- MWCryptRand: A new api for generating cryptographic randomness for security tokens. Uses whatever cryptographic source is available and if not falls back to using random state and clock drift.
- wfRandomString - A simple non-cryptographic pesudo-random string generation function to replace wfGenerateToken which was written pretending to be secure when it's really not.
- Core updates to use MWCryptRand in various places:
-- user_token generation (to do this we stop generating user_token implicitly and only generate it when needed to avoid depleting the system's entropy pool by reading random data we'll never use)
-- email confirmation token generation
-- password salt generation
-- temporary password generation
-- Generation of the automatic watchlist token
-- login and create user tokens
-- session ids when php's entropy sources are not set
-- the installer when generating wgSecretKey and the upgrade key
* Introduced Parser::killMarkers() based on the concept from StringFunctions. Used it in cases where markerStripCallback() doesn't make sense semantically, namely grammar, padleft, padright and anchorencode. Used markerStripCallback() in other cases.
* Changed headline unstrip order as suggested by P.Copp on bug 18295
* In CPF::lc() and CPF::uc(), removed the is_callable(). This was a temporary testing hack committed by me in r30109, which allowed me to do differential testing against a copy of the parser from before that revision.
* Add @since, fix indentation.
* Change default from 'all' to 'mw' as it's the most used (so default fetchLanguageNames() is equivalent to default getLanguageNames()).
* Add the include parameter also to fetchLanguageName() as it's needed in Parser: interlanguage links should only take into account mediawiki names. (Doesn't make a difference with how the functions are now, but could have been later.)
Introduce a global variable which causes language conversion to not be disabled in interface messages (as before r94279). Use $wgContLang for conversion (as before r97849) since $wgContLang is set to the base language (e.g. zh) on converter wikis, whereas a typical user language (e.g. zh-tw) only has a FakeConverter.
* The language object used for lc() in Parser::braceSubstitution() must match the one used in setFunctionHook() during firstCallInit(). It can't change depending on what message you are parsing. Use $wgContLang like before r97849.
Fixes the problems with r102179 and r102179, as there are
valid tags which begin the same, which meant they were not removed from
the TOC (the second regex, intended to remove tag parameters, then converted
<img or <blockquote> into <i> / <b>).
The same problem existed in the original regex, but as there are no valid
tags which begin with sup or sub, it never happened).
Added comment explaining the tocline regex, and added a bunch of parser tests.
{{NAMESPACE}} relative to the Title of the currently being parsed page.
Basically wfMsgForContent expands messages with wrong title while doing linksupdate stuff via job queue. For the broken file tracking category (r86534),Wikipedia folk want to sort the page into different categories based on namespace, and for some namespaces not categorize them at all (After all, a broken file link in a talk namespace is often not a bad thing).
Anyhow, explicitly set the title object for the message using wfMessage. There's probably deeper issues here in regards to why wfMsg et al is using wrong title, but this should fix the immediate issue.
So far I've encountered 2 extensions that give fatal errors from calling $wgParser->disableOutput() from hooks that are called at points where parsing is not taking place! Exception with a backtrace is much nicer than "Fatal error: Call to a member function disableCache() on a non-object..."
Just skip the whole replaceInternalLinks2 parser function whenever we hit
js/css pages. Previous patch r103476 only handled Category links which was
not enough.
This patch skip the [[Category:#]] parsing logic when the Title is in
NS_MEDIAWIKI and ends with .js or .css. This way the code is kept as is
and pages are no more categorized.
How to reproduce the issue:
$ echo 'var foo = "[[Category:bug32450]]"' \
| php maintenance/parse.php --title MediaWiki:Foobar.js
<p>var foo = ""
</p>
$
Note how the text got stripped.
After this patch:
$ echo 'var foo = "[[Category:bug32450]]"' \
| php maintenance/parse.php --title MediaWiki:Foobar.js
<p>var foo = "[[Category:bug32450]]"
</p>
$
TEST PLAN:
==========
$ php parserTests.php --quiet
This is MediaWiki version 1.19alpha (r103473).
Reading tests from "tests/parser/parserTests.txt"...
Reading tests from "tests/parser/extraParserTests.txt"...
Passed 654 of 654 tests (100%)... ALL TESTS PASSED!
$
* Missed one call to ParserOptions::getUserLang() in Parser
* Also convert RefreshLinksJob and RefreshLinksJob2 to use ParserOptions::newFromUserAndLang() and pass $wgContLang instead of whatever $wgLang could be
* ParserOptions::getUserLang() will still return a string for compatibility, added ParserOptions::getUserLangObj() to get the object
* Added ParserOptions::newFromUserAndLang() and ParserOptions::newFromContext() to easily get a ParserOptions object when a context is available or when someone wants to force the language
* Updated OutputPage and Preferences to use newFromContext() and WikiPage to use newFromUserAndLang()
* ParserOptions::setUserLang() still accepts either a string or a Language object, but changed the calls to pass an object instead of a string
* Changed Parser::getFunctionLang() to return the Language object from ParserOptions when parsing interface messages rather than $wgLang directly and updated the documentation to say that $wgLang should not be used directly (as $wgUser, $wgTitle and $wgRequest)
