pdf-for-wpforms / libs / HtmlWeb.php

HtmlWeb.php in PDF for WPForms + Drag and Drop Template Builder trunk, at libs/HtmlWeb.php

140 lines 3.2 KB

Raw Download Zip

1	<?php
2
3	namespace simplehtmldom;
4
5	if (! defined('ABSPATH')) exit; // Exit if accessed directly
6	/**
7	* Website: http://sourceforge.net/projects/simplehtmldom/
8	* Acknowledge: Jose Solorzano (https://sourceforge.net/projects/php-html/)
9	*
10	* Licensed under The MIT License
11	* See the LICENSE file in the project root for more information.
12	*
13	* Authors:
14	* S.C. Chen
15	* John Schlick
16	* Rus Carroll
17	* logmanoriginal
18	*
19	* Contributors:
20	* Yousuke Kumakura
21	* Vadim Voituk
22	* Antcs
23	*
24	* Version Rev. 2.0-RC2 (415)
25	*/
26
27	include_once 'HtmlDocument.php';
28
29	class HtmlWeb
30	{
31
32	/**
33	* @return HtmlDocument Returns the DOM for a webpage
34	* @return null Returns null if the cURL extension is not loaded and allow_url_fopen=Off
35	* @return null Returns null if the provided URL is invalid (not PHP_URL_SCHEME)
36	* @return null Returns null if the provided URL does not specify the HTTP or HTTPS protocol
37	*/
38	function load($url)
39	{
40	if (!filter_var($url, FILTER_VALIDATE_URL)) {
41	return null;
42	}
43
44	if ($scheme = parse_url($url, PHP_URL_SCHEME)) {
45	switch (strtolower($scheme)) {
46	case 'http':
47	case 'https':
48	break;
49	default:
50	return null;
51	}
52
53	if (extension_loaded('curl')) {
54	return $this->load_curl($url);
55	} elseif (ini_get('allow_url_fopen')) {
56	return $this->load_fopen($url);
57	} else {
58	//error_log(__FUNCTION__ . ' requires either the cURL extension or allow_url_fopen=On in php.ini');
59	}
60	}
61
62	return null;
63	}
64
65	/**
66	* cURL implementation of load
67	*/
68	private function load_curl($url)
69	{
70	$ch = curl_init();
71
72	curl_setopt($ch, CURLOPT_URL, $url);
73	curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
74	curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
75
76	// There is no guarantee this request will be fulfilled
77	// -- https://www.php.net/manual/en/function.curl-setopt.php
78	curl_setopt($ch, CURLOPT_BUFFERSIZE, MAX_FILE_SIZE);
79
80	// There is no guarantee this request will be fulfilled
81	$header = array(
82	'Accept: text/html', // Prefer HTML format
83	'Accept-Charset: utf-8', // Prefer UTF-8 encoding
84	);
85	curl_setopt($ch, CURLOPT_HTTPHEADER, $header);
86
87	$doc = curl_exec($ch);
88
89	if (curl_getinfo($ch, CURLINFO_RESPONSE_CODE) !== 200) {
90	return null;
91	}
92
93	curl_close($ch);
94
95	if (strlen($doc) > MAX_FILE_SIZE) {
96	return null;
97	}
98
99	return new HtmlDocument($doc);
100	}
101
102	/**
103	* fopen implementation of load
104	*/
105	private function load_fopen($url)
106	{
107	// There is no guarantee this request will be fulfilled
108	$context = stream_context_create(array('http' => array(
109	'header' => array(
110	'Accept: text/html', // Prefer HTML format
111	'Accept-Charset: utf-8', // Prefer UTF-8 encoding
112	),
113	'ignore_errors' => true // Always fetch content
114	)));
115
116	$doc = file_get_contents($url, false, $context, 0, MAX_FILE_SIZE + 1);
117
118	if (isset($http_response_header)) {
119	foreach ($http_response_header as $rh) {
120	// https://stackoverflow.com/a/1442526
121	$parts = explode(' ', $rh, 3);
122
123	if (preg_match('/HTTP\/\d\.\d/', $parts[0])) {
124	$code = $parts[1];
125	}
126	} // Last code is final status
127
128	if (!isset($code) \|\| $code !== '200') {
129	return null;
130	}
131	}
132
133	if (strlen($doc) > MAX_FILE_SIZE) {
134	return null;
135	}
136
137	return new HtmlDocument($doc);
138	}
139	}
140