1 [[!template id=plugin name=po core=0 author="[[intrigeri]]"]]
4 This plugin adds support for multi-lingual wikis, translated with
5 gettext, using [po4a](http://po4a.alioth.debian.org/).
7 It depends on the Perl `Locale::Po4a::Po` library (`apt-get install po4a`).
14 A language is chosen as the "master" one, and any other supported
15 language is a "slave" one.
17 A page written in the "master" language is a "master" page. It can be
18 of any page type supported by ikiwiki, except `po`. It does not have to be
19 named a special way: migration to this plugin does not imply any page
22 Example: `bla/page.mdwn` is a "master" Markdown page written in
23 English; if `usedirs` is enabled, it is rendered as
24 `bla/page/index.en.html`, else as `bla/page.en.html`.
26 Any translation of a "master" page into a "slave" language is called
27 a "slave" page; it is written in the gettext PO format. `po` is now
28 a page type supported by ikiwiki.
30 Example: `bla/page.fr.po` is the PO "message catalog" used to
31 translate `bla/page.mdwn` into French; if `usedirs` is enabled, it is
32 rendered as `bla/page/index.fr.html`, else as `bla/page.fr.html`
41 `po_master_language` is used to set the "master" language in
42 `ikiwiki.setup`, such as:
44 po_master_language => { 'code' => 'en', 'name' => 'English' }
46 `po_slave_languages` is used to set the list of supported "slave"
49 po_slave_languages => { 'fr' => 'Français',
54 Decide which pages are translatable
55 -----------------------------------
57 The `po_translatable_pages` setting configures what pages are
58 translatable. It is a [[ikiwiki/PageSpec]], so you have lots of
59 control over what kind of pages are translatable.
61 The `.po` files are not considered as being translatable, so you don't need to
62 worry about excluding them explicitly from this [[ikiwiki/PageSpec]].
67 The `po_link_to` option in `ikiwiki.setup` is used to decide how
68 internal links should be generated, depending on web server features
69 and site-specific preferences.
71 ### Default linking behavior
73 If `po_link_to` is unset, or set to `default`, ikiwiki's default
74 linking behavior is preserved: `\[[destpage]]` links to the master
77 ### Link to current language
79 If `po_link_to` is set to `current`, `\[[destpage]]` links to the
80 `destpage`'s version written in the current page's language, if
83 - `foo/destpage/index.LL.html` if `usedirs` is enabled
84 - `foo/destpage.LL.html` if `usedirs` is disabled
86 ### Link to negotiated language
88 If `po_link_to` is set to `negotiated`, `\[[page]]` links to the
89 negotiated preferred language, *i.e.* `foo/page/`.
91 (In)compatibility notes:
93 - if `usedirs` is disabled, it does not make sense to set `po_link_to`
94 to `negotiated`; this option combination is neither implemented
96 - if the web server does not support Content Negotiation, setting
97 `po_link_to` to `negotiated` will produce a unusable website.
106 Using Apache `mod_negotiation` makes it really easy to have Apache
107 serve any page in the client's preferred language, if available.
108 This is the default Debian Apache configuration.
110 When `usedirs` is enabled, one has to set `DirectoryIndex index` for
113 Setting `DefaultLanguage LL` (replace `LL` with your default MIME
114 language code) for the wiki context can help to ensure
115 `bla/page/index.en.html` is served as `Content-Language: LL`.
120 lighttpd unfortunately does not support content negotiation.
122 **FIXME**: does `mod_magnet` provide the functionality needed to
132 When `po_link_to` is not set to `negotiated`, one should replace some
133 occurrences of `BASEURL` with `HOMEPAGEURL` to get correct links to
136 The `ISTRANSLATION` and `ISTRANSLATABLE` variables can be used to
137 display things only on translatable or translation pages.
139 ### Display page's versions in other languages
141 The `OTHERLANGUAGES` loop provides ways to display other languages'
142 versions of the same page, and the translations' status.
144 One typically adds the following code to `templates/page.tmpl`:
146 <TMPL_IF NAME="OTHERLANGUAGES">
147 <div id="otherlanguages">
149 <TMPL_LOOP NAME="OTHERLANGUAGES">
151 <a href="<TMPL_VAR NAME="URL">"><TMPL_VAR NAME="LANGUAGE"></a>
152 <TMPL_UNLESS NAME="MASTER">
153 (<TMPL_VAR NAME="PERCENT"> %)
161 The following variables are available inside the loop (for every page in):
163 - `URL` - url to the page
164 - `CODE` - two-letters language code
165 - `LANGUAGE` - language name (as defined in `po_slave_languages`)
166 - `MASTER` - is true (1) if, and only if the page is a "master" page
167 - `PERCENT` - for "slave" pages, is set to the translation completeness, in percents
169 ### Display the current translation status
171 The `PERCENTTRANSLATED` variable is set to the translation
172 completeness, expressed in percent, on "slave" pages.
174 One can use it this way:
176 <TMPL_IF NAME="ISTRANSLATION">
177 <div id="percenttranslated">
178 <TMPL_VAR NAME="PERCENTTRANSLATED">
182 Additional PageSpec tests
183 -------------------------
185 This plugin enhances the regular [[ikiwiki/PageSpec]] syntax with some
186 additional tests that are documented [[here|ikiwiki/pagespec/po]].
188 Automatic PO file update
189 ------------------------
191 Committing changes to a "master" page:
193 1. updates the POT file and the PO files for the "slave" languages;
194 the updated PO files are then put under version control;
195 2. triggers a refresh of the corresponding HTML slave pages.
197 Also, when the plugin has just been enabled, or when a page has just
198 been declared as being translatable, the needed POT and PO files are
199 created, and the PO files are checked into version control.
201 Discussion pages and other sub-pages
202 ------------------------------------
204 Discussion should happen in the language in which the pages are
205 written for real, *i.e.* the "master" one. If discussion pages are
206 enabled, "slave" pages therefore link to the "master" page's
209 Likewise, "slave" pages are not supposed to have sub-pages;
210 [[WikiLinks|wikilink]] that appear on a "slave" page therefore link to
211 the master page's sub-pages.
216 One can edit the PO files using ikiwiki's CGI (a message-by-message
217 interface could also be implemented at some point).
219 If [[tips/untrusted_git_push]] is setup, one can edit the PO files in one's
220 preferred `$EDITOR`, without needing to be online.
230 The only past security issues I could find in GNU gettext and po4a
233 - [CVE-2004-0966](http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2004-0966),
234 *i.e.* [Debian bug #278283](http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=278283):
235 the autopoint and gettextize scripts in the GNU gettext package
236 1.14 and later versions, as used in Trustix Secure Linux 1.5
237 through 2.1 and other operating systems, allows local users to
238 overwrite files via a symlink attack on temporary files.
239 - [CVE-2007-4462](http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2007-4462):
240 `lib/Locale/Po4a/Po.pm` in po4a before 0.32 allows local users to
241 overwrite arbitrary files via a symlink attack on the
242 gettextization.failed.po temporary file.
244 **FIXME**: check whether this plugin would have been a possible attack
245 vector to exploit these vulnerabilities.
247 Depending on my mood, the lack of found security issues can either
248 indicate that there are none, or reveal that no-one ever bothered to
249 find (and publish) them.
253 Can any sort of directives be put in po files that will cause mischief
254 (ie, include other files, run commands, crash gettext, whatever)?
256 > No [documented](http://www.gnu.org/software/gettext/manual/gettext.html#PO-Files)
257 > directive is supposed to do so. [[--intrigeri]]
259 ### Running po4a on untrusted content
261 Are there any security issues on running po4a on untrusted content?
263 To say the least, this issue is not well covered, at least publicly:
265 - the documentation does not talk about it;
266 - grep'ing the source code for `security` or `trust` gives no answer.
268 On the other hand, a po4a developer answered my questions in
269 a convincing manner, stating that processing untrusted content was not
270 an initial goal, and analysing in detail the possible issues.
274 - the core (`Po.pm`, `Transtractor.pm`) should be safe
275 - po4a source code was fully checked for other potential symlink
276 attacks, after discovery of one such issue
277 - the only external program run by the core is `diff`, in `Po.pm` (in
278 parts of its code we don't use)
279 - `Locale::gettext`: only used to display translated error messages
280 - Nicolas François "hopes" `DynaLoader` is safe, and has "no reason to
281 think that `Encode` is not safe"
282 - Nicolas François has "no reason to think that `Encode::Guess` is not
283 safe". The po plugin nevertheless avoids using it by defining the
284 input charset (`file_in_charset`) before asking `Transtractor` to
285 read any file. NB: this hack depends on po4a internals to stay
290 ##### Locale::Po4a modules
292 The modules we want to use have to be checked, as not all are safe
293 (e.g. the LaTeX module's behaviour is changed by commands included in
294 the content); they may use regexps generated from the content.
296 `Chooser.pm` only loads the plugin we tell it too: currently, this
297 means the `Text` module only.
299 `Text` module (I checked the CVS version):
301 - it does not run any external program
302 - only `do_paragraph()` builds regexp's that expand untrusted
303 variables; they seem safe to me, but someone more expert than me
304 will need to check. Joey?
306 > Freaky code, but seems ok due to use of `quotementa`.
310 `Text::WrapI18N` can cause DoS (see the
311 [Debian bug #470250](http://bugs.debian.org/470250)), but it is
312 optional and we do not need the features it provides.
314 It is loaded if available by `Locale::Po4a::Common`; looking at the
315 code, I'm not sure we can prevent this at all, but maybe some symbol
316 table manipulation tricks could work; overriding
317 `Locale::Po4a::Common::wrapi18n` may be easier. I'm no expert at all
318 in this field. Joey? [[--intrigeri]]
320 > Update: Nicolas François suggests we add an option to po4a to
321 > disable it. It would do the trick, but only for people running
322 > a brand new po4a (probably too late for Lenny). Anyway, this option
323 > would have to take effect in a `BEGIN` / `eval` that I'm not
324 > familiar with. I can learn and do it, in case no Perl wizard
325 > volunteers to provide the po4a patch. [[--intrigeri]]
327 >> That doesn't really need to be in a BEGIN. This patch moves it to
328 >> `import`, and makes this disable wrap18n:
329 >> `use Locale::Po4a::Common q{nowrapi18n}` --[[Joey]]
332 --- /usr/share/perl5/Locale/Po4a/Common.pm 2008-07-21 14:54:52.000000000 -0400
333 +++ Common.pm 2008-11-11 18:27:34.000000000 -0500
339 - if (eval { require Text::WrapI18N }) {
343 + if ($_[0] eq 'nowrapi18n') {
347 + $class->export_to_level(1, $class, @_);
349 + if ($wrapi18n && eval { require Text::WrapI18N }) {
351 # Don't bother determining the wrap column if we cannot wrap.
352 my $col=$ENV{COLUMNS};
357 `Term::ReadKey` is not a hard dependency in our case, *i.e.* po4a
358 works nicely without it. But the po4a Debian package recommends
359 `libterm-readkey-perl`, so it will probably be installed on most
360 systems using the po plugin.
362 If `$ENV{COLUMNS}` is not set, `Locale::Po4a::Common` uses
363 `Term::ReadKey::GetTerminalSize()` to get the terminal size. How safe
366 Part of `Term::ReadKey` is written in C. Depending on the runtime
367 platform, this function use ioctl, environment, or C library function
368 calls, and may end up running the `resize` command (without
371 IMHO, using Term::ReadKey has too far reaching implications for us to
372 be able to guarantee anything wrt. security. Since it is anyway of no
373 use in our case, I suggest we define `ENV{COLUMNS}` before loading
374 `Locale::Po4a::Common`, just to be on the safe side. Joey?
377 > Update: adding an option to disable `Text::WrapI18N`, as Nicolas
378 > François suggested, would as a bonus disable `Term::ReadKey`
379 > as well. [[--intrigeri]]
383 `refreshpofiles()` runs this external program. A po4a developer
384 answered he does "not expect any security issues from it".
388 I was not able to find any public information about gettext or po4a
389 having been tested with a fuzzing program, such as `zzuf` or `fusil`.
390 Moreover, some gettext parsers seem to be quite
391 [easy to crash](http://fusil.hachoir.org/trac/browser/trunk/fuzzers/fusil-gettext),
392 so it might be useful to bang msgmerge/po4a's heads against such
393 a program in order to easily detect some of the most obvious DoS.
396 > po4a was not fuzzy-tested, but according to one of its developers,
397 > "it would be really appreciated". [[--intrigeri]]
401 - a 21M file containing 100 concatenated copies of all the files in my
402 `/usr/share/common-licenses/`; I had no existing PO file or
403 translated versions at hand, which renders these tests
405 - po4a was the Debian 0.34-2 package; the same tests were also run
406 after replacing the `Text` module with the CVS one (the core was not
407 changed in CVS since 0.34-2 was released), without any significant
408 difference in the results.
413 `po4a-gettextize` uses more or less the same po4a features as our
414 `refreshpot` function.
416 Without specifying an input charset, zzuf'ed `po4a-gettextize` quickly
417 errors out, complaining it was not able to detect the input charset;
418 it leaves no incomplete file on disk.
420 So I had to pretend the input was in UTF-8, as does the po plugin.
422 Two ways of crashing were revealed by this command-line:
424 zzuf -vc -s 0:100 -r 0.1:0.5 \
425 po4a-gettextize -f text -o markdown -M utf-8 -L utf-8 \
426 -m LICENSES >/dev/null
430 Malformed UTF-8 character (UTF-16 surrogate 0xdcc9) in substitution iterator at /usr/share/perl5/Locale/Po4a/Po.pm line 1443.
431 Malformed UTF-8 character (fatal) at /usr/share/perl5/Locale/Po4a/Po.pm line 1443.
435 Malformed UTF-8 character (UTF-16 surrogate 0xdcec) in substitution (s///) at /usr/share/perl5/Locale/Po4a/Po.pm line 1443.
436 Malformed UTF-8 character (fatal) at /usr/share/perl5/Locale/Po4a/Po.pm line 1443.
438 Perl seems to exit cleanly, and an incomplete PO file is written on
439 disk. I not sure whether if this is a bug in Perl or in `Po.pm`.
441 > It's fairly standard perl behavior when fed malformed utf-8. As long as it doesn't
442 > crash ikiwiki, it's probably acceptable. Ikiwiki can do some similar things itself when fed malformed utf-8 (doesn't crash tho) --[[Joey]]
446 `po4a-translate` uses more or less the same po4a features as our
449 Without specifying an input charset, same behaviour as
450 `po4a-gettextize`, so let's specify UTF-8 as input charset as of now.
453 po4a-translate -d -f text -o markdown -M utf-8 -L utf-8 \
454 -k 0 -m LICENSES -p LICENSES.fr.po -l test.fr
456 ... prints tons of occurences of the following error, but a complete
457 translated document is written (obviously with some weird chars
460 Use of uninitialized value in string ne at /usr/share/perl5/Locale/Po4a/TransTractor.pm line 854.
461 Use of uninitialized value in string ne at /usr/share/perl5/Locale/Po4a/TransTractor.pm line 840.
462 Use of uninitialized value in pattern match (m//) at /usr/share/perl5/Locale/Po4a/Po.pm line 1002.
466 zzuf -cv -s 0:10 -r 0.001:0.3 \
467 po4a-translate -d -f text -o markdown -M utf-8 -L utf-8 \
468 -k 0 -m LICENSES -p LICENSES.fr.po -l test.fr
470 ... seems to lose the fight, at the `readpo(LICENSES.fr.po)` step,
471 against some kind of infinite loop, deadlock, or any similar beast.
472 It does not seem to eat memory, though.
474 Whatever format module is used does not change anything. This is thus
475 probably a bug in po4a's core or in a lib it depends on.
477 The sub `read`, in `TransTractor.pm`, seems to be a good debugging
482 `msgmerge` is run in our `refreshpofiles` function. I did not manage
483 to crash it with `zzuf`.
485 gettext/po4a rough corners
486 --------------------------
488 - fix infinite loop when synchronizing two ikiwiki (when checkouts
489 live in different directories): say bla.fr.po has been updated in
490 repo2; pulling repo2 from repo1 seems to trigger a PO update, that
491 changes bla.fr.po in repo1; then pushing repo1 to repo2 triggers
492 a PO update, that changes bla.fr.po in repo2; etc.; quickly fixed in
493 `629968fc89bced6727981c0a1138072631751fee`, by disabling references
494 in Pot files. Using `Locale::Po4a::write_if_needed` might be
495 a cleaner solution. (warning: this function runs the external
496 `diff` program, have to check security)
497 - new translations created in the web interface must get proper
498 charset/encoding gettext metadata, else the next automatic PO update
499 removes any non-ascii chars; possible solution: put such metadata
500 into the Pot file, and let it propagate; should be fixed in
501 `773de05a7a1ee68d2bed173367cf5e716884945a`, time will tell.
506 ### Page title in links
509 [[bugs/pagetitle_function_does_not_respect_meta_titles]] from
510 intrigeri's `meta` branch, the generated links' text is based on the
511 page titles set with the [[meta|plugins/meta]] plugin. This has to be
512 merged upstream, though.
514 ### Translation status in links
516 Quoting [[contrib/po]]: a given translation's status (currently:
517 percent translated) could be displayed next to the link that leads to
518 it; a color code could as well be used ("just" a matter of adding
519 a CSS id or class to the links, depending on this variable).
523 They are not updated when the source page changes (e.g. meta title).
528 Markdown is well supported, great, but what about others?
530 The [[po|plugins/po]] uses `Locale::Po4a::Text` for every page format;
531 this can be expected to work out of the box with most other wiki-like
532 formats supported by ikiwiki. Some of their ad-hoc syntax might be
533 parsed in a strange way, but the worst problems I can imagine would be
534 wrapping issues; e.g. there is code in po4a dedicated to prevent
535 re-wrapping the underlined Markdown headers.
537 While it would be easy to better support formats such as [[html]] or
538 LaTeX, by using for each one the dedicated po4a module, this can be
539 problematic from a security point of view.
541 **TODO**: test the more popular formats and write proper documentation
544 Translation quality assurance
545 -----------------------------
547 Modifying a PO file via the CGI must be forbidden if the new version
548 is not a valid PO file. As a bonus, check that it provides a more
549 complete translation than the existing one.
551 A new `cansave` type of hook would be needed to implement this.
553 Note: committing to the underlying repository is a way to bypass
556 Creating new pages on the web
557 -----------------------------
559 See [[contrib/po|contrib/po]].
564 - Renaming a translation should be forbidden.
569 ### Disabling the plugin
571 - enabling the plugin with `po_translatable_pages` set
572 - enabling the plugin without `po_translatable_pages` set: **OK**
573 - disabling the plugin: **OK**
575 ### Changing the plugin config
577 - adding existing pages to `po_translatable_pages`: **OK**
578 - removing existing pages from `po_translatable_pages`: **OK**
579 - adding a language to `po_slave_languages`: **OK**
580 - removing a language from `po_slave_languages`: **OK**
581 - changing `po_master_language`: **OK**
582 - replacing `po_master_language` with a language previously part of
583 `po_slave_languages`: needs two rebuilds, but **OK** (this is quite
584 a perverse test actually)
588 - creating a master page via RCS: **OK**
589 - creating a master page via CGI: **OK**
593 - removing a master page via RCS: **OK**
594 - removing a translation via RCS: **OK**
595 - removing a master page via CGI: **OK**
596 - removing a translation via CGI: **OK**
600 - renaming a master page via RCS: **OK** (but the old translations
601 are lost, because not all RCS track file renaming)
602 - renaming a master page and its translations via RCS: **OK**
603 - renaming a master page via CGI: **OK**
604 - renaming a translation via RCS
605 - renaming a translation via CGI
609 - general test with `usedirs` disabled: **OK**
610 - general test with `indexpages` enabled
611 - general test with `po_link_to=default`
616 Maybe write separate documentation depending on the people it targets:
617 translators, wiki administrators, hackers. This plugin may be complex
618 enough to deserve this.