diff --git a/bin/queue_server.php b/bin/queue_server.php index adc829c1b..f826dc002 100755 --- a/bin/queue_server.php +++ b/bin/queue_server.php @@ -1095,7 +1095,7 @@ class QueueServer implements CrawlConstants, Join $site_parts = explode("#", $this->disallowed_sites[$i]); if(count($site_parts) > 1) { $quota = intval(array_pop($site_parts)); - if($quota <= 0) continue; + if($quota <= 0) { continue; } $this->disallowed_sites[$i] = false; $quota_site = implode("#", $site_parts); $active_quota_sites[] = $quota_site; diff --git a/controllers/components/crawl_component.php b/controllers/components/crawl_component.php index 89a624472..6c1eef8cc 100644 --- a/controllers/components/crawl_component.php +++ b/controllers/components/crawl_component.php @@ -215,7 +215,6 @@ class CrawlComponent extends Component implements CrawlConstants } } } - /* Write the new crawl parameters to the name server, so that it can pass them along in the case of a new archive @@ -225,7 +224,6 @@ class CrawlComponent extends Component implements CrawlConstants "/schedules/name_server_messages.txt"; file_put_contents($filename, serialize($crawl_params)); chmod($filename, 0777); - $crawl_model->sendStartCrawlMessage($crawl_params, $seed_info, $machine_urls); } diff --git a/tests/url_parser_test.php b/tests/url_parser_test.php index 54076bd24..c35800fbc 100644 --- a/tests/url_parser_test.php +++ b/tests/url_parser_test.php @@ -192,7 +192,10 @@ class UrlParserTest extends UnitTest "http://www.cs.sjsu.edu/faculty/pollett/*/*/", "http://www.bing.com/video/search?*&*&", "http://*.cool.*/a/*/", "domain:ucla.edu", - "domain:foodnetwork.com"); + "domain:foodnetwork.com", + "domain:.ottawa.ca", + "domain:.ottawa2.ca", + "http://ottawa2.ca/"); $test_urls = array( array("http://www.cs.sjsu.edu/faculty/pollett/", false, "regex url negative 1"), @@ -216,6 +219,14 @@ class UrlParserTest extends UnitTest "domain gopher stest positive"), array("http://www.foodnetworkstore.com/small-appliances/", false, "domain test negative"), + array("http://a.ottawa.ca/", true, + "domain starting dot test positive"), + array("http://ottawa.ca/", false, + "domain starting dot test negative"), + array("http://a.ottawa2.ca/", true, + "domain starting dot test positive 2"), + array("http://ottawa2.ca/", true, + "domain starting dot test positive 3"), ); foreach($test_urls as $test_url) { $result = UrlParser::urlMemberSiteArray($test_url[0], $sites,