENCODING_SET RB_ENCODING_SET
1028 #define RB_ENCODING_SET_INLINED RB_ENCODING_SET_INLINED
1029 #define rb_enc_asciicompat rb_enc_asciicompat
1030 #define rb_enc_code_to_mbclen rb_enc_code_to_mbclen
1031 #define rb_enc_codepoint rb_enc_codepoint
1032 #define rb_enc_left_char_head rb_enc_left_char_head
1033 #define rb_enc_mbc_to_codepoint rb_enc_mbc_to_codepoint
1034 #define rb_enc_mbcput rb_enc_mbcput
1035 #define rb_enc_mbmaxlen rb_enc_mbmaxlen
1036 #define rb_enc_mbminlen rb_enc_mbminlen
1037 #define rb_enc_name rb_enc_name
1038 #define rb_enc_prev_char rb_enc_prev_char
1039 #define rb_enc_right_char_head rb_enc_right_char_head
1040 #define rb_enc_step_back rb_enc_step_back
1041 #define rb_enc_str_asciicompat_p rb_enc_str_asciicompat_p
Routines for code ranges.
ruby_coderange_type
What rb_enc_str_coderange() returns.
static void RB_ENC_CODERANGE_SET(VALUE obj, enum ruby_coderange_type cr)
Destructively modifies the passed object so that its (inline) code range is the passed one.
Defines RBIMPL_ATTR_CONST.
Defines RBIMPL_ATTR_DEPRECATED.
Tweaking visibility of C variables/functions.
#define RUBY_EXTERN
Declaration of externally visible global variables.
#define RBIMPL_SYMBOL_EXPORT_END()
Counterpart of RBIMPL_SYMBOL_EXPORT_BEGIN.
#define RBIMPL_SYMBOL_EXPORT_BEGIN()
Shortcut macro equivalent to RUBY_SYMBOL_EXPORT_BEGIN extern "C" {.
Defines enum ruby_fl_type.
@ RUBY_FL_USHIFT
Number of bits in ruby_fl_type that are not open to users.
static VALUE RB_FL_TEST_RAW(VALUE obj, VALUE flags)
This is an implementation detail of RB_FL_TEST().
static void RB_FL_SET_RAW(VALUE obj, VALUE flags)
This is an implementation detail of RB_FL_SET().
static void RB_FL_UNSET_RAW(VALUE obj, VALUE flags)
This is an implementation detail of RB_FL_UNSET().
VALUE rb_cEncoding
Encoding class.
rb_encoding * rb_locale_encoding(void)
Queries the encoding that represents the current locale.
rb_encoding * rb_default_external_encoding(void)
Queries the "default external" encoding.
int rb_enc_dummy_p(rb_encoding *enc)
Queries if the passed encoding is dummy.
int rb_enc_precise_mbclen(const char *p, const char *e, rb_encoding *enc)
Queries the number of bytes of the character at the passed pointer.
int rb_enc_get_index(VALUE obj)
Queries the index of the encoding of the passed object, if any.
int rb_to_encoding_index(VALUE obj)
Obtains a encoding index from a wider range of objects (than rb_enc_find_index()).
int rb_filesystem_encindex(void)
Identical to rb_filesystem_encoding(), except it returns the encoding's index instead of the encoding...
VALUE rb_enc_associate(VALUE obj, rb_encoding *enc)
Identical to rb_enc_associate_index(), except it takes an encoding itself instead of its index.
rb_encoding * rb_usascii_encoding(void)
Queries the encoding that represents US-ASCII.
static void RB_ENCODING_SET_INLINED(VALUE obj, int encindex)
Destructively assigns the passed encoding to the passed object.
static bool RB_ENCODING_IS_ASCII8BIT(VALUE obj)
Queries if the passed object is in ascii 8bit (== binary) encoding.
int rb_enc_codelen(int code, rb_encoding *enc)
Queries the number of bytes requested to represent the passed code point using the passed encoding.
const OnigEncodingType rb_encoding
The type of encoding.
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.
void rb_enc_copy(VALUE dst, VALUE src)
Destructively copies the encoding of the latter object to that of former one.
int rb_utf8_encindex(void)
Identical to rb_utf8_encoding(), except it returns the encoding's index instead of the encoding itsel...
int rb_enc_fast_mbclen(const char *p, const char *e, rb_encoding *enc)
Identical to rb_enc_mbclen() unless the character at p overruns e.
static int RB_ENCODING_GET(VALUE obj)
Just another name of rb_enc_get_index.
int rb_ascii8bit_encindex(void)
Identical to rb_ascii8bit_encoding(), except it returns the encoding's index instead of the encoding ...
rb_encoding * rb_enc_compatible(VALUE str1, VALUE str2)
Look for the "common" encoding between the two.
unsigned int rb_enc_codepoint_len(const char *p, const char *e, int *len, rb_encoding *enc)
Queries the code point of character pointed by the passed pointer.
int rb_enc_unicode_p(rb_encoding *enc)
Queries if the passed encoding is either one of UTF-8/16/32.
rb_encoding * rb_default_internal_encoding(void)
Queries the "default internal" encoding.
int rb_enc_to_index(rb_encoding *enc)
Queries the index of the encoding.
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.
rb_encoding * rb_find_encoding(VALUE obj)
Identical to rb_to_encoding_index(), except the return type.
rb_encoding * rb_enc_find(const char *name)
Identical to rb_find_encoding(), except it takes a C's string instead of Ruby's.
rb_encoding * rb_to_encoding(VALUE obj)
Identical to rb_find_encoding(), except it raises an exception instead of returning NULL.
void rb_enc_set_index(VALUE obj, int encindex)
Destructively assigns an encoding (via its index) to an object.
VALUE rb_locale_charmap(VALUE klass)
Returns a platform-depended "charmap" of the current locale.
rb_encoding * rb_ascii8bit_encoding(void)
Queries the encoding that represents ASCII-8BIT a.k.a.
void rb_enc_set_default_internal(VALUE encoding)
Destructively assigns the passed encoding as the default internal encoding.
VALUE rb_enc_default_external(void)
Identical to rb_default_external_encoding(), except it returns the Ruby-level counterpart instance of...
rb_encoding * rb_enc_check(VALUE str1, VALUE str2)
Identical to rb_enc_compatible(), except it raises an exception instead of returning NULL.
VALUE rb_enc_from_encoding(rb_encoding *enc)
Queries the Ruby-level counterpart instance of rb_cEncoding that corresponds to the passed encoding.
static bool rb_enc_asciicompat(rb_encoding *enc)
Queries if the passed encoding is in some sense compatible with ASCII.
int rb_define_dummy_encoding(const char *name)
Creates a new "dummy" encoding.
rb_encoding * rb_utf8_encoding(void)
Queries the encoding that represents UTF-8.
static char * rb_enc_prev_char(const char *s, const char *p, const char *e, rb_encoding *enc)
Queries the previous (left) character.
static unsigned int rb_enc_codepoint(const char *p, const char *e, rb_encoding *enc)
Queries the code point of character pointed by the passed pointer.
static int rb_enc_mbcput(unsigned int c, void *buf, rb_encoding *enc)
Identical to rb_enc_uint_chr(), except it writes back to the passed buffer instead of allocating one.
int rb_locale_encindex(void)
Identical to rb_locale_encoding(), except it returns the encoding's index instead of the encoding its...
int rb_char_to_option_kcode(int c, int *option, int *kcode)
Converts a character option to its encoding.
rb_encoding * rb_enc_from_index(int idx)
Identical to rb_find_encoding(), except it takes an encoding index instead of a Ruby object.
static int rb_enc_mbmaxlen(rb_encoding *enc)
Queries the maximum number of bytes that the passed encoding needs to represent a character.
int rb_enc_mbclen(const char *p, const char *e, rb_encoding *enc)
Queries the number of bytes of the character at the passed pointer.
static void RB_ENCODING_SET(VALUE obj, int encindex)
Just another name of rb_enc_set_index.
int rb_enc_capable(VALUE obj)
Queries if the passed object can have its encoding.
static void RB_ENCODING_CODERANGE_SET(VALUE obj, int encindex, enum ruby_coderange_type cr)
This is RB_ENCODING_SET + RB_ENC_CODERANGE_SET combo.
VALUE rb_enc_default_internal(void)
Identical to rb_default_internal_encoding(), except it returns the Ruby-level counterpart instance of...
VALUE rb_enc_associate_index(VALUE obj, int encindex)
Identical to rb_enc_set_index(), except it additionally does contents fix-up depending on the passed ...
rb_encoding * rb_enc_get(VALUE obj)
Identical to rb_enc_get_index(), except the return type.
static int RB_ENCODING_GET_INLINED(VALUE obj)
Queries the encoding of the passed object.
static OnigCodePoint rb_enc_mbc_to_codepoint(const char *p, const char *e, rb_encoding *enc)
Identical to rb_enc_codepoint(), except it assumes the passed character is not broken.
static const char * rb_enc_name(rb_encoding *enc)
Queries the (canonical) name of the passed encoding.
void rb_enc_set_default_external(VALUE encoding)
Destructively assigns the passed encoding as the default external encoding.
static char * rb_enc_step_back(const char *s, const char *p, const char *e, int n, rb_encoding *enc)
Scans the string backwards for n characters.
int rb_enc_find_index(const char *name)
Queries the index of the encoding.
static int rb_enc_mbminlen(rb_encoding *enc)
Queries the minimum number of bytes that the passed encoding needs to represent a character.
int rb_enc_alias(const char *alias, const char *orig)
Registers an "alias" name.
static int rb_enc_code_to_mbclen(int c, rb_encoding *enc)
Identical to rb_enc_codelen(), except it returns 0 for invalid code points.
int rb_enc_ascget(const char *p, const char *e, int *len, rb_encoding *enc)
Queries the code point of character pointed by the passed pointer.
static bool rb_enc_str_asciicompat_p(VALUE str)
Queries if the passed string is in an ASCII-compatible encoding.
int rb_usascii_encindex(void)
Identical to rb_usascii_encoding(), except it returns the encoding's index instead of the encoding it...
rb_encoding * rb_filesystem_encoding(void)
Queries the "filesystem" encoding.
RBIMPL_ATTR_CONST() int rb_io_oflags_fmode(int oflags)
Converts an oflags (that rb_io_modestr_oflags() returns) to a fmode (that rb_io_mode_flags() returns)...
RBIMPL_ATTR_PURE() int rb_io_read_pending(rb_io_t *fptr)
Queries if the passed IO has any pending reads.
int len
Length of the buffer.
Defines RBIMPL_ATTR_NOALIAS.
#define RBIMPL_ATTR_NOALIAS()
Wraps (or simulates) __declspec((noalias))
#define inline
Old Visual Studio versions do not support the inline keyword, so we need to define it to be __inline.
Defines RBIMPL_ATTR_PURE.
Defines RBIMPL_ATTR_RETURNS_NONNULL.
#define RBIMPL_ATTR_RETURNS_NONNULL()
Wraps (or simulates) __attribute__((returns_nonnull))
uintptr_t VALUE
Type that represents a Ruby object.