Correctly parse 'redirect' XML tag during Special:Import.

Bug: 65481
Change-Id: Id9b3b7878b2e7b6fc7a06b163e5bac60e700490e

diff --git a/includes/Import.php b/includes/Import.php
index 4940c19..0f88082 100644 (file)
--- a/includes/Import.php
+++ b/includes/Import.php
@@ -393,6 +393,15 @@ class WikiImporter {
                }
        }
 
+       /**
+        * Retrieves the contents of the named attribute of the current element.
+        * @param string $attr the name of the attribute
+        * @return string the value of the attribute or an empty string if it is not set in the current element.
+        */
+       public function nodeAttribute( $attr ) {
+               return $this->reader->getAttribute( $attr );
+       }
+
        /**
         * Shouldn't something like this be built-in to XMLReader?
         * Fetches text contents of the current element, assuming
@@ -618,17 +627,28 @@ class WikiImporter {
                                                &$pageInfo ) ) ) {
                                // Do nothing
                        } elseif ( in_array( $tag, $normalFields ) ) {
-                               $pageInfo[$tag] = $this->nodeContents();
-                               if ( $tag == 'title' ) {
-                                       $title = $this->processTitle( $pageInfo['title'] );
+                               // An XML snippet:
+                               // <page>
+                               //     <id>123</id>
+                               //     <title>Page</title>
+                               //     <redirect title="NewTitle"/>
+                               //     ...
+                               // Because the redirect tag is built differently, we need special handling for that case.
+                               if ( $tag == 'redirect' ) {
+                                       $pageInfo[$tag] = $this->nodeAttribute( 'title' );
+                               } else {
+                                       $pageInfo[$tag] = $this->nodeContents();
+                                       if ( $tag == 'title' ) {
+                                               $title = $this->processTitle( $pageInfo['title'] );
 
-                                       if ( !$title ) {
-                                               $badTitle = true;
-                                               $skip = true;
-                                       }
+                                               if ( !$title ) {
+                                                       $badTitle = true;
+                                                       $skip = true;
+                                               }
 
-                                       $this->pageCallback( $title );
-                                       list( $pageInfo['_title'], $origTitle ) = $title;
+                                               $this->pageCallback( $title );
+                                               list( $pageInfo['_title'], $origTitle ) = $title;
+                                       }
                                }
                        } elseif ( $tag == 'revision' ) {
                                $this->handleRevision( $pageInfo );

diff --git a/tests/phpunit/includes/ImportTest.php b/tests/phpunit/includes/ImportTest.php
new file mode 100644 (file)
index 0000000..8895403
--- /dev/null
+++ b/tests/phpunit/includes/ImportTest.php
@@ -0,0 +1,100 @@
+<?php
+
+/**
+ * Test class for Import methods.
+ *
+ * @group Database
+ *
+ * @author Sebastian Brückner < sebastian.brueckner@student.hpi.uni-potsdam.de >
+ */
+class ImportTest extends MediaWikiLangTestCase {
+
+       private function getInputStreamSource( $xml ) {
+               $file = 'data:application/xml,' . $xml;
+               $status = ImportStreamSource::newFromFile( $file );
+               if ( !$status->isGood() ) {
+                       throw new MWException( "Cannot create InputStreamSource." );
+               }
+               return $status->value;
+       }
+
+       /**
+        * @covers WikiImporter::handlePage
+        * @dataProvider getRedirectXML
+        * @param string $xml
+        */
+       public function testHandlePageContainsRedirect( $xml, $redirectTitle ) {
+               $source = $this->getInputStreamSource( $xml );
+
+               $redirect = NULL;
+               $callback = function( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) use ( &$redirect ) {
+                       if ( array_key_exists( 'redirect', $pageInfo ) ) {
+                               $redirect = $pageInfo['redirect'];
+                       }
+               };
+
+               $importer = new WikiImporter( $source );
+               $importer->setPageOutCallback( $callback );
+               $importer->doImport();
+
+               $this->assertEquals( $redirectTitle, $redirect );
+       }
+
+       public function getRedirectXML() {
+               return array(
+                       array(
+                               <<< EOF
+<mediawiki>
+       <page>
+               <title>Test</title>
+               <ns>0</ns>
+               <id>21</id>
+               <redirect title="Test22"/>
+               <revision>
+                       <id>20</id>
+                       <timestamp>2014-05-27T10:00:00Z</timestamp>
+                       <contributor>
+                               <username>Admin</username>
+                               <id>10</id>
+                       </contributor>
+                       <comment>Admin moved page [[Test]] to [[Test22]]</comment>
+                       <text xml:space="preserve" bytes="20">#REDIRECT [[Test22]]</text>
+                       <sha1>tq456o9x3abm7r9ozi6km8yrbbc56o6</sha1>
+                       <model>wikitext</model>
+                       <format>text/x-wiki</format>
+               </revision>
+       </page>
+</mediawiki>
+EOF
+                       ,
+                               'Test22'
+                       ),
+                       array(
+                               <<< EOF
+<mediawiki>
+       <page>
+               <title>Test</title>
+               <ns>0</ns>
+               <id>42</id>
+               <revision>
+                       <id>421</id>
+                       <timestamp>2014-05-27T11:00:00Z</timestamp>
+                       <contributor>
+                               <username>Admin</username>
+                               <id>10</id>
+                       </contributor>
+                       <text xml:space="preserve" bytes="4">Abcd</text>
+                       <sha1>n7uomjq96szt60fy5w3x7ahf7q8m8rh</sha1>
+                       <model>wikitext</model>
+                       <format>text/x-wiki</format>
+               </revision>
+       </page>
+</mediawiki>
+EOF
+                       ,
+                               NULL
+                       ),
+               );
+       }
+
+}