Mercurial > vim

comparison runtime/tools/unicode.vim @ 2063:1378bc45ebe5 v7.2.348

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
updated for version 7.2.348 Problem: Unicode double-width characters are not up-to date. Solution: Produce the double-width table like the others.
author Bram Moolenaar <bram@zimbu.org>
date Wed, 27 Jan 2010 18:29:26 +0100
parents d5867fd6b2b7
children c9a5d51c9161
comparison
equal deleted inserted replaced
2062:dae4cd29a0b7 2063:1378bc45ebe5
185 call setline('$', getline('$')[:-2]) " remove last comma 185 call setline('$', getline('$')[:-2]) " remove last comma
186 call setline(line('$') + 1, " };") 186 call setline(line('$') + 1, " };")
187 wincmd p 187 wincmd p
188 endfunc 188 endfunc
189 189
190 " Build the ambiguous table in a new buffer. 190 " Build the double width or ambiguous width table in a new buffer.
191 " Uses s:widthprops and s:dataprops. 191 " Uses s:widthprops and s:dataprops.
192 func! BuildAmbiguousTable() 192 func! BuildWidthTable(pattern, tableName)
193 let start = -1 193 let start = -1
194 let end = -1 194 let end = -1
195 let ranges = [] 195 let ranges = []
196 let dataidx = 0 196 let dataidx = 0
197 for p in s:widthprops 197 for p in s:widthprops
198 if p[1][0] == 'A' 198 if p[1][0] =~ a:pattern
199 let n = ('0x' . p[0]) + 0 199 if p[0] =~ '\.\.'
200 " It is a range. we don't check for composing char then.
201 let rng = split(p[0], '\.\.')
202 if len(rng) != 2
203 echoerr "Cannot parse range: '" . p[0] . "' in width table"
204 endif
205 let n = ('0x' . rng[0]) + 0
206 let n_last = ('0x' . rng[1]) + 0
207 else
208 let n = ('0x' . p[0]) + 0
209 let n_last = n
210 endif
200 " Find this char in the data table. 211 " Find this char in the data table.
201 while 1 212 while 1
202 let dn = ('0x' . s:dataprops[dataidx][0]) + 0 213 let dn = ('0x' . s:dataprops[dataidx][0]) + 0
203 if dn >= n 214 if dn >= n
204 break 215 break
205 endif 216 endif
206 let dataidx += 1 217 let dataidx += 1
207 endwhile 218 endwhile
208 if dn != n 219 if dn != n && n_last == n
209 echoerr "Cannot find character " . n . " in data table" 220 echoerr "Cannot find character " . n . " in data table"
210 endif 221 endif
211 " Only use the char when it's not a composing char. 222 " Only use the char when it's not a composing char.
223 " But use all chars from a range.
212 let dp = s:dataprops[dataidx] 224 let dp = s:dataprops[dataidx]
213 if dp[2] != 'Mn' && dp[2] != 'Mc' && dp[2] != 'Me' 225 if n_last > n || (dp[2] != 'Mn' && dp[2] != 'Mc' && dp[2] != 'Me')
214 if start >= 0 && end + 1 == n 226 if start >= 0 && end + 1 == n
215 " continue with same range. 227 " continue with same range.
216 let end = n
217 else 228 else
218 if start >= 0 229 if start >= 0
219 " produce previous range 230 " produce previous range
220 call add(ranges, printf("\t{0x%04x, 0x%04x},", start, end)) 231 call add(ranges, printf("\t{0x%04x, 0x%04x},", start, end))
221 endif 232 endif
222 let start = n 233 let start = n
223 if p[0] =~ '\.\.' 234 endif
224 let end = ('0x' . substitute(p[0], '.*\.\.', '', '')) + 0 235 let end = n_last
225 else
226 let end = n
227 endif
228 endif
229 endif 236 endif
230 endif 237 endif
231 endfor 238 endfor
232 if start >= 0 239 if start >= 0
233 call add(ranges, printf("\t{0x%04x, 0x%04x},", start, end)) 240 call add(ranges, printf("\t{0x%04x, 0x%04x},", start, end))
234 endif 241 endif
235 242
236 " New buffer to put the result in. 243 " New buffer to put the result in.
237 new 244 new
238 file ambiguous 245 exe "file " . a:tableName
239 call setline(1, " static struct interval ambiguous[] =") 246 call setline(1, " static struct interval " . a:tableName . "[] =")
240 call setline(2, " {") 247 call setline(2, " {")
241 call append('$', ranges) 248 call append('$', ranges)
242 call setline('$', getline('$')[:-2]) " remove last comma 249 call setline('$', getline('$')[:-2]) " remove last comma
243 call setline(line('$') + 1, " };") 250 call setline(line('$') + 1, " };")
244 wincmd p 251 wincmd p
274 edit http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt 281 edit http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt
275 282
276 " Parse each line, create a list of lists. 283 " Parse each line, create a list of lists.
277 call ParseWidthProps() 284 call ParseWidthProps()
278 285
279 " Build the ambiguous table. 286 " Build the double width table.
280 call BuildAmbiguousTable() 287 call BuildWidthTable('[WF]', 'doublewidth')
288
289 " Build the ambiguous width table.
290 call BuildWidthTable('A', 'ambiguous')