PageRenderTime 121ms CodeModel.GetById 22ms RepoModel.GetById 1ms app.codeStats 0ms

/include/ruby/encoding.h

https://github.com/nazy/ruby
C Header | 332 lines | 223 code | 69 blank | 40 comment | 15 complexity | d6ebc9c97f3253f0f8fd3a27f69747d2 MD5 | raw file
Possible License(s): GPL-2.0, BSD-3-Clause, AGPL-3.0, 0BSD, Unlicense
  1. /**********************************************************************
  2. encoding.h -
  3. $Author: matz $
  4. created at: Thu May 24 11:49:41 JST 2007
  5. Copyright (C) 2007 Yukihiro Matsumoto
  6. **********************************************************************/
  7. #ifndef RUBY_ENCODING_H
  8. #define RUBY_ENCODING_H 1
  9. #if defined(__cplusplus)
  10. extern "C" {
  11. #if 0
  12. } /* satisfy cc-mode */
  13. #endif
  14. #endif
  15. #include <stdarg.h>
  16. #include "ruby/oniguruma.h"
  17. #if defined __GNUC__ && __GNUC__ >= 4
  18. #pragma GCC visibility push(default)
  19. #endif
  20. #define ENCODING_INLINE_MAX 1023
  21. #define ENCODING_SHIFT (FL_USHIFT+10)
  22. #define ENCODING_MASK (((VALUE)ENCODING_INLINE_MAX)<<ENCODING_SHIFT)
  23. #define ENCODING_SET_INLINED(obj,i) do {\
  24. RBASIC(obj)->flags &= ~ENCODING_MASK;\
  25. RBASIC(obj)->flags |= (VALUE)(i) << ENCODING_SHIFT;\
  26. } while (0)
  27. #define ENCODING_SET(obj,i) do {\
  28. VALUE rb_encoding_set_obj = (obj); \
  29. int encoding_set_enc_index = (i); \
  30. if (encoding_set_enc_index < ENCODING_INLINE_MAX) \
  31. ENCODING_SET_INLINED(rb_encoding_set_obj, encoding_set_enc_index); \
  32. else \
  33. rb_enc_set_index(rb_encoding_set_obj, encoding_set_enc_index); \
  34. } while (0)
  35. #define ENCODING_GET_INLINED(obj) (int)((RBASIC(obj)->flags & ENCODING_MASK)>>ENCODING_SHIFT)
  36. #define ENCODING_GET(obj) \
  37. (ENCODING_GET_INLINED(obj) != ENCODING_INLINE_MAX ? \
  38. ENCODING_GET_INLINED(obj) : \
  39. rb_enc_get_index(obj))
  40. #define ENCODING_IS_ASCII8BIT(obj) (ENCODING_GET_INLINED(obj) == 0)
  41. #define ENCODING_MAXNAMELEN 42
  42. #define ENC_CODERANGE_MASK ((int)(FL_USER8|FL_USER9))
  43. #define ENC_CODERANGE_UNKNOWN 0
  44. #define ENC_CODERANGE_7BIT ((int)FL_USER8)
  45. #define ENC_CODERANGE_VALID ((int)FL_USER9)
  46. #define ENC_CODERANGE_BROKEN ((int)(FL_USER8|FL_USER9))
  47. #define ENC_CODERANGE(obj) ((int)RBASIC(obj)->flags & ENC_CODERANGE_MASK)
  48. #define ENC_CODERANGE_ASCIIONLY(obj) (ENC_CODERANGE(obj) == ENC_CODERANGE_7BIT)
  49. #define ENC_CODERANGE_SET(obj,cr) (RBASIC(obj)->flags = \
  50. (RBASIC(obj)->flags & ~ENC_CODERANGE_MASK) | (cr))
  51. #define ENC_CODERANGE_CLEAR(obj) ENC_CODERANGE_SET(obj,0)
  52. /* assumed ASCII compatibility */
  53. #define ENC_CODERANGE_AND(a, b) \
  54. (a == ENC_CODERANGE_7BIT ? b : \
  55. a == ENC_CODERANGE_VALID ? (b == ENC_CODERANGE_7BIT ? ENC_CODERANGE_VALID : b) : \
  56. ENC_CODERANGE_UNKNOWN)
  57. #define ENCODING_CODERANGE_SET(obj, encindex, cr) \
  58. do { \
  59. VALUE rb_encoding_coderange_obj = (obj); \
  60. ENCODING_SET(rb_encoding_coderange_obj, (encindex)); \
  61. ENC_CODERANGE_SET(rb_encoding_coderange_obj, (cr)); \
  62. } while (0)
  63. typedef OnigEncodingType rb_encoding;
  64. int rb_char_to_option_kcode(int c, int *option, int *kcode);
  65. int rb_enc_replicate(const char *, rb_encoding *);
  66. int rb_define_dummy_encoding(const char *);
  67. #define rb_enc_to_index(enc) ((enc) ? ENC_TO_ENCINDEX(enc) : 0)
  68. int rb_enc_get_index(VALUE obj);
  69. void rb_enc_set_index(VALUE obj, int encindex);
  70. int rb_enc_find_index(const char *name);
  71. int rb_to_encoding_index(VALUE);
  72. rb_encoding* rb_to_encoding(VALUE);
  73. rb_encoding* rb_enc_get(VALUE);
  74. rb_encoding* rb_enc_compatible(VALUE,VALUE);
  75. rb_encoding* rb_enc_check(VALUE,VALUE);
  76. VALUE rb_enc_associate_index(VALUE, int);
  77. VALUE rb_enc_associate(VALUE, rb_encoding*);
  78. void rb_enc_copy(VALUE dst, VALUE src);
  79. VALUE rb_enc_str_new(const char*, long, rb_encoding*);
  80. VALUE rb_enc_reg_new(const char*, long, rb_encoding*, int);
  81. PRINTF_ARGS(VALUE rb_enc_sprintf(rb_encoding *, const char*, ...), 2, 3);
  82. VALUE rb_enc_vsprintf(rb_encoding *, const char*, va_list);
  83. long rb_enc_strlen(const char*, const char*, rb_encoding*);
  84. char* rb_enc_nth(const char*, const char*, long, rb_encoding*);
  85. VALUE rb_obj_encoding(VALUE);
  86. VALUE rb_enc_str_buf_cat(VALUE str, const char *ptr, long len, rb_encoding *enc);
  87. VALUE rb_external_str_new_with_enc(const char *ptr, long len, rb_encoding *);
  88. VALUE rb_str_export_to_enc(VALUE, rb_encoding *);
  89. VALUE rb_str_conv_enc(VALUE str, rb_encoding *from, rb_encoding *to);
  90. VALUE rb_str_conv_enc_opts(VALUE str, rb_encoding *from, rb_encoding *to, int ecflags, VALUE ecopts);
  91. /* index -> rb_encoding */
  92. rb_encoding* rb_enc_from_index(int idx);
  93. /* name -> rb_encoding */
  94. rb_encoding * rb_enc_find(const char *name);
  95. /* rb_encoding * -> name */
  96. #define rb_enc_name(enc) (enc)->name
  97. /* rb_encoding * -> minlen/maxlen */
  98. #define rb_enc_mbminlen(enc) (enc)->min_enc_len
  99. #define rb_enc_mbmaxlen(enc) (enc)->max_enc_len
  100. /* -> mbclen (no error notification: 0 < ret <= e-p, no exception) */
  101. int rb_enc_mbclen(const char *p, const char *e, rb_encoding *enc);
  102. /* -> mbclen (only for valid encoding) */
  103. int rb_enc_fast_mbclen(const char *p, const char *e, rb_encoding *enc);
  104. /* -> chlen, invalid or needmore */
  105. int rb_enc_precise_mbclen(const char *p, const char *e, rb_encoding *enc);
  106. #define MBCLEN_CHARFOUND_P(ret) ONIGENC_MBCLEN_CHARFOUND_P(ret)
  107. #define MBCLEN_CHARFOUND_LEN(ret) ONIGENC_MBCLEN_CHARFOUND_LEN(ret)
  108. #define MBCLEN_INVALID_P(ret) ONIGENC_MBCLEN_INVALID_P(ret)
  109. #define MBCLEN_NEEDMORE_P(ret) ONIGENC_MBCLEN_NEEDMORE_P(ret)
  110. #define MBCLEN_NEEDMORE_LEN(ret) ONIGENC_MBCLEN_NEEDMORE_LEN(ret)
  111. /* -> 0x00..0x7f, -1 */
  112. int rb_enc_ascget(const char *p, const char *e, int *len, rb_encoding *enc);
  113. /* -> code (and len) or raise exception */
  114. unsigned int rb_enc_codepoint_len(const char *p, const char *e, int *len, rb_encoding *enc);
  115. /* prototype for obsolete function */
  116. unsigned int rb_enc_codepoint(const char *p, const char *e, rb_encoding *enc);
  117. /* overriding macro */
  118. #define rb_enc_codepoint(p,e,enc) rb_enc_codepoint_len((p),(e),0,(enc))
  119. #define rb_enc_mbc_to_codepoint(p, e, enc) ONIGENC_MBC_TO_CODE(enc,(UChar*)(p),(UChar*)(e))
  120. /* -> codelen>0 or raise exception */
  121. int rb_enc_codelen(int code, rb_encoding *enc);
  122. /* code,ptr,encoding -> write buf */
  123. #define rb_enc_mbcput(c,buf,enc) ONIGENC_CODE_TO_MBC(enc,c,(UChar*)(buf))
  124. /* start, ptr, end, encoding -> prev_char */
  125. #define rb_enc_prev_char(s,p,e,enc) (char *)onigenc_get_prev_char_head(enc,(UChar*)(s),(UChar*)(p),(UChar*)(e))
  126. /* start, ptr, end, encoding -> next_char */
  127. #define rb_enc_left_char_head(s,p,e,enc) (char *)onigenc_get_left_adjust_char_head(enc,(UChar*)(s),(UChar*)(p),(UChar*)(e))
  128. #define rb_enc_right_char_head(s,p,e,enc) (char *)onigenc_get_right_adjust_char_head(enc,(UChar*)(s),(UChar*)(p),(UChar*)(e))
  129. /* ptr, ptr, encoding -> newline_or_not */
  130. #define rb_enc_is_newline(p,end,enc) ONIGENC_IS_MBC_NEWLINE(enc,(UChar*)(p),(UChar*)(end))
  131. #define rb_enc_isctype(c,t,enc) ONIGENC_IS_CODE_CTYPE(enc,c,t)
  132. #define rb_enc_isascii(c,enc) ONIGENC_IS_CODE_ASCII(c)
  133. #define rb_enc_isalpha(c,enc) ONIGENC_IS_CODE_ALPHA(enc,c)
  134. #define rb_enc_islower(c,enc) ONIGENC_IS_CODE_LOWER(enc,c)
  135. #define rb_enc_isupper(c,enc) ONIGENC_IS_CODE_UPPER(enc,c)
  136. #define rb_enc_ispunct(c,enc) ONIGENC_IS_CODE_PUNCT(enc,c)
  137. #define rb_enc_isalnum(c,enc) ONIGENC_IS_CODE_ALNUM(enc,c)
  138. #define rb_enc_isprint(c,enc) ONIGENC_IS_CODE_PRINT(enc,c)
  139. #define rb_enc_isspace(c,enc) ONIGENC_IS_CODE_SPACE(enc,c)
  140. #define rb_enc_isdigit(c,enc) ONIGENC_IS_CODE_DIGIT(enc,c)
  141. #define rb_enc_asciicompat(enc) (rb_enc_mbminlen(enc)==1 && !rb_enc_dummy_p(enc))
  142. int rb_enc_casefold(char *to, const char *p, const char *e, rb_encoding *enc);
  143. int rb_enc_toupper(int c, rb_encoding *enc);
  144. int rb_enc_tolower(int c, rb_encoding *enc);
  145. ID rb_intern3(const char*, long, rb_encoding*);
  146. ID rb_interned_id_p(const char *, long, rb_encoding *);
  147. int rb_enc_symname_p(const char*, rb_encoding*);
  148. int rb_enc_symname2_p(const char*, long, rb_encoding*);
  149. int rb_enc_str_coderange(VALUE);
  150. long rb_str_coderange_scan_restartable(const char*, const char*, rb_encoding*, int*);
  151. int rb_enc_str_asciionly_p(VALUE);
  152. #define rb_enc_str_asciicompat_p(str) rb_enc_asciicompat(rb_enc_get(str))
  153. VALUE rb_enc_from_encoding(rb_encoding *enc);
  154. int rb_enc_unicode_p(rb_encoding *enc);
  155. rb_encoding *rb_ascii8bit_encoding(void);
  156. rb_encoding *rb_utf8_encoding(void);
  157. rb_encoding *rb_usascii_encoding(void);
  158. rb_encoding *rb_locale_encoding(void);
  159. rb_encoding *rb_filesystem_encoding(void);
  160. rb_encoding *rb_default_external_encoding(void);
  161. rb_encoding *rb_default_internal_encoding(void);
  162. int rb_ascii8bit_encindex(void);
  163. int rb_utf8_encindex(void);
  164. int rb_usascii_encindex(void);
  165. int rb_locale_encindex(void);
  166. int rb_filesystem_encindex(void);
  167. VALUE rb_enc_default_external(void);
  168. VALUE rb_enc_default_internal(void);
  169. void rb_enc_set_default_external(VALUE encoding);
  170. void rb_enc_set_default_internal(VALUE encoding);
  171. VALUE rb_locale_charmap(VALUE klass);
  172. long rb_memsearch(const void*,long,const void*,long,rb_encoding*);
  173. RUBY_EXTERN VALUE rb_cEncoding;
  174. #define ENC_DUMMY_FLAG (1<<24)
  175. #define ENC_INDEX_MASK (~(~0U<<24))
  176. #define ENC_TO_ENCINDEX(enc) (int)((enc)->ruby_encoding_index & ENC_INDEX_MASK)
  177. #define ENC_DUMMY_P(enc) ((enc)->ruby_encoding_index & ENC_DUMMY_FLAG)
  178. #define ENC_SET_DUMMY(enc) ((enc)->ruby_encoding_index |= ENC_DUMMY_FLAG)
  179. static inline int
  180. rb_enc_dummy_p(rb_encoding *enc)
  181. {
  182. return ENC_DUMMY_P(enc) != 0;
  183. }
  184. /* econv stuff */
  185. typedef enum {
  186. econv_invalid_byte_sequence,
  187. econv_undefined_conversion,
  188. econv_destination_buffer_full,
  189. econv_source_buffer_empty,
  190. econv_finished,
  191. econv_after_output,
  192. econv_incomplete_input
  193. } rb_econv_result_t;
  194. typedef struct rb_econv_t rb_econv_t;
  195. VALUE rb_str_encode(VALUE str, VALUE to, int ecflags, VALUE ecopts);
  196. int rb_econv_has_convpath_p(const char* from_encoding, const char* to_encoding);
  197. int rb_econv_prepare_opts(VALUE opthash, VALUE *ecopts);
  198. rb_econv_t *rb_econv_open(const char *source_encoding, const char *destination_encoding, int ecflags);
  199. rb_econv_t *rb_econv_open_opts(const char *source_encoding, const char *destination_encoding, int ecflags, VALUE ecopts);
  200. rb_econv_result_t rb_econv_convert(rb_econv_t *ec,
  201. const unsigned char **source_buffer_ptr, const unsigned char *source_buffer_end,
  202. unsigned char **destination_buffer_ptr, unsigned char *destination_buffer_end,
  203. int flags);
  204. void rb_econv_close(rb_econv_t *ec);
  205. /* result: 0:success -1:failure */
  206. int rb_econv_set_replacement(rb_econv_t *ec, const unsigned char *str, size_t len, const char *encname);
  207. /* result: 0:success -1:failure */
  208. int rb_econv_decorate_at_first(rb_econv_t *ec, const char *decorator_name);
  209. int rb_econv_decorate_at_last(rb_econv_t *ec, const char *decorator_name);
  210. VALUE rb_econv_open_exc(const char *senc, const char *denc, int ecflags);
  211. /* result: 0:success -1:failure */
  212. int rb_econv_insert_output(rb_econv_t *ec,
  213. const unsigned char *str, size_t len, const char *str_encoding);
  214. /* encoding that rb_econv_insert_output doesn't need conversion */
  215. const char *rb_econv_encoding_to_insert_output(rb_econv_t *ec);
  216. /* raise an error if the last rb_econv_convert is error */
  217. void rb_econv_check_error(rb_econv_t *ec);
  218. /* returns an exception object or nil */
  219. VALUE rb_econv_make_exception(rb_econv_t *ec);
  220. int rb_econv_putbackable(rb_econv_t *ec);
  221. void rb_econv_putback(rb_econv_t *ec, unsigned char *p, int n);
  222. /* returns the corresponding ASCII compatible encoding for encname,
  223. * or NULL if encname is not ASCII incompatible encoding. */
  224. const char *rb_econv_asciicompat_encoding(const char *encname);
  225. VALUE rb_econv_str_convert(rb_econv_t *ec, VALUE src, int flags);
  226. VALUE rb_econv_substr_convert(rb_econv_t *ec, VALUE src, long byteoff, long bytesize, int flags);
  227. VALUE rb_econv_str_append(rb_econv_t *ec, VALUE src, VALUE dst, int flags);
  228. VALUE rb_econv_substr_append(rb_econv_t *ec, VALUE src, long byteoff, long bytesize, VALUE dst, int flags);
  229. void rb_econv_binmode(rb_econv_t *ec);
  230. /* flags for rb_econv_open */
  231. #define ECONV_ERROR_HANDLER_MASK 0x000000ff
  232. #define ECONV_INVALID_MASK 0x0000000f
  233. #define ECONV_INVALID_REPLACE 0x00000002
  234. #define ECONV_UNDEF_MASK 0x000000f0
  235. #define ECONV_UNDEF_REPLACE 0x00000020
  236. #define ECONV_UNDEF_HEX_CHARREF 0x00000030
  237. #define ECONV_DECORATOR_MASK 0x0000ff00
  238. #define ECONV_UNIVERSAL_NEWLINE_DECORATOR 0x00000100
  239. #define ECONV_CRLF_NEWLINE_DECORATOR 0x00001000
  240. #define ECONV_CR_NEWLINE_DECORATOR 0x00002000
  241. #define ECONV_XML_TEXT_DECORATOR 0x00004000
  242. #define ECONV_XML_ATTR_CONTENT_DECORATOR 0x00008000
  243. #define ECONV_STATEFUL_DECORATOR_MASK 0x00f00000
  244. #define ECONV_XML_ATTR_QUOTE_DECORATOR 0x00100000
  245. /* end of flags for rb_econv_open */
  246. /* flags for rb_econv_convert */
  247. #define ECONV_PARTIAL_INPUT 0x00010000
  248. #define ECONV_AFTER_OUTPUT 0x00020000
  249. /* end of flags for rb_econv_convert */
  250. #if defined __GNUC__ && __GNUC__ >= 4
  251. #pragma GCC visibility pop
  252. #endif
  253. #if defined(__cplusplus)
  254. #if 0
  255. { /* satisfy cc-mode */
  256. #endif
  257. } /* extern "C" { */
  258. #endif
  259. #endif /* RUBY_ENCODING_H */