Issue #2259751 by mikeryan: First stab at leveraging Migrate new namespace support

wordpress.inc

@@ -319,6 +319,9 @@ class WordPressBlog {
    *  The raw WXR file as uploaded.
    * @param $destination
    *  Filespec to which to write the cleaned-up WXR file.
+   *
+   * @return array
+   *  List of referenced namespaces, keyed by prefix.
    */
   static public function preprocessFile($sourcefile, $destination) {
     // Cleanup some stuff in the process of moving the file to its final
@@ -353,6 +356,12 @@ class WordPressBlog {
     $itunes_ns = 'xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"';
     $header = str_replace($content_ns, $itunes_ns . "\n\t" . $content_ns, $header);
 
+    preg_match_all('|xmlns:(.+?)="(.+?)"|i', $header, $matches, PREG_SET_ORDER);
+    $namespaces = array();
+    foreach ($matches as $index => $match) {
+      $namespaces[$match[1]] = $match[2];
+    }
+
     // Replace HTML entities with XML entities
     $header = strtr($header, self::$entityReplacements);
     fputs($dest_handle, $header);
@@ -374,6 +383,8 @@ class WordPressBlog {
     fclose($dest_handle);
     fclose($source_handle);
     unlink($sourcefile);
+
+    return $namespaces;
   }
 
   /**

wordpress_attachment.inc

@@ -60,13 +60,13 @@ class WordPressAttachment extends WordPressMigration {
       MigrateDestinationFile::getKeySchema()
     );
 
-    $this->addFieldMapping('value', 'attachment_url');
-    $this->addFieldMapping('destination_file', 'attachment_url')
+    $this->addFieldMapping('value', 'wp:attachment_url');
+    $this->addFieldMapping('destination_file', 'wp:attachment_url')
          ->callbacks(array($this, 'fixDestinationFile'));
-    $this->addFieldMapping('uid', 'creator')
+    $this->addFieldMapping('uid', 'dc:creator')
          ->description('Use matching username if any, otherwise current user');
-    $this->addFieldMapping('timestamp', 'post_date');
-    $this->addFieldMapping(NULL, 'post_parent')
+    $this->addFieldMapping('timestamp', 'wp:post_date');
+    $this->addFieldMapping(NULL, 'sp:post_parent')
          ->description('Mapping is handled dynamically');
 
     // Unmapped destination fields
@@ -74,14 +74,14 @@ class WordPressAttachment extends WordPressMigration {
                                      'file_replace', 'preserve_files', 'source_dir'));
 
     // Unmapped source fields
-    $this->addUnmigratedSources(array('link', 'guid', 'description', 'excerpt',
-      'post_id', 'comment_status', 'ping_status', 'post_name', 'status',
-      'menu_order', 'post_type', 'post_password', 'is_sticky', 'tag', 'category',
-      '_wp_attached_file', '_wp_attachment_metadata', 'title', 'content'));
+    $this->addUnmigratedSources(array('link', 'guid', 'description', 'excerpt:encoded',
+      'wp:post_id', 'wp:comment_status', 'wp:ping_status', 'wp:post_name', 'wp:status',
+      'wp:menu_order', 'wp:post_type', 'wp:post_password', 'wp:is_sticky', 'tag', 'category',
+      '_wp_attached_file', '_wp_attachment_metadata', 'title', 'content:encoded'));
     $this->addFieldMapping(NULL, 'pubDate')
          ->description('Use post_date')
          ->issueGroup(t('DNM'));
-    $this->addFieldMapping(NULL, 'post_date_gmt')
+    $this->addFieldMapping(NULL, 'wp:post_date_gmt')
          ->description('Use post_date')
          ->issueGroup(t('DNM'));
   }

wordpress_author.inc

@@ -29,12 +29,12 @@ class WordPressAuthor extends WordPressMigration {
     );
 
     $fields = array(
-      'author_id' => 'Unique WordPress ID of the author',
-      'author_login' => 'Username (for login) of the author',
-      'author_email' => 'Author email address',
-      'author_display_name' => 'Displayed author name',
-      'author_first_name' => 'Author first name',
-      'author_last_name' => 'Author last name',
+      'wp:author_id' => 'Unique WordPress ID of the author',
+      'wp:author_login' => 'Username (for login) of the author',
+      'wp:author_email' => 'Author email address',
+      'wp:author_display_name' => 'Displayed author name',
+      'wp:author_first_name' => 'Author first name',
+      'wp:author_last_name' => 'Author last name',
     );
 
     // Construct the source and destination objects.
@@ -43,13 +43,13 @@ class WordPressAuthor extends WordPressMigration {
       'cache_counts' => TRUE,
     );
     $this->source = new WordPressSourceXML($this->wxrFile, '/rss/channel/wp:author',
-      'wp:author_login', $fields, $source_options);
+      'wp:author_login', $fields, $source_options, $this->arguments['namespaces']);
     $this->destination = new WordPressAuthorDestination($arguments);
 
     // The basic mappings
-    $this->addFieldMapping('name', 'author_login')
+    $this->addFieldMapping('name', 'wp:author_login')
          ->dedupe('users', 'name');
-    $this->addFieldMapping('mail', 'author_email');
+    $this->addFieldMapping('mail', 'wp:author_email');
     $this->addFieldMapping('roles')
          ->defaultValue(DRUPAL_AUTHENTICATED_RID);
     $this->addFieldMapping('status')
@@ -64,8 +64,8 @@ class WordPressAuthor extends WordPressMigration {
     }
 
     // Unmapped source fields
-    $this->addUnmigratedSources(array('author_display_name', 'author_first_name',
-                                'author_last_name', 'author_id'));
+    $this->addUnmigratedSources(array('wp:author_display_name', 'wp:author_first_name',
+                                'wp:author_last_name', 'wp:author_id'));
   }
 }
 

wordpress_category.inc

@@ -43,13 +43,13 @@ class WordPressCategory extends WordPressMigration {
       'cache_counts' => TRUE,
     );
     $this->source = new WordPressSourceXML($this->wxrFile, '/rss/channel/category',
-      'wp:category_nicename', $fields, $source_options);
+      'wp:category_nicename', $fields, $source_options, $this->arguments['namespaces']);
     $this->destination = new MigrateDestinationTerm($this->arguments['category_vocabulary']);
 
     // The basic mappings
-    $this->addFieldMapping('name', 'cat_name');
-    $this->addFieldMapping('description', 'category_description');
-    $this->addFieldMapping('parent', 'category_parent')
+    $this->addFieldMapping('name', 'wp:cat_name');
+    $this->addFieldMapping('description', 'wp:category_description');
+    $this->addFieldMapping('parent', 'wp:category_parent')
          ->sourceMigration($this->machineName);
 
     // Unmapped destination fields

wordpress_comment.inc

@@ -132,19 +132,19 @@ class WordPressComment extends WordPressMigration {
     );
 
     $fields = array(
-        'post_id' => 'Unique ID of the item the comment is attached to',
-        'comment_id' => 'Unique ID of the comment',
-        'comment_author' => 'Name of comment author',
-        'comment_author_email' => 'Email address of comment author',
-        'comment_author_url' => 'URL of comment author',
-        'comment_author_IP' => 'IP address from which comment was posted',
-        'comment_date' => 'Date of comment (what timezone?)',
-        'comment_date_gmt' => 'Date of comment (GMT)',
-        'comment_content' => 'Body of comment',
-        'comment_approved' => '1/0/spam - spam comments will not be imported',
-        'comment_type' => '?',
-        'comment_parent' => 'comment_id (?) of parent comment',
-        'comment_user_id' => 'WordPress user ID of commenter (?)',
+        'wp:post_id' => 'Unique ID of the item the comment is attached to',
+        'wp:comment_id' => 'Unique ID of the comment',
+        'wp:comment_author' => 'Name of comment author',
+        'wp:comment_author_email' => 'Email address of comment author',
+        'wp:comment_author_url' => 'URL of comment author',
+        'wp:comment_author_IP' => 'IP address from which comment was posted',
+        'wp:comment_date' => 'Date of comment (what timezone?)',
+        'wp:comment_date_gmt' => 'Date of comment (GMT)',
+        'wp:comment_content' => 'Body of comment',
+        'wp:comment_approved' => '1/0/spam - spam comments will not be imported',
+        'wp:comment_type' => '?',
+        'wp:comment_parent' => 'comment_id (?) of parent comment',
+        'wp:comment_user_id' => 'WordPress user ID of commenter (?)',
       );
 
     $source_options = array(
@@ -152,44 +152,44 @@ class WordPressComment extends WordPressMigration {
       'cache_counts' => TRUE,
     );
     $this->source = new WordPressSourceXML($this->wxrFile, '/rss/channel/item/comment',
-      'wp:comment_id', $fields, $source_options);
+      'wp:comment_id', $fields, $source_options, $this->arguments['namespaces']);
     $this->destination = new MigrateDestinationComment('comment_node_' . $arguments['post_type']);
 
     // The basic mappings
-    $this->addFieldMapping('nid', 'post_id')
+    $this->addFieldMapping('nid', 'wp:post_id')
          ->sourceMigration($this->arguments['dependencies']);
-    $this->addFieldMapping('pid', 'comment_parent')
+    $this->addFieldMapping('pid', 'wp:comment_parent')
          ->sourceMigration($this->generateMachineName('WordPressComment'));
     $this->addFieldMapping('uid')
          ->description('Use email to match Drupal account; if no match, default to anonymous');
     $this->addFieldMapping('subject')
          ->description('No comment subjects in WP')
          ->issueGroup('DNM');
-    $this->addFieldMapping('hostname', 'comment_author_IP');
-    $this->addFieldMapping('created', 'comment_date');
-    $this->addFieldMapping('changed', 'comment_date');
-    $this->addFieldMapping(NULL, 'comment_date_gmt')
+    $this->addFieldMapping('hostname', 'wp:comment_author_IP');
+    $this->addFieldMapping('created', 'wp:comment_date');
+    $this->addFieldMapping('changed', 'wp:comment_date');
+    $this->addFieldMapping(NULL, 'wp:comment_date_gmt')
          ->description('Using comment_date')
          ->issueGroup('DNM');
-    $this->addFieldMapping('status', 'comment_approved')
+    $this->addFieldMapping('status', 'wp:comment_approved')
          ->description('Do not import those with values of "spam"');
     $this->addFieldMapping('thread')
          ->issueGroup('DNM');
-    $this->addFieldMapping('name', 'comment_author')
+    $this->addFieldMapping('name', 'wp:comment_author')
          ->callbacks(array($this, 'truncateName'));
-    $this->addFieldMapping('mail', 'comment_author_email');
-    $this->addFieldMapping('homepage', 'comment_author_url');
+    $this->addFieldMapping('mail', 'wp:comment_author_email');
+    $this->addFieldMapping('homepage', 'wp:comment_author_url');
     $this->addFieldMapping('language');
-    $this->addFieldMapping(NULL, 'comment_id')
+    $this->addFieldMapping(NULL, 'wp:comment_id')
          ->description('Source primary key')
          ->issueGroup('DNM');
-    $this->addFieldMapping(NULL, 'comment_user_id')
+    $this->addFieldMapping(NULL, 'wp:comment_user_id')
          ->description('Always 0?')
          ->issueGroup('DNM');
-    $this->addFieldMapping(NULL, 'comment_type')
+    $this->addFieldMapping(NULL, 'wp:comment_type')
          ->description('Always empty?')
          ->issueGroup('DNM');
-    $this->addFieldMapping('comment_body', 'comment_content');
+    $this->addFieldMapping('comment_body', 'wp:comment_content');
     $this->addFieldMapping('comment_body:format')
          ->defaultValue($arguments['text_format_comment']);
     $this->addFieldMapping('comment_body:language');

wordpress_item.inc

@@ -52,14 +52,15 @@ class WordPressItemSource extends WordPressSourceXML {
    * Simple initialization.
    *
    */
-  public function __construct($filename, $post_type, $cache_key) {
+  public function __construct($filename, $post_type, $cache_key, $namespaces = array()) {
     $source_options = array(
       'reader_class' => 'WordPressXMLReader',
       'cache_counts' => TRUE,
       'cache_key' => $cache_key,
     );
+    drush_print_r($namespaces);
     parent::__construct($filename, '/rss/channel/item', 'wp:post_id',
-      $this->fields, $source_options);
+      $this->fields, $source_options, $namespaces);
     $this->postType = $post_type;
   }
 
@@ -71,8 +72,10 @@ class WordPressItemSource extends WordPressSourceXML {
     foreach ($this->sourceUrls as $url) {
       $reader = new $this->readerClass($url, $this->elementQuery, $this->idQuery);
       foreach ($reader as $element) {
+//        drush_print_r($element);
         // Only count relevant postType
-        if ($element->post_type == $this->postType) {
+        $field = 'wp:post_type';
+        if ($element->$field == $this->postType) {
           $count++;
         }
       }
@@ -147,34 +150,34 @@ abstract class WordPressItemMigration extends WordPressMigration {
     );
 
     // Construct the source objects.
-    $this->source = new WordPressItemSource($this->wxrFile, $this->postType, $this->machineName);
+    $this->source = new WordPressItemSource($this->wxrFile, $this->postType, $this->machineName, $this->arguments['namespaces']);
     $this->destination = new MigrateDestinationNode($bundle);
 
     // Default mappings, applying to most or all migrations
     $this->addFieldMapping('title', 'title');
-    $this->addFieldMapping('created', 'post_date')
+    $this->addFieldMapping('created', 'wp:post_date')
          ->description('Empty dates handled in prepare()');
-    $this->addFieldMapping('changed', 'post_date')
+    $this->addFieldMapping('changed', 'wp:post_date')
          ->description('Empty dates handled in prepare()');
     // If we have a separate author migration, use it here
-    $uid_mapping = $this->addFieldMapping('uid', 'creator')
+    $uid_mapping = $this->addFieldMapping('uid', 'dc:creator')
          ->description('Use matching username if any, otherwise current user');
     if ($this->blog->getWxrVersion() != '1.0') {
       $uid_mapping->sourceMigration($this->group->getName() . 'Author');
     }
-    $this->addFieldMapping('body', 'content');
-    $this->addFieldMapping('body:summary', 'excerpt');
+    $this->addFieldMapping('body', 'content:encoded');
+    $this->addFieldMapping('body:summary', 'excerpt:encoded');
     $this->addFieldMapping('body:format')
          ->defaultValue($this->arguments['text_format']);
     if (module_exists('comment')) {
-      $this->addFieldMapping('comment', 'comment_status')
+      $this->addFieldMapping('comment', 'wp:comment_status')
            ->description('WP "open" mapped to Drupal COMMENT_NODE_OPEN');
     }
-    $this->addFieldMapping('status', 'status')
+    $this->addFieldMapping('status', 'wp:status')
          ->description('Set Drupal status to 1 iff wp:status=publish');
-    $this->addFieldMapping(NULL, 'post_parent')
+    $this->addFieldMapping(NULL, 'wp:post_parent')
          ->description('Only applies to attachments');
-    $this->addFieldMapping('sticky', 'is_sticky');
+    $this->addFieldMapping('sticky', 'wp:is_sticky');
     if (module_exists('path')) {
       switch ($this->arguments['path_action']) {
         // Do not set path aliases
@@ -257,7 +260,7 @@ abstract class WordPressItemMigration extends WordPressMigration {
       'promote', 'revision_uid', 'log', 'tnid', 'translate', 'body:language'));
 
     // Unmapped source fields
-    $this->addUnmigratedSources(array('post_id', 'menu_order', 'post_type'));
+    $this->addUnmigratedSources(array('wp:post_id', 'wp:menu_order', 'wp:post_type'));
     $this->addFieldMapping(NULL, 'guid')
          ->description('same as link, plus isPermaLink attribute?')
          ->issueGroup(t('DNM'));
@@ -267,17 +270,17 @@ abstract class WordPressItemMigration extends WordPressMigration {
     $this->addFieldMapping(NULL, 'pubDate')
          ->description('Use post_date')
          ->issueGroup(t('DNM'));
-    $this->addFieldMapping(NULL, 'post_date_gmt')
+    $this->addFieldMapping(NULL, 'wp:post_date_gmt')
          ->description('Use post_date')
          ->issueGroup(t('DNM'));
-    $this->addFieldMapping(NULL, 'ping_status')
+    $this->addFieldMapping(NULL, 'wp:ping_status')
          ->description('What does this mean?')
          ->issueGroup(t('Open issues'))
          ->issuePriority(MigrateFieldMapping::ISSUE_PRIORITY_MEDIUM);
-    $this->addFieldMapping(NULL, 'post_name')
+    $this->addFieldMapping(NULL, 'wp:post_name')
          ->description('Looks like last component of path')
          ->issueGroup(t('DNM'));
-    $this->addFieldMapping(NULL, 'post_password')
+    $this->addFieldMapping(NULL, 'wp:post_password')
          ->description('???')
          ->issueGroup(t('DNM'));
   }
@@ -453,13 +456,15 @@ abstract class WordPressItemMigration extends WordPressMigration {
       // Attempt to parse embedded media automatically through the media module.
       try {
         $uri = media_parse_to_uri($src);
+        if ($uri) {
           // Sometimes, at least with oembed, it doesn't like the /v/ form
           $uri = str_replace('youtube.com/v/', 'youtube.com/watch?v=', $uri);
         }
+      }
       catch (Exception $e) {
         return $result;
       }
-      if (!isset($uri)) {
+      if (empty($uri)) {
         return $result;
       }
 

wordpress_migrate.migrate.inc

@@ -33,6 +33,13 @@ class WordPressMigrateWizard extends MigrateUIWizard {
 
   protected $contentValues = array();
 
+  /**
+   * List of namespaces referenced by the WXR file, keyed by prefix.
+   *
+   * @var array
+   */
+  protected $namespaces = array();
+
   /**
    * Lay out the basic steps of the wizard.
    */
@@ -259,7 +266,7 @@ class WordPressMigrateWizard extends MigrateUIWizard {
       }
     }
 
-    WordPressBlog::preprocessFile($tmpfile, $this->filename);
+    $this->namespaces = WordPressBlog::preprocessFile($tmpfile, $this->filename);
 
     // Only include the author step if we have author data (which was introduced
     // in WXR 1.1)
@@ -628,6 +635,7 @@ class WordPressMigrateWizard extends MigrateUIWizard {
     $this->groupArguments = array(
       'filename' => $this->filename,
       'source_system' => $this->getSourceName(),
+      'namespaces' => $this->namespaces,
     );
 
     $message = t('<p>Please review your migration configuration. When you submit this

wordpress_source.inc

@@ -12,7 +12,9 @@ class WordPressXMLReader extends MigrateXMLReader {
    * @return void
    */
   public function next() {
+    $this->registerNamespaces($this->currentElement);
     parent::next();
+    return;
     if (is_a($this->currentElement, 'SimpleXMLElement')) {
       $item = $this->currentElement;
       $currentRow = new stdClass;
@@ -61,6 +63,18 @@ class WordPressXMLReader extends MigrateXMLReader {
       $this->currentElement = $currentRow;
     }
   }
+
+  /**
+   * Explicitly register namespaces on an XML element.
+   *
+   * @param SimpleXMLElement $xml
+   *   A SimpleXMLElement to register the namespaces on.
+   */
+  protected function registerNamespaces(SimpleXMLElement &$xml) {
+    foreach ($this->namespaces as $prefix => $namespace) {
+      $xml->registerXPathNamespace($prefix, $namespace);
+    }
+  }
 }
 
 /**

wordpress_tag.inc

@@ -31,9 +31,9 @@ class WordPressTag extends WordPressMigration {
     );
 
     $fields = array(
-        'tag_slug' => 'Unique "machine name" of the tag',
-        'tag_name' => 'User-friendly tag name',
-        'tag_description' => 'Description of tag',
+        'wp:tag_slug' => 'Unique "machine name" of the tag',
+        'wp:tag_name' => 'User-friendly tag name',
+        'wp:tag_description' => 'Description of tag',
     );
 
     // Construct the source and destination objects.
@@ -42,12 +42,12 @@ class WordPressTag extends WordPressMigration {
       'cache_counts' => TRUE,
     );
     $this->source = new WordPressSourceXML($this->wxrFile, '/rss/channel/tag',
-      'wp:tag_slug', $fields, $source_options);
+      'wp:tag_slug', $fields, $source_options, $this->arguments['namespaces']);
     $this->destination = new MigrateDestinationTerm($arguments['tag_vocabulary']);
 
     // The basic mappings
-    $this->addFieldMapping('name', 'tag_name');
-    $this->addFieldMapping('description', 'tag_description');
+    $this->addFieldMapping('name', 'wp:tag_name');
+    $this->addFieldMapping('description', 'wp:tag_description');
     $this->addFieldMapping('parent_name')
          ->issueGroup('DNM');