implement ttrss_feeds.cache_content

classes/api.php

@@ -280,7 +280,7 @@ class API extends Handler {
 
 		$article_id = join(",", array_filter(explode(",", db_escape_string($_REQUEST["article_id"])), is_numeric));
 
-		$query = "SELECT id,title,link,content,feed_id,comments,int_id,
+		$query = "SELECT id,title,link,content,cached_content,feed_id,comments,int_id,
 			marked,unread,published,
 			".SUBSTRING_FOR_DATE."(updated,1,16) as updated,
 			author
@@ -309,7 +309,7 @@ class API extends Handler {
 					"comments" => $line["comments"],
 					"author" => $line["author"],
 					"updated" => strtotime($line["updated"]),
-					"content" => $line["content"],
+					"content" => $line["cached_content"] != "" ? $line["cached_content"] : $line["content"],
 					"feed_id" => $line["feed_id"],
 					"attachments" => $attachments
 				);

classes/feeds.php

@@ -158,16 +158,20 @@ class Feeds extends Handler_Protected {
 			// Update the feed if required with some basic flood control
 
 			$result = db_query($this->link,
-				"SELECT cache_images,".SUBSTRING_FOR_DATE."(last_updated,1,19) AS last_updated
+				"SELECT cache_images,cache_content,".SUBSTRING_FOR_DATE."(last_updated,1,19) AS last_updated
 					FROM ttrss_feeds WHERE id = '$feed'");
 
 				if (db_num_rows($result) != 0) {
 					$last_updated = strtotime(db_fetch_result($result, 0, "last_updated"));
 					$cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
+					$cache_content = sql_bool_to_bool(db_fetch_result($result, 0, "cache_content"));
 
-					if (!$cache_images && time() - $last_updated > 120 || isset($_REQUEST['DevForceUpdate'])) {
+					if (!$cache_images && !$cache_content && time() - $last_updated > 120 || isset($_REQUEST['DevForceUpdate'])) {
 						include "rssfuncs.php";
 						update_rss_feed($this->link, $feed, true, true);
+					} else {
+						db_query($this->link, "UPDATE ttrss_feeds SET last_updated = '1970-01-01', last_update_started = '1970-01-01'
+							WHERE id = '$feed'");
 					}
 				}
 		}
@@ -234,6 +238,7 @@ class Feeds extends Handler_Protected {
 		$feed_title = $qfh_ret[1];
 		$feed_site_url = $qfh_ret[2];
 		$last_error = $qfh_ret[3];
+		$cache_content = true;
 
 		$vgroup_last_feed = $vgr_last_feed;
 
@@ -627,6 +632,10 @@ class Feeds extends Handler_Protected {
 
 					$feed_site_url = $line["site_url"];
 
+					if ($cache_content && $line["cached_content"] != "") {
+						$line["content_preview"] =& $line["cached_content"];
+					}
+
 					$article_content = sanitize($this->link, $line["content_preview"],
 							false, false, $feed_site_url);
 

classes/pref/feeds.php

@@ -653,6 +653,19 @@ class Pref_Feeds extends Handler_Protected {
 			$checked>&nbsp;<label for=\"cache_images\">".
 		__('Cache images locally')."</label>";
 
+		$cache_content = sql_bool_to_bool(db_fetch_result($result, 0, "cache_content"));
+
+		if ($cache_content) {
+			$checked = "checked=\"1\"";
+		} else {
+			$checked = "";
+		}
+
+		print "<hr/><input dojoType=\"dijit.form.CheckBox\" type=\"checkbox\" id=\"cache_content\"
+		name=\"cache_content\"
+			$checked>&nbsp;<label for=\"cache_content\">".
+		__('Cache content locally')."</label>";
+
 		$mark_unread_on_update = sql_bool_to_bool(db_fetch_result($result, 0, "mark_unread_on_update"));
 
 		if ($mark_unread_on_update) {
@@ -914,6 +927,8 @@ class Pref_Feeds extends Handler_Protected {
 			db_escape_string($_POST["include_in_digest"]));
 		$cache_images = checkbox_to_sql_bool(
 			db_escape_string($_POST["cache_images"]));
+		$cache_content = checkbox_to_sql_bool(
+			db_escape_string($_POST["cache_content"]));
 		$update_method = (int) db_escape_string($_POST["update_method"]);
 
 		$always_display_enclosures = checkbox_to_sql_bool(
@@ -938,8 +953,6 @@ class Pref_Feeds extends Handler_Protected {
 			$category_qpart_nocomma = "";
 		}
 
-		$cache_images_qpart = "cache_images = $cache_images,";
-
 		if (!$batch) {
 
 			$result = db_query($this->link, "UPDATE ttrss_feeds SET
@@ -951,7 +964,8 @@ class Pref_Feeds extends Handler_Protected {
 				auth_pass = '$auth_pass',
 				private = $private,
 				rtl_content = $rtl_content,
-				$cache_images_qpart
+				cache_images = $cache_images,
+				cache_content = $cache_content,
 				include_in_digest = $include_in_digest,
 				always_display_enclosures = $always_display_enclosures,
 				mark_unread_on_update = $mark_unread_on_update,
@@ -1023,6 +1037,10 @@ class Pref_Feeds extends Handler_Protected {
 						$qpart = "cache_images = $cache_images";
 						break;
 
+					case "cache_content":
+						$qpart = "cache_content = $cache_content";
+						break;
+
 					case "rtl_content":
 						$qpart = "rtl_content = $rtl_content";
 						break;

db-updater.php

@@ -1,5 +1,5 @@
 <?php
-	set_include_path(get_include_path() . PATH_SEPARATOR . 
+	set_include_path(get_include_path() . PATH_SEPARATOR .
 		dirname(__FILE__) . "/include");
 
 	require_once "functions.php";
@@ -125,6 +125,7 @@ function confirmOP() {
 		foreach (array_keys($update_versions) as $v) {
 			if ($v == $version + 1) {
 				print "<p>".T_sprintf("Updating to version %d...", $v)."</p>";
+				db_query($link, "BEGIN");
 				$fp = fopen($update_versions[$v], "r");
 				if ($fp) {
 					while (!feof($fp)) {
@@ -136,6 +137,7 @@ function confirmOP() {
 					}
 				}
 				fclose($fp);
+				db_query($link, "COMMIT");
 
 				print "<p>".__("Checking version... ");
 

include/functions.php

@@ -1,6 +1,6 @@
 <?php
 	define('EXPECTED_CONFIG_VERSION', 26);
-	define('SCHEMA_VERSION', 98);
+	define('SCHEMA_VERSION', 99);
 
 	$fetch_last_error = false;
 
@@ -2367,7 +2367,7 @@
 				}
 			}
 
-			$content_query_part = "content as content_preview,";
+			$content_query_part = "content as content_preview, cached_content, ";
 
 			if (is_numeric($feed)) {
 
@@ -3111,15 +3111,17 @@
 
 		//if (!$zoom_mode) { print "<article id='$id'><![CDATA["; };
 
-		$result = db_query($link, "SELECT rtl_content, always_display_enclosures FROM ttrss_feeds
+		$result = db_query($link, "SELECT rtl_content, always_display_enclosures, cache_content FROM ttrss_feeds
 			WHERE id = '$feed_id' AND owner_uid = $owner_uid");
 
 		if (db_num_rows($result) == 1) {
 			$rtl_content = sql_bool_to_bool(db_fetch_result($result, 0, "rtl_content"));
 			$always_display_enclosures = sql_bool_to_bool(db_fetch_result($result, 0, "always_display_enclosures"));
+			$cache_content = sql_bool_to_bool(db_fetch_result($result, 0, "cache_content"));
 		} else {
 			$rtl_content = false;
 			$always_display_enclosures = false;
+			$cache_content = false;
 		}
 
 		if ($rtl_content) {
@@ -3146,7 +3148,8 @@
 			tag_cache,
 			author,
 			orig_feed_id,
-			note
+			note,
+			cached_content
 			FROM ttrss_entries,ttrss_user_entries
 			WHERE	id = '$id' AND ref_id = id AND owner_uid = $owner_uid");
 
@@ -3344,6 +3347,10 @@
 				}
 			}
 
+			if ($cache_content && $line["cached_content"] != "") {
+				$line["content"] =& $line["cached_content"];
+			}
+
 			$article_content = sanitize($link, $line["content"], false, $owner_uid,
 				$feed_site_url);
 
@@ -4460,6 +4467,11 @@
 				}
 
 				if ($show_content) {
+
+					if ($line["cached_content"] != "") {
+						$line["content_preview"] =& $line["cached_content"];
+					}
+
 					if ($sanitize_content) {
 						$headline_row["content"] = sanitize($link,
 							$line["content_preview"], false, false, $line["site_url"]);

include/rssfuncs.php

@@ -157,6 +157,7 @@
 
 	} // function update_daemon_common
 
+	// ignore_daemon is not used
 	function update_rss_feed($link, $feed, $ignore_daemon = false, $no_cache = false,
 		$override_url = false) {
 
@@ -166,35 +167,15 @@
 
 		$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) {
+		$result = db_query($link, "SELECT id,update_interval,auth_login,
-
+			feed_url,auth_pass,cache_images,update_method,last_updated,cache_content,
-			if (DB_TYPE == "pgsql") {
+			mark_unread_on_update, owner_uid, update_on_checksum_change,
-					$updstart_thresh_qpart = "(ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
+			pubsub_state
-				} else {
+			FROM ttrss_feeds WHERE id = '$feed'");
-					$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,last_updated
-				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) {
@@ -240,6 +221,7 @@
 		}
 
 		$cache_images = sql_bool_to_bool(db_fetch_result($result, 0, "cache_images"));
+		$cache_content = sql_bool_to_bool(db_fetch_result($result, 0, "cache_content"));
 		$fetch_url = db_fetch_result($result, 0, "feed_url");
 
 		$feed = db_escape_string($feed);
@@ -624,6 +606,7 @@
 				}
 
 				$entry_content_unescaped = $entry_content;
+				$entry_cached_content = "";
 
 				if ($use_simplepie) {
 					$entry_comments = strip_tags($item->data["comments"]);
@@ -782,6 +765,20 @@
 						_debug("update_rss_feed: base guid not found");
 					}
 
+					if ($cache_content) {
+						if ($debug_enabled) {
+							_debug("update_rss_feed: caching content...");
+						}
+
+						$entry_cached_content = cache_content($link, $entry_link, $auth_login, $auth_pass);
+
+						if ($cache_images && is_writable(CACHE_DIR . '/images'))
+							$entry_cached_content = cache_images($entry_cached_content, $site_url, $debug_enabled);
+
+						$entry_cached_content = db_escape_string($entry_cached_content, false);
+
+					}
+
 					// base post entry does not exist, create it
 
 					$result = db_query($link,
@@ -792,6 +789,7 @@
 							updated,
 							content,
 							content_hash,
+							cached_content,
 							no_orig_date,
 							date_updated,
 							date_entered,
@@ -804,6 +802,7 @@
 							'$entry_link',
 							'$entry_timestamp_fmt',
 							'$entry_content',
+							'$entry_cached_content',
 							'$content_hash',
 							$no_orig_date,
 							NOW(),
@@ -996,6 +995,19 @@
 					if ($content_hash != $orig_content_hash) {
 						$post_needs_update = true;
 						$update_insignificant = false;
+
+						if ($cache_content) {
+							if ($debug_enabled) {
+								_debug("update_rss_feed: caching content because original checksum changed...");
+							}
+
+							$entry_cached_content = cache_content($link, $entry_link, $auth_login, $auth_pass);
+
+							if ($cache_images && is_writable(CACHE_DIR . '/images'))
+								$entry_cached_content = cache_images($entry_cached_content, $site_url, $debug_enabled);
+
+							$entry_cached_content = db_escape_string($entry_cached_content, false);
+						}
 					}
 
 					if (db_escape_string($orig_title) != $entry_title) {
@@ -1016,6 +1028,7 @@
 						db_query($link, "UPDATE ttrss_entries
 							SET title = '$entry_title', content = '$entry_content',
 								content_hash = '$content_hash',
+								cached_content = '$entry_cached_content',
 								updated = '$entry_timestamp_fmt',
 								num_comments = '$num_comments'
 							WHERE id = '$ref_id'");
@@ -1484,4 +1497,25 @@
 			}
 		}
 	}
+
+	function cache_content($link, $url, $login, $pass) {
+
+		$content = fetch_file_contents($url, $login, $pass);
+
+		if ($content) {
+			$doc = new DOMDocument();
+			@$doc->loadHTML($content);
+			$xpath = new DOMXPath($doc);
+
+			$node = $doc->getElementsByTagName('body')->item(0);
+
+			if ($node) {
+				$content = $doc->saveXML($node, LIBXML_NOEMPTYTAG);
+
+				return $content;
+			}
+		}
+
+		return "";
+	}
 ?>

schema/ttrss_schema_mysql.sql

@@ -116,6 +116,7 @@ create table ttrss_feeds (id integer not null auto_increment primary key,
 	hidden bool not null default false,
 	include_in_digest boolean not null default true,
 	cache_images boolean not null default false,
+	cache_content boolean not null default false,
 	auth_pass_encrypted boolean not null default false,
 	last_viewed datetime default null,
 	last_update_started datetime default null,
@@ -150,6 +151,7 @@ create table ttrss_entries (id integer not null primary key auto_increment,
 	updated datetime not null,
 	content longtext not null,
 	content_hash varchar(250) not null,
+	cached_content longtext,
 	no_orig_date bool not null default 0,
 	date_entered datetime not null,
 	date_updated datetime not null,
@@ -306,7 +308,7 @@ create table ttrss_tags (id integer primary key auto_increment,
 
 create table ttrss_version (schema_version int not null) ENGINE=InnoDB DEFAULT CHARSET=UTF8;
 
-insert into ttrss_version values (98);
+insert into ttrss_version values (99);
 
 create table ttrss_enclosures (id integer primary key auto_increment,
 	content_url text not null,

schema/ttrss_schema_pgsql.sql

@@ -78,6 +78,7 @@ create table ttrss_feeds (id serial not null primary key,
 	include_in_digest boolean not null default true,
 	rtl_content boolean not null default false,
 	cache_images boolean not null default false,
+	cache_content boolean not null default false,
 	last_viewed timestamp default null,
 	last_update_started timestamp default null,
 	update_method integer not null default 0,
@@ -130,6 +131,7 @@ create table ttrss_entries (id serial not null primary key,
 	updated timestamp not null,
 	content text not null,
 	content_hash varchar(250) not null,
+	cached_content text,
 	no_orig_date boolean not null default false,
 	date_entered timestamp not null,
 	date_updated timestamp not null,
@@ -254,7 +256,7 @@ create index ttrss_tags_post_int_id_idx on ttrss_tags(post_int_id);
 
 create table ttrss_version (schema_version int not null);
 
-insert into ttrss_version values (98);
+insert into ttrss_version values (99);
 
 create table ttrss_enclosures (id serial not null primary key,
 	content_url text not null,

schema/versions/mysql/99.sql

@@ -0,0 +1,12 @@
+begin;
+
+alter table ttrss_feeds add column cache_content bool;
+update ttrss_feeds set cache_content = false;
+alter table ttrss_feeds change cache_content cache_content bool not null;
+alter table ttrss_feeds alter column cache_content set default false;
+
+alter table ttrss_entries add column cached_content longtext;
+
+update ttrss_version set schema_version = 99;
+
+commit;

schema/versions/pgsql/99.sql

@@ -0,0 +1,12 @@
+begin;
+
+alter table ttrss_feeds add column cache_content boolean;
+update ttrss_feeds set cache_content = false;
+alter table ttrss_feeds alter column cache_content set not null;
+alter table ttrss_feeds alter column cache_content set default false;
+
+alter table ttrss_entries add column cached_content text;
+
+update ttrss_version set schema_version = 99;
+
+commit;