From 4fb059095f7ed058af0f77af5f3a68a401e71bc6 Mon Sep 17 00:00:00 2001
From: Michael Vogel <icarus@dabo.de>
Date: Sun, 21 Dec 2014 02:03:06 +0100
Subject: [PATCH] Performance stuff: Caching for getsiteinfo function, check
 before running onepoll.php

---
 include/items.php   |  7 ++++++-
 include/onepoll.php | 16 +++++++++-------
 include/poller.php  | 16 +++++++++++-----
 3 files changed, 26 insertions(+), 13 deletions(-)

diff --git a/include/items.php b/include/items.php
index bd1f48c5e..22ae2f60a 100644
--- a/include/items.php
+++ b/include/items.php
@@ -948,7 +948,12 @@ function add_page_info_data($data) {
 function add_page_info($url, $no_photos = false, $photo = "", $keywords = false, $keyword_blacklist = "") {
 	require_once("mod/parse_url.php");
 
-	$data = parseurl_getsiteinfo($url, true);
+	$data = Cache::get("parse_url:".$url);
+	if (is_null($data)){
+		$data = parseurl_getsiteinfo($url, true);
+		Cache::set("parse_url:".$url,serialize($data));
+	} else
+		$data = unserialize($data);
 
 	if ($photo != "")
 		$data["images"][0]["src"] = $photo;
diff --git a/include/onepoll.php b/include/onepoll.php
index 9052937fd..bb5b8905a 100644
--- a/include/onepoll.php
+++ b/include/onepoll.php
@@ -15,7 +15,7 @@ function onepoll_run(&$argv, &$argc){
 	if(is_null($a)) {
 		$a = new App;
 	}
-  
+
 	if(is_null($db)) {
 	    @include(".htconfig.php");
     	require_once("include/dba.php");
@@ -57,28 +57,30 @@ function onepoll_run(&$argv, &$argc){
 		return;
 	}
 
-	// Test
 	$lockpath = get_lockpath();
 	if ($lockpath != '') {
 		$pidfile = new pidfile($lockpath, 'onepoll'.$contact_id);
-		if($pidfile->is_already_running()) {
+		if ($pidfile->is_already_running()) {
 			logger("onepoll: Already running for contact ".$contact_id);
+			if ($pidfile->running_time() > 9*60) {
+				$pidfile->kill();
+				logger("killed stale process");
+			}
 			exit;
 		}
 	}
 
-
 	$d = datetime_convert();
 
 	// Only poll from those with suitable relationships,
-	// and which have a polling address and ignore Diaspora since 
+	// and which have a polling address and ignore Diaspora since
 	// we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 
-	$contacts = q("SELECT `contact`.* FROM `contact` 
+	$contacts = q("SELECT `contact`.* FROM `contact`
 		WHERE ( `rel` = %d OR `rel` = %d ) AND `poll` != ''
 		AND NOT `network` IN ( '%s', '%s', '%s' )
 		AND `contact`.`id` = %d
-		AND `self` = 0 AND `contact`.`blocked` = 0 AND `contact`.`readonly` = 0 
+		AND `self` = 0 AND `contact`.`blocked` = 0 AND `contact`.`readonly` = 0
 		AND `contact`.`archive` = 0 LIMIT 1",
 		intval(CONTACT_IS_SHARING),
 		intval(CONTACT_IS_FRIEND),
diff --git a/include/poller.php b/include/poller.php
index e94ab8746..0edda8170 100644
--- a/include/poller.php
+++ b/include/poller.php
@@ -240,14 +240,14 @@ function poller_run(&$argv, &$argc){
 				// We should be getting everything via a hub. But just to be sure, let's check once a day.
 				// (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 				// This also lets us update our subscription to the hub, and add or replace hubs in case it
-				// changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'. 
+				// changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 
 
 				if($contact['subhub']) {
 					$poll_interval = get_config('system','pushpoll_frequency');
 					$contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
 					$hub_update = false;
-	
+
 					if((datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day")) || $force)
 							$hub_update = true;
 				}
@@ -256,13 +256,13 @@ function poller_run(&$argv, &$argc){
 
 				/**
 				 * Based on $contact['priority'], should we poll this site now? Or later?
-				 */			
+				 */
 
 				switch ($contact['priority']) {
 					case 5:
 						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 							$update = true;
-						break;					
+						break;
 					case 4:
 						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 							$update = true;
@@ -285,7 +285,13 @@ function poller_run(&$argv, &$argc){
 					continue;
 			}
 
-			proc_run('php','include/onepoll.php',$contact['id']);
+			// Don't run onepoll.php if the contact isn't pollable
+			// This check also is inside the onepoll.php - but this will reduce the load
+			if (in_array($contact["rel"], array(CONTACT_IS_SHARING, CONTACT_IS_FRIEND)) AND ($contact["poll"] != "")
+				AND !in_array($contact['network'], array(NETWORK_DIASPORA, NETWORK_FACEBOOK, NETWORK_PUMPIO, NETWORK_TWITTER, NETWORK_APPNET))
+				AND !$contact["self"] AND !$contact["blocked"] AND !$contact["readonly"] AND !$contact["archive"])
+				proc_run('php','include/onepoll.php',$contact['id']);
+
 			if($interval)
 				@time_sleep_until(microtime(true) + (float) $interval);
 		}