yeemail / libs / HtmlWeb.php

HtmlWeb.php in YeeMail — Email Template Builder & Customizer 2.2.1, at libs/HtmlWeb.php

135 lines 3.1 KB

Raw Download Zip

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