Add skipping to nth page option/ability for dump importing process

Usage: php importDump.php --skip-to 271500 /path_to/dumpfile.xml.gz

When importing a database dump and the import process crashes
(for random reasons) after a certain number of pages, the
"--skip-to" parameter allows restarting the import process at
a certain page instead of starting the import from scratch.

Change-Id: Ib36063b69d6846fc197800bba44287493b0632c0

1 files changed, 21 insertions, 0 deletions

diff --git a/includes/import/WikiImporter.php b/includes/import/WikiImporter.php
index 06b579a7d9f7..2fc9f5e52740 100644
--- a/includes/import/WikiImporter.php
+++ b/includes/import/WikiImporter.php
@@ -39,6 +39,7 @@ class WikiImporter {
 	private $mNoticeCallback, $mDebug;
 	private $mImportUploads, $mImageBasePath;
 	private $mNoUpdates = false;
+	private $pageOffset = 0;
 	/** @var Config */
 	private $config;
 	/** @var ImportTitleFactory */
@@ -147,6 +148,16 @@ class WikiImporter {
 	}
 
 	/**
+	 * 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
 	 *
 	 * @param callable $callback
@@ -562,9 +573,19 @@ class WikiImporter {
 		$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 ] ) ) {