diff options
Diffstat (limited to 'src/phinde/Crawler.php')
| -rw-r--r-- | src/phinde/Crawler.php | 18 |
1 files changed, 13 insertions, 5 deletions
diff --git a/src/phinde/Crawler.php b/src/phinde/Crawler.php index 43d9459..6459fb4 100644 --- a/src/phinde/Crawler.php +++ b/src/phinde/Crawler.php @@ -84,11 +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); } } |
