rework crawler; add atom link extraction

1 files changed, 54 insertions, 0 deletions

diff --git a/src/phinde/Queue.php b/src/phinde/Queue.php
new file mode 100644
index 0000000..98f6462
--- /dev/null
+++ b/src/phinde/Queue.php
@@ -0,0 +1,54 @@
+<?php
+namespace phinde;
+
+class Queue
+{
+    protected $gmclient;
+
+    public function __construct()
+    {
+        $this->gmclient = new \GearmanClient();
+        $this->gmclient->addServer('127.0.0.1');
+    }
+
+    public function addToIndex($linkUrl, $linkTitle, $sourceUrl)
+    {
+        echo "Queuing for indexing: $linkUrl\n";
+        $this->gmclient->doBackground(
+            'phinde_index',
+            serialize(
+                array(
+                    'url'    => $linkUrl,
+                    'title'  => $linkTitle,
+                    'source' => $sourceUrl
+                )
+            )
+        );
+        if ($this->gmclient->returnCode() != GEARMAN_SUCCESS) {
+            echo 'Error queueing URL indexing for '
+                . $linkUrl . "\n"
+                . 'Error code: ' . $this->gmclient->returnCode() . "\n";
+            exit(2);
+        }
+    }
+
+    public function addToCrawl($linkUrl)
+    {
+        echo "Queuing for crawling: $linkUrl\n";
+        $this->gmclient->doBackground(
+            'phinde_crawl',
+            serialize(
+                array(
+                    'url' => $linkUrl
+                )
+            )
+        );
+        if ($this->gmclient->returnCode() != GEARMAN_SUCCESS) {
+            echo 'Error queueing URL crawling for '
+                . $linkUrl . "\n"
+                . 'Error code: ' . $this->gmclient->returnCode() . "\n";
+            exit(2);
+        }
+    }
+}
+?>