Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 1 | " Vim script for checking .po files. |
| 2 | " |
Bram Moolenaar | 595f51c | 2008-06-09 12:46:00 +0000 | [diff] [blame] | 3 | " Go through the file and verify that: |
| 4 | " - All %...s items in "msgid" are identical to the ones in "msgstr". |
| 5 | " - An error or warning code in "msgid" matches the one in "msgstr". |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 6 | |
| 7 | if 1 " Only execute this if the eval feature is available. |
| 8 | |
Bram Moolenaar | 78ee625 | 2023-05-28 18:39:55 +0100 | [diff] [blame] | 9 | " using line continuation |
| 10 | set cpo&vim |
| 11 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 12 | " Function to get a split line at the cursor. |
| 13 | " Used for both msgid and msgstr lines. |
| 14 | " Removes all text except % items and returns the result. |
| 15 | func! GetMline() |
| 16 | let idline = substitute(getline('.'), '"\(.*\)"$', '\1', '') |
| 17 | while line('.') < line('$') |
| 18 | + |
| 19 | let line = getline('.') |
| 20 | if line[0] != '"' |
| 21 | break |
| 22 | endif |
| 23 | let idline .= substitute(line, '"\(.*\)"$', '\1', '') |
| 24 | endwhile |
| 25 | |
Bram Moolenaar | a5792f5 | 2005-11-23 21:25:05 +0000 | [diff] [blame] | 26 | " remove '%', not used for formatting. |
| 27 | let idline = substitute(idline, "'%'", '', 'g') |
Bram Moolenaar | 0b8cf27 | 2020-08-30 19:42:06 +0200 | [diff] [blame] | 28 | let idline = substitute(idline, "%%", '', 'g') |
Bram Moolenaar | a5792f5 | 2005-11-23 21:25:05 +0000 | [diff] [blame] | 29 | |
Bram Moolenaar | 1c6136a | 2009-09-11 11:00:05 +0000 | [diff] [blame] | 30 | " remove '%' used for plural forms. |
| 31 | let idline = substitute(idline, '\\nPlural-Forms: .\+;\\n', '', '') |
| 32 | |
Christ van Willegen | 0c6181f | 2023-08-13 18:03:14 +0200 | [diff] [blame] | 33 | " remove duplicate positional format arguments |
| 34 | let idline2 = "" |
| 35 | while idline2 != idline |
| 36 | let idline2 = idline |
| 37 | let idline = substitute(idline, '%\([1-9][0-9]*\)\$\([-+ #''.*]*[0-9]*l\=[dsuxXpoc%]\)\(.*\)%\1$\([-+ #''.*]*\)\(l\=[dsuxXpoc%]\)', '%\1$\2\3\4', 'g') |
| 38 | endwhile |
| 39 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 40 | " remove everything but % items. |
Christ van Willegen | 0c6181f | 2023-08-13 18:03:14 +0200 | [diff] [blame] | 41 | return substitute(idline, '[^%]*\(%([1-9][0-9]*\$)\=[-+ #''.0-9*]*l\=[dsuxXpoc%]\)\=', '\1', 'g') |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 42 | endfunc |
| 43 | |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 44 | " This only works when 'wrapscan' is not set. |
Bram Moolenaar | a411e5d | 2010-08-04 15:47:08 +0200 | [diff] [blame] | 45 | let s:save_wrapscan = &wrapscan |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 46 | set nowrapscan |
Bram Moolenaar | a411e5d | 2010-08-04 15:47:08 +0200 | [diff] [blame] | 47 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 48 | " Start at the first "msgid" line. |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 49 | let wsv = winsaveview() |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 50 | 1 |
Bram Moolenaar | 8a37047 | 2022-05-22 15:28:31 +0100 | [diff] [blame] | 51 | keeppatterns /^msgid\> |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 52 | |
| 53 | " When an error is detected this is set to the line number. |
| 54 | " Note: this is used in the Makefile. |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 55 | let error = 0 |
| 56 | |
| 57 | while 1 |
Bram Moolenaar | e53ec39 | 2019-11-16 18:49:50 +0100 | [diff] [blame] | 58 | let lnum = line('.') |
| 59 | if getline(lnum) =~ 'msgid "Text;.*;"' |
| 60 | if getline(lnum + 1) !~ '^msgstr "\([^;]\+;\)\+"' |
| 61 | echomsg 'Mismatching ; in line ' . (lnum + 1) |
| 62 | echomsg 'Did you forget the trailing semicolon?' |
| 63 | if error == 0 |
| 64 | let error = lnum + 1 |
| 65 | endif |
| 66 | endif |
| 67 | endif |
| 68 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 69 | if getline(line('.') - 1) !~ "no-c-format" |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 70 | " go over the "msgid" and "msgid_plural" lines |
| 71 | let prevfromline = 'foobar' |
Bram Moolenaar | 78ee625 | 2023-05-28 18:39:55 +0100 | [diff] [blame] | 72 | let plural = 0 |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 73 | while 1 |
Bram Moolenaar | 78ee625 | 2023-05-28 18:39:55 +0100 | [diff] [blame] | 74 | if getline('.') =~ 'msgid_plural' |
| 75 | let plural += 1 |
| 76 | endif |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 77 | let fromline = GetMline() |
| 78 | if prevfromline != 'foobar' && prevfromline != fromline |
Bram Moolenaar | 78ee625 | 2023-05-28 18:39:55 +0100 | [diff] [blame] | 79 | \ && (plural != 1 |
| 80 | \ || count(prevfromline, '%') + 1 != count(fromline, '%')) |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 81 | echomsg 'Mismatching % in line ' . (line('.') - 1) |
| 82 | echomsg 'msgid: ' . prevfromline |
Bram Moolenaar | 78ee625 | 2023-05-28 18:39:55 +0100 | [diff] [blame] | 83 | echomsg 'msgid: ' . fromline |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 84 | if error == 0 |
| 85 | let error = line('.') |
| 86 | endif |
| 87 | endif |
| 88 | if getline('.') !~ 'msgid_plural' |
| 89 | break |
| 90 | endif |
| 91 | let prevfromline = fromline |
| 92 | endwhile |
| 93 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 94 | if getline('.') !~ '^msgstr' |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 95 | echomsg 'Missing "msgstr" in line ' . line('.') |
| 96 | if error == 0 |
| 97 | let error = line('.') |
| 98 | endif |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 99 | endif |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 100 | |
| 101 | " check all the 'msgstr' lines |
| 102 | while getline('.') =~ '^msgstr' |
| 103 | let toline = GetMline() |
| 104 | if fromline != toline |
Bram Moolenaar | 78ee625 | 2023-05-28 18:39:55 +0100 | [diff] [blame] | 105 | \ && (plural == 0 || count(fromline, '%') != count(toline, '%') + 1) |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 106 | echomsg 'Mismatching % in line ' . (line('.') - 1) |
| 107 | echomsg 'msgid: ' . fromline |
| 108 | echomsg 'msgstr: ' . toline |
| 109 | if error == 0 |
| 110 | let error = line('.') |
| 111 | endif |
| 112 | endif |
| 113 | if line('.') == line('$') |
| 114 | break |
| 115 | endif |
| 116 | endwhile |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 117 | endif |
| 118 | |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 119 | " Find next msgid. Quit when there is no more. |
| 120 | let lnum = line('.') |
Bram Moolenaar | 8a37047 | 2022-05-22 15:28:31 +0100 | [diff] [blame] | 121 | silent! keeppatterns /^msgid\> |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 122 | if line('.') == lnum |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 123 | break |
| 124 | endif |
| 125 | endwhile |
| 126 | |
Bram Moolenaar | 595f51c | 2008-06-09 12:46:00 +0000 | [diff] [blame] | 127 | " Check that error code in msgid matches the one in msgstr. |
| 128 | " |
| 129 | " Examples of mismatches found with msgid "E123: ..." |
| 130 | " - msgstr "E321: ..." error code mismatch |
| 131 | " - msgstr "W123: ..." warning instead of error |
| 132 | " - msgstr "E123 ..." missing colon |
| 133 | " - msgstr "..." missing error code |
| 134 | " |
| 135 | 1 |
| 136 | if search('msgid "\("\n"\)\?\([EW][0-9]\+:\).*\nmsgstr "\("\n"\)\?[^"]\@=\2\@!') > 0 |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 137 | echomsg 'Mismatching error/warning code in line ' . line('.') |
| 138 | if error == 0 |
| 139 | let error = line('.') |
| 140 | endif |
Bram Moolenaar | 595f51c | 2008-06-09 12:46:00 +0000 | [diff] [blame] | 141 | endif |
| 142 | |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 143 | func! CountNl(first, last) |
| 144 | let nl = 0 |
| 145 | for lnum in range(a:first, a:last) |
Bram Moolenaar | 9966b21 | 2017-07-28 16:46:57 +0200 | [diff] [blame] | 146 | let nl += count(getline(lnum), "\n") |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 147 | endfor |
| 148 | return nl |
| 149 | endfunc |
| 150 | |
Bram Moolenaar | 9966b21 | 2017-07-28 16:46:57 +0200 | [diff] [blame] | 151 | " Check that the \n at the end of the msgid line is also present in the msgstr |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 152 | " line. Skip over the header. |
Bram Moolenaar | b07bbb0 | 2018-04-30 15:45:17 +0200 | [diff] [blame] | 153 | 1 |
Bram Moolenaar | 8a37047 | 2022-05-22 15:28:31 +0100 | [diff] [blame] | 154 | keeppatterns /^"MIME-Version: |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 155 | while 1 |
| 156 | let lnum = search('^msgid\>') |
| 157 | if lnum <= 0 |
| 158 | break |
| 159 | endif |
| 160 | let strlnum = search('^msgstr\>') |
| 161 | let end = search('^$') |
| 162 | if end <= 0 |
| 163 | let end = line('$') + 1 |
| 164 | endif |
| 165 | let origcount = CountNl(lnum, strlnum - 1) |
| 166 | let transcount = CountNl(strlnum, end - 1) |
| 167 | " Allow for a few more or less line breaks when there are 2 or more |
| 168 | if origcount != transcount && (origcount <= 2 || transcount <= 2) |
Bram Moolenaar | 9966b21 | 2017-07-28 16:46:57 +0200 | [diff] [blame] | 169 | echomsg 'Mismatching "\n" in line ' . line('.') |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 170 | if error == 0 |
| 171 | let error = lnum |
| 172 | endif |
| 173 | endif |
| 174 | endwhile |
| 175 | |
Bram Moolenaar | aaef1ba | 2017-08-01 17:40:23 +0200 | [diff] [blame] | 176 | " Check that the file is well formed according to msgfmts understanding |
| 177 | if executable("msgfmt") |
| 178 | let filename = expand("%") |
Bram Moolenaar | 01164a6 | 2017-11-02 22:58:42 +0100 | [diff] [blame] | 179 | " Newer msgfmt does not take OLD_PO_FILE_INPUT argument, must be in |
| 180 | " environment. |
| 181 | let $OLD_PO_FILE_INPUT = 'yes' |
| 182 | let a = system("msgfmt --statistics " . filename) |
Bram Moolenaar | aaef1ba | 2017-08-01 17:40:23 +0200 | [diff] [blame] | 183 | if v:shell_error != 0 |
| 184 | let error = matchstr(a, filename.':\zs\d\+\ze:')+0 |
| 185 | for line in split(a, '\n') | echomsg line | endfor |
| 186 | endif |
| 187 | endif |
| 188 | |
Bram Moolenaar | 9cfc7d8 | 2018-05-13 22:37:03 +0200 | [diff] [blame] | 189 | " Check that the plural form is properly initialized |
| 190 | 1 |
| 191 | let plural = search('^msgid_plural ', 'n') |
| 192 | if (plural && search('^"Plural-Forms: ', 'n') == 0) || (plural && search('^msgstr\[0\] ".\+"', 'n') != plural + 1) |
| 193 | if search('^"Plural-Forms: ', 'n') == 0 |
| 194 | echomsg "Missing Plural header" |
| 195 | if error == 0 |
| 196 | let error = search('\(^"[A-Za-z-_]\+: .*\\n"\n\)\+\zs', 'n') - 1 |
| 197 | endif |
| 198 | elseif error == 0 |
| 199 | let error = plural |
| 200 | endif |
| 201 | elseif !plural && search('^"Plural-Forms: ', 'n') |
| 202 | " We allow for a stray plural header, msginit adds one. |
| 203 | endif |
| 204 | |
Bram Moolenaar | d1d037e | 2018-06-24 18:04:50 +0200 | [diff] [blame] | 205 | " Check that 8bit encoding is used instead of 8-bit |
| 206 | let cte = search('^"Content-Transfer-Encoding:\s\+8-bit', 'n') |
| 207 | let ctc = search('^"Content-Type:.*;\s\+\<charset=[iI][sS][oO]_', 'n') |
| 208 | let ctu = search('^"Content-Type:.*;\s\+\<charset=utf-8', 'n') |
| 209 | if cte |
| 210 | echomsg "Content-Transfer-Encoding should be 8bit instead of 8-bit" |
| 211 | " TODO: make this an error |
| 212 | " if error == 0 |
| 213 | " let error = cte |
| 214 | " endif |
| 215 | elseif ctc |
| 216 | echomsg "Content-Type charset should be 'ISO-...' instead of 'ISO_...'" |
| 217 | " TODO: make this an error |
| 218 | " if error == 0 |
| 219 | " let error = ct |
| 220 | " endif |
| 221 | elseif ctu |
| 222 | echomsg "Content-Type charset should be 'UTF-8' instead of 'utf-8'" |
| 223 | " TODO: make this an error |
| 224 | " if error == 0 |
| 225 | " let error = ct |
| 226 | " endif |
| 227 | endif |
| 228 | |
Christian Brabandt | 32a5faa | 2024-05-30 09:51:47 +0200 | [diff] [blame] | 229 | " Check that no lines are longer than 80 chars (except comments) |
| 230 | let overlong = search('^[^#]\%>80v', 'n') |
Christian Brabandt | 6f585da | 2024-04-11 22:02:28 +0200 | [diff] [blame] | 231 | if overlong > 0 |
| 232 | echomsg "Lines should be wrapped at 80 columns" |
Christian Brabandt | c858382 | 2024-04-11 23:10:54 +0200 | [diff] [blame] | 233 | " TODO: make this an error |
| 234 | " if error == 0 |
| 235 | " let error = overlong |
| 236 | " endif |
Christian Brabandt | 6f585da | 2024-04-11 22:02:28 +0200 | [diff] [blame] | 237 | endif |
Bram Moolenaar | 9cfc7d8 | 2018-05-13 22:37:03 +0200 | [diff] [blame] | 238 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 239 | if error == 0 |
Bram Moolenaar | 7f93703 | 2017-07-19 14:34:42 +0200 | [diff] [blame] | 240 | " If all was OK restore the view. |
| 241 | call winrestview(wsv) |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 242 | echomsg "OK" |
| 243 | else |
Bram Moolenaar | d1d037e | 2018-06-24 18:04:50 +0200 | [diff] [blame] | 244 | " Put the cursor on the line with the error. |
Bram Moolenaar | ec42059 | 2016-08-26 15:51:53 +0200 | [diff] [blame] | 245 | exe error |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 246 | endif |
| 247 | |
Bram Moolenaar | a411e5d | 2010-08-04 15:47:08 +0200 | [diff] [blame] | 248 | let &wrapscan = s:save_wrapscan |
| 249 | unlet s:save_wrapscan |
| 250 | |
Bram Moolenaar | 1d94f9b | 2005-08-04 21:29:45 +0000 | [diff] [blame] | 251 | endif |