friendica/include/Scrape.php

110 lines
2.1 KiB
PHP
Raw Normal View History

2010-07-02 01:48:07 +02:00
<?php
require_once('library/HTML5/Parser.php');
if(! function_exists('attribute_contains')) {
function attribute_contains($attr,$s) {
$a = explode(' ', $attr);
if(count($a) && in_array($s,$a))
return true;
return false;
}}
if(! function_exists('scrape_dfrn')) {
function scrape_dfrn($url) {
$ret = array();
$s = fetch_url($url);
if(! $s)
return $ret;
$dom = HTML5_Parser::parse($s);
if(! $dom)
return $ret;
$items = $dom->getElementsByTagName('link');
// get DFRN link elements
foreach($items as $item) {
$x = $item->getAttribute('rel');
if(substr($x,0,5) == "dfrn-")
$ret[$x] = $item->getAttribute('href');
if($x === 'lrdd') {
$decoded = urldecode($item->getAttribute('href'));
if(preg_match('/acct:([^@]*)@/',$decoded,$matches))
$ret['nick'] = $matches[1];
}
2010-07-02 01:48:07 +02:00
}
// Pull out hCard profile elements
$items = $dom->getElementsByTagName('*');
foreach($items as $item) {
if(attribute_contains($item->getAttribute('class'), 'vcard')) {
$level2 = $item->getElementsByTagName('*');
foreach($level2 as $x) {
if(attribute_contains($x->getAttribute('class'),'fn'))
$ret['fn'] = $x->textContent;
if(attribute_contains($x->getAttribute('class'),'photo'))
$ret['photo'] = $x->getAttribute('src');
if(attribute_contains($x->getAttribute('class'),'key'))
$ret['key'] = $x->textContent;
}
}
}
return $ret;
}}
if(! function_exists('validate_dfrn')) {
function validate_dfrn($a) {
$errors = 0;
if(! x($a,'key'))
$errors ++;
if(! x($a,'dfrn-request'))
$errors ++;
if(! x($a,'dfrn-confirm'))
$errors ++;
if(! x($a,'dfrn-notify'))
$errors ++;
if(! x($a,'dfrn-poll'))
$errors ++;
return $errors;
}}
if(! function_exists('scrape_meta')) {
function scrape_meta($url) {
2010-07-02 01:48:07 +02:00
$ret = array();
$s = fetch_url($url);
if(! $s)
return $ret;
$dom = HTML5_Parser::parse($s);
if(! $dom)
return $ret;
2010-07-02 01:48:07 +02:00
$items = $dom->getElementsByTagName('meta');
// get DFRN link elements
foreach($items as $item) {
$x = $item->getAttribute('name');
if(substr($x,0,5) == "dfrn-")
$ret[$x] = $item->getAttribute('content');
}
return $ret;
}}