src/intl/charsets.h

   1 #ifndef EL__INTL_CHARSETS_H
   2 #define EL__INTL_CHARSETS_H
   3
   4 struct hash;
   5 typedef uint32_t unicode_val_T;
   6
   7 /* U+0020 SPACE.  Normally the same as ' ' or L' ' but perhaps ELinks
   8  * shouldn't rely on that.  */
   9 #define UCS_SPACE ((unicode_val_T) 0x0020)
  10
  11 /* U+00A0 NO-BREAK SPACE.  */
  12 #define UCS_NO_BREAK_SPACE ((unicode_val_T) 0x00A0)
  13
  14 /* U+00AD SOFT HYPHEN.  */
  15 #define UCS_SOFT_HYPHEN ((unicode_val_T) 0x00AD)
  16
  17 /* U+FFFD REPLACEMENT CHARACTER.  Used when no Unicode mapping is
  18  * known for a byte in a codepage, or when invalid UTF-8 is received
  19  * from a terminal.  After generating the character, ELinks then
  20  * treats it like any other Unicode character.  The user can also type
  21  * this character directly, and it can occur in documents.  */
  22 #define UCS_REPLACEMENT_CHARACTER ((unicode_val_T) 0xFFFD)
  23
  24 /* A special value that fits in unicode_val_T but is outside the range
  25  * of Unicode characters.  utf8_to_unicode and cp_to_unicode return
  26  * this if the input is too short.  This is also used as a placeholder
  27  * for the second cell of a double-cell character.  */
  28 #define UCS_NO_CHAR ((unicode_val_T) 0xFFFFFFFD)
  29
  30 #ifdef CONFIG_COMBINE
  31 #define UCS_END_COMBINED ((unicode_val_T) 0xFFFFFFFC)
  32
  33 #define UCS_BEGIN_COMBINED ((unicode_val_T) (UCS_END_COMBINED - (unicode_val_T) 10000))
  34
  35 /* Base character and up to 5 combining characters. */
  36 #define UCS_MAX_LENGTH_COMBINED 6
  37 #endif /* CONFIG_COMBINE */
  38
  39 /* If ELinks should display a double-cell character but there is only
  40  * one cell available, it displays this character instead.  This must
  41  * be a single-cell character but need not be unique.  Possible values
  42  * might be U+0020 SPACE or U+303F IDEOGRAPHIC HALF FILL SPACE.
  43  *
  44  * Some BFU widgets (at least input fields and list boxes) currently
  45  * ignore this setting and use U+0020 instead.  (They first draw spaces
  46  * and then overwrite with text; look for utf8_cells2bytes calls.)
  47  * We should fix that if we ever change the value.  */
  48 #define UCS_ORPHAN_CELL ((unicode_val_T) 0x20)
  49
  50 /* &nbsp; replacement character. See u2cp().
  51  * UTF-8 strings should use the encoding of U+00A0 instead. */
  52 #define NBSP_CHAR ((unsigned char) 1)
  53 #define NBSP_CHAR_STRING "\001"
  54
  55 /* How to convert a byte from a source charset.  This is used in an
  56  * array (struct conv_table[256]) indexed by the byte value.  */
  57 struct conv_table {
  58         /* 0 if this is the final byte of a character, or 1 if more
  59          * bytes are needed.  */
  60         int t;
  61         union {
  62                 /* If @t==0: a null-terminated string that is the
  63                  * corresponding character in the target charset.
  64                  * Normally, the string is statically allocated.
  65                  * However, if the conversion table is to UTF-8, then
  66                  * the strings in elements 0x80 to 0xFF are allocated
  67                  * with @mem_alloc and owned by the table.  */
  68                 const unsigned char *str;
  69                 /* If @t==1: a pointer to a nested conversion table
  70                  * (with 256 elements) that describes how to convert
  71                  * each possible subsequent byte.  The conversion
  72                  * table owns the nested conversion table.  */
  73                 struct conv_table *tbl;
  74         } u;
  75 };
  76
  77 enum convert_string_mode {
  78         CSM_DEFAULT, /* Convert any char. */
  79         CSM_QUERY, /* Special handling of '&' and '=' chars. */
  80         CSM_FORM, /* Special handling of '&' and '=' chars in forms. */
  81         CSM_NONE, /* Convert nothing. */
  82 };
  83
  84 /* How to translate U+00A0 NO-BREAK SPACE.  If u2cp_ is converting to
  85  * UTF-8, it ignores this choice and just encodes the U+00A0.  */
  86 enum nbsp_mode {
  87         /* Convert to NBSP_CHAR.  This lets the HTML renderer
  88          * recognize nbsp even if the codepage doesn't support
  89          * nbsp.  (VISCII doesn't.)  */
  90         NBSP_MODE_HACK = 0,
  91
  92         /* Convert to normal ASCII space.  */
  93         NBSP_MODE_ASCII = 1
  94 };
  95
  96 struct conv_table *get_translation_table(int, int);
  97 const unsigned char *get_entity_string(const unsigned char *str,
  98                                        const int strlen, int encoding);
  99
 100 /* The convert_string() name is also used by Samba (version 3.0.3), which
 101  * provides libnss_wins.so.2, which is called somewhere inside
 102  * _nss_wins_gethostbyname_r(). This name clash causes the elinks hostname
 103  * lookup thread to crash so we need to rename the symbol. */
 104 /* Reported by Derek Poon and filed as bug 453 */
 105 #undef convert_string
 106 #define convert_string convert_string_elinks
 107
 108 /* This routine converts a string from one charset to another according to the
 109  * passed @convert_table, potentially also decoding SGML (HTML) entities along
 110  * the way (according to @mode). It either returns dynamically allocated
 111  * converted string of length @length, or if the @callback is non-NULL it calls
 112  * it each few bytes instead and always returns NULL (@length is undefined).
 113  * Note that it's ok not to care and pass NULL as @length. */
 114 unsigned char *convert_string(struct conv_table *convert_table,
 115                               unsigned char *chars, int charslen, int cp,
 116                               enum convert_string_mode mode, int *length,
 117                               void (*callback)(void *data, unsigned char *buf, int buflen),
 118                               void *callback_data);
 119
 120 int get_cp_index(const unsigned char *);
 121 unsigned char *get_cp_name(int);
 122 unsigned char *get_cp_config_name(int);
 123 unsigned char *get_cp_mime_name(int);
 124 int is_cp_utf8(int);
 125 void free_conv_table(void);
 126 #ifdef CONFIG_UTF8
 127 inline unsigned char *encode_utf8(unicode_val_T);
 128 inline unsigned char *utf8_prevchar(unsigned char *, int, unsigned char *);
 129 inline int utf8charlen(const unsigned char *);
 130 int utf8_char2cells(unsigned char *, unsigned char *);
 131 int utf8_ptr2cells(unsigned char *, unsigned char *);
 132 int utf8_ptr2chars(unsigned char *, unsigned char *);
 133 int utf8_cells2bytes(unsigned char *, int, unsigned char *);
 134 /* How utf8_step_forward and utf8_step_backward count steps.  */
 135 enum utf8_step {
 136         /* Each step is one character, even if it is a combining or
 137          * double-width character.  */
 138         UTF8_STEP_CHARACTERS,
 139
 140         /* Each step is one cell.  If the specified number of steps
 141          * would end in the middle of a double-width character, do not
 142          * include the character.  */
 143         UTF8_STEP_CELLS_FEWER,
 144
 145         /* Each step is one cell.  If the specified number of steps
 146          * would end in the middle of a double-width character,
 147          * include the whole character.  */
 148         UTF8_STEP_CELLS_MORE
 149 };
 150 unsigned char *utf8_step_forward(unsigned char *, unsigned char *,
 151                                  int, enum utf8_step, int *);
 152 unsigned char *utf8_step_backward(unsigned char *, unsigned char *,
 153                                   int, enum utf8_step, int *);
 154 inline int unicode_to_cell(unicode_val_T);
 155 unicode_val_T unicode_fold_label_case(unicode_val_T);
 156 inline int strlen_utf8(unsigned char **);
 157 inline unicode_val_T utf8_to_unicode(unsigned char **, const unsigned char *);
 158 unicode_val_T cp_to_unicode(int, unsigned char **, unsigned char *);
 159 #endif /* CONFIG_UTF8 */
 160
 161 #ifdef CONFIG_COMBINE
 162 extern unicode_val_T last_combined;
 163 extern unicode_val_T **combined;
 164 extern struct hash *combined_hash;
 165 unicode_val_T get_combined(unicode_val_T *, int);
 166 void free_combined();
 167 #endif /* CONFIG_COMBINE */
 168
 169 unicode_val_T cp2u(int, unsigned char);
 170 const unsigned char *cp2utf8(int, int);
 171
 172 const unsigned char *u2cp_(unicode_val_T, int, enum nbsp_mode);
 173 #define u2cp(u, to) u2cp_(u, to, NBSP_MODE_HACK)
 174 #define u2cp_no_nbsp(u, to) u2cp_(u, to, NBSP_MODE_ASCII)
 175
 176 void init_charsets_lookup(void);
 177 void free_charsets_lookup(void);
 178
 179 /* UTF-16 encodes each Unicode character U+0000...U+FFFF as a single
 180  * 16-bit code unit, and each character U+10000...U+10FFFF as a pair
 181  * of two code units: a high surrogate followed by a low surrogate.
 182  * The range U+D800...U+DFFF is reserved for these surrogates.  */
 183 #define is_utf16_surrogate(u)           (((u) & 0xFFFFF800) == 0xD800)
 184 #define is_utf16_high_surrogate(u)      (((u) & 0xFFFFFC00) == 0xD800)
 185 #define is_utf16_low_surrogate(u)       (((u) & 0xFFFFFC00) == 0xDC00)
 186 #define join_utf16_surrogates(high,low) (0x10000 + (((high) - 0xD800L) << 10) + ((low) - 0xDC00))
 187 #define needs_utf16_surrogates(u)       ((uint32_t) ((u) - 0x10000) < 0x100000)
 188 #define get_utf16_high_surrogate(u)     (0xD800 + (((u) - 0x10000) >> 10))
 189 #define get_utf16_low_surrogate(u)      (0xDC00 + ((u) & 0x3FF))
 190
 191 #endif