diff options
| author | Christian Weiske <cweiske@cweiske.de> | 2016-02-04 23:46:45 +0100 |
|---|---|---|
| committer | Christian Weiske <cweiske@cweiske.de> | 2016-02-04 23:46:45 +0100 |
| commit | 9f0bdf5bf6d5a40c3673647c5861d91ccd2f9225 (patch) | |
| tree | 24524bb0582bb366a40dfbbec8357748f34da28a /bin/index.php | |
| parent | 1de186bb0fbe997a1595b0cdeac146f9a3647368 (diff) | |
| download | phinde-9f0bdf5bf6d5a40c3673647c5861d91ccd2f9225.tar.gz phinde-9f0bdf5bf6d5a40c3673647c5861d91ccd2f9225.zip | |
remove multiple tags
Diffstat (limited to 'bin/index.php')
| -rwxr-xr-x | bin/index.php | 22 |
1 files changed, 14 insertions, 8 deletions
diff --git a/bin/index.php b/bin/index.php index 374923c..04cc9ac 100755 --- a/bin/index.php +++ b/bin/index.php @@ -14,6 +14,16 @@ if ($argc < 2) { exit(1); } +function removeTags($doc, $tag) { + $elems = array(); + foreach ($doc->getElementsbyTagName($tag) as $elem) { + $elems[] = $elem; + } + foreach ($elems as $elem) { + $elem->parentNode->removeChild($elem); + } +} + $es = new Elasticsearch($GLOBALS['phinde']['elasticsearch']); $url = $argv[1]; @@ -61,13 +71,9 @@ $doc = new \DOMDocument(); $dx = new \DOMXPath($doc); //remove script tags -$elems = array(); -foreach ($doc->getElementsbyTagName('script') as $elem) { - $elems[] = $elem; -} -foreach ($elems as $elem) { - $elem->parentNode->removeChild($elem); -} +removeTags($doc, 'script'); +removeTags($doc, 'style'); +removeTags($doc, 'nav'); //default content: <body> $xpContext = $doc->getElementsByTagName('body')->item(0); @@ -128,7 +134,7 @@ foreach (array('h1', 'h2', 'h3', 'h4', 'h5', 'h6') as $headlinetype) { } } -//FIXME: limit to h-entry e-content +//FIXME: split paragraphs //FIXME: insert space after br $indexDoc->text = array(); $indexDoc->text[] = trim( |