X-Git-Url: https://git.cweiske.de/phinde.git/blobdiff_plain/cd02bac646f42a0cb402ff2dc8240aa01f1f0fb8..087e616b4eed27573d5a00b725eca02c1584fa4f:/bin/crawl.php

diff --git a/bin/crawl.php b/bin/crawl.php
index e9a6218..0d57bb3 100755
--- a/bin/crawl.php
+++ b/bin/crawl.php
@@ -3,12 +3,33 @@
 namespace phinde;
 require_once __DIR__ . '/../src/init.php';
 
-if ($argc < 2) {
-    echo "No URL given\n";
-    exit(1);
+$cc = new \Console_CommandLine();
+$cc->description = 'phinde URL crawler';
+$cc->version = '0.0.1';
+$cc->addOption(
+    'showLinksOnly',
+    array(
+        'short_name'  => '-s',
+        'long_name'   => '--show-links',
+        'description' => 'Only show which URLs were found',
+        'action'      => 'StoreTrue',
+        'default'     => false
+    )
+);
+$cc->addArgument(
+    'url',
+    array(
+        'description' => 'URL to crawl',
+        'multiple'    => false
+    )
+);
+try {
+    $res = $cc->parse();
+} catch (\Exception $e) {
+    $cc->displayError($e->getMessage());
 }
 
-$url = $argv[1];
+$url = $res->args['url'];
 $url = Helper::addSchema($url);
 if (!Helper::isUrlAllowed($url)) {
     echo "Domain is not allowed; not crawling\n";
@@ -17,6 +38,7 @@ if (!Helper::isUrlAllowed($url)) {
 
 try {
     $crawler = new Crawler();
+    $crawler->setShowLinksOnly($res->options['showLinksOnly']);
     $crawler->crawl($url);
 } catch (\Exception $e) {
     echo $e->getMessage() . "\n";