1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Gentics Aloha Editor AJAX Gateway |
4
|
|
|
* Copyright (c) 2010 Gentics Software GmbH |
5
|
|
|
* Licensed unter the terms of http://www.aloha-editor.com/license.html |
6
|
|
|
* [email protected] |
7
|
|
|
* Author Haymo Meran [email protected] |
8
|
|
|
* Author Johannes Schüth [email protected] |
9
|
|
|
* Author Tobias Steiner [email protected] |
10
|
|
|
* |
11
|
|
|
* Testing from the command line: |
12
|
|
|
* function getallheaders(){return array('X-Gentics' => 'X');}; |
13
|
|
|
* https url example: https://google.com/adsense |
14
|
|
|
* |
15
|
|
|
*/ |
16
|
|
|
|
17
|
|
|
// for debugging |
18
|
|
|
//$_SERVER['SERVER_PROTOCOL'] = 'HTTP/1.0'; |
19
|
|
|
//$_SERVER['REQUEST_METHOD'] = 'HEAD'; |
20
|
|
|
//error_reporting(E_ALL); |
21
|
|
|
|
22
|
|
|
$request = [ |
23
|
|
|
'method' => $_SERVER['REQUEST_METHOD'], |
24
|
|
|
'protocol' => $_SERVER['SERVER_PROTOCOL'], |
25
|
|
|
'headers' => getallheaders(), |
26
|
|
|
//TODO: multipart/form-data is not handled by php://input. there |
27
|
|
|
//doesn't seem to be a generic way to get at the raw post data for |
28
|
|
|
//that content-type. |
29
|
|
|
'payload' => file_get_contents('php://input'), |
30
|
|
|
]; |
31
|
|
|
|
32
|
|
|
// read url parameter |
33
|
|
|
if (array_key_exists('url', $_GET)) { |
34
|
|
|
$request['url'] = urldecode($_GET['url']); |
35
|
|
|
} else { |
36
|
|
|
header('HTTP/1.0 400 Bad Request'); |
37
|
|
|
echo 'Aloha Editor AJAX Gateway failed because parameter url is missing.'; |
38
|
|
|
exit(); |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
// check if link exists |
42
|
|
|
$response = http_request($request); |
43
|
|
|
|
44
|
|
|
// Note HEAD does not always work even if specified... |
45
|
|
|
// We use HEAD for Linkchecking so we do a 2nd request. |
46
|
|
|
if (!array_key_exists('method', $response)) { |
|
|
|
|
47
|
|
|
$response['method'] = false; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
if ('HEAD' == strtoupper($response['method']) && (int)$response['status'] >= 400) { |
51
|
|
|
$request['method'] = 'GET'; |
52
|
|
|
$response = http_request($request); |
53
|
|
|
|
54
|
|
|
//since we handle a HEAD, we don't need to proxy any contents |
55
|
|
|
fclose($response['socket']); |
56
|
|
|
$response['socket'] = null; |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
// forward each returned header... |
60
|
|
|
foreach ($response['headers'] as $key => $value) { |
61
|
|
|
if ('content-length' == strtolower($key)) { |
62
|
|
|
//there is no need to specify a content length since we don't do keep |
63
|
|
|
//alive, and this can cause problems for integration (e.g. gzip output, |
64
|
|
|
//which would change the content length) |
65
|
|
|
//Note: overriding with header('Content-length:') will set |
66
|
|
|
//the content-length to zero for some reason |
67
|
|
|
continue; |
68
|
|
|
} |
69
|
|
|
header("$key: $value"); |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
header('Connection: close'); |
73
|
|
|
|
74
|
|
|
// output the contents if any |
75
|
|
|
if (null !== $response['socket']) { |
76
|
|
|
fpassthru($response['socket']); |
77
|
|
|
fclose($response['socket']); |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
exit; |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Query an HTTP(S) URL with the given request parameters and return the |
84
|
|
|
* response headers and status code. The socket is returned as well and |
85
|
|
|
* will point to the begining of the response payload (after all headers |
86
|
|
|
* have been read), and must be closed with fclose(). |
87
|
|
|
* @param $url the request URL |
88
|
|
|
* @param $request the request method may optionally be overridden. |
89
|
|
|
* @param $timeout connection and read timeout in seconds |
|
|
|
|
90
|
|
|
*/ |
91
|
|
|
function http_request($request, $timeout = 5) |
92
|
|
|
{ |
93
|
|
|
$url = $request['url']; |
94
|
|
|
// Extract the hostname from url |
95
|
|
|
$parts = parse_url($url); |
96
|
|
|
if (array_key_exists('host', $parts)) { |
97
|
|
|
$remote = $parts['host']; |
98
|
|
|
} else { |
99
|
|
|
return myErrorHandler("url ($url) has no host. Is it relative?"); |
|
|
|
|
100
|
|
|
} |
101
|
|
|
if (array_key_exists('port', $parts)) { |
102
|
|
|
$port = $parts['port']; |
103
|
|
|
} else { |
104
|
|
|
$port = 0; |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
// Beware that RFC2616 (HTTP/1.1) defines header fields as case-insensitive entities. |
108
|
|
|
$request_headers = ''; |
109
|
|
|
foreach ($request['headers'] as $name => $value) { |
110
|
|
|
switch (strtolower($name)) { |
111
|
|
|
//omit some headers |
112
|
|
|
case 'keep-alive': |
113
|
|
|
case 'connection': |
114
|
|
|
case 'cookie': |
115
|
|
|
//TODO: we don't handle any compression encodings. compression |
116
|
|
|
//can cause a problem if client communication is already being |
117
|
|
|
//compressed by the server/app that integrates this script |
118
|
|
|
//(which would double compress the content, once from the remote |
119
|
|
|
//server to us, and once from us to the client, but the client |
120
|
|
|
//would de-compress only once). |
121
|
|
|
case 'accept-encoding': |
122
|
|
|
break; |
123
|
|
|
// correct the host parameter |
124
|
|
|
case 'host': |
125
|
|
|
$host_info = $remote; |
126
|
|
|
if ($port) { |
127
|
|
|
$host_info .= ':' . $port; |
128
|
|
|
} |
129
|
|
|
$request_headers .= "$name: $host_info\r\n"; |
130
|
|
|
break; |
131
|
|
|
// forward all other headers |
132
|
|
|
default: |
133
|
|
|
$request_headers .= "$name: $value\r\n"; |
134
|
|
|
break; |
135
|
|
|
} |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
//set fsockopen transport scheme, and the default port |
139
|
|
|
switch (strtolower($parts['scheme'])) { |
140
|
|
|
case 'https': |
141
|
|
|
$scheme = 'ssl://'; |
142
|
|
|
if (!$port) { |
143
|
|
|
$port = 443; |
144
|
|
|
} |
145
|
|
|
break; |
146
|
|
|
case 'http': |
147
|
|
|
$scheme = ''; |
148
|
|
|
if (!$port) { |
149
|
|
|
$port = 80; |
150
|
|
|
} |
151
|
|
|
break; |
152
|
|
|
default: |
153
|
|
|
//some other transports are available but not really supported |
154
|
|
|
//by this script: http://php.net/manual/en/transports.inet.php |
155
|
|
|
$scheme = $parts['scheme'] . '://'; |
156
|
|
|
if (!$port) { |
157
|
|
|
return myErrorHandler("Unknown scheme ($scheme) and no port."); |
|
|
|
|
158
|
|
|
} |
159
|
|
|
break; |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
//we make the request with socket operations since we don't want to |
163
|
|
|
//depend on the curl extension, and the higher level wrappers don't |
164
|
|
|
//give us usable error information. |
165
|
|
|
$sock = @fsockopen("$scheme$remote", $port, $errno, $errstr, $timeout); |
166
|
|
|
if (!$sock) { |
|
|
|
|
167
|
|
|
return myErrorHandler("Unable to open URL ($url): $errstr"); |
|
|
|
|
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
//the timeout in fsockopen is only for the connection, the following |
171
|
|
|
//is for reading the content |
172
|
|
|
stream_set_timeout($sock, $timeout); |
173
|
|
|
|
174
|
|
|
//an absolute url should only be specified for proxy requests |
175
|
|
|
if (array_key_exists('path', $parts)) { |
176
|
|
|
$path_info = $parts['path']; |
177
|
|
|
} else { |
178
|
|
|
$path_info = '/'; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
if (array_key_exists('query', $parts)) { |
182
|
|
|
$path_info .= '?' . $parts['query']; |
183
|
|
|
} |
184
|
|
|
if (array_key_exists('fragment', $parts)) { |
185
|
|
|
$path_info .= '#' . $parts['fragment']; |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
$out = $request['method'] . ' ' . $path_info . ' ' . $request['protocol'] . "\r\n" . $request_headers . "Connection: close\r\n\r\n"; |
189
|
|
|
fwrite($sock, $out); |
190
|
|
|
fwrite($sock, $request['payload']); |
191
|
|
|
|
192
|
|
|
$header_str = stream_get_line($sock, 1024 * 16, "\r\n\r\n"); |
193
|
|
|
$headers = http_parse_headers($header_str); |
194
|
|
|
$status_line = array_shift($headers); |
195
|
|
|
|
196
|
|
|
// get http status |
197
|
|
|
preg_match('|HTTP/\d+\.\d+\s+(\d+)\s+.*|i', $status_line, $match); |
198
|
|
|
$status = $match[1]; |
199
|
|
|
|
200
|
|
|
return ['headers' => $headers, 'socket' => $sock, 'status' => $status]; |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
/** |
204
|
|
|
* Parses a string containing multiple HTTP header lines into an array |
205
|
|
|
* of key => values. |
206
|
|
|
* Inspired by HTTP::Daemon (CPAN). |
207
|
|
|
*/ |
208
|
|
|
function http_parse_headers($header_str) |
209
|
|
|
{ |
210
|
|
|
$headers = []; |
211
|
|
|
|
212
|
|
|
//ignore leading blank lines |
213
|
|
|
$header_str = preg_replace("/^(?:\x0D?\x0A)+/", '', $header_str); |
214
|
|
|
|
215
|
|
|
while (preg_match("/^([^\x0A]*?)\x0D?(?:\x0A|\$)/", $header_str, $matches)) { |
216
|
|
|
$header_str = substr($header_str, strlen($matches[0])); |
217
|
|
|
$status_line = $matches[1]; |
218
|
|
|
|
219
|
|
|
if (empty($headers)) { |
220
|
|
|
// the status line |
221
|
|
|
$headers[] = $status_line; |
222
|
|
|
} elseif (preg_match('/^([^:\s]+)\s*:\s*(.*)/', $status_line, $matches)) { |
223
|
|
|
if (isset($key)) { |
224
|
|
|
//previous header is finished (was potentially multi-line) |
225
|
|
|
$headers[$key] = $val; |
226
|
|
|
} |
227
|
|
|
list(, $key, $val) = $matches; |
228
|
|
|
} elseif (preg_match('/^\s+(.*)/', $status_line, $matches)) { |
229
|
|
|
//continue a multi-line header |
230
|
|
|
$val .= ' ' . $matches[1]; |
231
|
|
|
} else { |
232
|
|
|
//empty (possibly malformed) header signals the end of all headers |
233
|
|
|
break; |
234
|
|
|
} |
235
|
|
|
} |
236
|
|
|
if (isset($key)) { |
237
|
|
|
$headers[$key] = $val; |
238
|
|
|
} |
239
|
|
|
return $headers; |
240
|
|
|
} |
241
|
|
|
|
242
|
|
|
function myErrorHandler($msg) |
243
|
|
|
{ |
244
|
|
|
// 500 could be misleading... |
245
|
|
|
// Should we return a special Error when a proxy error occurs? |
246
|
|
|
header('HTTP/1.0 500 Internal Error'); |
247
|
|
|
echo "Gentics Aloha Editor AJAX Gateway Error: $msg"; |
248
|
|
|
exit(); |
|
|
|
|
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
//EOF |
252
|
|
|
|