summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorAndrew Dolgov <[email protected]>2011-12-13 16:06:25 +0400
committerAndrew Dolgov <[email protected]>2011-12-13 16:06:25 +0400
commit2c08214a7f4dd693cb401714412b7a1ad635ed41 (patch)
tree1660ccc500ddcb309fe50ad91e604661488cdcc5
parentde8260cb10da0ac4e49e4b3fc3a42c3578eff007 (diff)
split rss updating stuff into separate include file
-rw-r--r--include/functions.php1272
-rw-r--r--include/rssfuncs.php1271
-rwxr-xr-xupdate.php1
-rwxr-xr-xupdate_daemon2.php1
4 files changed, 1278 insertions, 1267 deletions
diff --git a/include/functions.php b/include/functions.php
index 7bd64cc5b..2246e6bf3 100644
--- a/include/functions.php
+++ b/include/functions.php
@@ -389,1117 +389,6 @@
}
}
- function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false) {
-
- global $memcache;
-
- /* Update all feeds with the same URL to utilize memcache */
-
- if ($memcache) {
- $result = db_query($link, "SELECT f1.id
- FROM ttrss_feeds AS f1, ttrss_feeds AS f2
- WHERE f2.feed_url = f1.feed_url AND f2.id = '$feed'");
-
- while ($line = db_fetch_assoc($result)) {
- update_rss_feed_real($link, $line["id"], $ignore_daemon, $no_cache);
- }
- } else {
- update_rss_feed_real($link, $feed, $ignore_daemon, $no_cache);
- }
- }
-
- function update_rss_feed_real($link, $feed, $ignore_daemon = false, $no_cache = false,
- $override_url = false) {
-
- require_once "lib/simplepie/simplepie.inc";
- require_once "lib/magpierss/rss_fetch.inc";
- require_once 'lib/magpierss/rss_utils.inc';
-
- global $memcache;
-
- $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
-
- if (!$_REQUEST["daemon"] && !$ignore_daemon) {
- return false;
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: start");
- }
-
- if (!$ignore_daemon) {
-
- if (DB_TYPE == "pgsql") {
- $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
- } else {
- $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))";
- }
-
- $result = db_query($link, "SELECT id,update_interval,auth_login,
- auth_pass,cache_images,update_method
- FROM ttrss_feeds WHERE id = '$feed' AND $updstart_thresh_qpart");
-
- } else {
-
- $result = db_query($link, "SELECT id,update_interval,auth_login,
- feed_url,auth_pass,cache_images,update_method,last_updated,
- mark_unread_on_update, owner_uid, update_on_checksum_change,
- pubsub_state
- FROM ttrss_feeds WHERE id = '$feed'");
-
- }
-
- if (db_num_rows($result) == 0) {
- if ($debug_enabled) {
- _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
- }
- return false;
- }
-
- $update_method = db_fetch_result($result, 0, "update_method");
- $last_updated = db_fetch_result($result, 0, "last_updated");
- $owner_uid = db_fetch_result($result, 0, "owner_uid");
- $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
- 0, "mark_unread_on_update"));
- $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result,
- 0, "update_on_checksum_change"));
- $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
-
- db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
- WHERE id = '$feed'");
-
- $auth_login = db_fetch_result($result, 0, "auth_login");
- $auth_pass = db_fetch_result($result, 0, "auth_pass");
-
- if ($update_method == 0)
- $update_method = DEFAULT_UPDATE_METHOD + 1;
-
- // 1 - Magpie
- // 2 - SimplePie
- // 3 - Twitter OAuth
-
- if ($update_method == 2)
- $use_simplepie = true;
- else
- $use_simplepie = false;
-
- if ($debug_enabled) {
- _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n");
- }
-
- if ($update_method == 1) {
- $auth_login = urlencode($auth_login);
- $auth_pass = urlencode($auth_pass);
- }
-
- $update_interval = db_fetch_result($result, 0, "update_interval");
- $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
- $fetch_url = db_fetch_result($result, 0, "feed_url");
-
- if ($update_interval < 0) { return false; }
-
- $feed = db_escape_string($feed);
-
- if ($auth_login && $auth_pass ){
- $url_parts = array();
- preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts);
-
- if ($url_parts[1] && $url_parts[2]) {
- $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2];
- }
-
- }
-
- if ($override_url)
- $fetch_url = $override_url;
-
- if ($debug_enabled) {
- _debug("update_rss_feed: fetching [$fetch_url]...");
- }
-
- $obj_id = md5("FDATA:$use_simplepie:$fetch_url");
-
- if ($memcache && $obj = $memcache->get($obj_id)) {
-
- if ($debug_enabled) {
- _debug("update_rss_feed: data found in memcache.");
- }
-
- $rss = $obj;
-
- } else {
-
- if ($update_method == 3) {
- $rss = fetch_twitter_rss($link, $fetch_url, $owner_uid);
- } else if ($update_method == 1) {
-
- define('MAGPIE_CACHE_AGE', get_feed_update_interval($link, $feed) * 60);
- define('MAGPIE_CACHE_ON', !$no_cache);
- define('MAGPIE_FETCH_TIME_OUT', 60);
- define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie");
-
- $rss = @fetch_rss($fetch_url);
- } else {
- $simplepie_cache_dir = CACHE_DIR . "/simplepie";
-
- if (!is_dir($simplepie_cache_dir)) {
- mkdir($simplepie_cache_dir);
- }
-
- $rss = new SimplePie();
- $rss->set_useragent(SELF_USER_AGENT);
- # $rss->set_timeout(10);
- $rss->set_feed_url($fetch_url);
- $rss->set_output_encoding('UTF-8');
- $rss->force_feed(true);
-
- if (SIMPLEPIE_CACHE_IMAGES && $cache_images) {
-
- if ($debug_enabled) {
- _debug("enabling image cache");
- }
-
- $rss->set_image_handler("image.php", 'i');
- }
-
- if ($debug_enabled) {
- _debug("feed update interval (sec): " .
- get_feed_update_interval($link, $feed)*60);
- }
-
- $rss->enable_cache(!$no_cache);
-
- if (!$no_cache) {
- $rss->set_cache_location($simplepie_cache_dir);
- $rss->set_cache_duration(get_feed_update_interval($link, $feed) * 60);
- }
-
- $rss->init();
- }
-
- if ($memcache && $rss) $memcache->add($obj_id, $rss, 0, 300);
- }
-
-// print_r($rss);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: fetch done, parsing...");
- }
-
- $feed = db_escape_string($feed);
-
- if ($update_method == 2) {
- $fetch_ok = !$rss->error();
- } else {
- $fetch_ok = !!$rss;
- }
-
- if ($fetch_ok) {
-
- if ($debug_enabled) {
- _debug("update_rss_feed: processing feed data...");
- }
-
-// db_query($link, "BEGIN");
-
- $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid
- FROM ttrss_feeds WHERE id = '$feed'");
-
- $registered_title = db_fetch_result($result, 0, "title");
- $orig_icon_url = db_fetch_result($result, 0, "icon_url");
- $orig_site_url = db_fetch_result($result, 0, "site_url");
-
- $owner_uid = db_fetch_result($result, 0, "owner_uid");
-
- if ($use_simplepie) {
- $site_url = $rss->get_link();
- } else {
- $site_url = $rss->channel["link"];
- }
-
- $site_url = rewrite_relative_url($fetch_url, $site_url);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: checking favicon...");
- }
-
- check_feed_favicon($site_url, $feed, $link);
-
- if (!$registered_title || $registered_title == "[Unknown]") {
-
- if ($use_simplepie) {
- $feed_title = db_escape_string($rss->get_title());
- } else {
- $feed_title = db_escape_string($rss->channel["title"]);
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: registering title: $feed_title");
- }
-
- db_query($link, "UPDATE ttrss_feeds SET
- title = '$feed_title' WHERE id = '$feed'");
- }
-
- // weird, weird Magpie
- if (!$use_simplepie) {
- if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]);
- }
-
- if ($site_url && $orig_site_url != db_escape_string($site_url)) {
- db_query($link, "UPDATE ttrss_feeds SET
- site_url = '$site_url' WHERE id = '$feed'");
- }
-
-// print "I: " . $rss->channel["image"]["url"];
-
- if (!$use_simplepie) {
- $icon_url = db_escape_string($rss->image["url"]);
- } else {
- $icon_url = db_escape_string($rss->get_image_url());
- }
-
- $icon_url = substr($icon_url, 0, 250);
-
- if ($icon_url && $orig_icon_url != $icon_url) {
- db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'");
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: loading filters...");
- }
-
- $filters = load_filters($link, $feed, $owner_uid);
-
-// if ($debug_enabled) {
-// print_r($filters);
-// }
-
- if ($use_simplepie) {
- $iterator = $rss->get_items();
- } else {
- $iterator = $rss->items;
- if (!$iterator || !is_array($iterator)) $iterator = $rss->entries;
- if (!$iterator || !is_array($iterator)) $iterator = $rss;
- }
-
- if (!is_array($iterator)) {
- /* db_query($link, "UPDATE ttrss_feeds
- SET last_error = 'Parse error: can\'t find any articles.'
- WHERE id = '$feed'"); */
-
- // clear any errors and mark feed as updated if fetched okay
- // even if it's blank
-
- if ($debug_enabled) {
- _debug("update_rss_feed: entry iterator is not an array, no articles?");
- }
-
- db_query($link, "UPDATE ttrss_feeds
- SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
-
- return; // no articles
- }
-
- if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
-
- if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
-
- $feed_hub_url = false;
- if ($use_simplepie) {
- $links = $rss->get_links('hub');
-
- if ($links && is_array($links)) {
- foreach ($links as $l) {
- $feed_hub_url = $l;
- break;
- }
- }
-
- } else {
- $atom = $rss->channel['atom'];
-
- if ($atom) {
- if ($atom['link@rel'] == 'hub') {
- $feed_hub_url = $atom['link@href'];
- }
-
- if (!$feed_hub_url && $atom['link#'] > 1) {
- for ($i = 2; $i <= $atom['link#']; $i++) {
- if ($atom["link#$i@rel"] == 'hub') {
- $feed_hub_url = $atom["link#$i@href"];
- break;
- }
- }
- }
- } else {
- $feed_hub_url = $rss->channel['link_hub'];
- }
- }
-
- if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
-
- if ($feed_hub_url && function_exists('curl_init') &&
- !ini_get("open_basedir")) {
-
- require_once 'lib/pubsubhubbub/subscriber.php';
-
- $callback_url = get_self_url_prefix() .
- "/public.php?op=pubsub&id=$feed";
-
- $s = new Subscriber($feed_hub_url, $callback_url);
-
- $rc = $s->subscribe($fetch_url);
-
- if ($debug_enabled)
- _debug("update_rss_feed: feed hub url found, subscribe request sent.");
-
- db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
- WHERE id = '$feed'");
- }
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: processing articles...");
- }
-
- foreach ($iterator as $item) {
-
- if ($_REQUEST['xdebug'] == 2) {
- print_r($item);
- }
-
- if ($use_simplepie) {
- $entry_guid = $item->get_id();
- if (!$entry_guid) $entry_guid = $item->get_link();
- if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
-
- } else {
-
- $entry_guid = $item["id"];
-
- if (!$entry_guid) $entry_guid = $item["guid"];
- if (!$entry_guid) $entry_guid = $item["about"];
- if (!$entry_guid) $entry_guid = $item["link"];
- if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]);
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: guid $entry_guid");
- }
-
- if (!$entry_guid) continue;
-
- $entry_timestamp = "";
-
- if ($use_simplepie) {
- $entry_timestamp = strtotime($item->get_date());
- } else {
- $rss_2_date = $item['pubdate'];
- $rss_1_date = $item['dc']['date'];
- $atom_date = $item['issued'];
- if (!$atom_date) $atom_date = $item['updated'];
-
- if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date);
- if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date);
- if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date);
-
- }
-
- if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) {
- $entry_timestamp = time();
- $no_orig_date = 'true';
- } else {
- $no_orig_date = 'false';
- }
-
- $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
- }
-
- if ($use_simplepie) {
- $entry_title = $item->get_title();
- } else {
- $entry_title = trim(strip_tags($item["title"]));
- }
-
- if ($use_simplepie) {
- $entry_link = $item->get_link();
- } else {
- // strange Magpie workaround
- $entry_link = $item["link_"];
- if (!$entry_link) $entry_link = $item["link"];
- }
-
- $entry_link = rewrite_relative_url($site_url, $entry_link);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: title $entry_title");
- _debug("update_rss_feed: link $entry_link");
- }
-
- if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
-
- $entry_link = strip_tags($entry_link);
-
- if ($use_simplepie) {
- $entry_content = $item->get_content();
- if (!$entry_content) $entry_content = $item->get_description();
- } else {
- $entry_content = $item["content:escaped"];
-
- if (!$entry_content) $entry_content = $item["content:encoded"];
- if (!$entry_content) $entry_content = $item["content"]["encoded"];
- if (!$entry_content) $entry_content = $item["content"];
-
- if (is_array($entry_content)) $entry_content = $entry_content[0];
-
- // Magpie bugs are getting ridiculous
- if (trim($entry_content) == "Array") $entry_content = false;
-
- if (!$entry_content) $entry_content = $item["atom_content"];
- if (!$entry_content) $entry_content = $item["summary"];
-
- if (!$entry_content ||
- strlen($entry_content) < strlen($item["description"])) {
- $entry_content = $item["description"];
- };
-
- // WTF
- if (is_array($entry_content)) {
- $entry_content = $entry_content["encoded"];
- if (!$entry_content) $entry_content = $entry_content["escaped"];
- }
- }
-
- if ($_REQUEST["xdebug"] == 2) {
- print "update_rss_feed: content: ";
- print_r(htmlspecialchars($entry_content));
- }
-
- $entry_content_unescaped = $entry_content;
-
- if ($use_simplepie) {
- $entry_comments = strip_tags($item->data["comments"]);
- if ($item->get_author()) {
- $entry_author_item = $item->get_author();
- $entry_author = $entry_author_item->get_name();
- if (!$entry_author) $entry_author = $entry_author_item->get_email();
-
- $entry_author = db_escape_string($entry_author);
- }
- } else {
- $entry_comments = strip_tags($item["comments"]);
-
- $entry_author = db_escape_string(strip_tags($item['dc']['creator']));
-
- if ($item['author']) {
-
- if (is_array($item['author'])) {
-
- if (!$entry_author) {
- $entry_author = db_escape_string(strip_tags($item['author']['name']));
- }
-
- if (!$entry_author) {
- $entry_author = db_escape_string(strip_tags($item['author']['email']));
- }
- }
-
- if (!$entry_author) {
- $entry_author = db_escape_string(strip_tags($item['author']));
- }
- }
- }
-
- if (preg_match('/^[\t\n\r ]*$/', $entry_author)) $entry_author = '';
-
- $entry_guid = db_escape_string(strip_tags($entry_guid));
- $entry_guid = mb_substr($entry_guid, 0, 250);
-
- $result = db_query($link, "SELECT id FROM ttrss_entries
- WHERE guid = '$entry_guid'");
-
- $entry_content = db_escape_string($entry_content, false);
-
- $content_hash = "SHA1:" . sha1(strip_tags($entry_content));
-
- $entry_title = db_escape_string($entry_title);
- $entry_link = db_escape_string($entry_link);
- $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250);
- $entry_author = mb_substr($entry_author, 0, 250);
-
- if ($use_simplepie) {
- $num_comments = 0; #FIXME#
- } else {
- $num_comments = db_escape_string($item["slash"]["comments"]);
- }
-
- if (!$num_comments) $num_comments = 0;
-
- if ($debug_enabled) {
- _debug("update_rss_feed: looking for tags [1]...");
- }
-
- // parse <category> entries into tags
-
- $additional_tags = array();
-
- if ($use_simplepie) {
-
- $additional_tags_src = $item->get_categories();
-
- if (is_array($additional_tags_src)) {
- foreach ($additional_tags_src as $tobj) {
- array_push($additional_tags, $tobj->get_term());
- }
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: category tags:");
- print_r($additional_tags);
- }
-
- } else {
-
- $t_ctr = $item['category#'];
-
- if ($t_ctr == 0) {
- $additional_tags = array();
- } else if ($t_ctr > 0) {
- $additional_tags = array($item['category']);
-
- if ($item['category@term']) {
- array_push($additional_tags, $item['category@term']);
- }
-
- for ($i = 0; $i <= $t_ctr; $i++ ) {
- if ($item["category#$i"]) {
- array_push($additional_tags, $item["category#$i"]);
- }
-
- if ($item["category#$i@term"]) {
- array_push($additional_tags, $item["category#$i@term"]);
- }
- }
- }
-
- // parse <dc:subject> elements
-
- $t_ctr = $item['dc']['subject#'];
-
- if ($t_ctr > 0) {
- array_push($additional_tags, $item['dc']['subject']);
-
- for ($i = 0; $i <= $t_ctr; $i++ ) {
- if ($item['dc']["subject#$i"]) {
- array_push($additional_tags, $item['dc']["subject#$i"]);
- }
- }
- }
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: looking for tags [2]...");
- }
-
- /* taaaags */
- // <a href="..." rel="tag">Xorg</a>, //
-
- $entry_tags = null;
-
- preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i",
- $entry_content_unescaped, $entry_tags);
-
- $entry_tags = $entry_tags[1];
-
- $entry_tags = array_merge($entry_tags, $additional_tags);
- $entry_tags = array_unique($entry_tags);
-
- for ($i = 0; $i < count($entry_tags); $i++)
- $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
-
- if ($debug_enabled) {
- _debug("update_rss_feed: unfiltered tags found:");
- print_r($entry_tags);
- }
-
- # sanitize content
-
- $entry_content = sanitize_article_content($entry_content);
- $entry_title = sanitize_article_content($entry_title);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: done collecting data [TITLE:$entry_title]");
- }
-
- db_query($link, "BEGIN");
-
- if (db_num_rows($result) == 0) {
-
- if ($debug_enabled) {
- _debug("update_rss_feed: base guid not found");
- }
-
- // base post entry does not exist, create it
-
- $result = db_query($link,
- "INSERT INTO ttrss_entries
- (title,
- guid,
- link,
- updated,
- content,
- content_hash,
- no_orig_date,
- date_updated,
- date_entered,
- comments,
- num_comments,
- author)
- VALUES
- ('$entry_title',
- '$entry_guid',
- '$entry_link',
- '$entry_timestamp_fmt',
- '$entry_content',
- '$content_hash',
- $no_orig_date,
- NOW(),
- NOW(),
- '$entry_comments',
- '$num_comments',
- '$entry_author')");
- } else {
- // we keep encountering the entry in feeds, so we need to
- // update date_updated column so that we don't get horrible
- // dupes when the entry gets purged and reinserted again e.g.
- // in the case of SLOW SLOW OMG SLOW updating feeds
-
- $base_entry_id = db_fetch_result($result, 0, "id");
-
- db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
- WHERE id = '$base_entry_id'");
- }
-
- // now it should exist, if not - bad luck then
-
- $result = db_query($link, "SELECT
- id,content_hash,no_orig_date,title,
- ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
- ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
- num_comments
- FROM
- ttrss_entries
- WHERE guid = '$entry_guid'");
-
- $entry_ref_id = 0;
- $entry_int_id = 0;
-
- if (db_num_rows($result) == 1) {
-
- if ($debug_enabled) {
- _debug("update_rss_feed: base guid found, checking for user record");
- }
-
- // this will be used below in update handler
- $orig_content_hash = db_fetch_result($result, 0, "content_hash");
- $orig_title = db_fetch_result($result, 0, "title");
- $orig_num_comments = db_fetch_result($result, 0, "num_comments");
- $orig_date_updated = strtotime(db_fetch_result($result,
- 0, "date_updated"));
-
- $ref_id = db_fetch_result($result, 0, "id");
- $entry_ref_id = $ref_id;
-
- // check for user post link to main table
-
- // do we allow duplicate posts with same GUID in different feeds?
- if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
- $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
- } else {
- $dupcheck_qpart = "";
- }
-
- /* Collect article tags here so we could filter by them: */
-
- $article_filters = get_article_filters($filters, $entry_title,
- $entry_content, $entry_link, $entry_timestamp, $entry_author,
- $entry_tags);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: article filters: ");
- if (count($article_filters) != 0) {
- print_r($article_filters);
- }
- }
-
- if (find_article_filter($article_filters, "filter")) {
- db_query($link, "COMMIT"); // close transaction in progress
- continue;
- }
-
- $score = calculate_article_score($article_filters);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: initial score: $score");
- }
-
- $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
- ref_id = '$ref_id' AND owner_uid = '$owner_uid'
- $dupcheck_qpart";
-
-// if ($_REQUEST["xdebug"]) print "$query\n";
-
- $result = db_query($link, $query);
-
- // okay it doesn't exist - create user entry
- if (db_num_rows($result) == 0) {
-
- if ($debug_enabled) {
- _debug("update_rss_feed: user record not found, creating...");
- }
-
- if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
- $unread = 'true';
- $last_read_qpart = 'NULL';
- } else {
- $unread = 'false';
- $last_read_qpart = 'NOW()';
- }
-
- if (find_article_filter($article_filters, 'mark') || $score > 1000) {
- $marked = 'true';
- } else {
- $marked = 'false';
- }
-
- if (find_article_filter($article_filters, 'publish')) {
- $published = 'true';
- } else {
- $published = 'false';
- }
-
- $result = db_query($link,
- "INSERT INTO ttrss_user_entries
- (ref_id, owner_uid, feed_id, unread, last_read, marked,
- published, score, tag_cache, label_cache, uuid)
- VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
- $last_read_qpart, $marked, $published, '$score', '', '', '')");
-
- if (PUBSUBHUBBUB_HUB && $published == 'true') {
- $rss_link = get_self_url_prefix() .
- "/public.php?op=rss&id=-2&key=" .
- get_feed_access_key($link, -2, false, $owner_uid);
-
- $p = new Publisher(PUBSUBHUBBUB_HUB);
-
- $pubsub_result = $p->publish_update($rss_link);
- }
-
- $result = db_query($link,
- "SELECT int_id FROM ttrss_user_entries WHERE
- ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
- feed_id = '$feed' LIMIT 1");
-
- if (db_num_rows($result) == 1) {
- $entry_int_id = db_fetch_result($result, 0, "int_id");
- }
- } else {
- if ($debug_enabled) {
- _debug("update_rss_feed: user record FOUND");
- }
-
- $entry_ref_id = db_fetch_result($result, 0, "ref_id");
- $entry_int_id = db_fetch_result($result, 0, "int_id");
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
- }
-
- $post_needs_update = false;
- $update_insignificant = false;
-
- if ($orig_num_comments != $num_comments) {
- $post_needs_update = true;
- $update_insignificant = true;
- }
-
- if ($content_hash != $orig_content_hash) {
- $post_needs_update = true;
- $update_insignificant = false;
- }
-
- if (db_escape_string($orig_title) != $entry_title) {
- $post_needs_update = true;
- $update_insignificant = false;
- }
-
- // if post needs update, update it and mark all user entries
- // linking to this post as updated
- if ($post_needs_update) {
-
- if (defined('DAEMON_EXTENDED_DEBUG')) {
- _debug("update_rss_feed: post $entry_guid needs update...");
- }
-
-// print "<!-- post $orig_title needs update : $post_needs_update -->";
-
- db_query($link, "UPDATE ttrss_entries
- SET title = '$entry_title', content = '$entry_content',
- content_hash = '$content_hash',
- updated = '$entry_timestamp_fmt',
- num_comments = '$num_comments'
- WHERE id = '$ref_id'");
-
- if (!$update_insignificant) {
- if ($mark_unread_on_update) {
- db_query($link, "UPDATE ttrss_user_entries
- SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
- } else if ($update_on_checksum_change) {
- db_query($link, "UPDATE ttrss_user_entries
- SET last_read = null WHERE ref_id = '$ref_id'
- AND unread = false");
- }
- }
- }
- }
-
- db_query($link, "COMMIT");
-
- if ($debug_enabled) {
- _debug("update_rss_feed: assigning labels...");
- }
-
- assign_article_to_labels($link, $entry_ref_id, $article_filters,
- $owner_uid);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: looking for enclosures...");
- }
-
- // enclosures
-
- $enclosures = array();
-
- if ($use_simplepie) {
- $encs = $item->get_enclosures();
-
- if (is_array($encs)) {
- foreach ($encs as $e) {
- $e_item = array(
- $e->link, $e->type, $e->length);
-
- array_push($enclosures, $e_item);
- }
- }
-
- } else {
- // <enclosure>
-
- $e_ctr = $item['enclosure#'];
-
- if ($e_ctr > 0) {
- $e_item = array($item['enclosure@url'],
- $item['enclosure@type'],
- $item['enclosure@length']);
-
- array_push($enclosures, $e_item);
-
- for ($i = 0; $i <= $e_ctr; $i++ ) {
-
- if ($item["enclosure#$i@url"]) {
- $e_item = array($item["enclosure#$i@url"],
- $item["enclosure#$i@type"],
- $item["enclosure#$i@length"]);
- array_push($enclosures, $e_item);
- }
- }
- }
-
- // <media:content>
- // can there be many of those? yes -fox
-
- $m_ctr = $item['media']['content#'];
-
- if ($m_ctr > 0) {
- $e_item = array($item['media']['content@url'],
- $item['media']['content@medium'],
- $item['media']['content@length']);
-
- array_push($enclosures, $e_item);
-
- for ($i = 0; $i <= $m_ctr; $i++ ) {
-
- if ($item["media"]["content#$i@url"]) {
- $e_item = array($item["media"]["content#$i@url"],
- $item["media"]["content#$i@medium"],
- $item["media"]["content#$i@length"]);
- array_push($enclosures, $e_item);
- }
- }
-
- }
- }
-
-
- if ($debug_enabled) {
- _debug("update_rss_feed: article enclosures:");
- print_r($enclosures);
- }
-
- db_query($link, "BEGIN");
-
- foreach ($enclosures as $enc) {
- $enc_url = db_escape_string($enc[0]);
- $enc_type = db_escape_string($enc[1]);
- $enc_dur = db_escape_string($enc[2]);
-
- $result = db_query($link, "SELECT id FROM ttrss_enclosures
- WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
-
- if (db_num_rows($result) == 0) {
- db_query($link, "INSERT INTO ttrss_enclosures
- (content_url, content_type, title, duration, post_id) VALUES
- ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
- }
- }
-
- db_query($link, "COMMIT");
-
- // check for manual tags (we have to do it here since they're loaded from filters)
-
- foreach ($article_filters as $f) {
- if ($f[0] == "tag") {
-
- $manual_tags = trim_array(explode(",", $f[1]));
-
- foreach ($manual_tags as $tag) {
- if (tag_is_valid($tag)) {
- array_push($entry_tags, $tag);
- }
- }
- }
- }
-
- // Skip boring tags
-
- $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
- 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
-
- $filtered_tags = array();
- $tags_to_cache = array();
-
- if ($entry_tags && is_array($entry_tags)) {
- foreach ($entry_tags as $tag) {
- if (array_search($tag, $boring_tags) === false) {
- array_push($filtered_tags, $tag);
- }
- }
- }
-
- $filtered_tags = array_unique($filtered_tags);
-
- if ($debug_enabled) {
- _debug("update_rss_feed: filtered article tags:");
- print_r($filtered_tags);
- }
-
- // Save article tags in the database
-
- if (count($filtered_tags) > 0) {
-
- db_query($link, "BEGIN");
-
- foreach ($filtered_tags as $tag) {
-
- $tag = sanitize_tag($tag);
- $tag = db_escape_string($tag);
-
- if (!tag_is_valid($tag)) continue;
-
- $result = db_query($link, "SELECT id FROM ttrss_tags
- WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
- owner_uid = '$owner_uid' LIMIT 1");
-
- if ($result && db_num_rows($result) == 0) {
-
- db_query($link, "INSERT INTO ttrss_tags
- (owner_uid,tag_name,post_int_id)
- VALUES ('$owner_uid','$tag', '$entry_int_id')");
- }
-
- array_push($tags_to_cache, $tag);
- }
-
- /* update the cache */
-
- $tags_to_cache = array_unique($tags_to_cache);
-
- $tags_str = db_escape_string(join(",", $tags_to_cache));
-
- db_query($link, "UPDATE ttrss_user_entries
- SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
- AND owner_uid = $owner_uid");
-
- db_query($link, "COMMIT");
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: article processed");
- }
- }
-
- if (!$last_updated) {
- if ($debug_enabled) {
- _debug("update_rss_feed: new feed, catching it up...");
- }
- catchup_feed($link, $feed, false, $owner_uid);
- }
-
- if ($debug_enabled) {
- _debug("purging feed...");
- }
-
- purge_feed($link, $feed, 0, $debug_enabled);
-
- db_query($link, "UPDATE ttrss_feeds
- SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
-
-// db_query($link, "COMMIT");
-
- } else {
-
- if ($use_simplepie) {
- $error_msg = mb_substr($rss->error(), 0, 250);
- } else {
- $error_msg = mb_substr(magpie_error(), 0, 250);
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: error fetching feed: $error_msg");
- }
-
- $error_msg = db_escape_string($error_msg);
-
- db_query($link,
- "UPDATE ttrss_feeds SET last_error = '$error_msg',
- last_updated = NOW() WHERE id = '$feed'");
- }
-
- if ($use_simplepie) {
- unset($rss);
- }
-
- if ($debug_enabled) {
- _debug("update_rss_feed: done");
- }
-
- }
-
function print_select($id, $default, $values, $attributes = "") {
print "<select name=\"$id\" id=\"$id\" $attributes>";
foreach ($values as $v) {
@@ -2904,6 +1793,8 @@
function subscribe_to_feed($link, $url, $cat_id = 0,
$auth_login = '', $auth_pass = '') {
+ require_once "include/rssfuncs.php";
+
$url = fix_url($url);
if (!$url || !validate_feed_url($url)) return 2;
@@ -4908,9 +3799,9 @@
catchupArticlesById($link, $ids, $cmode);
} */
- if ($method == "ForceUpdate" && $feed && is_numeric($feed) > 0) {
- update_rss_feed($link, $feed, true);
- }
+ //if ($method == "ForceUpdate" && $feed && is_numeric($feed) > 0) {
+ // update_rss_feed($link, $feed, true);
+ //}
if ($method == "MarkAllRead") {
catchup_feed($link, $feed, $cat_view);
@@ -5657,111 +4548,6 @@
}
} // function encrypt_password
- /**
- * Update a feed batch.
- * Used by daemons to update n feeds by run.
- * Only update feed needing a update, and not being processed
- * by another process.
- *
- * @param mixed $link Database link
- * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
- * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
- * @param boolean $debug Set to false to disable debug output. Default to true.
- * @return void
- */
- function update_daemon_common($link, $limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
- // Process all other feeds using last_updated and interval parameters
-
- // Test if the user has loggued in recently. If not, it does not update its feeds.
- if (DAEMON_UPDATE_LOGIN_LIMIT > 0) {
- if (DB_TYPE == "pgsql") {
- $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
- } else {
- $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
- }
- } else {
- $login_thresh_qpart = "";
- }
-
- // Test if the feed need a update (update interval exceded).
- if (DB_TYPE == "pgsql") {
- $update_limit_qpart = "AND ((
- ttrss_feeds.update_interval = 0
- AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
- ) OR (
- ttrss_feeds.update_interval > 0
- AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
- ) OR ttrss_feeds.last_updated IS NULL)";
- } else {
- $update_limit_qpart = "AND ((
- ttrss_feeds.update_interval = 0
- AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
- ) OR (
- ttrss_feeds.update_interval > 0
- AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
- ) OR ttrss_feeds.last_updated IS NULL)";
- }
-
- // Test if feed is currently being updated by another process.
- if (DB_TYPE == "pgsql") {
- $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')";
- } else {
- $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))";
- }
-
- // Test if there is a limit to number of updated feeds
- $query_limit = "";
- if($limit) $query_limit = sprintf("LIMIT %d", $limit);
-
- $random_qpart = sql_random_function();
-
- // We search for feed needing update.
- $result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id, ttrss_feeds.owner_uid,
- ".SUBSTRING_FOR_DATE."(ttrss_feeds.last_updated,1,19) AS last_updated,
- ttrss_feeds.update_interval
- FROM
- ttrss_feeds, ttrss_users, ttrss_user_prefs
- WHERE
- ttrss_feeds.owner_uid = ttrss_users.id
- AND ttrss_users.id = ttrss_user_prefs.owner_uid
- AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
- $login_thresh_qpart $update_limit_qpart
- $updstart_thresh_qpart
- ORDER BY $random_qpart $query_limit");
-
- $user_prefs_cache = array();
-
- if($debug) _debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result)));
-
- // Here is a little cache magic in order to minimize risk of double feed updates.
- $feeds_to_update = array();
- while ($line = db_fetch_assoc($result)) {
- $feeds_to_update[$line['id']] = $line;
- }
-
- // We update the feed last update started date before anything else.
- // There is no lag due to feed contents downloads
- // It prevent an other process to update the same feed.
- $feed_ids = array_keys($feeds_to_update);
- if($feed_ids) {
- db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
- WHERE id IN (%s)", implode(',', $feed_ids)));
- }
-
- // For each feed, we call the feed update function.
- while ($line = array_pop($feeds_to_update)) {
-
- if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
-
- update_rss_feed($link, $line["id"], true);
-
- sleep(1); // prevent flood (FIXME make this an option?)
- }
-
- // Send feed digests by email if needed.
- if (DAEMON_SENDS_DIGESTS) send_headlines_digests($link);
-
- } // function update_daemon_common
function sanitize_article_content($text) {
# we don't support CDATA sections in articles, they break our own escaping
@@ -7212,54 +5998,6 @@
}
- function fetch_twitter_rss($link, $url, $owner_uid) {
-
- require_once 'lib/tmhoauth/tmhOAuth.php';
-
- $result = db_query($link, "SELECT twitter_oauth FROM ttrss_users
- WHERE id = $owner_uid");
-
- $access_token = json_decode(db_fetch_result($result, 0, 'twitter_oauth'), true);
- $url_escaped = db_escape_string($url);
-
- if ($access_token) {
-
- $tmhOAuth = new tmhOAuth(array(
- 'consumer_key' => CONSUMER_KEY,
- 'consumer_secret' => CONSUMER_SECRET,
- 'user_token' => $access_token['oauth_token'],
- 'user_secret' => $access_token['oauth_token_secret'],
- ));
-
- $code = $tmhOAuth->request('GET', $url);
-
- if ($code == 200) {
-
- $content = $tmhOAuth->response['response'];
-
- define('MAGPIE_CACHE_ON', false);
-
- $rss = new MagpieRSS($content, MAGPIE_OUTPUT_ENCODING,
- MAGPIE_INPUT_ENCODING, MAGPIE_DETECT_ENCODING );
-
- return $rss;
-
- } else {
-
- db_query($link, "UPDATE ttrss_feeds
- SET last_error = 'OAuth authorization failed ($code).'
- WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid");
- }
-
- } else {
-
- db_query($link, "UPDATE ttrss_feeds
- SET last_error = 'OAuth information not found.'
- WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid");
-
- return false;
- }
- }
function print_user_stylesheet($link) {
$value = get_pref($link, 'USER_STYLESHEET');
diff --git a/include/rssfuncs.php b/include/rssfuncs.php
new file mode 100644
index 000000000..8846d985f
--- /dev/null
+++ b/include/rssfuncs.php
@@ -0,0 +1,1271 @@
+<?php
+ /**
+ * Update a feed batch.
+ * Used by daemons to update n feeds by run.
+ * Only update feed needing a update, and not being processed
+ * by another process.
+ *
+ * @param mixed $link Database link
+ * @param integer $limit Maximum number of feeds in update batch. Default to DAEMON_FEED_LIMIT.
+ * @param boolean $from_http Set to true if you call this function from http to disable cli specific code.
+ * @param boolean $debug Set to false to disable debug output. Default to true.
+ * @return void
+ */
+ function update_daemon_common($link, $limit = DAEMON_FEED_LIMIT, $from_http = false, $debug = true) {
+ // Process all other feeds using last_updated and interval parameters
+
+ // Test if the user has loggued in recently. If not, it does not update its feeds.
+ if (DAEMON_UPDATE_LOGIN_LIMIT > 0) {
+ if (DB_TYPE == "pgsql") {
+ $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
+ } else {
+ $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
+ }
+ } else {
+ $login_thresh_qpart = "";
+ }
+
+ // Test if the feed need a update (update interval exceded).
+ if (DB_TYPE == "pgsql") {
+ $update_limit_qpart = "AND ((
+ ttrss_feeds.update_interval = 0
+ AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_user_prefs.value || ' minutes') AS INTERVAL)
+ ) OR (
+ ttrss_feeds.update_interval > 0
+ AND ttrss_feeds.last_updated < NOW() - CAST((ttrss_feeds.update_interval || ' minutes') AS INTERVAL)
+ ) OR ttrss_feeds.last_updated IS NULL)";
+ } else {
+ $update_limit_qpart = "AND ((
+ ttrss_feeds.update_interval = 0
+ AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL CONVERT(ttrss_user_prefs.value, SIGNED INTEGER) MINUTE)
+ ) OR (
+ ttrss_feeds.update_interval > 0
+ AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ttrss_feeds.update_interval MINUTE)
+ ) OR ttrss_feeds.last_updated IS NULL)";
+ }
+
+ // Test if feed is currently being updated by another process.
+ if (DB_TYPE == "pgsql") {
+ $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '5 minutes')";
+ } else {
+ $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 5 MINUTE))";
+ }
+
+ // Test if there is a limit to number of updated feeds
+ $query_limit = "";
+ if($limit) $query_limit = sprintf("LIMIT %d", $limit);
+
+ $random_qpart = sql_random_function();
+
+ // We search for feed needing update.
+ $result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id, ttrss_feeds.owner_uid,
+ ".SUBSTRING_FOR_DATE."(ttrss_feeds.last_updated,1,19) AS last_updated,
+ ttrss_feeds.update_interval
+ FROM
+ ttrss_feeds, ttrss_users, ttrss_user_prefs
+ WHERE
+ ttrss_feeds.owner_uid = ttrss_users.id
+ AND ttrss_users.id = ttrss_user_prefs.owner_uid
+ AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
+ $login_thresh_qpart $update_limit_qpart
+ $updstart_thresh_qpart
+ ORDER BY $random_qpart $query_limit");
+
+ $user_prefs_cache = array();
+
+ if($debug) _debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result)));
+
+ // Here is a little cache magic in order to minimize risk of double feed updates.
+ $feeds_to_update = array();
+ while ($line = db_fetch_assoc($result)) {
+ $feeds_to_update[$line['id']] = $line;
+ }
+
+ // We update the feed last update started date before anything else.
+ // There is no lag due to feed contents downloads
+ // It prevent an other process to update the same feed.
+ $feed_ids = array_keys($feeds_to_update);
+ if($feed_ids) {
+ db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
+ WHERE id IN (%s)", implode(',', $feed_ids)));
+ }
+
+ // For each feed, we call the feed update function.
+ while ($line = array_pop($feeds_to_update)) {
+
+ if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
+
+ update_rss_feed($link, $line["id"], true);
+
+ sleep(1); // prevent flood (FIXME make this an option?)
+ }
+
+ // Send feed digests by email if needed.
+ if (DAEMON_SENDS_DIGESTS) send_headlines_digests($link);
+
+ } // function update_daemon_common
+
+ function fetch_twitter_rss($link, $url, $owner_uid) {
+
+ require_once 'lib/tmhoauth/tmhOAuth.php';
+
+ $result = db_query($link, "SELECT twitter_oauth FROM ttrss_users
+ WHERE id = $owner_uid");
+
+ $access_token = json_decode(db_fetch_result($result, 0, 'twitter_oauth'), true);
+ $url_escaped = db_escape_string($url);
+
+ if ($access_token) {
+
+ $tmhOAuth = new tmhOAuth(array(
+ 'consumer_key' => CONSUMER_KEY,
+ 'consumer_secret' => CONSUMER_SECRET,
+ 'user_token' => $access_token['oauth_token'],
+ 'user_secret' => $access_token['oauth_token_secret'],
+ ));
+
+ $code = $tmhOAuth->request('GET', $url);
+
+ if ($code == 200) {
+
+ $content = $tmhOAuth->response['response'];
+
+ define('MAGPIE_CACHE_ON', false);
+
+ $rss = new MagpieRSS($content, MAGPIE_OUTPUT_ENCODING,
+ MAGPIE_INPUT_ENCODING, MAGPIE_DETECT_ENCODING );
+
+ return $rss;
+
+ } else {
+
+ db_query($link, "UPDATE ttrss_feeds
+ SET last_error = 'OAuth authorization failed ($code).'
+ WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid");
+ }
+
+ } else {
+
+ db_query($link, "UPDATE ttrss_feeds
+ SET last_error = 'OAuth information not found.'
+ WHERE feed_url = '$url_escaped' AND owner_uid = $owner_uid");
+
+ return false;
+ }
+ }
+
+ function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false) {
+
+ global $memcache;
+
+ /* Update all feeds with the same URL to utilize memcache */
+
+ if ($memcache) {
+ $result = db_query($link, "SELECT f1.id
+ FROM ttrss_feeds AS f1, ttrss_feeds AS f2
+ WHERE f2.feed_url = f1.feed_url AND f2.id = '$feed'");
+
+ while ($line = db_fetch_assoc($result)) {
+ update_rss_feed_real($link, $line["id"], $ignore_daemon, $no_cache);
+ }
+ } else {
+ update_rss_feed_real($link, $feed, $ignore_daemon, $no_cache);
+ }
+ }
+
+ function update_rss_feed_real($link, $feed, $ignore_daemon = false, $no_cache = false,
+ $override_url = false) {
+
+ require_once "lib/simplepie/simplepie.inc";
+ require_once "lib/magpierss/rss_fetch.inc";
+ require_once 'lib/magpierss/rss_utils.inc';
+
+ global $memcache;
+
+ $debug_enabled = defined('DAEMON_EXTENDED_DEBUG') || $_REQUEST['xdebug'];
+
+ if (!$_REQUEST["daemon"] && !$ignore_daemon) {
+ return false;
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: start");
+ }
+
+ if (!$ignore_daemon) {
+
+ if (DB_TYPE == "pgsql") {
+ $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
+ } else {
+ $updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))";
+ }
+
+ $result = db_query($link, "SELECT id,update_interval,auth_login,
+ auth_pass,cache_images,update_method
+ FROM ttrss_feeds WHERE id = '$feed' AND $updstart_thresh_qpart");
+
+ } else {
+
+ $result = db_query($link, "SELECT id,update_interval,auth_login,
+ feed_url,auth_pass,cache_images,update_method,last_updated,
+ mark_unread_on_update, owner_uid, update_on_checksum_change,
+ pubsub_state
+ FROM ttrss_feeds WHERE id = '$feed'");
+
+ }
+
+ if (db_num_rows($result) == 0) {
+ if ($debug_enabled) {
+ _debug("update_rss_feed: feed $feed NOT FOUND/SKIPPED");
+ }
+ return false;
+ }
+
+ $update_method = db_fetch_result($result, 0, "update_method");
+ $last_updated = db_fetch_result($result, 0, "last_updated");
+ $owner_uid = db_fetch_result($result, 0, "owner_uid");
+ $mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result,
+ 0, "mark_unread_on_update"));
+ $update_on_checksum_change = sql_bool_to_bool(db_fetch_result($result,
+ 0, "update_on_checksum_change"));
+ $pubsub_state = db_fetch_result($result, 0, "pubsub_state");
+
+ db_query($link, "UPDATE ttrss_feeds SET last_update_started = NOW()
+ WHERE id = '$feed'");
+
+ $auth_login = db_fetch_result($result, 0, "auth_login");
+ $auth_pass = db_fetch_result($result, 0, "auth_pass");
+
+ if ($update_method == 0)
+ $update_method = DEFAULT_UPDATE_METHOD + 1;
+
+ // 1 - Magpie
+ // 2 - SimplePie
+ // 3 - Twitter OAuth
+
+ if ($update_method == 2)
+ $use_simplepie = true;
+ else
+ $use_simplepie = false;
+
+ if ($debug_enabled) {
+ _debug("update method: $update_method (feed setting: $update_method) (use simplepie: $use_simplepie)\n");
+ }
+
+ if ($update_method == 1) {
+ $auth_login = urlencode($auth_login);
+ $auth_pass = urlencode($auth_pass);
+ }
+
+ $update_interval = db_fetch_result($result, 0, "update_interval");
+ $cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
+ $fetch_url = db_fetch_result($result, 0, "feed_url");
+
+ if ($update_interval < 0) { return false; }
+
+ $feed = db_escape_string($feed);
+
+ if ($auth_login && $auth_pass ){
+ $url_parts = array();
+ preg_match("/(^[^:]*):\/\/(.*)/", $fetch_url, $url_parts);
+
+ if ($url_parts[1] && $url_parts[2]) {
+ $fetch_url = $url_parts[1] . "://$auth_login:$auth_pass@" . $url_parts[2];
+ }
+
+ }
+
+ if ($override_url)
+ $fetch_url = $override_url;
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: fetching [$fetch_url]...");
+ }
+
+ $obj_id = md5("FDATA:$use_simplepie:$fetch_url");
+
+ if ($memcache && $obj = $memcache->get($obj_id)) {
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: data found in memcache.");
+ }
+
+ $rss = $obj;
+
+ } else {
+
+ if ($update_method == 3) {
+ $rss = fetch_twitter_rss($link, $fetch_url, $owner_uid);
+ } else if ($update_method == 1) {
+
+ define('MAGPIE_CACHE_AGE', get_feed_update_interval($link, $feed) * 60);
+ define('MAGPIE_CACHE_ON', !$no_cache);
+ define('MAGPIE_FETCH_TIME_OUT', 60);
+ define('MAGPIE_CACHE_DIR', CACHE_DIR . "/magpie");
+
+ $rss = @fetch_rss($fetch_url);
+ } else {
+ $simplepie_cache_dir = CACHE_DIR . "/simplepie";
+
+ if (!is_dir($simplepie_cache_dir)) {
+ mkdir($simplepie_cache_dir);
+ }
+
+ $rss = new SimplePie();
+ $rss->set_useragent(SELF_USER_AGENT);
+ # $rss->set_timeout(10);
+ $rss->set_feed_url($fetch_url);
+ $rss->set_output_encoding('UTF-8');
+ $rss->force_feed(true);
+
+ if (SIMPLEPIE_CACHE_IMAGES && $cache_images) {
+
+ if ($debug_enabled) {
+ _debug("enabling image cache");
+ }
+
+ $rss->set_image_handler("image.php", 'i');
+ }
+
+ if ($debug_enabled) {
+ _debug("feed update interval (sec): " .
+ get_feed_update_interval($link, $feed)*60);
+ }
+
+ $rss->enable_cache(!$no_cache);
+
+ if (!$no_cache) {
+ $rss->set_cache_location($simplepie_cache_dir);
+ $rss->set_cache_duration(get_feed_update_interval($link, $feed) * 60);
+ }
+
+ $rss->init();
+ }
+
+ if ($memcache && $rss) $memcache->add($obj_id, $rss, 0, 300);
+ }
+
+// print_r($rss);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: fetch done, parsing...");
+ }
+
+ $feed = db_escape_string($feed);
+
+ if ($update_method == 2) {
+ $fetch_ok = !$rss->error();
+ } else {
+ $fetch_ok = !!$rss;
+ }
+
+ if ($fetch_ok) {
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: processing feed data...");
+ }
+
+// db_query($link, "BEGIN");
+
+ $result = db_query($link, "SELECT title,icon_url,site_url,owner_uid
+ FROM ttrss_feeds WHERE id = '$feed'");
+
+ $registered_title = db_fetch_result($result, 0, "title");
+ $orig_icon_url = db_fetch_result($result, 0, "icon_url");
+ $orig_site_url = db_fetch_result($result, 0, "site_url");
+
+ $owner_uid = db_fetch_result($result, 0, "owner_uid");
+
+ if ($use_simplepie) {
+ $site_url = $rss->get_link();
+ } else {
+ $site_url = $rss->channel["link"];
+ }
+
+ $site_url = rewrite_relative_url($fetch_url, $site_url);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: checking favicon...");
+ }
+
+ check_feed_favicon($site_url, $feed, $link);
+
+ if (!$registered_title || $registered_title == "[Unknown]") {
+
+ if ($use_simplepie) {
+ $feed_title = db_escape_string($rss->get_title());
+ } else {
+ $feed_title = db_escape_string($rss->channel["title"]);
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: registering title: $feed_title");
+ }
+
+ db_query($link, "UPDATE ttrss_feeds SET
+ title = '$feed_title' WHERE id = '$feed'");
+ }
+
+ // weird, weird Magpie
+ if (!$use_simplepie) {
+ if (!$site_url) $site_url = db_escape_string($rss->channel["link_"]);
+ }
+
+ if ($site_url && $orig_site_url != db_escape_string($site_url)) {
+ db_query($link, "UPDATE ttrss_feeds SET
+ site_url = '$site_url' WHERE id = '$feed'");
+ }
+
+// print "I: " . $rss->channel["image"]["url"];
+
+ if (!$use_simplepie) {
+ $icon_url = db_escape_string($rss->image["url"]);
+ } else {
+ $icon_url = db_escape_string($rss->get_image_url());
+ }
+
+ $icon_url = substr($icon_url, 0, 250);
+
+ if ($icon_url && $orig_icon_url != $icon_url) {
+ db_query($link, "UPDATE ttrss_feeds SET icon_url = '$icon_url' WHERE id = '$feed'");
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: loading filters...");
+ }
+
+ $filters = load_filters($link, $feed, $owner_uid);
+
+// if ($debug_enabled) {
+// print_r($filters);
+// }
+
+ if ($use_simplepie) {
+ $iterator = $rss->get_items();
+ } else {
+ $iterator = $rss->items;
+ if (!$iterator || !is_array($iterator)) $iterator = $rss->entries;
+ if (!$iterator || !is_array($iterator)) $iterator = $rss;
+ }
+
+ if (!is_array($iterator)) {
+ /* db_query($link, "UPDATE ttrss_feeds
+ SET last_error = 'Parse error: can\'t find any articles.'
+ WHERE id = '$feed'"); */
+
+ // clear any errors and mark feed as updated if fetched okay
+ // even if it's blank
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: entry iterator is not an array, no articles?");
+ }
+
+ db_query($link, "UPDATE ttrss_feeds
+ SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
+
+ return; // no articles
+ }
+
+ if ($pubsub_state != 2 && PUBSUBHUBBUB_ENABLED) {
+
+ if ($debug_enabled) _debug("update_rss_feed: checking for PUSH hub...");
+
+ $feed_hub_url = false;
+ if ($use_simplepie) {
+ $links = $rss->get_links('hub');
+
+ if ($links && is_array($links)) {
+ foreach ($links as $l) {
+ $feed_hub_url = $l;
+ break;
+ }
+ }
+
+ } else {
+ $atom = $rss->channel['atom'];
+
+ if ($atom) {
+ if ($atom['link@rel'] == 'hub') {
+ $feed_hub_url = $atom['link@href'];
+ }
+
+ if (!$feed_hub_url && $atom['link#'] > 1) {
+ for ($i = 2; $i <= $atom['link#']; $i++) {
+ if ($atom["link#$i@rel"] == 'hub') {
+ $feed_hub_url = $atom["link#$i@href"];
+ break;
+ }
+ }
+ }
+ } else {
+ $feed_hub_url = $rss->channel['link_hub'];
+ }
+ }
+
+ if ($debug_enabled) _debug("update_rss_feed: feed hub url: $feed_hub_url");
+
+ if ($feed_hub_url && function_exists('curl_init') &&
+ !ini_get("open_basedir")) {
+
+ require_once 'lib/pubsubhubbub/subscriber.php';
+
+ $callback_url = get_self_url_prefix() .
+ "/public.php?op=pubsub&id=$feed";
+
+ $s = new Subscriber($feed_hub_url, $callback_url);
+
+ $rc = $s->subscribe($fetch_url);
+
+ if ($debug_enabled)
+ _debug("update_rss_feed: feed hub url found, subscribe request sent.");
+
+ db_query($link, "UPDATE ttrss_feeds SET pubsub_state = 1
+ WHERE id = '$feed'");
+ }
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: processing articles...");
+ }
+
+ foreach ($iterator as $item) {
+
+ if ($_REQUEST['xdebug'] == 2) {
+ print_r($item);
+ }
+
+ if ($use_simplepie) {
+ $entry_guid = $item->get_id();
+ if (!$entry_guid) $entry_guid = $item->get_link();
+ if (!$entry_guid) $entry_guid = make_guid_from_title($item->get_title());
+
+ } else {
+
+ $entry_guid = $item["id"];
+
+ if (!$entry_guid) $entry_guid = $item["guid"];
+ if (!$entry_guid) $entry_guid = $item["about"];
+ if (!$entry_guid) $entry_guid = $item["link"];
+ if (!$entry_guid) $entry_guid = make_guid_from_title($item["title"]);
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: guid $entry_guid");
+ }
+
+ if (!$entry_guid) continue;
+
+ $entry_timestamp = "";
+
+ if ($use_simplepie) {
+ $entry_timestamp = strtotime($item->get_date());
+ } else {
+ $rss_2_date = $item['pubdate'];
+ $rss_1_date = $item['dc']['date'];
+ $atom_date = $item['issued'];
+ if (!$atom_date) $atom_date = $item['updated'];
+
+ if ($atom_date != "") $entry_timestamp = parse_w3cdtf($atom_date);
+ if ($rss_1_date != "") $entry_timestamp = parse_w3cdtf($rss_1_date);
+ if ($rss_2_date != "") $entry_timestamp = strtotime($rss_2_date);
+
+ }
+
+ if ($entry_timestamp == "" || $entry_timestamp == -1 || !$entry_timestamp) {
+ $entry_timestamp = time();
+ $no_orig_date = 'true';
+ } else {
+ $no_orig_date = 'false';
+ }
+
+ $entry_timestamp_fmt = strftime("%Y/%m/%d %H:%M:%S", $entry_timestamp);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: date $entry_timestamp [$entry_timestamp_fmt]");
+ }
+
+ if ($use_simplepie) {
+ $entry_title = $item->get_title();
+ } else {
+ $entry_title = trim(strip_tags($item["title"]));
+ }
+
+ if ($use_simplepie) {
+ $entry_link = $item->get_link();
+ } else {
+ // strange Magpie workaround
+ $entry_link = $item["link_"];
+ if (!$entry_link) $entry_link = $item["link"];
+ }
+
+ $entry_link = rewrite_relative_url($site_url, $entry_link);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: title $entry_title");
+ _debug("update_rss_feed: link $entry_link");
+ }
+
+ if (!$entry_title) $entry_title = date("Y-m-d H:i:s", $entry_timestamp);;
+
+ $entry_link = strip_tags($entry_link);
+
+ if ($use_simplepie) {
+ $entry_content = $item->get_content();
+ if (!$entry_content) $entry_content = $item->get_description();
+ } else {
+ $entry_content = $item["content:escaped"];
+
+ if (!$entry_content) $entry_content = $item["content:encoded"];
+ if (!$entry_content) $entry_content = $item["content"]["encoded"];
+ if (!$entry_content) $entry_content = $item["content"];
+
+ if (is_array($entry_content)) $entry_content = $entry_content[0];
+
+ // Magpie bugs are getting ridiculous
+ if (trim($entry_content) == "Array") $entry_content = false;
+
+ if (!$entry_content) $entry_content = $item["atom_content"];
+ if (!$entry_content) $entry_content = $item["summary"];
+
+ if (!$entry_content ||
+ strlen($entry_content) < strlen($item["description"])) {
+ $entry_content = $item["description"];
+ };
+
+ // WTF
+ if (is_array($entry_content)) {
+ $entry_content = $entry_content["encoded"];
+ if (!$entry_content) $entry_content = $entry_content["escaped"];
+ }
+ }
+
+ if ($_REQUEST["xdebug"] == 2) {
+ print "update_rss_feed: content: ";
+ print_r(htmlspecialchars($entry_content));
+ }
+
+ $entry_content_unescaped = $entry_content;
+
+ if ($use_simplepie) {
+ $entry_comments = strip_tags($item->data["comments"]);
+ if ($item->get_author()) {
+ $entry_author_item = $item->get_author();
+ $entry_author = $entry_author_item->get_name();
+ if (!$entry_author) $entry_author = $entry_author_item->get_email();
+
+ $entry_author = db_escape_string($entry_author);
+ }
+ } else {
+ $entry_comments = strip_tags($item["comments"]);
+
+ $entry_author = db_escape_string(strip_tags($item['dc']['creator']));
+
+ if ($item['author']) {
+
+ if (is_array($item['author'])) {
+
+ if (!$entry_author) {
+ $entry_author = db_escape_string(strip_tags($item['author']['name']));
+ }
+
+ if (!$entry_author) {
+ $entry_author = db_escape_string(strip_tags($item['author']['email']));
+ }
+ }
+
+ if (!$entry_author) {
+ $entry_author = db_escape_string(strip_tags($item['author']));
+ }
+ }
+ }
+
+ if (preg_match('/^[\t\n\r ]*$/', $entry_author)) $entry_author = '';
+
+ $entry_guid = db_escape_string(strip_tags($entry_guid));
+ $entry_guid = mb_substr($entry_guid, 0, 250);
+
+ $result = db_query($link, "SELECT id FROM ttrss_entries
+ WHERE guid = '$entry_guid'");
+
+ $entry_content = db_escape_string($entry_content, false);
+
+ $content_hash = "SHA1:" . sha1(strip_tags($entry_content));
+
+ $entry_title = db_escape_string($entry_title);
+ $entry_link = db_escape_string($entry_link);
+ $entry_comments = mb_substr(db_escape_string($entry_comments), 0, 250);
+ $entry_author = mb_substr($entry_author, 0, 250);
+
+ if ($use_simplepie) {
+ $num_comments = 0; #FIXME#
+ } else {
+ $num_comments = db_escape_string($item["slash"]["comments"]);
+ }
+
+ if (!$num_comments) $num_comments = 0;
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: looking for tags [1]...");
+ }
+
+ // parse <category> entries into tags
+
+ $additional_tags = array();
+
+ if ($use_simplepie) {
+
+ $additional_tags_src = $item->get_categories();
+
+ if (is_array($additional_tags_src)) {
+ foreach ($additional_tags_src as $tobj) {
+ array_push($additional_tags, $tobj->get_term());
+ }
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: category tags:");
+ print_r($additional_tags);
+ }
+
+ } else {
+
+ $t_ctr = $item['category#'];
+
+ if ($t_ctr == 0) {
+ $additional_tags = array();
+ } else if ($t_ctr > 0) {
+ $additional_tags = array($item['category']);
+
+ if ($item['category@term']) {
+ array_push($additional_tags, $item['category@term']);
+ }
+
+ for ($i = 0; $i <= $t_ctr; $i++ ) {
+ if ($item["category#$i"]) {
+ array_push($additional_tags, $item["category#$i"]);
+ }
+
+ if ($item["category#$i@term"]) {
+ array_push($additional_tags, $item["category#$i@term"]);
+ }
+ }
+ }
+
+ // parse <dc:subject> elements
+
+ $t_ctr = $item['dc']['subject#'];
+
+ if ($t_ctr > 0) {
+ array_push($additional_tags, $item['dc']['subject']);
+
+ for ($i = 0; $i <= $t_ctr; $i++ ) {
+ if ($item['dc']["subject#$i"]) {
+ array_push($additional_tags, $item['dc']["subject#$i"]);
+ }
+ }
+ }
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: looking for tags [2]...");
+ }
+
+ /* taaaags */
+ // <a href="..." rel="tag">Xorg</a>, //
+
+ $entry_tags = null;
+
+ preg_match_all("/<a.*?rel=['\"]tag['\"].*?\>([^<]+)<\/a>/i",
+ $entry_content_unescaped, $entry_tags);
+
+ $entry_tags = $entry_tags[1];
+
+ $entry_tags = array_merge($entry_tags, $additional_tags);
+ $entry_tags = array_unique($entry_tags);
+
+ for ($i = 0; $i < count($entry_tags); $i++)
+ $entry_tags[$i] = mb_strtolower($entry_tags[$i], 'utf-8');
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: unfiltered tags found:");
+ print_r($entry_tags);
+ }
+
+ # sanitize content
+
+ $entry_content = sanitize_article_content($entry_content);
+ $entry_title = sanitize_article_content($entry_title);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: done collecting data [TITLE:$entry_title]");
+ }
+
+ db_query($link, "BEGIN");
+
+ if (db_num_rows($result) == 0) {
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: base guid not found");
+ }
+
+ // base post entry does not exist, create it
+
+ $result = db_query($link,
+ "INSERT INTO ttrss_entries
+ (title,
+ guid,
+ link,
+ updated,
+ content,
+ content_hash,
+ no_orig_date,
+ date_updated,
+ date_entered,
+ comments,
+ num_comments,
+ author)
+ VALUES
+ ('$entry_title',
+ '$entry_guid',
+ '$entry_link',
+ '$entry_timestamp_fmt',
+ '$entry_content',
+ '$content_hash',
+ $no_orig_date,
+ NOW(),
+ NOW(),
+ '$entry_comments',
+ '$num_comments',
+ '$entry_author')");
+ } else {
+ // we keep encountering the entry in feeds, so we need to
+ // update date_updated column so that we don't get horrible
+ // dupes when the entry gets purged and reinserted again e.g.
+ // in the case of SLOW SLOW OMG SLOW updating feeds
+
+ $base_entry_id = db_fetch_result($result, 0, "id");
+
+ db_query($link, "UPDATE ttrss_entries SET date_updated = NOW()
+ WHERE id = '$base_entry_id'");
+ }
+
+ // now it should exist, if not - bad luck then
+
+ $result = db_query($link, "SELECT
+ id,content_hash,no_orig_date,title,
+ ".SUBSTRING_FOR_DATE."(date_updated,1,19) as date_updated,
+ ".SUBSTRING_FOR_DATE."(updated,1,19) as updated,
+ num_comments
+ FROM
+ ttrss_entries
+ WHERE guid = '$entry_guid'");
+
+ $entry_ref_id = 0;
+ $entry_int_id = 0;
+
+ if (db_num_rows($result) == 1) {
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: base guid found, checking for user record");
+ }
+
+ // this will be used below in update handler
+ $orig_content_hash = db_fetch_result($result, 0, "content_hash");
+ $orig_title = db_fetch_result($result, 0, "title");
+ $orig_num_comments = db_fetch_result($result, 0, "num_comments");
+ $orig_date_updated = strtotime(db_fetch_result($result,
+ 0, "date_updated"));
+
+ $ref_id = db_fetch_result($result, 0, "id");
+ $entry_ref_id = $ref_id;
+
+ // check for user post link to main table
+
+ // do we allow duplicate posts with same GUID in different feeds?
+ if (get_pref($link, "ALLOW_DUPLICATE_POSTS", $owner_uid, false)) {
+ $dupcheck_qpart = "AND (feed_id = '$feed' OR feed_id IS NULL)";
+ } else {
+ $dupcheck_qpart = "";
+ }
+
+ /* Collect article tags here so we could filter by them: */
+
+ $article_filters = get_article_filters($filters, $entry_title,
+ $entry_content, $entry_link, $entry_timestamp, $entry_author,
+ $entry_tags);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: article filters: ");
+ if (count($article_filters) != 0) {
+ print_r($article_filters);
+ }
+ }
+
+ if (find_article_filter($article_filters, "filter")) {
+ db_query($link, "COMMIT"); // close transaction in progress
+ continue;
+ }
+
+ $score = calculate_article_score($article_filters);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: initial score: $score");
+ }
+
+ $query = "SELECT ref_id, int_id FROM ttrss_user_entries WHERE
+ ref_id = '$ref_id' AND owner_uid = '$owner_uid'
+ $dupcheck_qpart";
+
+// if ($_REQUEST["xdebug"]) print "$query\n";
+
+ $result = db_query($link, $query);
+
+ // okay it doesn't exist - create user entry
+ if (db_num_rows($result) == 0) {
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: user record not found, creating...");
+ }
+
+ if ($score >= -500 && !find_article_filter($article_filters, 'catchup')) {
+ $unread = 'true';
+ $last_read_qpart = 'NULL';
+ } else {
+ $unread = 'false';
+ $last_read_qpart = 'NOW()';
+ }
+
+ if (find_article_filter($article_filters, 'mark') || $score > 1000) {
+ $marked = 'true';
+ } else {
+ $marked = 'false';
+ }
+
+ if (find_article_filter($article_filters, 'publish')) {
+ $published = 'true';
+ } else {
+ $published = 'false';
+ }
+
+ $result = db_query($link,
+ "INSERT INTO ttrss_user_entries
+ (ref_id, owner_uid, feed_id, unread, last_read, marked,
+ published, score, tag_cache, label_cache, uuid)
+ VALUES ('$ref_id', '$owner_uid', '$feed', $unread,
+ $last_read_qpart, $marked, $published, '$score', '', '', '')");
+
+ if (PUBSUBHUBBUB_HUB && $published == 'true') {
+ $rss_link = get_self_url_prefix() .
+ "/public.php?op=rss&id=-2&key=" .
+ get_feed_access_key($link, -2, false, $owner_uid);
+
+ $p = new Publisher(PUBSUBHUBBUB_HUB);
+
+ $pubsub_result = $p->publish_update($rss_link);
+ }
+
+ $result = db_query($link,
+ "SELECT int_id FROM ttrss_user_entries WHERE
+ ref_id = '$ref_id' AND owner_uid = '$owner_uid' AND
+ feed_id = '$feed' LIMIT 1");
+
+ if (db_num_rows($result) == 1) {
+ $entry_int_id = db_fetch_result($result, 0, "int_id");
+ }
+ } else {
+ if ($debug_enabled) {
+ _debug("update_rss_feed: user record FOUND");
+ }
+
+ $entry_ref_id = db_fetch_result($result, 0, "ref_id");
+ $entry_int_id = db_fetch_result($result, 0, "int_id");
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: RID: $entry_ref_id, IID: $entry_int_id");
+ }
+
+ $post_needs_update = false;
+ $update_insignificant = false;
+
+ if ($orig_num_comments != $num_comments) {
+ $post_needs_update = true;
+ $update_insignificant = true;
+ }
+
+ if ($content_hash != $orig_content_hash) {
+ $post_needs_update = true;
+ $update_insignificant = false;
+ }
+
+ if (db_escape_string($orig_title) != $entry_title) {
+ $post_needs_update = true;
+ $update_insignificant = false;
+ }
+
+ // if post needs update, update it and mark all user entries
+ // linking to this post as updated
+ if ($post_needs_update) {
+
+ if (defined('DAEMON_EXTENDED_DEBUG')) {
+ _debug("update_rss_feed: post $entry_guid needs update...");
+ }
+
+// print "<!-- post $orig_title needs update : $post_needs_update -->";
+
+ db_query($link, "UPDATE ttrss_entries
+ SET title = '$entry_title', content = '$entry_content',
+ content_hash = '$content_hash',
+ updated = '$entry_timestamp_fmt',
+ num_comments = '$num_comments'
+ WHERE id = '$ref_id'");
+
+ if (!$update_insignificant) {
+ if ($mark_unread_on_update) {
+ db_query($link, "UPDATE ttrss_user_entries
+ SET last_read = null, unread = true WHERE ref_id = '$ref_id'");
+ } else if ($update_on_checksum_change) {
+ db_query($link, "UPDATE ttrss_user_entries
+ SET last_read = null WHERE ref_id = '$ref_id'
+ AND unread = false");
+ }
+ }
+ }
+ }
+
+ db_query($link, "COMMIT");
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: assigning labels...");
+ }
+
+ assign_article_to_labels($link, $entry_ref_id, $article_filters,
+ $owner_uid);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: looking for enclosures...");
+ }
+
+ // enclosures
+
+ $enclosures = array();
+
+ if ($use_simplepie) {
+ $encs = $item->get_enclosures();
+
+ if (is_array($encs)) {
+ foreach ($encs as $e) {
+ $e_item = array(
+ $e->link, $e->type, $e->length);
+
+ array_push($enclosures, $e_item);
+ }
+ }
+
+ } else {
+ // <enclosure>
+
+ $e_ctr = $item['enclosure#'];
+
+ if ($e_ctr > 0) {
+ $e_item = array($item['enclosure@url'],
+ $item['enclosure@type'],
+ $item['enclosure@length']);
+
+ array_push($enclosures, $e_item);
+
+ for ($i = 0; $i <= $e_ctr; $i++ ) {
+
+ if ($item["enclosure#$i@url"]) {
+ $e_item = array($item["enclosure#$i@url"],
+ $item["enclosure#$i@type"],
+ $item["enclosure#$i@length"]);
+ array_push($enclosures, $e_item);
+ }
+ }
+ }
+
+ // <media:content>
+ // can there be many of those? yes -fox
+
+ $m_ctr = $item['media']['content#'];
+
+ if ($m_ctr > 0) {
+ $e_item = array($item['media']['content@url'],
+ $item['media']['content@medium'],
+ $item['media']['content@length']);
+
+ array_push($enclosures, $e_item);
+
+ for ($i = 0; $i <= $m_ctr; $i++ ) {
+
+ if ($item["media"]["content#$i@url"]) {
+ $e_item = array($item["media"]["content#$i@url"],
+ $item["media"]["content#$i@medium"],
+ $item["media"]["content#$i@length"]);
+ array_push($enclosures, $e_item);
+ }
+ }
+
+ }
+ }
+
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: article enclosures:");
+ print_r($enclosures);
+ }
+
+ db_query($link, "BEGIN");
+
+ foreach ($enclosures as $enc) {
+ $enc_url = db_escape_string($enc[0]);
+ $enc_type = db_escape_string($enc[1]);
+ $enc_dur = db_escape_string($enc[2]);
+
+ $result = db_query($link, "SELECT id FROM ttrss_enclosures
+ WHERE content_url = '$enc_url' AND post_id = '$entry_ref_id'");
+
+ if (db_num_rows($result) == 0) {
+ db_query($link, "INSERT INTO ttrss_enclosures
+ (content_url, content_type, title, duration, post_id) VALUES
+ ('$enc_url', '$enc_type', '', '$enc_dur', '$entry_ref_id')");
+ }
+ }
+
+ db_query($link, "COMMIT");
+
+ // check for manual tags (we have to do it here since they're loaded from filters)
+
+ foreach ($article_filters as $f) {
+ if ($f[0] == "tag") {
+
+ $manual_tags = trim_array(explode(",", $f[1]));
+
+ foreach ($manual_tags as $tag) {
+ if (tag_is_valid($tag)) {
+ array_push($entry_tags, $tag);
+ }
+ }
+ }
+ }
+
+ // Skip boring tags
+
+ $boring_tags = trim_array(explode(",", mb_strtolower(get_pref($link,
+ 'BLACKLISTED_TAGS', $owner_uid, ''), 'utf-8')));
+
+ $filtered_tags = array();
+ $tags_to_cache = array();
+
+ if ($entry_tags && is_array($entry_tags)) {
+ foreach ($entry_tags as $tag) {
+ if (array_search($tag, $boring_tags) === false) {
+ array_push($filtered_tags, $tag);
+ }
+ }
+ }
+
+ $filtered_tags = array_unique($filtered_tags);
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: filtered article tags:");
+ print_r($filtered_tags);
+ }
+
+ // Save article tags in the database
+
+ if (count($filtered_tags) > 0) {
+
+ db_query($link, "BEGIN");
+
+ foreach ($filtered_tags as $tag) {
+
+ $tag = sanitize_tag($tag);
+ $tag = db_escape_string($tag);
+
+ if (!tag_is_valid($tag)) continue;
+
+ $result = db_query($link, "SELECT id FROM ttrss_tags
+ WHERE tag_name = '$tag' AND post_int_id = '$entry_int_id' AND
+ owner_uid = '$owner_uid' LIMIT 1");
+
+ if ($result && db_num_rows($result) == 0) {
+
+ db_query($link, "INSERT INTO ttrss_tags
+ (owner_uid,tag_name,post_int_id)
+ VALUES ('$owner_uid','$tag', '$entry_int_id')");
+ }
+
+ array_push($tags_to_cache, $tag);
+ }
+
+ /* update the cache */
+
+ $tags_to_cache = array_unique($tags_to_cache);
+
+ $tags_str = db_escape_string(join(",", $tags_to_cache));
+
+ db_query($link, "UPDATE ttrss_user_entries
+ SET tag_cache = '$tags_str' WHERE ref_id = '$entry_ref_id'
+ AND owner_uid = $owner_uid");
+
+ db_query($link, "COMMIT");
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: article processed");
+ }
+ }
+
+ if (!$last_updated) {
+ if ($debug_enabled) {
+ _debug("update_rss_feed: new feed, catching it up...");
+ }
+ catchup_feed($link, $feed, false, $owner_uid);
+ }
+
+ if ($debug_enabled) {
+ _debug("purging feed...");
+ }
+
+ purge_feed($link, $feed, 0, $debug_enabled);
+
+ db_query($link, "UPDATE ttrss_feeds
+ SET last_updated = NOW(), last_error = '' WHERE id = '$feed'");
+
+// db_query($link, "COMMIT");
+
+ } else {
+
+ if ($use_simplepie) {
+ $error_msg = mb_substr($rss->error(), 0, 250);
+ } else {
+ $error_msg = mb_substr(magpie_error(), 0, 250);
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: error fetching feed: $error_msg");
+ }
+
+ $error_msg = db_escape_string($error_msg);
+
+ db_query($link,
+ "UPDATE ttrss_feeds SET last_error = '$error_msg',
+ last_updated = NOW() WHERE id = '$feed'");
+ }
+
+ if ($use_simplepie) {
+ unset($rss);
+ }
+
+ if ($debug_enabled) {
+ _debug("update_rss_feed: done");
+ }
+
+ }
+
+
+
+
+?>
diff --git a/update.php b/update.php
index 2fa2e2f54..8cca3daa0 100755
--- a/update.php
+++ b/update.php
@@ -7,6 +7,7 @@
chdir(dirname(__FILE__));
require_once "functions.php";
+ require_once "rssfuncs.php";
require_once "sanity_check.php";
require_once "config.php";
require_once "db.php";
diff --git a/update_daemon2.php b/update_daemon2.php
index 27b4c35d6..6b9d9246f 100755
--- a/update_daemon2.php
+++ b/update_daemon2.php
@@ -32,6 +32,7 @@
require_once "db.php";
require_once "db-prefs.php";
require_once "functions.php";
+ require_once "rssfuncs.php";
require_once "lib/magpierss/rss_fetch.inc";
$children = array();