<?php
/**
- * @package MediaWiki
- * @subpackage Maintenance
+ * @addtogroup Maintenance
*/
define( 'REPORTING_INTERVAL', 10 );
# Destination directory
var $dest;
+ # Skip existing files
+ var $noOverwrite = false;
+
# Show interlanguage links?
var $interwiki = true;
-
+
# Depth of HTML directory tree
var $depth = 3;
# Directory that commons images are copied into
- var $sharedStaticPath;
-
+ var $sharedStaticDirectory;
+
+ # Directory that the images are in, after copying
+ var $destUploadDirectory;
+
# Relative path to image directory
var $imageRel = 'upload';
# Copy commons images instead of symlinking
var $forceCopy = false;
- # Make links assuming the script path is in the same directory as
+ # Make a copy of all images encountered
+ var $makeSnapshot = false;
+
+ # Don't image description pages in doEverything()
+ var $noSharedDesc = false;
+
+ # Make links assuming the script path is in the same directory as
# the destination
var $alternateScriptPath = false;
# Has setupGlobals been called?
var $setupDone = false;
+ # Has to compress html pages
+ var $compress = false;
+
# List of raw pages used in the current article
var $rawPages;
-
- function DumpHTML( $settings ) {
+
+ # Skin to use
+ var $skin = 'htmldump';
+
+ # Checkpoint stuff
+ var $checkpointFile = false, $checkpoints = false;
+
+ var $startID = 1, $endID = false;
+
+ var $sliceNumerator = 1, $sliceDenominator = 1;
+
+ # Max page ID, lazy initialised
+ var $maxPageID = false;
+
+ # UDP profiling
+ var $udpProfile, $udpProfileCounter = 0, $udpProfileInit = false;
+
+ function DumpHTML( $settings = array() ) {
foreach ( $settings as $var => $value ) {
$this->$var = $value;
}
}
- /**
- * Write a set of articles specified by start and end page_id
+ function loadCheckpoints() {
+ if ( $this->checkpoints !== false ) {
+ return true;
+ } elseif ( !$this->checkpointFile ) {
+ return false;
+ } else {
+ $lines = @file( $this->checkpointFile );
+ if ( $lines === false ) {
+ print "Starting new checkpoint file \"{$this->checkpointFile}\"\n";
+ $this->checkpoints = array();
+ } else {
+ $lines = array_map( 'trim', $lines );
+ $this->checkpoints = array();
+ foreach ( $lines as $line ) {
+ list( $name, $value ) = explode( '=', $line, 2 );
+ $this->checkpoints[$name] = $value;
+ }
+ }
+ return true;
+ }
+ }
+
+ function getCheckpoint( $type, $defValue = false ) {
+ if ( !$this->loadCheckpoints() ) {
+ return false;
+ }
+ if ( !isset( $this->checkpoints[$type] ) ) {
+ return false;
+ } else {
+ return $this->checkpoints[$type];
+ }
+ }
+
+ function setCheckpoint( $type, $value ) {
+ if ( !$this->checkpointFile ) {
+ return;
+ }
+ $this->checkpoints[$type] = $value;
+ $blob = '';
+ foreach ( $this->checkpoints as $type => $value ) {
+ $blob .= "$type=$value\n";
+ }
+ file_put_contents( $this->checkpointFile, $blob );
+ }
+
+ function doEverything() {
+ if ( $this->getCheckpoint( 'everything' ) == 'done' ) {
+ print "Checkpoint says everything is already done\n";
+ return;
+ }
+ $this->doArticles();
+ $this->doCategories();
+ $this->doRedirects();
+ if ( $this->sliceNumerator == 1 ) {
+ $this->doSpecials();
+ }
+ $this->doLocalImageDescriptions();
+
+ if ( !$this->noSharedDesc ) {
+ $this->doSharedImageDescriptions();
+ }
+
+ $this->setCheckpoint( 'everything', 'done' );
+ }
+
+ /**
+ * Write a set of articles specified by start and end page_id
* Skip categories and images, they will be done separately
*/
- function doArticles( $start, $end = false ) {
- $fname = 'DumpHTML::doArticles';
-
- $this->setupGlobals();
+ function doArticles() {
+ if ( $this->endID === false ) {
+ $end = $this->getMaxPageID();
+ } else {
+ $end = $this->endID;
+ }
+ $start = $this->startID;
- if ( $end === false ) {
- $dbr =& wfGetDB( DB_SLAVE );
- $end = $dbr->selectField( 'page', 'max(page_id)', false, $fname );
+ # Start from the checkpoint
+ $cp = $this->getCheckpoint( 'article' );
+ if ( $cp == 'done' ) {
+ print "Articles already done\n";
+ return;
+ } elseif ( $cp !== false ) {
+ $start = $cp;
+ print "Resuming article dump from checkpoint at page_id $start of $end\n";
+ } else {
+ print "Starting from page_id $start of $end\n";
}
+ # Move the start point to the correct slice if it isn't there already
+ $start = $this->modSliceStart( $start );
+
+ $this->setupGlobals();
+
$mainPageObj = Title::newMainPage();
$mainPage = $mainPageObj->getPrefixedDBkey();
-
- for ($id = $start; $id <= $end; $id++) {
+ for ( $id = $start, $i = 0; $id <= $end; $id += $this->sliceDenominator, $i++ ) {
wfWaitForSlaves( 20 );
- if ( !($id % REPORTING_INTERVAL) ) {
+ if ( !( $i % REPORTING_INTERVAL) ) {
print "Processing ID: $id\r";
+ $this->setCheckpoint( 'article', $id );
}
- if ( !($id % (REPORTING_INTERVAL*10) ) ) {
+ if ( !($i % (REPORTING_INTERVAL*10) ) ) {
print "\n";
}
$title = Title::newFromID( $id );
if ( $title ) {
$ns = $title->getNamespace() ;
- if ( $ns != NS_CATEGORY && $title->getPrefixedDBkey() != $mainPage ) {
+ if ( $ns != NS_CATEGORY && $ns != NS_MEDIAWIKI &&
+ $title->getPrefixedDBkey() != $mainPage ) {
$this->doArticle( $title );
}
}
}
+ $this->setCheckpoint( 'article', 'done' );
print "\n";
- }
+ }
function doSpecials() {
$this->doMainPage();
$this->setupGlobals();
print "Special:Categories...";
- $this->doArticle( Title::makeTitle( NS_SPECIAL, 'Categories' ) );
+ $this->doArticle( SpecialPage::getTitleFor( 'Categories' ) );
print "\n";
}
$title = Title::newMainPage();
$text = $this->getArticleHTML( $title );
+
+ # Parse the XHTML to find the images
+ $images = $this->findImages( $text );
+ $this->copyImages( $images );
+
$file = fopen( "{$this->dest}/index.html", "w" );
if ( !$file ) {
print "\nCan't open index.html for writing\n";
}
function doImageDescriptions() {
- global $wgSharedUploadDirectory;
-
- $fname = 'DumpHTML::doImageDescriptions';
+ $this->doLocalImageDescriptions();
+ if ( !$this->noSharedDesc ) {
+ $this->doSharedImageDescriptions();
+ }
+ }
+
+ /**
+ * Dump image description pages that don't have an associated article, but do
+ * have a local image
+ */
+ function doLocalImageDescriptions() {
+ $chunkSize = 1000;
+
+ $dbr = wfGetDB( DB_SLAVE );
+ $cp = $this->getCheckpoint( 'local image' );
+ if ( $cp == 'done' ) {
+ print "Local image descriptions already done\n";
+ return;
+ } elseif ( $cp !== false ) {
+ print "Writing image description pages starting from $cp\n";
+ $conds = array( 'img_name >= ' . $dbr->addQuotes( $cp ) );
+ } else {
+ print "Writing image description pages for local images\n";
+ $conds = false;
+ }
+
$this->setupGlobals();
+ $i = 0;
- /**
- * Dump image description pages that don't have an associated article, but do
- * have a local image
- */
- $dbr =& wfGetDB( DB_SLAVE );
- extract( $dbr->tableNames( 'image', 'page' ) );
- $res = $dbr->select( 'image', array( 'img_name' ), false, $fname );
+ do {
+ $res = $dbr->select( 'image', array( 'img_name' ), $conds, __METHOD__,
+ array( 'ORDER BY' => 'img_name', 'LIMIT' => $chunkSize ) );
+ $numRows = $dbr->numRows( $res );
+
+ while ( $row = $dbr->fetchObject( $res ) ) {
+ # Update conds for the next chunk query
+ $conds = array( 'img_name > ' . $dbr->addQuotes( $row->img_name ) );
+
+ // Slice the result set with a filter
+ if ( !$this->sliceFilter( $row->img_name ) ) {
+ continue;
+ }
- $i = 0;
- print "Writing image description pages for local images\n";
- $num = $dbr->numRows( $res );
- while ( $row = $dbr->fetchObject( $res ) ) {
- wfWaitForSlaves( 10 );
- if ( !( ++$i % REPORTING_INTERVAL ) ) {
- print "Done $i of $num\r";
- }
- $title = Title::makeTitle( NS_IMAGE, $row->img_name );
- if ( $title->getArticleID() ) {
- // Already done by dumpHTML
- continue;
+ wfWaitForSlaves( 10 );
+ if ( !( ++$i % REPORTING_INTERVAL ) ) {
+ print "{$row->img_name}\n";
+ if ( $row->img_name !== 'done' ) {
+ $this->setCheckpoint( 'local image', $row->img_name );
+ }
+ }
+ $title = Title::makeTitle( NS_IMAGE, $row->img_name );
+ if ( $title->getArticleID() ) {
+ // Already done by dumpHTML
+ continue;
+ }
+ $this->doArticle( $title );
}
- $this->doArticle( $title );
- }
+ $dbr->freeResult( $res );
+ } while ( $numRows );
+
+ $this->setCheckpoint( 'local image', 'done' );
print "\n";
+ }
- /**
- * Dump images which only have a real description page on commons
- */
- print "Writing description pages for commons images\n";
- $i = 0;
- for ( $hash = 0; $hash < 256; $hash++ ) {
- $dir = sprintf( "%01x/%02x", intval( $hash / 16 ), $hash );
- $paths = array_merge( glob( "{$this->sharedStaticPath}/$dir/*" ),
- glob( "{$this->sharedStaticPath}/thumb/$dir/*" ) );
+ /**
+ * Dump images which only have a real description page on commons
+ */
+ function doSharedImageDescriptions() {
+ list( $start, $end ) = $this->sliceRange( 0, 255 );
- foreach ( $paths as $path ) {
- $file = basename( $path );
+ $cp = $this->getCheckpoint( 'shared image' );
+ if ( $cp == 'done' ) {
+ print "Shared description pages already done\n";
+ return;
+ } elseif ( $cp !== false ) {
+ print "Writing description pages for commons images starting from directory $cp/255\n";
+ $start = $cp;
+ } else {
+ print "Writing description pages for commons images\n";
+ }
+
+ $this->setupGlobals();
+ $i = 0;
+ for ( $hash = $start; $hash <= $end; $hash++ ) {
+ $this->setCheckpoint( 'shared image', $hash );
+
+ $dir = sprintf( "%s/%01x/%02x", $this->sharedStaticDirectory,
+ intval( $hash / 16 ), $hash );
+ $handle = @opendir( $dir );
+ while ( $handle && $file = readdir( $handle ) ) {
+ if ( $file[0] == '.' ) {
+ continue;
+ }
if ( !(++$i % REPORTING_INTERVAL ) ) {
print "$i\r";
}
- $title = Title::makeTitle( NS_IMAGE, $file );
+ $title = Title::makeTitleSafe( NS_IMAGE, $file );
$this->doArticle( $title );
}
+ if ( $handle ) {
+ closedir( $handle );
+ }
}
+ $this->setCheckpoint( 'shared image', 'done' );
print "\n";
}
function doCategories() {
- $fname = 'DumpHTML::doCategories';
+ $chunkSize = 1000;
+
$this->setupGlobals();
+ $dbr = wfGetDB( DB_SLAVE );
+
+ $cp = $this->getCheckpoint( 'category' );
+ if ( $cp == 'done' ) {
+ print "Category pages already done\n";
+ return;
+ } elseif ( $cp !== false ) {
+ print "Resuming category page dump from $cp\n";
+ $conds = array( 'cl_to >= ' . $dbr->addQuotes( $cp ) );
+ } else {
+ print "Starting category pages\n";
+ $conds = false;
+ }
- $dbr =& wfGetDB( DB_SLAVE );
- $categorylinks = $dbr->tableName( 'categorylinks' );
- print "Selecting categories...";
- $sql = 'SELECT DISTINCT cl_to FROM categorylinks';
- $res = $dbr->query( $sql, $fname );
-
- print "\nWriting " . $dbr->numRows( $res ). " category pages\n";
$i = 0;
- while ( $row = $dbr->fetchObject( $res ) ) {
- wfWaitForSlaves( 10 );
- if ( !(++$i % REPORTING_INTERVAL ) ) {
- print "$i\r";
+ do {
+ $res = $dbr->select( 'categorylinks', 'DISTINCT cl_to', $conds, __METHOD__,
+ array( 'ORDER BY' => 'cl_to', 'LIMIT' => $chunkSize ) );
+ $numRows = $dbr->numRows( $res );
+
+ while ( $row = $dbr->fetchObject( $res ) ) {
+ // Set conditions for next chunk
+ $conds = array( 'cl_to > ' . $dbr->addQuotes( $row->cl_to ) );
+
+ // Filter pages from other slices
+ if ( !$this->sliceFilter( $row->cl_to ) ) {
+ continue;
+ }
+
+ wfWaitForSlaves( 10 );
+ if ( !(++$i % REPORTING_INTERVAL ) ) {
+ print "{$row->cl_to}\n";
+ if ( $row->cl_to != 'done' ) {
+ $this->setCheckpoint( 'category', $row->cl_to );
+ }
+ }
+ $title = Title::makeTitle( NS_CATEGORY, $row->cl_to );
+ $this->doArticle( $title );
}
- $title = Title::makeTitle( NS_CATEGORY, $row->cl_to );
- $this->doArticle( $title );
- }
+ $dbr->freeResult( $res );
+ } while ( $numRows );
+
+ $this->setCheckpoint( 'category', 'done' );
print "\n";
}
function doRedirects() {
print "Doing redirects...\n";
- $fname = 'DumpHTML::doRedirects';
- $this->setupGlobals();
- $dbr =& wfGetDB( DB_SLAVE );
- $res = $dbr->select( 'page', array( 'page_namespace', 'page_title' ),
- array( 'page_is_redirect' => 1 ), $fname );
- $num = $dbr->numRows( $res );
- print "$num redirects to do...\n";
+ $chunkSize = 10000;
+ $end = $this->getMaxPageID();
+ $cp = $this->getCheckpoint( 'redirect' );
+ if ( $cp == 'done' ) {
+ print "Redirects already done\n";
+ return;
+ } elseif ( $cp !== false ) {
+ print "Resuming redirect generation from page_id $cp\n";
+ $start = intval( $cp );
+ } else {
+ $start = 1;
+ }
+
+ $this->setupGlobals();
+ $dbr = wfGetDB( DB_SLAVE );
$i = 0;
- while ( $row = $dbr->fetchObject( $res ) ) {
- $title = Title::makeTitle( $row->page_namespace, $row->page_title );
- if ( !(++$i % (REPORTING_INTERVAL*10) ) ) {
- print "Done $i of $num\n";
- }
- $this->doArticle( $title );
+
+ for ( $chunkStart = $start; $chunkStart <= $end; $chunkStart += $chunkSize ) {
+ $chunkEnd = min( $end, $chunkStart + $chunkSize - 1 );
+ $conds = array(
+ 'page_is_redirect' => 1,
+ "page_id BETWEEN $chunkStart AND $chunkEnd"
+ );
+ # Modulo slicing in SQL
+ if ( $this->sliceDenominator != 1 ) {
+ $n = intval( $this->sliceNumerator );
+ $m = intval( $this->sliceDenominator );
+ $conds[] = "page_id % $m = $n";
+ }
+ $res = $dbr->select( 'page', array( 'page_id', 'page_namespace', 'page_title' ),
+ $conds, __METHOD__ );
+
+ while ( $row = $dbr->fetchObject( $res ) ) {
+ $title = Title::makeTitle( $row->page_namespace, $row->page_title );
+ if ( !(++$i % (REPORTING_INTERVAL*10) ) ) {
+ printf( "Done %d redirects (%2.3f%%)\n", $i, $row->page_id / $end * 100 );
+ $this->setCheckpoint( 'redirect', $row->page_id );
+ }
+ $this->doArticle( $title );
+ }
+ $dbr->freeResult( $res );
}
+ $this->setCheckpoint( 'redirect', 'done' );
}
/** Write an article specified by title */
function doArticle( $title ) {
- global $wgTitle, $wgSharedUploadPath, $wgSharedUploadDirectory;
- global $wgUploadDirectory;
+ if ( $this->noOverwrite ) {
+ $fileName = $this->dest.'/'.$this->getHashedFilename( $title );
+ if ( file_exists( $fileName ) ) {
+ return;
+ }
+ }
+
+ $this->profile();
$this->rawPages = array();
$text = $this->getArticleHTML( $title );
print "Writing $file\n";
$file = fopen( $path, 'w' );
if ( !$file ) {
- print("Can't open file $fullName for writing\n");
+ print("Can't open file $path for writing\n");
continue;
}
fwrite( $file, $text );
fclose( $file );
}
}
+
+ wfIncrStats( 'dumphtml_article' );
}
/** Write the given text to the file identified by the given title object */
- function writeArticle( &$title, $text ) {
+ function writeArticle( $title, $text ) {
$filename = $this->getHashedFilename( $title );
+
+ # Temporary hack for current dump, this should be moved to
+ # getFriendlyName() at the earliest opportunity.
+ #
+ # Limit filename length to 255 characters, so it works on ext3.
+ # Titles are in fact limited to 255 characters, but dumpHTML
+ # adds a suffix which may put them over the limit.
+ $length = strlen( $filename );
+ if ( $length > 255 ) {
+ print "Warning: Filename too long ($length bytes). Skipping.\n";
+ return;
+ }
+
$fullName = "{$this->dest}/$filename";
$fullDir = dirname( $fullName );
+ if ( $this->compress ) {
+ $fullName .= ".gz";
+ $text = gzencode( $text, 9 );
+ }
+
wfMkdirParents( $fullDir, 0755 );
+ wfSuppressWarnings();
$file = fopen( $fullName, 'w' );
+ wfRestoreWarnings();
+
if ( !$file ) {
- print("Can't open file $fullName for writing\n");
+ die("Can't open file '$fullName' for writing.\nCheck permissions or use another destination (-d).\n");
return;
}
-
+
fwrite( $file, $text );
fclose( $file );
}
/** Set up globals required for parsing */
function setupGlobals( $currentDepth = NULL ) {
- global $wgUser, $wgTitle, $wgStylePath, $wgArticlePath;
+ global $wgUser, $wgStylePath, $wgArticlePath, $wgMathPath;
global $wgUploadPath, $wgLogo, $wgMaxCredits, $wgSharedUploadPath;
global $wgHideInterlanguageLinks, $wgUploadDirectory, $wgThumbnailScriptPath;
global $wgSharedThumbnailScriptPath, $wgEnableParserCache, $wgHooks, $wgServer;
- global $wgRightsUrl, $wgRightsText, $wgCopyrightIcon;
+ global $wgRightsUrl, $wgRightsText, $wgCopyrightIcon, $wgEnableSidebarCache;
+ global $wgGenerateThumbnailOnParse;
static $oldLogo = NULL;
-
+
if ( !$this->setupDone ) {
$wgHooks['GetLocalURL'][] =& $this;
$wgHooks['GetFullURL'][] =& $this;
+ $wgHooks['SiteNoticeBefore'][] =& $this;
+ $wgHooks['SiteNoticeAfter'][] =& $this;
$this->oldArticlePath = $wgServer . $wgArticlePath;
}
if ( is_null( $currentDepth ) ) {
$currentDepth = $this->depth;
}
-
+
if ( $this->alternateScriptPath ) {
if ( $currentDepth == 0 ) {
$wgScriptPath = '.';
if ( !$this->setupDone ) {
$this->oldCopyrightIcon = $wgCopyrightIcon;
}
- $wgCopyrightIcon = str_replace( 'src="/images',
+ $wgCopyrightIcon = str_replace( 'src="/images',
'src="' . htmlspecialchars( $wgScriptPath ) . '/images', $this->oldCopyrightIcon );
-
-
$wgStylePath = "$wgScriptPath/skins";
$wgUploadPath = "$wgScriptPath/{$this->imageRel}";
$wgSharedUploadPath = "$wgUploadPath/shared";
$wgMaxCredits = -1;
- $wgHideInterlangageLinks = !$this->interwiki;
+ $wgHideInterlanguageLinks = !$this->interwiki;
$wgThumbnailScriptPath = $wgSharedThumbnailScriptPath = false;
$wgEnableParserCache = false;
$wgMathPath = "$wgScriptPath/math";
-
+ $wgEnableSidebarCache = false;
+ $wgGenerateThumbnailOnParse = true;
+
if ( !empty( $wgRightsText ) ) {
$wgRightsUrl = "$wgScriptPath/COPYING.html";
}
$wgUser = new User;
- $wgUser->setOption( 'skin', 'htmldump' );
+ $wgUser->setOption( 'skin', $this->skin );
$wgUser->setOption( 'editsection', 0 );
- $this->sharedStaticPath = "$wgUploadDirectory/shared";
+ $this->destUploadDirectory = "{$this->dest}/{$this->imageRel}";
+ if ( realpath( $this->destUploadDirectory ) == realpath( $wgUploadDirectory ) ) {
+ print "Disabling image snapshot because the destination is the same as the source\n";
+ $this->makeSnapshot = false;
+ }
+ $this->sharedStaticDirectory = "{$this->destUploadDirectory}/shared";
$this->setupDone = true;
}
/** Reads the content of a title object, executes the skin and captures the result */
- function getArticleHTML( &$title ) {
+ function getArticleHTML( $title ) {
global $wgOut, $wgTitle, $wgArticle, $wgUser;
-
+
$linkCache =& LinkCache::singleton();
$linkCache->clear();
$wgTitle = $title;
if ( is_null( $wgTitle ) ) {
return false;
}
-
+
$ns = $wgTitle->getNamespace();
if ( $ns == NS_SPECIAL ) {
$wgOut = new OutputPage;
$wgOut->setParserOptions( new ParserOptions );
SpecialPage::executePath( $wgTitle );
} else {
+ /** @todo merge with Wiki.php code */
if ( $ns == NS_IMAGE ) {
$wgArticle = new ImagePage( $wgTitle );
} elseif ( $ns == NS_CATEGORY ) {
} else {
$wgOut = new OutputPage;
$wgOut->setParserOptions( new ParserOptions );
-
+
$wgArticle->view();
}
}
+
$sk =& $wgUser->getSkin();
ob_start();
$sk->outputPage( $wgOut );
global $wgOutputEncoding, $wgDumpImages;
$parser = xml_parser_create( $wgOutputEncoding );
xml_set_element_handler( $parser, 'wfDumpStartTagHandler', 'wfDumpEndTagHandler' );
-
+
$wgDumpImages = array();
xml_parse( $parser, $text );
xml_parser_free( $parser );
return $wgDumpImages;
}
+ /**
+ * Copy a file specified by a URL to a given directory
+ *
+ * @param string $srcPath The source URL
+ * @param string $srcPathBase The base directory of the source URL
+ * @param string $srcDirBase The base filesystem directory of the source URL
+ * @param string $destDirBase The base filesystem directory of the destination URL
+ */
+ function relativeCopy( $srcPath, $srcPathBase, $srcDirBase, $destDirBase ) {
+ $rel = substr( $srcPath, strlen( $srcPathBase ) + 1 ); // +1 for slash
+ $sourceLoc = "$srcDirBase/$rel";
+ $destLoc = "$destDirBase/$rel";
+ #print "Copying $sourceLoc to $destLoc\n";
+ if ( !file_exists( $destLoc ) ) {
+ wfMkdirParents( dirname( $destLoc ), 0755 );
+ if ( function_exists( 'symlink' ) && !$this->forceCopy ) {
+ if ( !symlink( $sourceLoc, $destLoc ) ) {
+ print "Warning: unable to create symlink at $destLoc\n";
+ }
+ } else {
+ if ( !copy( $sourceLoc, $destLoc ) ) {
+ print "Warning: unable to copy $sourceLoc to $destLoc\n";
+ }
+ }
+ }
+ }
+
+ /**
+ * Copy an image, and if it is a thumbnail, copy its parent image too
+ */
+ function copyImage( $srcPath, $srcPathBase, $srcDirBase, $destDirBase ) {
+ global $wgUploadPath, $wgUploadDirectory, $wgSharedUploadPath;
+ $this->relativeCopy( $srcPath, $srcPathBase, $srcDirBase, $destDirBase );
+ if ( substr( $srcPath, strlen( $srcPathBase ) + 1, 6 ) == 'thumb/' ) {
+ # The image was a thumbnail
+ # Copy the source image as well
+ $rel = substr( $srcPath, strlen( $srcPathBase ) + 1 );
+ $parts = explode( '/', $rel );
+ $rel = "{$parts[1]}/{$parts[2]}/{$parts[3]}";
+ $newSrc = "$srcPathBase/$rel";
+ $this->relativeCopy( $newSrc, $srcPathBase, $srcDirBase, $destDirBase );
+ }
+ }
+
/**
* Copy images (or create symlinks) from commons to a static directory.
* This is necessary even if you intend to distribute all of commons, because
* the directory contents is used to work out which image description pages
* are needed.
- *
- * Also copies math images
+ *
+ * Also copies math images, and full-sized images if the makeSnapshot option
+ * is specified.
*
*/
function copyImages( $images ) {
- global $wgSharedUploadPath, $wgSharedUploadDirectory, $wgMathPath, $wgMathDirectory;
+ global $wgUploadPath, $wgUploadDirectory, $wgSharedUploadPath, $wgSharedUploadDirectory,
+ $wgMathPath, $wgMathDirectory;
# Find shared uploads and copy them into the static directory
- $sharedPathLength = strlen( $wgSharedUploadPath );
+ $sharedPathLength = strlen( $wgSharedUploadPath );
$mathPathLength = strlen( $wgMathPath );
+ $uploadPathLength = strlen( $wgUploadPath );
foreach ( $images as $escapedImage => $dummy ) {
$image = urldecode( $escapedImage );
- # Is it shared?
if ( substr( $image, 0, $sharedPathLength ) == $wgSharedUploadPath ) {
- # Reconstruct full filename
- $rel = substr( $image, $sharedPathLength + 1 ); // +1 for slash
- $sourceLoc = "$wgSharedUploadDirectory/$rel";
- $staticLoc = "{$this->sharedStaticPath}/$rel";
- #print "Copying $sourceLoc to $staticLoc\n";
- # Copy to static directory
- if ( !file_exists( $staticLoc ) ) {
- wfMkdirParents( dirname( $staticLoc ), 0755 );
- if ( function_exists( 'symlink' ) && !$this->forceCopy ) {
- symlink( $sourceLoc, $staticLoc );
- } else {
- copy( $sourceLoc, $staticLoc );
- }
- }
-
- if ( substr( $rel, 0, 6 ) == 'thumb/' ) {
- # That was a thumbnail
- # We will also copy the real image
- $parts = explode( '/', $rel );
- $rel = "{$parts[1]}/{$parts[2]}/{$parts[3]}";
- $sourceLoc = "$wgSharedUploadDirectory/$rel";
- $staticLoc = "{$this->sharedStaticPath}/$rel";
- #print "Copying $sourceLoc to $staticLoc\n";
- if ( !file_exists( $staticLoc ) ) {
- wfMkdirParents( dirname( $staticLoc ), 0755 );
- if ( function_exists( 'symlink' ) && !$this->forceCopy ) {
- symlink( $sourceLoc, $staticLoc );
- } else {
- copy( $sourceLoc, $staticLoc );
- }
- }
- }
- } else
- # Is it math?
- if ( substr( $image, 0, $mathPathLength ) == $wgMathPath ) {
- $rel = substr( $image, $mathPathLength + 1 ); // +1 for slash
- $source = "$wgMathDirectory/$rel";
- $dest = "{$this->dest}/math/$rel";
- @mkdir( "{$this->dest}/math", 0755 );
- if ( !file_exists( $dest ) ) {
- copy( $source, $dest );
- }
+ $this->copyImage( $image, $wgSharedUploadPath, $wgSharedUploadDirectory, $this->sharedStaticDirectory );
+ } elseif ( substr( $image, 0, $mathPathLength ) == $wgMathPath ) {
+ $this->relativeCopy( $image, $wgMathPath, $wgMathDirectory, "{$this->dest}/math" );
+ } elseif ( $this->makeSnapshot && substr( $image, 0, $uploadPathLength ) == $wgUploadPath ) {
+ $this->copyImage( $image, $wgUploadPath, $wgUploadDirectory, $this->destUploadDirectory );
}
}
}
if ( $title->getDBkey() == '' ) {
$url = str_replace( '$1', "../$iw/index.html", $wgArticlePath );
} else {
- $url = str_replace( '$1', "../$iw/" . wfUrlencode( $this->getHashedFilename( $title ) ),
+ $url = str_replace( '$1', "../$iw/" . wfUrlencode( $this->getHashedFilename( $title ) ),
$wgArticlePath );
}
+ $url .= $this->compress ? ".gz" : "";
return false;
} else {
return true;
}
}
-
+
function onGetLocalURL( &$title, &$url, $query ) {
global $wgArticlePath;
$url = false;
if ( $query != '' ) {
+ $params = array();
parse_str( $query, $params );
if ( isset($params['action']) && $params['action'] == 'raw' ) {
if ( $params['gen'] == 'css' || $params['gen'] == 'js' ) {
} else {
$file = $this->getFriendlyName( $title->getPrefixedDBkey() );
// Clean up Monobook.css etc.
+ $matches = array();
if ( preg_match( '/^(.*)\.(css|js)_[0-9a-f]{4}$/', $file, $matches ) ) {
$file = $matches[1] . '.' . $matches[2];
}
if ( $url === false ) {
$url = str_replace( '$1', wfUrlencode( $this->getHashedFilename( $title ) ), $wgArticlePath );
}
-
+ $url .= $this->compress ? ".gz" : "";
return false;
}
return 'index.html';
}
- return $this->getHashedDirectory( $title ) . '/' .
+ return $this->getHashedDirectory( $title ) . '/' .
$this->getFriendlyName( $dbkey ) . '.html';
}
return $friendlyName;
}
-
+
/**
* Get a relative directory for putting a title into
*/
}
# Split into characters
+ $m = array();
preg_match_all( '/./us', $dbk, $m );
$chars = $m[0];
return $dir;
}
+ /**
+ * Calculate the start end end of a job based on the current slice
+ * @param integer $start
+ * @param integer $end
+ * @return array of integers
+ */
+ function sliceRange( $start, $end ) {
+ $count = $end - $start + 1;
+ $each = $count / $this->sliceDenominator;
+ $sliceStart = $start + intval( $each * ( $this->sliceNumerator - 1 ) );
+ if ( $this->sliceNumerator == $this->sliceDenominator ) {
+ $sliceEnd = $end;
+ } else {
+ $sliceEnd = $start + intval( $each * $this->sliceNumerator ) - 1;
+ }
+ return array( $sliceStart, $sliceEnd );
+ }
+
+ /**
+ * Adjust a start point so that it belongs to the current slice, where slices are defined by integer modulo
+ * @param integer $start
+ * @param integer $base The true start of the range; the minimum start
+ */
+ function modSliceStart( $start, $base = 1 ) {
+ return $start - ( $start % $this->sliceDenominator ) + $this->sliceNumerator - 1 + $base;
+ }
+
+ /**
+ * Determine whether a string belongs to the current slice, based on hash
+ */
+ function sliceFilter( $s ) {
+ return crc32( $s ) % $this->sliceDenominator == $this->sliceNumerator - 1;
+ }
+
+ /**
+ * No site notice
+ */
+ function onSiteNoticeBefore( &$text ) {
+ $text = '';
+ return false;
+ }
+ function onSiteNoticeAfter( &$text ) {
+ $text = '';
+ return false;
+ }
+
+ function getMaxPageID() {
+ if ( $this->maxPageID === false ) {
+ $dbr = wfGetDB( DB_SLAVE );
+ $this->maxPageID = $dbr->selectField( 'page', 'max(page_id)', false, __METHOD__ );
+ }
+ return $this->maxPageID;
+ }
+
+ function profile() {
+ global $wgProfiler;
+
+ if ( !$this->udpProfile ) {
+ return;
+ }
+ if ( !$this->udpProfileInit ) {
+ $this->udpProfileInit = true;
+ } elseif ( $this->udpProfileCounter == 1 % $this->udpProfile ) {
+ $wgProfiler->getFunctionReport();
+ $wgProfiler = new DumpHTML_ProfilerStub;
+ }
+ if ( $this->udpProfileCounter == 0 ) {
+ $wgProfiler = new ProfilerSimpleUDP;
+ $wgProfiler->setProfileID( 'dumpHTML' );
+ }
+ $this->udpProfileCounter = ( $this->udpProfileCounter + 1 ) % $this->udpProfile;
+ }
+}
+
+class DumpHTML_ProfilerStub {
+ function profileIn() {}
+ function profileOut() {}
+ function getOutput() {}
+ function close() {}
+ function getFunctionReport() {}
}
/** XML parser callback */