Fix notice

[phinde.git] / src / phinde / Crawler.php

diff --git a/src/phinde/Crawler.php b/src/phinde/Crawler.php
index 9b148789437bac97dbdf897048487e48b2ca0121..a63815de69f93e008e5ceed04b4b0958f8f0e83c 100644 (file)
--- a/src/phinde/Crawler.php
+++ b/src/phinde/Crawler.php
@@ -25,7 +25,11 @@ class Crawler
 
     public function crawl($url)
     {
-        $res       = $this->fetch($url);
+        $res = $this->fetch($url);
+        if ($res === false) {
+            return;
+        }
+
         $linkInfos = $this->extractLinks($res);
         if ($this->showLinksOnly) {
             $this->showLinks($linkInfos);
@@ -36,13 +40,23 @@ class Crawler
 
     protected function fetch($url)
     {
+        $existingDoc = $this->es->get($url);
+
         $req = new HttpRequest($url);
         $req->setHeader(
             'accept',
             implode(',', array_keys(static::$supportedIndexTypes))
         );
+        if ($existingDoc && isset($existingDoc->modate)) {
+            $nMoDate = strtotime($existingDoc->modate);
+            $req->setHeader('If-Modified-Since: ' . date('r', $nMoDate));
+        }
+
         $res = $req->send();
-        if ($res->getStatus() !== 200) {
+        if ($res->getStatus() === 304) {
+            //not modified since last time, so don't crawl again
+            return false;
+        } else if ($res->getStatus() !== 200) {
             throw new \Exception(
                 "Response code is not 200 but "
                 . $res->getStatus() . ", stopping"
@@ -70,11 +84,19 @@ class Crawler
             if ($this->es->isKnown($linkInfo->url)) {
                 continue;
             }
-            $this->es->markQueued($linkInfo->url);
-            $this->queue->addToIndex(
-                $linkInfo->url, $linkInfo->title, $linkInfo->source
-            );
-            if (Helper::isUrlAllowed($linkInfo->url)) {
+            $allowed = Helper::isUrlAllowed($linkInfo->url);
+            $crawl   = $allowed;
+            $index   = $GLOBALS['phinde']['indexNonAllowed'] || $allowed;
+
+            if ($crawl || $index) {
+                $this->es->markQueued($linkInfo->url);
+            }
+            if ($index) {
+                $this->queue->addToIndex(
+                    $linkInfo->url, $linkInfo->title, $linkInfo->source
+                );
+            }
+            if ($allowed) {
                 $this->queue->addToCrawl($linkInfo->url);
             }
         }