Club Drupal

<?php

/**

 * @file

 * Tests for FeedsXPathParser.inc.

 */

/**

 * Test single feeds.

 */

class FeedsXPathParseHTMLTestCase extends FeedsXPathParserWebTestCase {

  /**

   * Describe this test.

   */

  public static function getInfo() {

    return array(

      'name' => 'HTML Parser',

      'description' => 'Regression tests for Feeds XPath HTML parser.',

      'group' => 'Feeds XPath Parser',

    );

  }

  /**

   * Run tests.

   */

  public function test() {

    $this->createImporterConfiguration('XPath', 'xpath');

    $this->setPlugin('xpath', 'FeedsXPathParserHTML');

    $this->addMappings('xpath', array(

      0 => array(

        'source' => 'xpathparser:0',

        'target' => 'title',

        'unique' => FALSE,

      ),

      1 => array(

        'source' => 'xpathparser:1',

        'target' => 'url',

        'unique' => TRUE,

      ),

    ));

    // Set importer default settings.

    $importer_url = $this->feeds_base . '/xpath/settings/FeedsXPathParserHTML';

    $edit = array(

      'xpath[context]' => '//tr[starts-with(@class, "odd ") or starts-with(@class, "even ")]',

      'xpath[sources][xpathparser:0]' => 'td[1]/a',

      'xpath[sources][xpathparser:1]' => 'td[1]/a/@href',

    );

    $this->postAndCheck($importer_url, $edit, t('Save'), t('Your changes have been saved.'));

    // Test import.

    // Set batch limit to 5 to force batching.

    variable_set('feeds_process_limit', 5);

    $path = $GLOBALS['base_url'] . '/' . drupal_get_path('module', 'feeds_xpathparser') . '/tests/feeds_xpathparser/';

    $nid = $this->createFeedNode('xpath', $path . 'issues_drupal.org.htm', 'Testing XPath HTML Parser');

    $feed_node_edit_url = 'node/' . $nid . '/edit';

    $this->assertText(t('Created 29 nodes'));

    // Import again, this verifies url field was mapped correctly.

    $this->drupalPost('node/' . $nid . '/import', array(), t('Import'));

    $this->assertText(t('There are no new nodes'));

    // Assert accuracy of aggregated content. I find humor in using our own

    // issue queue to run tests against.

    $this->drupalGet('node');

    $this->assertText('Xpath Functions');

    $this->assertText('Unable to upload .html files');

    $this->assertText('Import to multiple content types');

    $this->assertText('Parser includes tags in mapped output');

    $this->assertText('Errors');

    $this->assertText('Loop through HTML - all data is in one node?');

    $this->assertText('Patch: add encoding options for PHP tidy feature');

    $this->assertText('Import and Maintain 1300+ Node Items');

    $this->assertText('Documentation update');

    $this->assertText('An HTTP error 404 occured');

    $this->assertText('Does it work with Feeds Image Grabber');

    $this->assertText('Node published date not being correctly mapped (set to 1 Jan 1970)');

    $this->assertText('fields to fill xpath not displayed in importer interface except for "body"');

    $this->assertText('parsing link field');

    $this->assertText('Error when switching to XML Parser');

    $this->assertText('Duplicate content even if "unique target" is set');

    $this->assertText('Labels/field names become meaningless with Data Processor');

    $this->assertText('Xpath namespace help');

    $this->assertText('warning: mysql_real_escape_string()');

    $this->assertText('Feeds XPath Parser: warning: Invalid argument');

    $this->assertText('What am I missing? FeedsXPathParser: No mappings are defined.');

    $this->assertText('CDATA in tag not producing text');

    $this->assertText('Cant map empty fields');

    $this->assertText('Support literal XPath expressions');

    $this->assertText('adding a prefix to a parsed xml value.');

    $this->assertText('Mapping on import');

    $this->assertText('Feeds XPath Parser: HTML parser example for number expressions');

    $this->assertText("I dont want to define any field queries");

    $this->assertText("Document // and other syntax for this module a little better");

    // Test debugging.

    $edit = array(

      'feeds[FeedsXPathParserHTML][xpath][exp][debug][xpathparser:0]' => TRUE,

    );

    $this->postAndCheck($feed_node_edit_url, $edit, t('Save'), 'Basic page Testing XPath HTML Parser has been updated.');

    $this->drupalPost('node/' . $nid . '/import', array(), t('Import'));

    $this->assertText('<a href="http://drupal.org/node/976478">Xpath Functions</a>');

    $this->assertText('<a href="http://drupal.org/node/1048030">Unable to upload .html files</a>');

    $this->assertText('<a href="http://drupal.org/node/1050310">Import to multiple content types</a>');

    $this->assertText('<a href="http://drupal.org/node/1047788">Parser includes tags in mapped output</a>');

    $this->assertText('<a href="http://drupal.org/node/1043608">Errors</a>');

    $this->assertText('<a href="http://drupal.org/node/1044546">Loop through HTML - all data is in one node?</a>');

    $this->assertText('<a href="http://drupal.org/node/1043728">Patch: add encoding options for PHP tidy feature</a>');

    $this->assertText('<a href="http://drupal.org/node/1040132">Import and Maintain 1300+ Node Items</a>');

    $this->assertText('<a href="http://drupal.org/node/1043604">Documentation update</a>');

    $this->assertText('<a href="http://drupal.org/node/1039492">An HTTP error 404 occured</a>');

    $this->assertText('<a href="http://drupal.org/node/1042048">Does it work with Feeds Image Grabber</a>');

    $this->assertText('<a href="http://drupal.org/node/961158">Node published date not being correctly mapped (set to 1 Jan 1970)</a>');

    $this->assertText('<a href="http://drupal.org/node/1021474">fields to fill xpath not displayed in importer interface except for "body"</a>');

    $this->assertText('<a href="http://drupal.org/node/1040530">parsing link field</a>');

    $this->assertText('<a href="http://drupal.org/node/1038912">Error when switching to XML Parser</a>');

    $this->assertText('<a href="http://drupal.org/node/1032340">Duplicate content even if "unique target" is set</a>');

    $this->assertText('<a href="http://drupal.org/node/982102">Labels/field names become meaningless with Data Processor</a>');

    $this->assertText('<a href="http://drupal.org/node/1034758">Xpath namespace help</a>');

    $this->assertText('<a href="http://drupal.org/node/908458">warning: mysql_real_escape_string()</a>');

    $this->assertText('<a href="http://drupal.org/node/869076">Feeds XPath Parser: warning: Invalid argument</a>');

    $this->assertText('<a href="http://drupal.org/node/991386">What am I missing? FeedsXPathParser: No mappings are defined.</a>');

    $this->assertText('<a href="http://drupal.org/node/990972">CDATA in tag not producing text</a>');

    $this->assertText('<a href="http://drupal.org/node/989948">Cant map empty fields</a>');

    $this->assertText('<a href="http://drupal.org/node/973324">Support literal XPath expressions</a>');

    $this->assertText('<a href="http://drupal.org/node/958344">adding a prefix to a parsed xml value.</a>');

    $this->assertText('<a href="http://drupal.org/node/914216">Mapping on import</a>');

    $this->assertText('<a href="http://drupal.org/node/863714">Feeds XPath Parser: HTML parser example for number expressions</a>');

    $this->assertText('<a href="http://drupal.org/node/915856">I dont want to define any field queries</a>');

    $this->assertText('<a href="http://drupal.org/node/950150">Document // and other syntax for this module a little better</a>');

    $this->assertText(t('There are no new nodes'));

    // Turn debugging off.

    $edit = array(

      'feeds[FeedsXPathParserHTML][xpath][exp][debug][xpathparser:0]' => FALSE,

    );

    $this->postAndCheck($feed_node_edit_url, $edit, t('Save'), 'Basic page Testing XPath HTML Parser has been updated.');

    // Test that overriding default settings works.

    $edit = array(

      'feeds[FeedsXPathParserHTML][xpath][context]' => '/foo',

      'feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:0]' => 'bar',

      'feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:1]' => 'baz',

    );

    $this->postAndCheck($feed_node_edit_url, $edit, t('Save'), 'Basic page Testing XPath HTML Parser has been updated.');

    // Assert the we don't create an empty node when XPath values don't return anything.

    // That happened at one point.

    $this->drupalPost('node/' . $nid . '/import', array(), t('Import'));

    $this->assertText(t('There are no new nodes'));

    // Test that validation works.

    $edit = array(

      'feeds[FeedsXPathParserHTML][xpath][context]' => 'sdf asf',

      'feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:0]' => 'asdf[sadfas asdf]',

    );

    $this->drupalPost($feed_node_edit_url, $edit, 'Save');

    // Check for valid error messages.

    $this->assertText('There was an error with the XPath selector: Invalid expression');

    $this->assertText('There was an error with the XPath selector: Invalid predicate');

    // Make sure the fields are errored out correctly. I.e. we have red outlines.

    $this->assertFieldByXPath('//input[@id="edit-feeds-feedsxpathparserhtml-xpath-context"][1]/@class', 'form-text required error');

    $this->assertFieldByXPath('//input[@id="edit-feeds-feedsxpathparserhtml-xpath-sources-xpathparser0"][1]/@class', 'form-text error');

    // Put the values back so we can test inheritance if the form was changed

    // and then changed back.

    $edit = array(

      'feeds[FeedsXPathParserHTML][xpath][context]' => '//tr[starts-with(@class, "odd ") or starts-with(@class, "even ")]',

      'feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:0]' => 'td[1]/a',

      'feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:1]' => 'td[1]/a/@href',

    );

    $this->postAndCheck($feed_node_edit_url, $edit, t('Save'), t('Basic page Testing XPath HTML Parser has been updated.'));

    // Change importer defaults.

    $edit = array(

      'xpath[context]' => '//tr',

      'xpath[sources][xpathparser:0]' => 'booya',

      'xpath[sources][xpathparser:1]' => 'boyz',

    );

    $this->postAndCheck($importer_url, $edit, t('Save'), t('Your changes have been saved.'));

    // Make sure the changes propigated.

    $this->drupalGet($feed_node_edit_url);

    $this->assertFieldByName('feeds[FeedsXPathParserHTML][xpath][context]', '//tr');

    $this->assertFieldByName('feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:0]', 'booya');

    $this->assertFieldByName('feeds[FeedsXPathParserHTML][xpath][sources][xpathparser:1]', 'boyz');

    //Cleanup

    $this->drupalPost("node/$nid/delete-items", array(), t('Delete'));

    $this->assertText(t('Deleted 29 nodes'));

    $this->_testGetRaw($importer_url);

  }

  public function _testGetRaw($importer_url) {

    $this->addMappings('xpath', array(

      2 => array(

        'source' => 'xpathparser:2',

        'target' => 'body',

      ),

    ));

    // Change importer defaults.

    $edit = array(

      'xpath[context]' => '/html',

      'xpath[sources][xpathparser:0]' => 'head/title',

      'xpath[sources][xpathparser:2]' => 'body',

      'xpath[rawXML][xpathparser:2]' => TRUE,

    );

    $this->postAndCheck($importer_url, $edit, t('Save'), t('Your changes have been saved.'));

    $path = $GLOBALS['base_url'] . '/' . drupal_get_path('module', 'feeds_xpathparser') . '/tests/feeds_xpathparser/';

    $nid = $this->createFeedNode('xpath', $path . 'simple.html', 'Testing GetRaw');

    $feed_node_edit_url = "node/$nid/edit";

    $this->assertText(t('Created 1 node'));

    $url = 'node/' . ++$nid . '/edit';

    $this->drupalGet($url);

    $this->assertFieldByName('body[und][0][value]', '<body><div>bla bla</div></body>');

  }

}