ext/hunspell/csutil.hxx

   1 #ifndef __CSUTILHXX__
   2 #define __CSUTILHXX__
   3
   4 #include "hunvisapi.h"
   5
   6 // First some base level utility routines
   7
   8 #include <string.h>
   9 #include "w_char.hxx"
  10 #include "htypes.hxx"
  11
  12 #ifdef MOZILLA_CLIENT
  13 #include "nscore.h" // for mozalloc headers
  14 #endif
  15
  16 // casing
  17 #define NOCAP   0
  18 #define INITCAP 1
  19 #define ALLCAP  2
  20 #define HUHCAP  3
  21 #define HUHINITCAP  4
  22
  23 // default encoding and keystring
  24 #define SPELL_ENCODING  "ISO8859-1"
  25 #define SPELL_KEYSTRING "qwertyuiop|asdfghjkl|zxcvbnm"
  26
  27 // default morphological fields
  28 #define MORPH_STEM        "st:"
  29 #define MORPH_ALLOMORPH   "al:"
  30 #define MORPH_POS         "po:"
  31 #define MORPH_DERI_PFX    "dp:"
  32 #define MORPH_INFL_PFX    "ip:"
  33 #define MORPH_TERM_PFX    "tp:"
  34 #define MORPH_DERI_SFX    "ds:"
  35 #define MORPH_INFL_SFX    "is:"
  36 #define MORPH_TERM_SFX    "ts:"
  37 #define MORPH_SURF_PFX    "sp:"
  38 #define MORPH_FREQ        "fr:"
  39 #define MORPH_PHON        "ph:"
  40 #define MORPH_HYPH        "hy:"
  41 #define MORPH_PART        "pa:"
  42 #define MORPH_FLAG        "fl:"
  43 #define MORPH_HENTRY      "_H:"
  44 #define MORPH_TAG_LEN     strlen(MORPH_STEM)
  45
  46 #define MSEP_FLD ' '
  47 #define MSEP_REC '\n'
  48 #define MSEP_ALT '\v'
  49
  50 // default flags
  51 #define DEFAULTFLAGS   65510
  52 #define FORBIDDENWORD  65510
  53 #define ONLYUPCASEFLAG 65511
  54
  55 // convert UTF-16 characters to UTF-8
  56 LIBHUNSPELL_DLL_EXPORTED char * u16_u8(char * dest, int size, const w_char * src, int srclen);
  57
  58 // convert UTF-8 characters to UTF-16
  59 LIBHUNSPELL_DLL_EXPORTED int u8_u16(w_char * dest, int size, const char * src);
  60
  61 // sort 2-byte vector
  62 LIBHUNSPELL_DLL_EXPORTED void flag_qsort(unsigned short flags[], int begin, int end);
  63
  64 // binary search in 2-byte vector
  65 LIBHUNSPELL_DLL_EXPORTED int flag_bsearch(unsigned short flags[], unsigned short flag, int right);
  66
  67 // remove end of line char(s)
  68 LIBHUNSPELL_DLL_EXPORTED void mychomp(char * s);
  69
  70 // duplicate string
  71 LIBHUNSPELL_DLL_EXPORTED char * mystrdup(const char * s);
  72
  73 // strcat for limited length destination string
  74 LIBHUNSPELL_DLL_EXPORTED char * mystrcat(char * dest, const char * st, int max);
  75
  76 // duplicate reverse of string
  77 LIBHUNSPELL_DLL_EXPORTED char * myrevstrdup(const char * s);
  78
  79 // parse into tokens with char delimiter
  80 LIBHUNSPELL_DLL_EXPORTED char * mystrsep(char ** sptr, const char delim);
  81 // parse into tokens with char delimiter
  82 LIBHUNSPELL_DLL_EXPORTED char * mystrsep2(char ** sptr, const char delim);
  83
  84 // parse into tokens with char delimiter
  85 LIBHUNSPELL_DLL_EXPORTED char * mystrrep(char *, const char *, const char *);
  86
  87 // append s to ends of every lines in text
  88 LIBHUNSPELL_DLL_EXPORTED void strlinecat(char * lines, const char * s);
  89
  90 // tokenize into lines with new line
  91 LIBHUNSPELL_DLL_EXPORTED int line_tok(const char * text, char *** lines, char breakchar);
  92
  93 // tokenize into lines with new line and uniq in place
  94 LIBHUNSPELL_DLL_EXPORTED char * line_uniq(char * text, char breakchar);
  95 LIBHUNSPELL_DLL_EXPORTED char * line_uniq_app(char ** text, char breakchar);
  96
  97 // change oldchar to newchar in place
  98 LIBHUNSPELL_DLL_EXPORTED char * tr(char * text, char oldc, char newc);
  99
 100 // reverse word
 101 LIBHUNSPELL_DLL_EXPORTED int reverseword(char *);
 102
 103 // reverse word
 104 LIBHUNSPELL_DLL_EXPORTED int reverseword_utf(char *);
 105
 106 // remove duplicates
 107 LIBHUNSPELL_DLL_EXPORTED int uniqlist(char ** list, int n);
 108
 109 // free character array list
 110 LIBHUNSPELL_DLL_EXPORTED void freelist(char *** list, int n);
 111
 112 // character encoding information
 113 struct cs_info {
 114   unsigned char ccase;
 115   unsigned char clower;
 116   unsigned char cupper;
 117 };
 118
 119 LIBHUNSPELL_DLL_EXPORTED int initialize_utf_tbl();
 120 LIBHUNSPELL_DLL_EXPORTED void free_utf_tbl();
 121 LIBHUNSPELL_DLL_EXPORTED unsigned short unicodetoupper(unsigned short c, int langnum);
 122 LIBHUNSPELL_DLL_EXPORTED unsigned short unicodetolower(unsigned short c, int langnum);
 123 LIBHUNSPELL_DLL_EXPORTED int unicodeisalpha(unsigned short c);
 124
 125 LIBHUNSPELL_DLL_EXPORTED struct cs_info * get_current_cs(const char * es);
 126
 127 // get language identifiers of language codes
 128 LIBHUNSPELL_DLL_EXPORTED int get_lang_num(const char * lang);
 129
 130 // get characters of the given 8bit encoding with lower- and uppercase forms
 131 LIBHUNSPELL_DLL_EXPORTED char * get_casechars(const char * enc);
 132
 133 // convert null terminated string to all caps using encoding
 134 LIBHUNSPELL_DLL_EXPORTED void enmkallcap(char * d, const char * p, const char * encoding);
 135
 136 // convert null terminated string to all little using encoding
 137 LIBHUNSPELL_DLL_EXPORTED void enmkallsmall(char * d, const char * p, const char * encoding);
 138
 139 // convert null terminated string to have initial capital using encoding
 140 LIBHUNSPELL_DLL_EXPORTED void enmkinitcap(char * d, const char * p, const char * encoding);
 141
 142 // convert null terminated string to all caps
 143 LIBHUNSPELL_DLL_EXPORTED void mkallcap(char * p, const struct cs_info * csconv);
 144
 145 // convert null terminated string to all little
 146 LIBHUNSPELL_DLL_EXPORTED void mkallsmall(char * p, const struct cs_info * csconv);
 147
 148 // convert null terminated string to have initial capital
 149 LIBHUNSPELL_DLL_EXPORTED void mkinitcap(char * p, const struct cs_info * csconv);
 150
 151 // convert first nc characters of UTF-8 string to little
 152 LIBHUNSPELL_DLL_EXPORTED void mkallsmall_utf(w_char * u, int nc, int langnum);
 153
 154 // convert first nc characters of UTF-8 string to capital
 155 LIBHUNSPELL_DLL_EXPORTED void mkallcap_utf(w_char * u, int nc, int langnum);
 156
 157 // get type of capitalization
 158 LIBHUNSPELL_DLL_EXPORTED int get_captype(char * q, int nl, cs_info *);
 159
 160 // get type of capitalization (UTF-8)
 161 LIBHUNSPELL_DLL_EXPORTED int get_captype_utf8(w_char * q, int nl, int langnum);
 162
 163 // strip all ignored characters in the string
 164 LIBHUNSPELL_DLL_EXPORTED void remove_ignored_chars_utf(char * word, unsigned short ignored_chars[], int ignored_len);
 165
 166 // strip all ignored characters in the string
 167 LIBHUNSPELL_DLL_EXPORTED void remove_ignored_chars(char * word, char * ignored_chars);
 168
 169 LIBHUNSPELL_DLL_EXPORTED int parse_string(char * line, char ** out, int ln);
 170
 171 LIBHUNSPELL_DLL_EXPORTED int parse_array(char * line, char ** out, unsigned short ** out_utf16,
 172     int * out_utf16_len, int utf8, int ln);
 173
 174 LIBHUNSPELL_DLL_EXPORTED int fieldlen(const char * r);
 175 LIBHUNSPELL_DLL_EXPORTED char * copy_field(char * dest, const char * morph, const char * var);
 176
 177 LIBHUNSPELL_DLL_EXPORTED int morphcmp(const char * s, const char * t);
 178
 179 LIBHUNSPELL_DLL_EXPORTED int get_sfxcount(const char * morph);
 180
 181 // conversion function for protected memory
 182 LIBHUNSPELL_DLL_EXPORTED void store_pointer(char * dest, char * source);
 183
 184 // conversion function for protected memory
 185 LIBHUNSPELL_DLL_EXPORTED char * get_stored_pointer(const char * s);
 186
 187 // hash entry macros
 188 LIBHUNSPELL_DLL_EXPORTED inline char* HENTRY_DATA(struct hentry *h)
 189 {
 190     char *ret;
 191     if (!h->var)
 192         ret = NULL;
 193     else if (h->var & H_OPT_ALIASM)
 194         ret = get_stored_pointer(HENTRY_WORD(h) + h->blen + 1);
 195     else
 196         ret = HENTRY_WORD(h) + h->blen + 1;
 197     return ret;
 198 }
 199
 200 // NULL-free version for warning-free OOo build
 201 LIBHUNSPELL_DLL_EXPORTED inline const char* HENTRY_DATA2(const struct hentry *h)
 202 {
 203     const char *ret;
 204     if (!h->var)
 205         ret = "";
 206     else if (h->var & H_OPT_ALIASM)
 207         ret = get_stored_pointer(HENTRY_WORD(h) + h->blen + 1);
 208     else
 209         ret = HENTRY_WORD(h) + h->blen + 1;
 210     return ret;
 211 }
 212
 213 LIBHUNSPELL_DLL_EXPORTED inline char* HENTRY_FIND(struct hentry *h, const char *p)
 214 {
 215     return (HENTRY_DATA(h) ? strstr(HENTRY_DATA(h), p) : NULL);
 216 }
 217
 218 #define w_char_eq(a,b) (((a).l == (b).l) && ((a).h == (b).h))
 219
 220 #endif