- Patch #16282 by Arancaytar, mustafau, keith.smith, et al: add OPML import...

- Patch #16282 by Arancaytar, mustafau, keith.smith, et al: add OPML import functionality for RSS feeds.  Woot.

CHANGELOG.txt

@@ -19,6 +19,8 @@ Drupal 7.0, xxxx-xx-xx (development version)
       slimmed down install profile for developers.
     * Image toolkits are added by modules instead of being copied to
       the includes directory.
+- News aggregator:
+    * Added OPML import functionality for RSS feeds.
 - Search:
     * Added support for language-aware searches.
 - Testing:

modules/aggregator/aggregator.admin.inc

@@ -213,6 +213,151 @@ function aggregator_admin_remove_feed_submit($form, &$form_state) {
   $form_state['redirect'] = 'admin/content/aggregator';
 }
 
+/**
+ * Form builder; Generate a form to import feeds from OPML.
+ *
+ * @ingroup forms
+ * @see aggregator_form_opml_validate()
+ * @see aggregator_form_opml_submit()
+ */
+function aggregator_form_opml(&$form_state) {
+  $period = drupal_map_assoc(array(900, 1800, 3600, 7200, 10800, 21600, 32400, 43200, 64800, 86400, 172800, 259200, 604800, 1209600, 2419200), 'format_interval');
+
+  $form['#attributes'] = array('enctype' => "multipart/form-data");
+
+  $form['upload'] = array(
+    '#type' => 'file',
+    '#title' => t('OPML File'),
+    '#description' => t('Upload an OPML file containing a list of feeds to be imported.'),
+  );
+  $form['remote'] = array(
+    '#type' => 'textfield',
+    '#title' => t('OPML Remote URL'),
+    '#description' => t('Enter the URL of an OPML file. This file will be downloaded and processed only once on submission of the form.'),
+  );
+  $form['refresh'] = array(
+    '#type' => 'select',
+    '#title' => t('Update interval'),
+    '#default_value' => 3600,
+    '#options' => $period,
+    '#description' => t('The length of time between feed updates. (Requires a correctly configured <a href="@cron">cron maintenance task</a>.)', array('@cron' => url('admin/reports/status'))),
+  );
+
+  // Handling of categories.
+  $options = array();
+  $categories = db_query('SELECT cid, title FROM {aggregator_category} ORDER BY title');
+  while ($category = db_fetch_object($categories)) {
+    $options[$category->cid] = check_plain($category->title);
+  }
+  if ($options) {
+    $form['category'] = array(
+      '#type' => 'checkboxes',
+      '#title' => t('Categorize news items'),
+      '#options' => $options,
+      '#description' => t('New feed items are automatically filed in the checked categories.'),
+    );
+  }
+  $form['submit'] = array(
+    '#type' => 'submit',
+    '#value' => t('Import')
+  );
+
+  return $form;
+}
+
+/**
+ * Validate aggregator_form_opml form submissions.
+ */
+function aggregator_form_opml_validate($form, &$form_state) {
+  // If both fields are empty or filled, cancel.
+  if (empty($form_state['values']['remote']) == empty($_FILES['files']['name']['upload'])) {
+    form_set_error('remote', t('You must <em>either</em> upload a file or enter a URL.'));
+  }
+
+  // Validate the URL, if one was entered.
+  if (!empty($form_state['values']['remote']) && !valid_url($form_state['values']['remote'], TRUE)) {
+    form_set_error('remote', t('This URL is not valid.'));
+  }
+}
+
+/**
+ * Process aggregator_form_opml form submissions.
+ */
+function aggregator_form_opml_submit($form, &$form_state) {
+  $data = '';
+  if ($file = file_save_upload('upload')) {
+    $data = file_get_contents($file->filepath);
+  }
+  else {
+    $response = drupal_http_request($form_state['values']['remote']);
+    if (!isset($response->error)) {
+      $data = $response->data;
+    }
+  }
+
+  $feeds = _aggregator_parse_opml($data);
+  if (empty($feeds)) {
+    drupal_set_message(t('No new feed has been added.'));
+    return;
+  }
+
+  $form_state['values']['op'] = t('Save');
+
+  foreach ($feeds as $feed) {
+    $result = db_query("SELECT title, url FROM {aggregator_feed} WHERE title = '%s' OR url = '%s'", $feed['title'], $feed['url']);
+    $duplicate = FALSE;
+    while ($old = db_fetch_object($result)) {
+      if (strcasecmp($old->title, $feed['title']) == 0) {
+        drupal_set_message(t('A feed named %title already exists.', array('%title' => $old->title)), 'warning');
+        $duplicate = TRUE;
+        continue;
+      }
+      if (strcasecmp($old->url, $feed['url']) == 0) {
+        drupal_set_message(t('A feed with the URL %url already exists.', array('%url' => $old->url)), 'warning');
+        $duplicate = TRUE;
+        continue;
+      }
+    }
+    if (!$duplicate) {
+      $form_state['values']['title'] = $feed['title'];
+      $form_state['values']['url'] = $feed['url'];
+      drupal_execute('aggregator_form_feed', $form_state);
+    }
+  }
+}
+
+/**
+ * Parse an OPML file.
+ *
+ * Feeds are recognized as <outline> elements with the attributes
+ * <em>text</em> and <em>xmlurl</em> set.
+ *
+ * @param $opml
+ *   The complete contents of an OPML document.
+ * @return
+ *   An array of feeds, each an associative array with a <em>title</em> and
+ *   a <em>url</em> element, or NULL if the OPML document failed to be parsed.
+ *   An empty array will be returned if the document is valid but contains
+ *   no feeds, as some OPML documents do.
+ */
+function _aggregator_parse_opml($opml) {
+  $feeds = array();
+  $xml_parser = drupal_xml_parser_create($opml);
+  if (xml_parse_into_struct($xml_parser, $opml, $values)) {
+    foreach ($values as $entry) {
+      if ($entry['tag'] == 'OUTLINE' && isset($entry['attributes'])) {
+        $item = $entry['attributes'];
+        if (!empty($item['XMLURL'])) {
+          $feeds[] = array('title' => $item['TEXT'], 'url' => $item['XMLURL']);
+        }
+      }
+    }
+  }
+  xml_parser_free($xml_parser);
+
+  return $feeds;
+}
+
 /**
  * Menu callback; refreshes a feed, then redirects to the overview page.
  *

modules/aggregator/aggregator.module

@@ -24,6 +24,8 @@ function aggregator_help($path, $arg) {
       return '<p>' . t('Add a feed in RSS, RDF or Atom format. A feed may only have one entry.') . '</p>';
     case 'admin/content/aggregator/add/category':
       return '<p>' . t('Categories allow feed items from different feeds to be grouped together. For example, several sport-related feeds may belong to a category named <em>Sports</em>. Feed items may be grouped automatically (by selecting a category when creating or editing a feed) or manually (via the <em>Categorize</em> page available from feed item listings). Each category provides its own feed page and block.') . '</p>';
+    case 'admin/content/aggregator/add/opml':
+      return '<p>' . t('<acronym title="Outline Processor Markup Language">OPML</acronym> is an XML format used to exchange multiple feeds between aggregators. A single OPML document may contain a collection of many feeds. Drupal can parse such a file and import all feeds at once, saving you the effort of adding them manually. You may either upload a local file from your computer or enter a URL where Drupal can download it.') . '</p>';
   }
 }
 
@@ -101,6 +103,14 @@ function aggregator_menu() {
     'type' => MENU_LOCAL_TASK,
     'parent' => 'admin/content/aggregator',
   );
+  $items['admin/content/aggregator/add/opml'] = array(
+    'title' => 'Import OPML',
+    'page callback' => 'drupal_get_form',
+    'page arguments' => array('aggregator_form_opml'),
+    'access arguments' => array('administer news feeds'),
+    'type' => MENU_LOCAL_TASK,
+    'parent' => 'admin/content/aggregator',
+  );
   $items['admin/content/aggregator/remove/%aggregator_feed'] = array(
     'title' => 'Remove items',
     'page callback' => 'drupal_get_form',

modules/aggregator/aggregator.test

@@ -116,6 +116,85 @@ class AggregatorTestCase extends DrupalWebTestCase {
     $result = db_result(db_query("SELECT count(*) FROM {aggregator_feed} WHERE title = '%s' AND url='%s'", $feed_name, $feed_url));
     return (1 == $result);
   }
+
+  /**
+   * Create a valid OPML file from an array of feeds.
+   *
+   * @param $feeds
+   *   An array of feeds.
+   * @return
+   *   Path to valid OPML file.
+   */
+  function getValidOpml($feeds) {
+    /**
+     * Does not have an XML declaration, must pass the parser.
+     */
+    $opml = <<<EOF
+<opml version="1.0">
+  <head></head>
+  <body>
+    <!-- First feed to be imported. -->
+    <outline text="{$feeds[0]['title']}" xmlurl="{$feeds[0]['url']}" />
+
+    <!-- Second feed. Test string delimitation and attribute order. -->
+    <outline xmlurl='{$feeds[1]['url']}' text='{$feeds[1]['title']}'/>
+
+    <!-- Test for duplicate URL and title. -->
+    <outline xmlurl="{$feeds[0]['url']}" text="Duplicate URL"/>
+    <outline xmlurl="http://duplicate.title" text="{$feeds[1]['title']}"/>
+
+    <!-- Test that feeds are only added with required attributes. -->
+    <outline text="{$feeds[2]['title']}" />
+    <outline xmlurl="{$feeds[2]['url']}" />
+  </body>
+</opml>
+EOF;
+
+    $path = file_directory_path() . '/valid-opml.xml';
+    file_save_data($opml, $path);
+    return $path;
+  }
+
+  /**
+   * Create an invalid OPML file.
+   *
+   * @return
+   *   Path to invalid OPML file.
+   */
+  function getInvalidOpml() {
+    $opml = <<<EOF
+<opml>
+  <invalid>
+</opml>
+EOF;
+
+    $path = file_directory_path() . '/invalid-opml.xml';
+    file_save_data($opml, $path);
+    return $path;
+  }
+
+  /**
+   * Create a valid but empty OPML file.
+   *
+   * @return
+   *   Path to empty OPML file.
+   */
+  function getEmptyOpml() {
+    $opml = <<<EOF
+<?xml version="1.0" encoding="utf-8"?>
+<opml version="1.0">
+  <head></head>
+  <body>
+    <outline text="Sample text" />
+    <outline text="Sample text" url="Sample URL" />
+  </body>
+</opml>
+EOF;
+
+    $path = file_directory_path() . '/empty-opml.xml';
+    file_save_data($opml, $path);
+    return $path;
+  }
 }
 
 class AddFeedTestCase extends AggregatorTestCase {
@@ -312,3 +391,121 @@ class CategorizeFeedItemTestCase extends AggregatorTestCase {
     $this->deleteFeed($feed);
   }
 }
+
+class ImportOPMLTestCase extends AggregatorTestCase {
+  private static $prefix = 'simpletest_aggregator_';
+
+  /**
+   * Implementation of getInfo().
+   */
+  function getInfo() {
+    return array(
+      'name' => t('Import feeds from OPML functionality'),
+      'description' => t('Test OPML import.'),
+      'group' => t('Aggregator'),
+    );
+  }
+
+  /**
+   * Open OPML import form.
+   */
+  function openImportForm() {
+    db_query('TRUNCATE {aggregator_category}');
+
+    $category = $this->randomName(10, self::$prefix);
+    db_query("INSERT INTO {aggregator_category} (cid, title, description) VALUES (%d, '%s', '%s')", 1, $category, '');
+
+    $this->drupalGet('admin/content/aggregator/add/opml');
+    $this->assertText('A single OPML document may contain a collection of many feeds.', t('Looking for help text.'));
+    $this->assertFieldByName('files[upload]', '', t('Looking for file upload field.'));
+    $this->assertFieldByName('remote', '', t('Looking for remote URL field.'));
+    $this->assertFieldByName('refresh', '', t('Looking for refresh field.'));
+    $this->assertFieldByName('category[1]', '1', t('Looking for category field.'));
+  }
+
+  /**
+   * Submit form filled with invalid fields.
+   */
+  function validateImportFormFields() {
+    $before = db_result(db_query('SELECT COUNT(*) FROM {aggregator_feed}'));
+
+    $form = array();
+    $this->drupalPost('admin/content/aggregator/add/opml', $form, t('Import'));
+    $this->assertRaw(t('You must <em>either</em> upload a file or enter a URL.'), t('Error if no fields are filled.'));
+
+    $path = $this->getEmptyOpml();
+    $form = array(
+      'files[upload]' => $path,
+      'remote' => file_create_url($path),
+    );
+    $this->drupalPost('admin/content/aggregator/add/opml', $form, t('Import'));
+    $this->assertRaw(t('You must <em>either</em> upload a file or enter a URL.'), t('Error if both fields are filled.'));
+
+    $form = array('remote' => 'invalidUrl://empty');
+    $this->drupalPost('admin/content/aggregator/add/opml', $form, t('Import'));
+    $this->assertText(t('This URL is not valid.'), t('Error if the URL is invalid.'));
+
+    $after = db_result(db_query('SELECT COUNT(*) FROM {aggregator_feed}'));
+    $this->assertEqual($before, $after, t('No feeds were added during the three last form submissions.'));
+  }
+
+  /**
+   * Submit form with invalid, empty and valid OPML files.
+   */
+  function submitImportForm() {
+    $before = db_result(db_query('SELECT COUNT(*) FROM {aggregator_feed}'));
+
+    $form['files[upload]'] = $this->getInvalidOpml();
+    $this->drupalPost('admin/content/aggregator/add/opml', $form, t('Import'));
+    $this->assertText(t('No new feed has been added.'), t('Attempting to upload invalid XML.'));
+
+    $form = array('remote' => file_create_url($this->getEmptyOpml()));
+    $this->drupalPost('admin/content/aggregator/add/opml', $form, t('Import'));
+    $this->assertText(t('No new feed has been added.'), t('Attempting to load empty OPML from remote URL.'));
+
+    $after = db_result(db_query('SELECT COUNT(*) FROM {aggregator_feed}'));
+    $this->assertEqual($before, $after, t('No feeds were added during the two last form submissions.'));
+
+    db_query('TRUNCATE {aggregator_feed}');
+    db_query('TRUNCATE {aggregator_category}');
+    db_query('TRUNCATE {aggregator_category_feed}');
+
+    $category = $this->randomName(10, self::$prefix);
+    db_query("INSERT INTO {aggregator_category} (cid, title, description) VALUES (%d, '%s', '%s')", 1, $category, '');
+
+    $feeds[0] = $this->getFeedEditArray();
+    $feeds[1] = $this->getFeedEditArray();
+    $feeds[2] = $this->getFeedEditArray();
+    $form = array(
+      'files[upload]' => $this->getValidOpml($feeds),
+      'refresh'       => '900',
+      'category[1]'   => $category,
+    );
+    $this->drupalPost('admin/content/aggregator/add/opml', $form, t('Import'));
+    $this->assertRaw(t('A feed with the URL %url already exists.', array('%url' => $feeds[0]['url'])), t('Verifying that a duplicate URL was identified'));
+    $this->assertRaw(t('A feed named %title already exists.', array('%title' => $feeds[1]['title'])), t('Verifying that a duplicate title was identified'));
+
+    $after = db_result(db_query('SELECT COUNT(*) FROM {aggregator_feed}'));
+    $this->assertEqual($after, 2, t('Verifying that two distinct feeds were added.'));
+
+    $feeds_from_db = db_query("SELECT f.title, f.url, f.refresh, cf.cid FROM {aggregator_feed} f LEFT JOIN {aggregator_category_feed} cf ON f.fid = cf.fid");
+    $refresh = $category = TRUE;
+    while ($feed = db_fetch_array($feeds_from_db)) {
+      $title[$feed['url']] = $feed['title'];
+      $url[$feed['title']] = $feed['url'];
+      $category = $category && $feed['cid'] == 1;
+      $refresh = $refresh && $feed['refresh'] == 900;
+    }
+
+    $this->assertEqual($title[$feeds[0]['url']], $feeds[0]['title'], t('First feed was added correctly.'));
+    $this->assertEqual($url[$feeds[1]['title']], $feeds[1]['url'], t('Second feed was added correctly.'));
+    $this->assertTrue($refresh, t('Refresh times are correct.'));
+    $this->assertTrue($category, t('Categories are correct.'));
+  }
+
+  function testOPMLImport() {
+    $this->openImportForm();
+    $this->validateImportFormFields();
+    $this->submitImportForm();
+  }
+}