aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorChristian Weiske <cweiske@cweiske.de>2016-08-29 22:59:16 +0200
committerChristian Weiske <cweiske@cweiske.de>2016-08-29 22:59:16 +0200
commit686f1cec3fd35782c30d20f891fec2f434e5d02f (patch)
tree3cfa6087c00690534353840295e6ea0646c2b0b6
parent90d2e0f8c2a80f4d5db5ae20248337d3a0594411 (diff)
downloadphinde-686f1cec3fd35782c30d20f891fec2f434e5d02f.tar.gz
phinde-686f1cec3fd35782c30d20f891fec2f434e5d02f.zip
Respect <meta name="robots" content="noindex"/>
Fixes: #1
-rwxr-xr-xbin/index.php11
1 files changed, 11 insertions, 0 deletions
diff --git a/bin/index.php b/bin/index.php
index 5a55427..5985a3e 100755
--- a/bin/index.php
+++ b/bin/index.php
@@ -90,6 +90,17 @@ if ($xbase) {
);
}
+$meta = $dx->evaluate('/html/head/meta[@name="robots" and @content]')
+ ->item(0);
+if ($meta) {
+ $robots = $meta->attributes->getNamedItem('content')->textContent;
+ foreach (explode(',', $robots) as $value) {
+ if (trim($value) == 'noindex') {
+ echo "URL does not want to be indexed: $url\n";
+ exit(0);
+ }
+ }
+}
//remove script tags
removeTags($doc, 'script');