/**
* @dataProvider firefoxtests
*/
public function test_from_file($data)
{
$locator = new SimplePie_Locator($data, 0, null, 'MockSimplePie_File', false);
$expected = SimplePie_Misc::get_element('link', $data->body);
$feed = $locator->find(SIMPLEPIE_LOCATOR_ALL, $all);
$this->assertFalse($locator->is_feed($data), 'HTML document not be a feed itself');
$this->assertInstanceOf('MockSimplePie_File', $feed);
$expected = array_map(array(get_class(), 'map_url_attrib'), $expected);
$success = array_filter($expected, array(get_class(), 'filter_success'));
$found = array_map(array(get_class(), 'map_url_file'), $all);
$this->assertEquals($success, $found);
}
/**
* Additional validation includes checking URL and tags
*/
public function validation($data, $files)
{
global $CFG;
$errors = parent::validation($data, $files);
require_once $CFG->libdir . '/simplepie/moodle_simplepie.php';
$rssfile = new moodle_simplepie_file($data['url']);
$filetest = new SimplePie_Locator($rssfile);
if (!$filetest->is_feed($rssfile)) {
$errors['url'] = get_string('feedisinvalid', 'blog');
} else {
$rss = new moodle_simplepie($data['url']);
if (!$rss->init()) {
$errors['url'] = get_string('emptyrssfeed', 'blog');
}
}
return $errors;
}
static function getFeedlinks($url)
{
if (!is_string($url)) {
return false;
}
$parsed = @parse_url($url);
if (empty($parsed['scheme'])) {
return false;
}
if (!in_array($parsed['scheme'], array('http', 'https'))) {
return false;
}
$timeout = 10;
$redirects = 5;
$headers = null;
$useragent = null;
$file = new SimplePie_File($url, $timeout, $redirects, $headers, $useragent);
$locator = new SimplePie_Locator($file, $timeout, $useragent);
$r = $locator->find();
return !empty($r->url) ? array($r->url) : false;
}
/**
* Given a record in the {blog_external} table, checks the blog's URL
* for new entries not yet copied into Moodle.
* Also attempts to identify and remove deleted blog entries
*
* @param object $externalblog
* @return boolean False if the Feed is invalid
*/
function blog_sync_external_entries($externalblog)
{
global $CFG, $DB;
require_once $CFG->libdir . '/simplepie/moodle_simplepie.php';
$rssfile = new moodle_simplepie_file($externalblog->url);
$filetest = new SimplePie_Locator($rssfile);
if (!$filetest->is_feed($rssfile)) {
$externalblog->failedlastsync = 1;
$DB->update_record('blog_external', $externalblog);
return false;
} else {
if (!empty($externalblog->failedlastsync)) {
$externalblog->failedlastsync = 0;
$DB->update_record('blog_external', $externalblog);
}
}
$rss = new moodle_simplepie($externalblog->url);
if (empty($rss->data)) {
return null;
}
//used to identify blog posts that have been deleted from the source feed
$oldesttimestamp = null;
$uniquehashes = array();
foreach ($rss->get_items() as $entry) {
// If filtertags are defined, use them to filter the entries by RSS category
if (!empty($externalblog->filtertags)) {
$containsfiltertag = false;
$categories = $entry->get_categories();
$filtertags = explode(',', $externalblog->filtertags);
$filtertags = array_map('trim', $filtertags);
$filtertags = array_map('strtolower', $filtertags);
foreach ($categories as $category) {
if (in_array(trim(strtolower($category->term)), $filtertags)) {
$containsfiltertag = true;
}
}
if (!$containsfiltertag) {
continue;
}
}
$uniquehashes[] = $entry->get_permalink();
$newentry = new stdClass();
$newentry->userid = $externalblog->userid;
$newentry->module = 'blog_external';
$newentry->content = $externalblog->id;
$newentry->uniquehash = $entry->get_permalink();
$newentry->publishstate = 'site';
$newentry->format = FORMAT_HTML;
// Clean subject of html, just in case
$newentry->subject = clean_param($entry->get_title(), PARAM_TEXT);
// Observe 128 max chars in DB
// TODO: +1 to raise this to 255
if (textlib::strlen($newentry->subject) > 128) {
$newentry->subject = textlib::substr($newentry->subject, 0, 125) . '...';
}
$newentry->summary = $entry->get_description();
//used to decide whether to insert or update
//uses enty permalink plus creation date if available
$existingpostconditions = array('uniquehash' => $entry->get_permalink());
//our DB doesnt allow null creation or modified timestamps so check the external blog supplied one
$entrydate = $entry->get_date('U');
if (!empty($entrydate)) {
$existingpostconditions['created'] = $entrydate;
}
//the post ID or false if post not found in DB
$postid = $DB->get_field('post', 'id', $existingpostconditions);
$timestamp = null;
if (empty($entrydate)) {
$timestamp = time();
} else {
$timestamp = $entrydate;
}
//only set created if its a new post so we retain the original creation timestamp if the post is edited
if ($postid === false) {
$newentry->created = $timestamp;
}
$newentry->lastmodified = $timestamp;
if (empty($oldesttimestamp) || $timestamp < $oldesttimestamp) {
//found an older post
$oldesttimestamp = $timestamp;
}
if (textlib::strlen($newentry->uniquehash) > 255) {
// The URL for this item is too long for the field. Rather than add
// the entry without the link we will skip straight over it.
// RSS spec says recommended length 500, we use 255.
debugging('External blog entry skipped because of oversized URL', DEBUG_DEVELOPER);
continue;
}
if ($postid === false) {
$id = $DB->insert_record('post', $newentry);
// Set tags
if ($tags = tag_get_tags_array('blog_external', $externalblog->id)) {
tag_set('post', $id, $tags);
}
} else {
$newentry->id = $postid;
$DB->update_record('post', $newentry);
}
}
// Look at the posts we have in the database to check if any of them have been deleted from the feed.
// Only checking posts within the time frame returned by the rss feed. Older items may have been deleted or
// may just not be returned anymore. We can't tell the difference so we leave older posts alone.
$sql = "SELECT id, uniquehash\n FROM {post}\n WHERE module = 'blog_external'\n AND " . $DB->sql_compare_text('content') . " = " . $DB->sql_compare_text(':blogid') . "\n AND created > :ts";
$dbposts = $DB->get_records_sql($sql, array('blogid' => $externalblog->id, 'ts' => $oldesttimestamp));
$todelete = array();
foreach ($dbposts as $dbpost) {
if (!in_array($dbpost->uniquehash, $uniquehashes)) {
$todelete[] = $dbpost->id;
}
}
$DB->delete_records_list('post', 'id', $todelete);
$DB->update_record('blog_external', array('id' => $externalblog->id, 'timefetched' => time()));
}
/**
* @dataProvider firefoxtests
*/
public function test_from_file($data)
{
$locator = new SimplePie_Locator($data, 0, null, false);
$registry = new SimplePie_Registry();
$registry->register('File', 'MockSimplePie_File');
$locator->set_registry($registry);
$expected = array();
$document = new DOMDocument();
$document->loadHTML($data->body);
$xpath = new DOMXPath($document);
foreach ($xpath->query('//link') as $element) {
$expected[] = 'http://example.com' . $element->getAttribute('href');
}
//$expected = SimplePie_Misc::get_element('link', $data->body);
$feed = $locator->find(SIMPLEPIE_LOCATOR_ALL, $all);
$this->assertFalse($locator->is_feed($data), 'HTML document not be a feed itself');
$this->assertInstanceOf('MockSimplePie_File', $feed);
$success = array_filter($expected, array(get_class(), 'filter_success'));
$found = array_map(array(get_class(), 'map_url_file'), $all);
$this->assertEquals($success, $found);
}
public static function autoDiscovery($url)
{
$parts = parse_url($url);
if (empty($parts['path'])) {
$url .= '/';
}
$result = '';
$proxy = null;
if (sfConfig::get('op_http_proxy')) {
$proxy = sfConfig::get('op_http_proxy');
}
$file = @new SimplePie_File($url, 10, 5, null, null, false, $proxy);
$locator = new SimplePie_Locator($file, 10, null, 'SimplePie_File', 10, $proxy);
$feedUrl = $locator->find();
if (SimplePie_Misc::is_a($feedUrl, 'SimplePie_File')) {
$result = $feedUrl->url;
}
return $result;
}
function check_feed($url)
{
require_once dirname(__FILE__) . '/lib/simplepie_1.3.compiled.php';
$file = new SimplePie_File($url);
$test = new SimplePie_Locator($file);
$test->set_registry(new SimplePie_Registry());
if ($test->is_feed($file)) {
return true;
} else {
return false;
}
}
/**
* RSS/Atom Auto-Discovery に対応したlinkタグからURLを抽出する(static)
*/
function auto_discovery($url)
{
// path 未指定の場合は「/」に設定する
$parts = parse_url($url);
if (empty($parts['path'])) {
$url .= '/';
}
$result = '';
if (OPENPNE_USE_HTTP_PROXY) {
$proxy = OPENPNE_HTTP_PROXY_HOST . ":" . OPENPNE_HTTP_PROXY_PORT;
}
$file = @new SimplePie_File($url, 10, 5, null, null, false, $proxy);
$locator = new SimplePie_Locator($file, 10, null, 'SimplePie_File', 10, $proxy);
$feed_url = $locator->find();
if (SimplePie_Misc::is_a($feed_url, 'SimplePie_File')) {
$result = $feed_url->url;
}
return $result;
}
/**
* Given a record in the {blog_external} table, checks the blog's URL
* for new entries not yet copied into Moodle.
*
* @param object $externalblog
* @return boolean False if the Feed is invalid
*/
function blog_sync_external_entries($externalblog)
{
global $CFG, $DB;
require_once $CFG->libdir . '/simplepie/moodle_simplepie.php';
$rssfile = new moodle_simplepie_file($externalblog->url);
$filetest = new SimplePie_Locator($rssfile);
if (!$filetest->is_feed($rssfile)) {
$externalblog->failedlastsync = 1;
$DB->update_record('blog_external', $externalblog);
return false;
} else {
if (!empty($externalblog->failedlastsync)) {
$externalblog->failedlastsync = 0;
$DB->update_record('blog_external', $externalblog);
}
}
// Delete all blog entries associated with this external blog
blog_delete_external_entries($externalblog);
$rss = new moodle_simplepie($externalblog->url);
if (empty($rss->data)) {
return null;
}
foreach ($rss->get_items() as $entry) {
// If filtertags are defined, use them to filter the entries by RSS category
if (!empty($externalblog->filtertags)) {
$containsfiltertag = false;
$categories = $entry->get_categories();
$filtertags = explode(',', $externalblog->filtertags);
$filtertags = array_map('trim', $filtertags);
$filtertags = array_map('strtolower', $filtertags);
foreach ($categories as $category) {
if (in_array(trim(strtolower($category->term)), $filtertags)) {
$containsfiltertag = true;
}
}
if (!$containsfiltertag) {
continue;
}
}
$newentry = new stdClass();
$newentry->userid = $externalblog->userid;
$newentry->module = 'blog_external';
$newentry->content = $externalblog->id;
$newentry->uniquehash = $entry->get_permalink();
$newentry->publishstate = 'site';
$newentry->format = FORMAT_HTML;
$newentry->subject = $entry->get_title();
$newentry->summary = $entry->get_description();
//our DB doesnt allow null creation or modified timestamps so check the external blog didnt supply one
$entrydate = $entry->get_date('U');
if (empty($entrydate)) {
$newentry->created = time();
$newentry->lastmodified = time();
} else {
$newentry->created = $entrydate;
$newentry->lastmodified = $entrydate;
}
$textlib = textlib_get_instance();
if ($textlib->strlen($newentry->uniquehash) > 255) {
// The URL for this item is too long for the field. Rather than add
// the entry without the link we will skip straight over it.
// RSS spec says recommended length 500, we use 255.
debugging('External blog entry skipped because of oversized URL', DEBUG_DEVELOPER);
continue;
}
$id = $DB->insert_record('post', $newentry);
// Set tags
if ($tags = tag_get_tags_array('blog_external', $externalblog->id)) {
tag_set('post', $id, $tags);
}
}
$DB->update_record('blog_external', array('id' => $externalblog->id, 'timefetched' => mktime()));
}