From d44bf90fe3e6da8f18ecf035480d84f3ae781154 Mon Sep 17 00:00:00 2001
From: yggverse <yggverse@bot>
Date: Sun, 24 Mar 2024 18:15:26 +0200
Subject: [PATCH] stop crawler on network connection lost #11

---
 composer.json              |  3 +-
 example/config.json        | 12 ++++++++
 src/cli/document/crawl.php | 59 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 73 insertions(+), 1 deletion(-)

diff --git a/composer.json b/composer.json
index 012f838..9e09263 100644
--- a/composer.json
+++ b/composer.json
@@ -19,6 +19,7 @@
         "symfony/dom-crawler": "^6.3",
         "jdenticon/jdenticon": "^1.0",
         "yggverse/ftp": "^1.0",
-        "gregwar/captcha": "^1.2"
+        "gregwar/captcha": "^1.2",
+        "yggverse/net": "^1.2"
     }
 }
diff --git a/example/config.json b/example/config.json
index fc3fde0..ea4b119 100644
--- a/example/config.json
+++ b/example/config.json
@@ -350,5 +350,17 @@
                 ]
             }
         }
+    },
+    "network":
+    {
+        "check":
+        {
+            "enabled":true,
+            "timeout":5,
+            "socket":
+            {
+                "201:5eb5:f061:678e:7565:6338:c02c:5251":80
+            }
+        }
     }
 }
\ No newline at end of file
diff --git a/src/cli/document/crawl.php b/src/cli/document/crawl.php
index fe7605a..6c161c5 100644
--- a/src/cli/document/crawl.php
+++ b/src/cli/document/crawl.php
@@ -65,6 +65,65 @@ if (false === sem_acquire($semaphore, true))
     exit;
 }
 
+// Check network connection
+if ($config->network->check->enabled)
+{
+    $network = false;
+
+    foreach ($config->network->check->socket as $host => $port)
+    {
+        if ($config->cli->document->crawl->debug->level->notice)
+        {
+            echo sprintf(
+                _('[%s] [notice] check network connection to socket "%s" port "%d"...') . PHP_EOL,
+                date('c'),
+                $host,
+                $port
+            );
+        }
+
+        if (\Yggverse\Net\Socket::isOpen($host, $port, $config->network->check->timeout))
+        {
+            if ($config->cli->document->crawl->debug->level->notice)
+            {
+                echo sprintf(
+                    _('[%s] [notice] network connection test successful') . PHP_EOL,
+                    date('c')
+                );
+            }
+
+            $network = true;
+            break;
+        }
+
+        else
+        {
+            if ($config->cli->document->crawl->debug->level->warning)
+            {
+                echo sprintf(
+                    _('[%s] [warning] could not connect to socket "%s" port "%d"...') . PHP_EOL,
+                    date('c'),
+                    $host,
+                    $port
+                );
+            }
+        }
+    }
+
+    if (!$network)
+    {
+        if ($config->cli->document->crawl->debug->level->error)
+        {
+            echo sprintf(
+                _('[%s] [error] network unreachable!') . PHP_EOL,
+                date('c')
+            );
+        }
+
+        exit;
+    }
+}
+
 // Init client
 try {