Context Navigation

source: lib/http_client.c @ 06b5893

Last change on this file since 06b5893 was bd31661, checked in by Wilmer van der Gaast <wilmer@…>, at 2011-12-02T10:32:54Z
Somewhat improve debug logging of HTTP/SOAP stuff.
Property mode set to `100644`
File size: 11.7 KB

Line
1	/********************************************************************\
2	* BitlBee -- An IRC to other IM-networks gateway *
3	* *
4	* Copyright 2002-2011 Wilmer van der Gaast and others *
5	\********************************************************************/
6
7	/* HTTP(S) module */
8
9	/*
10	This program is free software; you can redistribute it and/or modify
11	it under the terms of the GNU General Public License as published by
12	the Free Software Foundation; either version 2 of the License, or
13	(at your option) any later version.
14
15	This program is distributed in the hope that it will be useful,
16	but WITHOUT ANY WARRANTY; without even the implied warranty of
17	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18	GNU General Public License for more details.
19
20	You should have received a copy of the GNU General Public License with
21	the Debian GNU/Linux distribution in /usr/share/common-licenses/GPL;
22	if not, write to the Free Software Foundation, Inc., 59 Temple Place,
23	Suite 330, Boston, MA 02111-1307 USA
24	*/
25
26	#include <string.h>
27	#include <stdio.h>
28
29	#include "http_client.h"
30	#include "url.h"
31	#include "sock.h"
32
33
34	static gboolean http_connected( gpointer data, int source, b_input_condition cond );
35	static gboolean http_ssl_connected( gpointer data, void *source, b_input_condition cond );
36	static gboolean http_incoming_data( gpointer data, int source, b_input_condition cond );
37	static void http_free( struct http_request *req );
38
39
40	struct http_request http_dorequest( char host, int port, int ssl, char *request, http_input_function func, gpointer data )
41	{
42	struct http_request *req;
43	int error = 0;
44
45	req = g_new0( struct http_request, 1 );
46
47	if( ssl )
48	{
49	req->ssl = ssl_connect( host, port, http_ssl_connected, req );
50	if( req->ssl == NULL )
51	error = 1;
52	}
53	else
54	{
55	req->fd = proxy_connect( host, port, http_connected, req );
56	if( req->fd < 0 )
57	error = 1;
58	}
59
60	if( error )
61	{
62	http_free( req );
63	return NULL;
64	}
65
66	req->func = func;
67	req->data = data;
68	req->request = g_strdup( request );
69	req->request_length = strlen( request );
70	req->redir_ttl = 3;
71
72	return( req );
73	}
74
75	struct http_request http_dorequest_url( char url_string, http_input_function func, gpointer data )
76	{
77	url_t *url = g_new0( url_t, 1 );
78	char *request;
79	void *ret;
80
81	if( !url_set( url, url_string ) )
82	{
83	g_free( url );
84	return NULL;
85	}
86
87	if( url->proto != PROTO_HTTP && url->proto != PROTO_HTTPS )
88	{
89	g_free( url );
90	return NULL;
91	}
92
93	request = g_strdup_printf( "GET %s HTTP/1.0\r\n"
94	"Host: %s\r\n"
95	"Connection: close\r\n"
96	"User-Agent: BitlBee " BITLBEE_VERSION " " ARCH "/" CPU "\r\n"
97	"\r\n", url->file, url->host );
98
99	ret = http_dorequest( url->host, url->port,
100	url->proto == PROTO_HTTPS, request, func, data );
101
102	g_free( url );
103	g_free( request );
104	return ret;
105	}
106
107	/* This one is actually pretty simple... Might get more calls if we can't write
108	the whole request at once. */
109	static gboolean http_connected( gpointer data, int source, b_input_condition cond )
110	{
111	struct http_request *req = data;
112	int st;
113
114	if( source < 0 )
115	goto error;
116
117	if( req->inpa > 0 )
118	b_event_remove( req->inpa );
119
120	sock_make_nonblocking( req->fd );
121
122	if( req->ssl )
123	{
124	st = ssl_write( req->ssl, req->request + req->bytes_written,
125	req->request_length - req->bytes_written );
126	if( st < 0 )
127	{
128	if( ssl_errno != SSL_AGAIN )
129	{
130	ssl_disconnect( req->ssl );
131	goto error;
132	}
133	}
134	}
135	else
136	{
137	st = write( source, req->request + req->bytes_written,
138	req->request_length - req->bytes_written );
139	if( st < 0 )
140	{
141	if( !sockerr_again() )
142	{
143	closesocket( req->fd );
144	goto error;
145	}
146	}
147	}
148
149	if( st > 0 )
150	req->bytes_written += st;
151
152	if( req->bytes_written < req->request_length )
153	req->inpa = b_input_add( source,
154	req->ssl ? ssl_getdirection( req->ssl ) : B_EV_IO_WRITE,
155	http_connected, req );
156	else
157	req->inpa = b_input_add( source, B_EV_IO_READ, http_incoming_data, req );
158
159	return FALSE;
160
161	error:
162	req->status_string = g_strdup( "Error while writing HTTP request" );
163
164	req->func( req );
165	http_free( req );
166	return FALSE;
167	}
168
169	static gboolean http_ssl_connected( gpointer data, void *source, b_input_condition cond )
170	{
171	struct http_request *req = data;
172
173	if( source == NULL )
174	return http_connected( data, -1, cond );
175
176	req->fd = ssl_getfd( source );
177
178	return http_connected( data, req->fd, cond );
179	}
180
181	static gboolean http_incoming_data( gpointer data, int source, b_input_condition cond )
182	{
183	struct http_request *req = data;
184	int evil_server = 0;
185	char buffer[2048];
186	char end1, end2;
187	int st;
188
189	if( req->inpa > 0 )
190	b_event_remove( req->inpa );
191
192	if( req->ssl )
193	{
194	st = ssl_read( req->ssl, buffer, sizeof( buffer ) );
195	if( st < 0 )
196	{
197	if( ssl_errno != SSL_AGAIN )
198	{
199	/* goto cleanup; */
200
201	/* YAY! We have to deal with crappy Microsoft
202	servers that LOVE to send invalid TLS
203	packets that abort connections! \o/ */
204
205	goto got_reply;
206	}
207	}
208	else if( st == 0 )
209	{
210	goto got_reply;
211	}
212	}
213	else
214	{
215	st = read( req->fd, buffer, sizeof( buffer ) );
216	if( st < 0 )
217	{
218	if( !sockerr_again() )
219	{
220	req->status_string = g_strdup( strerror( errno ) );
221	goto cleanup;
222	}
223	}
224	else if( st == 0 )
225	{
226	goto got_reply;
227	}
228	}
229
230	if( st > 0 )
231	{
232	req->reply_headers = g_realloc( req->reply_headers, req->bytes_read + st + 1 );
233	memcpy( req->reply_headers + req->bytes_read, buffer, st );
234	req->bytes_read += st;
235	}
236
237	/* There will be more! */
238	req->inpa = b_input_add( req->fd,
239	req->ssl ? ssl_getdirection( req->ssl ) : B_EV_IO_READ,
240	http_incoming_data, req );
241
242	if( ssl_pending( req->ssl ) )
243	return http_incoming_data( data, source, cond );
244	else
245	return FALSE;
246
247	got_reply:
248	/* Maybe if the webserver is overloaded, or when there's bad SSL
249	support... */
250	if( req->bytes_read == 0 )
251	{
252	req->status_string = g_strdup( "Empty HTTP reply" );
253	goto cleanup;
254	}
255
256	/* Zero termination is very convenient. */
257	req->reply_headers[req->bytes_read] = 0;
258
259	/* Find the separation between headers and body, and keep stupid
260	webservers in mind. */
261	end1 = strstr( req->reply_headers, "\r\n\r\n" );
262	end2 = strstr( req->reply_headers, "\n\n" );
263
264	if( end2 && end2 < end1 )
265	{
266	end1 = end2 + 1;
267	evil_server = 1;
268	}
269	else if( end1 )
270	{
271	end1 += 2;
272	}
273	else
274	{
275	req->status_string = g_strdup( "Malformed HTTP reply" );
276	goto cleanup;
277	}
278
279	*end1 = 0;
280
281	if( getenv( "BITLBEE_DEBUG" ) )
282	printf( "HTTP response headers:\n%s", req->reply_headers );
283
284	if( evil_server )
285	req->reply_body = end1 + 1;
286	else
287	req->reply_body = end1 + 2;
288
289	req->body_size = req->reply_headers + req->bytes_read - req->reply_body;
290
291	if( ( end1 = strchr( req->reply_headers, ' ' ) ) != NULL )
292	{
293	if( sscanf( end1 + 1, "%d", &req->status_code ) != 1 )
294	{
295	req->status_string = g_strdup( "Can't parse status code" );
296	req->status_code = -1;
297	}
298	else
299	{
300	char *eol;
301
302	if( evil_server )
303	eol = strchr( end1, '\n' );
304	else
305	eol = strchr( end1, '\r' );
306
307	req->status_string = g_strndup( end1 + 1, eol - end1 - 1 );
308
309	/* Just to be sure... */
310	if( ( eol = strchr( req->status_string, '\r' ) ) )
311	*eol = 0;
312	if( ( eol = strchr( req->status_string, '\n' ) ) )
313	*eol = 0;
314	}
315	}
316	else
317	{
318	req->status_string = g_strdup( "Can't locate status code" );
319	req->status_code = -1;
320	}
321
322	if( ( req->status_code >= 301 && req->status_code <= 303 ) && req->redir_ttl-- > 0 )
323	{
324	char loc, new_request, *new_host;
325	int error = 0, new_port, new_proto;
326
327	/* We might fill it again, so let's not leak any memory. */
328	g_free( req->status_string );
329	req->status_string = NULL;
330
331	loc = strstr( req->reply_headers, "\nLocation: " );
332	if( loc == NULL ) /* We can't handle this redirect... */
333	{
334	req->status_string = g_strdup( "Can't locate Location: header" );
335	goto cleanup;
336	}
337
338	loc += 11;
339	while( *loc == ' ' )
340	loc ++;
341
342	/* TODO/FIXME: Possibly have to handle relative redirections,
343	and rewrite Host: headers. Not necessary for now, it's
344	enough for passport authentication like this. */
345
346	if( *loc == '/' )
347	{
348	/* Just a different pathname... */
349
350	/* Since we don't cache the servername, and since we
351	don't need this yet anyway, I won't implement it. */
352
353	req->status_string = g_strdup( "Can't handle recursive redirects" );
354
355	goto cleanup;
356	}
357	else
358	{
359	/* A whole URL */
360	url_t *url;
361	char *s;
362	const char *new_method;
363
364	s = strstr( loc, "\r\n" );
365	if( s == NULL )
366	goto cleanup;
367
368	url = g_new0( url_t, 1 );
369	*s = 0;
370
371	if( !url_set( url, loc ) )
372	{
373	req->status_string = g_strdup( "Malformed redirect URL" );
374	g_free( url );
375	goto cleanup;
376	}
377
378	/* Find all headers and, if necessary, the POST request contents.
379	Skip the old Host: header though. This crappy code here means
380	anything using this http_client MUST put the Host: header at
381	the top. */
382	if( !( ( s = strstr( req->request, "\r\nHost: " ) ) &&
383	( s = strstr( s + strlen( "\r\nHost: " ), "\r\n" ) ) ) )
384	{
385	req->status_string = g_strdup( "Error while rebuilding request string" );
386	g_free( url );
387	goto cleanup;
388	}
389
390	/* More or less HTTP/1.0 compliant, from my reading of RFC 2616.
391	Always perform a GET request unless we received a 301. 303 was
392	meant for this but it's HTTP/1.1-only and we're specifically
393	speaking HTTP/1.0. */
394	new_method = req->status_code != 301 \|\| req->request[0] == 'G' ? "GET" : "POST";
395
396	/* Okay, this isn't fun! We have to rebuild the request... :-( */
397	new_request = g_strdup_printf( "%s %s HTTP/1.0\r\nHost: %s%s",
398	new_method, url->file, url->host, s );
399
400	new_host = g_strdup( url->host );
401	new_port = url->port;
402	new_proto = url->proto;
403
404	/* If we went from POST to GET, truncate the request content. */
405	if( new_request[0] != req->request[0] && new_request[0] == 'G' &&
406	( s = strstr( new_request, "\r\n\r\n" ) ) )
407	s[4] = '\0';
408
409	g_free( url );
410	}
411
412	if( req->ssl )
413	ssl_disconnect( req->ssl );
414	else
415	closesocket( req->fd );
416
417	req->fd = -1;
418	req->ssl = NULL;
419
420	if( getenv( "BITLBEE_DEBUG" ) )
421	printf( "New headers for redirected HTTP request:\n%s", new_request );
422
423	if( new_proto == PROTO_HTTPS )
424	{
425	req->ssl = ssl_connect( new_host, new_port, http_ssl_connected, req );
426	if( req->ssl == NULL )
427	error = 1;
428	}
429	else
430	{
431	req->fd = proxy_connect( new_host, new_port, http_connected, req );
432	if( req->fd < 0 )
433	error = 1;
434	}
435	g_free( new_host );
436
437	if( error )
438	{
439	req->status_string = g_strdup( "Connection problem during redirect" );
440	g_free( new_request );
441	goto cleanup;
442	}
443
444	g_free( req->request );
445	g_free( req->reply_headers );
446	req->request = new_request;
447	req->request_length = strlen( new_request );
448	req->bytes_read = req->bytes_written = req->inpa = 0;
449	req->reply_headers = req->reply_body = NULL;
450
451	return FALSE;
452	}
453
454	/* Assume that a closed connection means we're finished, this indeed
455	breaks with keep-alive connections and faulty connections. */
456	req->finished = 1;
457
458	cleanup:
459	if( req->ssl )
460	ssl_disconnect( req->ssl );
461	else
462	closesocket( req->fd );
463
464	if( getenv( "BITLBEE_DEBUG" ) && req )
465	printf( "Finishing HTTP request with status: %s",
466	req->status_string ? req->status_string : "NULL" );
467
468	req->func( req );
469	http_free( req );
470	return FALSE;
471	}
472
473	static void http_free( struct http_request *req )
474	{
475	g_free( req->request );
476	g_free( req->reply_headers );
477	g_free( req->status_string );
478	g_free( req );
479	}
480

Note: See TracBrowser for help on using the repository browser.

Download in other formats: