src/intl/charsets.h

   1 #ifndef EL__INTL_CHARSETS_H
   2 #define EL__INTL_CHARSETS_H
   3
   4 typedef uint32_t unicode_val_T;
   5
   6 /* UCS/Unicode replacement character. */
   7 #define UCS_NO_CHAR ((unicode_val_T) 0xFFFFFFFD)
   8
   9 /* &nbsp; replacement character. See u2cp(). */
  10 #define NBSP_CHAR ((unsigned char) 1)
  11 #define NBSP_CHAR_STRING "\001"
  12
  13 struct conv_table {
  14         int t;
  15         union {
  16                 unsigned char *str;
  17                 struct conv_table *tbl;
  18         } u;
  19 };
  20
  21 enum convert_string_mode {
  22         CSM_DEFAULT, /* Convert any char. */
  23         CSM_QUERY, /* Special handling of '&' and '=' chars. */
  24         CSM_FORM, /* Special handling of '&' and '=' chars in forms. */
  25         CSM_NONE, /* Convert nothing. */
  26 };
  27
  28 struct conv_table *get_translation_table(int, int);
  29 unsigned char *get_entity_string(const unsigned char *str, const int strlen, int encoding);
  30
  31 /* The convert_string() name is also used by Samba (version 3.0.3), which
  32  * provides libnss_wins.so.2, which is called somewhere inside
  33  * _nss_wins_gethostbyname_r(). This name clash causes the elinks hostname
  34  * lookup thread to crash so we need to rename the symbol. */
  35 /* Reported by Derek Poon and filed as bug 453 */
  36 #undef convert_string
  37 #define convert_string convert_string_elinks
  38
  39 /* This routine converts a string from one charset to another according to the
  40  * passed @convert_table, potentially also decoding SGML (HTML) entities along
  41  * the way (according to @mode). It either returns dynamically allocated
  42  * converted string of length @length, or if the @callback is non-NULL it calls
  43  * it each few bytes instead and always returns NULL (@length is undefined).
  44  * Note that it's ok not to care and pass NULL as @length. */
  45 unsigned char *convert_string(struct conv_table *convert_table,
  46                               unsigned char *chars, int charslen, int cp,
  47                               enum convert_string_mode mode, int *length,
  48                               void (*callback)(void *data, unsigned char *buf, int buflen),
  49                               void *callback_data);
  50
  51 int get_cp_index(unsigned char *);
  52 unsigned char *get_cp_name(int);
  53 unsigned char *get_cp_mime_name(int);
  54 int is_cp_utf8(int);
  55 void free_conv_table(void);
  56 #ifdef CONFIG_UTF_8
  57 inline unsigned char *encode_utf_8(unicode_val_T);
  58 inline unsigned char *utf8_prevchar(unsigned char *, int, unsigned char *);
  59 inline int utf8charlen(const unsigned char *);
  60 int utf8_char2cells(unsigned char *, unsigned char *);
  61 int utf8_ptr2cells(unsigned char *, unsigned char *);
  62 int utf8_ptr2chars(unsigned char *, unsigned char *);
  63 int utf8_cells2bytes(unsigned char *, int, unsigned char *);
  64 inline int unicode_to_cell(unicode_val_T);
  65 unicode_val_T unicode_fold_label_case(unicode_val_T);
  66 inline int strlen_utf8(unsigned char **);
  67 inline unicode_val_T utf_8_to_unicode(unsigned char **, unsigned char *);
  68 unicode_val_T cp2u(int, unsigned char);
  69 #endif /* CONFIG_UTF_8 */
  70
  71 unsigned char *cp2utf_8(int, int);
  72
  73 unsigned char *u2cp_(unicode_val_T, int, int no_nbsp_hack);
  74 #define u2cp(u, to) u2cp_(u, to, 0)
  75 #define u2cp_no_nbsp(u, to) u2cp_(u, to, 1)
  76
  77 void init_charsets_lookup(void);
  78 void free_charsets_lookup(void);
  79
  80 #endif