etc/zebradb/etc/word-phrase-utf.chr

   1 # Generic search equivanence character map for Latin languages (English, French, etc.)
   2
   3 # Define the basic value-set. *Beware* of changing this without re-indexing
   4 # your databases.
   5 encoding utf-8
   6
   7 # basic character set
   8 lowercase {0-9}{a-z}
   9 uppercase {0-9}{A-Z}
  10
  11 # Breaking characters
  12 space {\001-\040}!"#$%&'\()*+,-./:;<=>?@\[\\]^_`\{|}~’
  13
  14 # Characters to be considered equivalent for sorting purposes
  15 equivalent aáàãåâăąȧǎȁȃ
  16 equivalent ӕä(ae)
  17 equivalent ā(aa)
  18 equivalent iíìîịĩĭįǐȉȋ
  19 equivalent ï(ie)
  20 equivalent ī(ii)
  21 equivalent uúùûũŭųűǔȕȗ
  22 equivalent ü(ue)
  23 equivalent ū(uu)
  24 equivalent eéèêẽĕęėěȅȇ
  25 equivalent ëē(ee)
  26 equivalent oóòõôŏǫȯőǒȍȏ
  27 equivalent Œœöø(oe)
  28 equivalent ō(oo)
  29
  30 # Supplemental mappings
  31 # mapping diacritics to plain ASCII
  32 # A
  33 map ā                  a
  34 map Ā                  a
  35 map á                  a
  36 map Á                  a
  37 map à                  a
  38 map À                  a
  39 map ã                  a
  40 map Ã                  a
  41 map å                  a
  42 map Å                  a
  43 map â                  a
  44 map Â                  a
  45 map ä                  a
  46 map Ä                  a
  47 map ă                  a
  48 map Ă                  a
  49 map ą                  a
  50 map Ą                  a
  51 map ȧ                  a
  52 map Ȧ                  a
  53 map ǎ                  a
  54 map Ǎ                  a
  55 map ȁ                  a
  56 map Ȁ                  a
  57 map ȃ                  a
  58 map Ȃ                  a
  59 map ⱥ                 a
  60 map Ⱥ                  a
  61 # AE
  62 map ӕ                  (ae)
  63 map Ӕ                  (ae)
  64 # I
  65 map ī                  i
  66 map Ī                  i
  67 map ï                  i
  68 map Ï                  i
  69 map î                  i
  70 map Î                  i
  71 map í                  i
  72 map Í                  i
  73 map ì                  i
  74 map Ì                  i
  75 map ị                 i
  76 map Ị                 i
  77 map ĩ                  i
  78 map Ĩ                  i
  79 map ĭ                  i
  80 map Ĭ                  i
  81 map į                  i
  82 map Į                  i
  83 map ǐ                  i
  84 map Ǐ                  i
  85 map ȉ                  i
  86 map Ȉ                  i
  87 map ȋ                  i
  88 map Ȋ                  i
  89 map ɨ                  i
  90 map Ɨ                  i
  91 # U
  92 map ū                  u
  93 map Ū                  u
  94 map ü                  u
  95 map Ü                  u
  96 map ù                  u
  97 map Ù                  u
  98 map ú                  u
  99 map Ú                  u
 100 map û                  u
 101 map Û                  u
 102 map ũ                  u
 103 map Ũ                  u
 104 map ŭ                  u
 105 map Ŭ                  u
 106 map ų                  u
 107 map Ų                  u
 108 map ű                  u
 109 map Ű                  u
 110 map ǔ                  u
 111 map Ǔ                  u
 112 map ȕ                  u
 113 map Ȕ                  u
 114 map ȗ                  u
 115 map Ȗ                  u
 116 map ʉ                  u
 117 map Ʉ                  u
 118 map ů                   u
 119 map Ů                   u
 120 # E
 121 map ê                  e
 122 map Ê                  e
 123 map ë                  e
 124 map Ë                  e
 125 map é                  e
 126 map É                  e
 127 map è                  e
 128 map È                  e
 129 map ē                  e
 130 map Ē                  e
 131 map ẽ                 e
 132 map Ẽ                 e
 133 map ĕ                  e
 134 map Ĕ                  e
 135 map ę                  e
 136 map Ę                  e
 137 map ė                  e
 138 map Ė                  e
 139 map ě                  e
 140 map Ě                  e
 141 map ȅ                  e
 142 map Ȅ                  e
 143 map ȇ                  e
 144 map Ȇ                  e
 145 map ɇ                  e
 146 map Ɇ                  e
 147 # O
 148 map ô                  o
 149 map Ô                  o
 150 map ó                  o
 151 map Ó                  o
 152 map ò                  o
 153 map Ò                  o
 154 map ö                  o
 155 map Ö                  o
 156 map õ                  o
 157 map Õ                  o
 158 map ø                  o
 159 map Ø                  o
 160 map ō                  o
 161 map Ō                  o
 162 map ǒ                  o
 163 map Ǒ                  o
 164 map ŏ                  o
 165 map Ŏ                  o
 166 map ǫ                  o
 167 map Ǫ                  o
 168 map ȯ                  o
 169 map Ȯ                  o
 170 map ő                  o
 171 map Ő                  o
 172 map ȍ                  o
 173 map Ȍ                  o
 174 map ȏ                  o
 175 map Ȏ                  o
 176 map Œ          (oe)
 177 map œ          (oe)
 178 # R
 179 map ṛ                 r
 180 map Ṛ                 r
 181 map r̥                 r
 182 map R̥                 r
 183 map ṝ                 r
 184 map Ṝ                 r
 185 map ř                  r
 186 map Ř                  r
 187 map ṟ                 r
 188 map ɍ                  r
 189 map Ɍ                  r
 190 # L
 191 map ḷ                 l
 192 map Ḷ                 l
 193 map ḹ                 l
 194 map Ḹ                 l
 195 map ḻ                 l
 196 map Ḻ                 l
 197 map ł                  l
 198 map Ł                  l
 199
 200 # N
 201 map ṅ                 n
 202 map Ṅ                 n
 203 map ñ                  n
 204 map Ñ                  n
 205 map ǹ                  n
 206 map Ǹ                  n
 207 map ń                  n
 208 map Ń                  n
 209 map ṉ                 n
 210 map Ṉ                 n
 211 map ṇ                 n
 212 map Ṇ                 n
 213 # T
 214 map ṭ                 t
 215 map Ṭ                 t
 216 map ŧ                  t
 217 map Ŧ                  t
 218 # D
 219 map ḍ                 d
 220 map Ḍ                 d
 221 map đ                  d
 222 map Đ                  d
 223 # S
 224 map ś                  s
 225 map Ś                  s
 226 map ṣ                 s
 227 map Ṣ                 s
 228 map ş                  s
 229 map Ş                  s
 230 map š                  s
 231 map Š                  s
 232 map ṡ                 s
 233 map Ṡ                 s
 234 # M
 235 map ṁ                 m
 236 map Ṁ                 m
 237 map ṃ                 m
 238 map Ṃ                 m
 239 # H
 240 map ḥ                 h
 241 map Ḥ                 h
 242 map ẖ                 h
 243 map ħ                  h
 244 map Ħ                  h
 245 # Z
 246 map ź                  z
 247 map Ź                  z
 248 map ẓ                 z
 249 map Ẓ                 z
 250 map ż                  z
 251 map Ż                  z
 252 map ƶ                  z
 253 map Ƶ                  z
 254 #C
 255 map ç                  c
 256 map Ç                  c
 257 map ć                  c
 258 map Ć                  c
 259 map č                  c
 260 map Č                  c
 261 map ȼ                  c
 262 map Ȼ                  c
 263 # G
 264 map ǥ                  g
 265 map Ǥ                  g
 266 map ġ                  g
 267 map Ġ                  g
 268 map ǧ                  g
 269 map Ǧ                  g
 270 map ğ                  g
 271 map Ğ                  g
 272 # K
 273 map ḵ                 k
 274 map Ḵ                 k
 275 # Y
 276 map ý                  y
 277 map Ý                  y
 278 map ɏ                  y
 279 map Ɏ                  y
 280
 281 # mapping blank string
 282 map (^impr\s)       @
 283 map (^impr\.)       @
 284 map (^Impr\s)       @
 285 map (^Impr\.)       @
 286 map (^IMPR\s)       @
 287 map (^IMPR\.)       @
 288 map (^DL\s)         @
 289 map (^DL\.)         @
 290 map (^Dl\s)         @
 291 map (^Dl\.)         @
 292 map (^dl\s)         @
 293 map (^dl\.)         @
 294 map (^cop\s)        @
 295 map (^cop\.)        @
 296 map (^Cop\s)        @
 297 map (^Cop\.)        @
 298 map (^COP\s)        @
 299 map (^COP\.)        @
 300 map (^c\s)          @
 301 map (^c\.)          @
 302 map (^C\s)          @
 303 map (^C\.)          @
 304 map (^\[sd\])       @
 305 map (^\[s\.d\.\])   @
 306 map (^\[Sd\])       @
 307 map (^\[S\.d\.\])   @
 308 map (^\[SD\])       @
 309 map (^\[S\.D\.\])   @
 310 map (\x88.*\x89))   @
 311