r54706 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r54705‎ \| r54706 \| r54707 >
Date:	10:25, 10 August 2009
Author:	nikerabbit
Status:	ok
Tags:
Comment:	Read part of GettextFFS implemented
Modified paths:	/trunk/extensions/Translate/_autoload.php (modified) (history) /trunk/extensions/Translate/ffs/Gettext.php (modified) (history)

Diff [purge]

Index: trunk/extensions/Translate/_autoload.php
—	—	@@ -116,3 +116,4 @@
117	117	$wgAutoloadClasses['JavaFFS'] = $dir . 'FFS.php';
118	118	$wgAutoloadClasses['YamlFFS'] = $dir . 'FFS.php';
119	119	$wgAutoloadClasses['JavaScriptFFS'] = $dir . 'FFS.php';
	120	+$wgAutoloadClasses['GettextFFS'] = $dir . '/ffs/Gettext.php';
Index: trunk/extensions/Translate/ffs/Gettext.php
—	—	@@ -70,7 +70,7 @@
71	71	$matches = array();
72	72	if ( preg_match( "/^msgctxt\s($poformat)/mx", $section, $matches ) ) {
73	73	// Remove quoting
74		~~- $item['ctxt'] = self::formatForWiki( $matches[1] );~~
	74	+ $item['ctxt'] = GettextFFS::formatForWiki( $matches[1] );
75	75	} elseif ( $useCtxtAsKey ) {
76	76	// Invalid message
77	77	continue;
—	—	@@ -78,7 +78,7 @@
79	79
80	80	$matches = array();
81	81	if ( preg_match( "/^msgid\s($poformat)/mx", $section, $matches ) ) {
82		~~- $item['id'] = self::formatForWiki( $matches[1] );~~
	82	+ $item['id'] = GettextFFS::formatForWiki( $matches[1] );
83	83	} else {
84	84	# echo "Definition not found!\n$section";
85	85	continue;
—	—	@@ -88,7 +88,7 @@
89	89	$matches = array();
90	90	if ( preg_match( "/^msgid_plural\s($poformat)/mx", $section, $matches ) ) {
91	91	$pluralMessage = true;
92		~~- $plural = self::formatForWiki( $matches[1] );~~
	92	+ $plural = GettextFFS::formatForWiki( $matches[1] );
93	93	$item['id'] = "{{PLURAL:GETTEXT\|{$item['id']}\|$plural}}";
94	94	}
95	95
—	—	@@ -98,7 +98,7 @@
99	99	for ( $i = 0; $i < $pluralForms[1]; $i++ ) {
100	100	$matches = array();
101	101	if ( preg_match( "/^msgstr\[$i\]\s($poformat)/mx", $section, $matches ) ) {
102		~~- $actualForms[] = self::formatForWiki( $matches[1] );~~
	102	+ $actualForms[] = GettextFFS::formatForWiki( $matches[1] );
103	103	} else {
104	104	throw new MWException( "Plural not found, expecting $i" );
105	105	}
—	—	@@ -109,7 +109,7 @@
110	110
111	111	$matches = array();
112	112	if ( preg_match( "/^msgstr\s($poformat)/mx", $section, $matches ) ) {
113		~~- $item['str'] = self::formatForWiki( $matches[1] );~~
	113	+ $item['str'] = GettextFFS::formatForWiki( $matches[1] );
114	114	} else {
115	115	# echo "Translation not found!\n";
116	116	continue;
—	—	@@ -142,15 +142,7 @@
143	143	if ( $useCtxtAsKey ) {
144	144	$key = $item['ctxt'];
145	145	} else {
146		~~- global $wgLegalTitleChars;~~
147		~~- $hash = sha1( $item['ctxt'] . $item['id'] );~~
148		~~- $snippet = $item['id'];~~
149		~~- $snippet = preg_replace( "/[^$wgLegalTitleChars]/", ' ', $snippet );~~
150		~~- $snippet = preg_replace( "/[:&%\/_]/", ' ', $snippet );~~
151		~~- $snippet = preg_replace( "/ {2,}/", ' ', $snippet );~~
152		~~- $snippet = $lang->truncate( $snippet, 30, '' );~~
153		~~- $snippet = str_replace( ' ', '_', trim( $snippet ) );~~
154		~~- $key = $this->prefix . $hash . '-' . $snippet;~~
	146	+ $key = GettextFFS::generateKeyFromItem( $item );
155	147	}
156	148
157	149	$changes[$key] = $item;
—	—	@@ -160,15 +152,6 @@
161	153	return $changes;
162	154	}
163	155
164		~~- public static function formatForWiki( $data ) {~~
165		~~- $quotePattern = '/(^"\|"$\n?)/m';~~
166		~~- $data = preg_replace( $quotePattern, '', $data );~~
167		~~- $data = stripcslashes( $data );~~
168		~~- if ( preg_match( '/\s$/', $data ) ) {~~
169		~~- $data .= '\\';~~
170		~~- }~~
171		~~- return $data;~~
172		~~- }~~
173	156
174	157	public function parseMessages( StringMangler $mangler ) {
175	158	$defs = $this->parseFile();
—	—	@@ -392,4 +375,219 @@
393	376
394	377	return $splitPlurals;
395	378	}
	379	+}
	380	+
	381	+class GettextFFS extends SimpleFFS {
	382	+
	383	+ //
	384	+ // READ
	385	+ //
	386	+
	387	+ public function readFromVariable( $data ) {
	388	+ $authors = $messages = array();
	389	+
	390	+ # Authors first
	391	+ $matches = array();
	392	+ preg_match_all( '/^#\sAuthor:\s(.*)$/m', $data, $matches );
	393	+ $authors = $matches[1];
	394	+
	395	+ # Then messages and everything else
	396	+ $parsedData = $this->parseGettext( $data );
	397	+ $parsedData['MESSAGES'] = $this->group->getMangler()->mangle( $parsedData['MESSAGES'] );
	398	+ $parsedData['AUTHORS'] = $authors;
	399	+
	400	+ return $parsedData;
	401	+ }
	402	+
	403	+ public function parseGettext( $data ) {
	404	+ $data = str_replace( "\r\n", "\n", $data );
	405	+ $messages = $template = $metadata = array();
	406	+
	407	+ // Defined only once. Be sure to not use it without match, or you might get old data
	408	+ $matches = array();
	409	+
	410	+ if ( preg_match( '/X-Language-Code:\s+([a-zA-Z-_]+)/', $data, $matches ) ) {
	411	+ $metadata['code'] = $matches[1];
	412	+ }
	413	+
	414	+ if ( preg_match( '/X-Message-Group:\s+([a-zA-Z0-9-_]+)/', $data, $matches ) ) {
	415	+ $metadata['group'] = $matches[1];
	416	+ }
	417	+
	418	+ $pluralForms = false;
	419	+ if ( preg_match( '/Plural-Forms:\s+nplurals=([0-9]+).*;/', $data, $matches ) ) {
	420	+ $metadata['plurals'] = $matches;
	421	+ $pluralForms = $matches;
	422	+ }
	423	+
	424	+ $useCtxtAsKey = isset($this->extra['CtxtAsKey']) && $this->extra['CtxtAsKey'];
	425	+
	426	+ $poformat = '"."\n?(^"."$\n?)*';
	427	+ $quotePattern = '/(^"\|"$\n?)/m';
	428	+
	429	+ $sections = preg_split( '/\n{2,}/', $data );
	430	+ array_shift( $sections ); // First isn't an actual message
	431	+
	432	+ foreach ( $sections as $section ) {
	433	+ if ( trim( $section ) === '' ) continue;
	434	+
	435	+ $item = array(
	436	+ 'ctxt' => '',
	437	+ 'id' => '',
	438	+ 'str' => '',
	439	+ 'flags' => array(),
	440	+ 'comments' => array(),
	441	+ );
	442	+
	443	+ $matches = array();
	444	+ if ( preg_match( "/^msgid\s($poformat)/mx", $section, $matches ) ) {
	445	+ $item['id'] = self::formatForWiki( $matches[1] );
	446	+ } else {
	447	+ throw new MWException( "Unable to parse msgid:\n\n$section" );
	448	+ }
	449	+
	450	+ if ( preg_match( "/^msgctxt\s($poformat)/mx", $section, $matches ) ) {
	451	+ $item['ctxt'] = self::formatForWiki( $matches[1] );
	452	+ } elseif ( $useCtxtAsKey ) { // Invalid message
	453	+ $metadata['warnings'][] = "Ctxt missing for {$item['id']}";
	454	+ }
	455	+
	456	+
	457	+ $pluralMessage = false;
	458	+ if ( preg_match( "/^msgid_plural\s($poformat)/mx", $section, $matches ) ) {
	459	+ $pluralMessage = true;
	460	+ $plural = self::formatForWiki( $matches[1] );
	461	+ $item['id'] = "{{PLURAL:GETTEXT\|{$item['id']}\|$plural}}";
	462	+ }
	463	+
	464	+ if ( $pluralMessage ) {
	465	+
	466	+ $actualForms = array();
	467	+ for ( $i = 0; $i < $pluralForms[1]; $i++ ) {
	468	+ if ( preg_match( "/^msgstr\[$i\]\s($poformat)/mx", $section, $matches ) ) {
	469	+ $actualForms[] = self::formatForWiki( $matches[1] );
	470	+ } else {
	471	+ throw new MWException( "Plural not found, expecting $i" );
	472	+ }
	473	+ }
	474	+
	475	+ $item['str'] = '{{PLURAL:GETTEXT\|' . implode( '\|', $actualForms ) . '}}';
	476	+ } else {
	477	+
	478	+ $matches = array();
	479	+ if ( preg_match( "/^msgstr\s($poformat)/mx", $section, $matches ) ) {
	480	+ $item['str'] = self::formatForWiki( $matches[1] );
	481	+ } else {
	482	+ throw new MWException( "Unable to parse msgstr:\n\n$section" );
	483	+ }
	484	+ }
	485	+
	486	+ // Parse flags
	487	+ $matches = array();
	488	+ if ( preg_match( '/^#,(.*)$/mu', $section, $matches ) ) {
	489	+ $flags = array_map( 'trim', explode( ',', $matches[1] ) );
	490	+ foreach ( $flags as $key => $flag ) {
	491	+ if ( $flag === 'fuzzy' ) {
	492	+ $item['str'] = TRANSLATE_FUZZY . $item['str'];
	493	+ unset( $flags[$key] );
	494	+ }
	495	+ }
	496	+ $item['flags'] = $flags;
	497	+ }
	498	+
	499	+ // Rest of the comments
	500	+ $matches = array();
	501	+ if ( preg_match_all( '/^#(.?) (.*)$/m', $section, $matches, PREG_SET_ORDER ) ) {
	502	+ foreach ( $matches as $match ) {
	503	+ if ( $match[1] !== ',' ) {
	504	+ $item['comments'][$match[1]][] = $match[2];
	505	+ }
	506	+ }
	507	+ }
	508	+
	509	+ if ( $useCtxtAsKey ) {
	510	+ $key = $item['ctxt'];
	511	+ } else {
	512	+ $key = self::generateKeyFromItem( $item );
	513	+ }
	514	+
	515	+ $messages[$key] = $item['str'];
	516	+ $template[$key] = $item;
	517	+
	518	+ }
	519	+
	520	+ return array(
	521	+ 'MESSAGES' => $messages,
	522	+ 'TEMPLATE' => $template,
	523	+ 'METADATA' => $metadata,
	524	+ );
	525	+ }
	526	+
	527	+ public static function generateKeyFromItem( $item ) {
	528	+ $lang = Language::factory( 'en' );
	529	+ global $wgLegalTitleChars;
	530	+ $hash = sha1( $item['ctxt'] . $item['id'] );
	531	+ $snippet = $item['id'];
	532	+ $snippet = preg_replace( "/[^$wgLegalTitleChars]/", ' ', $snippet );
	533	+ $snippet = preg_replace( "/[:&%\/_]/", ' ', $snippet );
	534	+ $snippet = preg_replace( "/ {2,}/", ' ', $snippet );
	535	+ $snippet = $lang->truncate( $snippet, 30, '' );
	536	+ $snippet = str_replace( ' ', '_', trim( $snippet ) );
	537	+ return "$hash-$snippet";
	538	+ }
	539	+
	540	+ public static function formatForWiki( $data ) {
	541	+ $quotePattern = '/(^"\|"$\n?)/m';
	542	+ $data = preg_replace( $quotePattern, '', $data );
	543	+ $data = stripcslashes( $data );
	544	+ if ( preg_match( '/\s$/', $data ) ) {
	545	+ $data .= '\\';
	546	+ }
	547	+ return $data;
	548	+ }
	549	+
	550	+ //
	551	+ // WRITE
	552	+ //
	553	+
	554	+ protected function writeReal( MessageCollection $collection ) {
	555	+ throw new MWException( 'Not implemented' );
	556	+ $output = $this->doHeader( $collection );
	557	+ $output .= $this->doAuthors( $collection );
	558	+
	559	+ $mangler = $this->group->getMangler();
	560	+
	561	+ $messages = array();
	562	+ foreach ( $collection as $key => $m ) {
	563	+ $key = $mangler->unmangle( $key );
	564	+ $value = $m->translation();
	565	+ $value = str_replace( TRANSLATE_FUZZY, '', $value );
	566	+ if ( $value === '' ) continue;
	567	+
	568	+ $messages[$key] = $value;
	569	+ }
	570	+ $output .= TranslateSpyc::dump( $messages );
	571	+ return $output;
	572	+ }
	573	+
	574	+ protected function doHeader( MessageCollection $collection ) {
	575	+ global $wgSitename;
	576	+ $code = $collection->code;
	577	+ $name = TranslateUtils::getLanguageName( $code );
	578	+ $native = TranslateUtils::getLanguageName( $code, true );
	579	+ $output = "# Messages for $name ($native)\n";
	580	+ $output .= "# Exported from $wgSitename\n";
	581	+ return $output;
	582	+ }
	583	+
	584	+ protected function doAuthors( MessageCollection $collection ) {
	585	+ $output = '';
	586	+ $authors = $collection->getAuthors();
	587	+ $authors = $this->filterAuthors( $authors, $collection->code );
	588	+ foreach ( $authors as $author ) {
	589	+ $output .= "# Author: $author\n";
	590	+ }
	591	+ return $output;
	592	+ }
	593	+
396	594	}
\ No newline at end of file

Status & tagging log

18:30, 16 August 2009 Siebrand (talk | contribs) changed the status of r54706 [removed: new added: ok]