update_rss_feed: bailout if cache is considered valid, but older than last_updated

[tt-rss.git] / include / rssfuncs.php

diff --git a/include/rssfuncs.php b/include/rssfuncs.php
index 6429a2edaec5d5fbe36f10536ac2b3d794f8f96d..49d9e6009c448a8ab7d7e33281602b37df9a701c 100644 (file)
--- a/include/rssfuncs.php
+++ b/include/rssfuncs.php
@@ -108,9 +108,7 @@
                $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
+               $result = db_query($link, "SELECT DISTINCT ttrss_feeds.feed_url
                        FROM
                                ttrss_feeds, ttrss_users, ttrss_user_prefs
                        WHERE
@@ -118,8 +116,8 @@
                                AND ttrss_users.id = ttrss_user_prefs.owner_uid
                                AND ttrss_user_prefs.pref_name = 'DEFAULT_UPDATE_INTERVAL'
                                $login_thresh_qpart $update_limit_qpart
-                        $updstart_thresh_qpart
-                       ORDER BY feed_url,$random_qpart $query_limit");
+                               $updstart_thresh_qpart
+                       ORDER BY feed_url $query_limit");
 
                $user_prefs_cache = array();
 
@@ -128,27 +126,49 @@
                // 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']] = $line;
+                       array_push($feeds_to_update, db_escape_string($link, $line['feed_url']));
                }
 
                // 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($feed_ids) {
+
+               if(count($feeds_to_update) > 0) {
+                       $feeds_quoted = array();
+
+                       foreach ($feeds_to_update as $feed) {
+                               array_push($feeds_quoted, "'" . db_escape_string($link, $feed) . "'");
+                       }
+
                        db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
-                               WHERE id IN (%s)", implode(',', $feed_ids)));
+                               WHERE feed_url IN (%s)", implode(',', $feeds_quoted)));
                }
 
                expire_cached_files($debug);
                expire_lock_files($debug);
 
                // For each feed, we call the feed update function.
-               while ($line = array_pop($feeds_to_update)) {
+               foreach ($feeds_to_update as $feed) {
+                       if($debug) _debug("Base feed: $feed");
+
+                       //update_rss_feed($link, $line["id"], true);
 
-                       if($debug) _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
+                       // since we have the data cached, we can deal with other feeds with the same url
 
-                       update_rss_feed($link, $line["id"], true);
+                       $tmp_result = db_query($link, "SELECT ttrss_feeds.feed_url,ttrss_feeds.id,last_updated
+                       FROM ttrss_feeds, ttrss_users WHERE
+                               ttrss_users.id = ttrss_feeds.owner_uid AND
+                               feed_url = '".db_escape_string($link, $feed)."' AND
+                               ttrss_feeds.update_interval != -1
+                               $login_thresh_qpart
+                       ORDER BY feed_url $query_limit");
+
+                       if (db_num_rows($tmp_result) > 0) {
+                               while ($tline = db_fetch_assoc($tmp_result)) {
+                                       if($debug) _debug(" => " . $tline["last_updated"] . ", " . $tline["id"]);
+                                       update_rss_feed($link, $tline["id"], true);
+                               }
+                       }
                }
 
                require_once "digest.php";
@@ -219,6 +239,8 @@
 
                $rss = false;
                $rss_hash = false;
+               $cache_timestamp = file_exists($cache_filename) ? filemtime($cache_filename) : 0;
+               $last_updated_timestamp = strtotime($last_updated);
 
                if (file_exists($cache_filename) &&
                        is_readable($cache_filename) &&
@@ -229,11 +251,18 @@
                                        _debug("update_rss_feed: using local cache.");
                                }
 
-                               @$rss_data = file_get_contents($cache_filename);
+                               if ($cache_timestamp > $last_updated_timestamp) {
+                                       @$rss_data = file_get_contents($cache_filename);
 
-                               if ($rss_data) {
-                                       $rss_hash = sha1($rss_data);
-                                       @$rss = unserialize($rss_data);
+                                       if ($rss_data) {
+                                               $rss_hash = sha1($rss_data);
+                                               @$rss = unserialize($rss_data);
+                                       }
+                               } else {
+                                       if ($debug_enabled) {
+                                               _debug("update_rss_feed: local cache valid and older than last_updated, nothing to do.");
+                                       }
+                                       return;
                                }
                }
 
@@ -241,11 +270,12 @@
 
                        if (!$feed_data) {
                                if ($debug_enabled) {
-                                       _debug("update_rss_feed: fetching [$fetch_url]...");
+                                       _debug("update_rss_feed: fetching [$fetch_url] (ts: $cache_timestamp/$last_updated_timestamp)");
                                }
 
                                $feed_data = fetch_file_contents($fetch_url, false,
-                                       $auth_login, $auth_pass, false, $no_cache ? 15 : 45);
+                                       $auth_login, $auth_pass, false, $no_cache ? 15 : 45,
+                                       max($last_updated_timestamp, $cache_timestamp));
 
                                if ($debug_enabled) {
                                        _debug("update_rss_feed: fetch done.");
@@ -255,12 +285,22 @@
 
                        if (!$feed_data) {
                                global $fetch_last_error;
+                               global $fetch_last_error_code;
 
                                if ($debug_enabled) {
-                                       _debug("update_rss_feed: unable to fetch: $fetch_last_error");
+                                       _debug("update_rss_feed: unable to fetch: $fetch_last_error [$fetch_last_error_code]");
                                }
 
-                               $error_escaped = db_escape_string($link, $fetch_last_error);
+                               $error_escaped = '';
+
+                               // If-Modified-Since
+                               if ($fetch_last_error_code != 304) {
+                                       $error_escaped = db_escape_string($link, $fetch_last_error);
+                               } else {
+                                       if ($debug_enabled) {
+                                               _debug("update_rss_feed: source claims data not modified, nothing to do.");
+                                       }
+                               }
 
                                db_query($link,
                                        "UPDATE ttrss_feeds SET last_error = '$error_escaped',