private $mNoticeCallback, $mDebug;
private $mImportUploads, $mImageBasePath;
private $mNoUpdates = false;
+ private $pageOffset = 0;
/** @var Config */
private $config;
/** @var ImportTitleFactory */
private $importTitleFactory;
/** @var array */
private $countableCache = [];
+ /** @var bool */
+ private $disableStatisticsUpdate = false;
/**
* Creates an ImportXMLReader drawing from the source provided
$this->mNoUpdates = $noupdates;
}
+ /**
+ * Sets 'pageOffset' value. So it will skip the first n-1 pages
+ * and start from the nth page. It's 1-based indexing.
+ * @param int $nthPage
+ * @since 1.29
+ */
+ function setPageOffset( $nthPage ) {
+ $this->pageOffset = $nthPage;
+ }
+
/**
* Set a callback that displays notice messages
*
$this->mImportUploads = $import;
}
+ /**
+ * Statistics update can cause a lot of time
+ * @since 1.29
+ */
+ public function disableStatisticsUpdate() {
+ $this->disableStatisticsUpdate = true;
+ }
+
/**
* Default per-page callback. Sets up some things related to site statistics
* @param array $titleAndForeignTitle Two-element array, with Title object at
// suffers from issues of replica DB lag. We let WikiPage handle the total page
// and revision count, and we implement our own custom logic for the
// article (content page) count.
- $page = WikiPage::factory( $title );
- $page->loadPageData( 'fromdbmaster' );
- $content = $page->getContent();
- if ( $content === null ) {
- wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
- ' because WikiPage::getContent() returned null' );
- } else {
- $editInfo = $page->prepareContentForEdit( $content );
- $countKey = 'title_' . $title->getPrefixedText();
- $countable = $page->isCountable( $editInfo );
- if ( array_key_exists( $countKey, $this->countableCache ) &&
- $countable != $this->countableCache[$countKey] ) {
- DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
- 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
- ] ) );
+ if ( !$this->disableStatisticsUpdate ) {
+ $page = WikiPage::factory( $title );
+ $page->loadPageData( 'fromdbmaster' );
+ $content = $page->getContent();
+ if ( $content === null ) {
+ wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
+ ' because WikiPage::getContent() returned null' );
+ } else {
+ $editInfo = $page->prepareContentForEdit( $content );
+ $countKey = 'title_' . $title->getPrefixedText();
+ $countable = $page->isCountable( $editInfo );
+ if ( array_key_exists( $countKey, $this->countableCache ) &&
+ $countable != $this->countableCache[$countKey] ) {
+ DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
+ 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
+ ] ) );
+ }
}
}
public function doImport() {
// Calls to reader->read need to be wrapped in calls to
// libxml_disable_entity_loader() to avoid local file
- // inclusion attacks (bug 46932).
+ // inclusion attacks (T48932).
$oldDisable = libxml_disable_entity_loader( true );
$this->reader->read();
$keepReading = $this->reader->read();
$skip = false;
$rethrow = null;
+ $pageCount = 0;
try {
while ( $keepReading ) {
$tag = $this->reader->localName;
+ if ( $this->pageOffset ) {
+ if ( $tag === 'page' ) {
+ $pageCount++;
+ }
+ if ( $pageCount < $this->pageOffset ) {
+ $keepReading = $this->reader->next();
+ continue;
+ }
+ }
$type = $this->reader->nodeType;
if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {