Context Navigation

source: lib/http_client.c @ 4c9d377

Last change on this file since 4c9d377 was 11ec078, checked in by Wilmer van der Gaast <wilmer@…>, at 2012-09-22T12:44:47Z

Since I can't figure out where the stalls are coming from at this point, at
least have a work-around for it.

After hitting the Twitter timer a few times while a previous fetch still
seems to be running, close the connection. Auto-reconnect will do the rest.

Property mode set to 100644

File size: 13.0 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, int returncode, 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, TRUE, 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	if( getenv( "BITLBEE_DEBUG" ) )
73	printf( "About to send HTTP request:\n%s\n", req->request );
74
75	return req;
76	}
77
78	struct http_request http_dorequest_url( char url_string, http_input_function func, gpointer data )
79	{
80	url_t *url = g_new0( url_t, 1 );
81	char *request;
82	void *ret;
83
84	if( !url_set( url, url_string ) )
85	{
86	g_free( url );
87	return NULL;
88	}
89
90	if( url->proto != PROTO_HTTP && url->proto != PROTO_HTTPS )
91	{
92	g_free( url );
93	return NULL;
94	}
95
96	request = g_strdup_printf( "GET %s HTTP/1.0\r\n"
97	"Host: %s\r\n"
98	"Connection: close\r\n"
99	"User-Agent: BitlBee " BITLBEE_VERSION " " ARCH "/" CPU "\r\n"
100	"\r\n", url->file, url->host );
101
102	ret = http_dorequest( url->host, url->port,
103	url->proto == PROTO_HTTPS, request, func, data );
104
105	g_free( url );
106	g_free( request );
107	return ret;
108	}
109
110	/* This one is actually pretty simple... Might get more calls if we can't write
111	the whole request at once. */
112	static gboolean http_connected( gpointer data, int source, b_input_condition cond )
113	{
114	struct http_request *req = data;
115	int st;
116
117	if( source < 0 )
118	goto error;
119
120	if( req->inpa > 0 )
121	b_event_remove( req->inpa );
122
123	sock_make_nonblocking( req->fd );
124
125	if( req->ssl )
126	{
127	st = ssl_write( req->ssl, req->request + req->bytes_written,
128	req->request_length - req->bytes_written );
129	if( st < 0 )
130	{
131	if( ssl_errno != SSL_AGAIN )
132	{
133	ssl_disconnect( req->ssl );
134	goto error;
135	}
136	}
137	}
138	else
139	{
140	st = write( source, req->request + req->bytes_written,
141	req->request_length - req->bytes_written );
142	if( st < 0 )
143	{
144	if( !sockerr_again() )
145	{
146	closesocket( req->fd );
147	goto error;
148	}
149	}
150	}
151
152	if( st > 0 )
153	req->bytes_written += st;
154
155	if( req->bytes_written < req->request_length )
156	req->inpa = b_input_add( source,
157	req->ssl ? ssl_getdirection( req->ssl ) : B_EV_IO_WRITE,
158	http_connected, req );
159	else
160	req->inpa = b_input_add( source, B_EV_IO_READ, http_incoming_data, req );
161
162	return FALSE;
163
164	error:
165	if( req->status_string == NULL )
166	req->status_string = g_strdup( "Error while writing HTTP request" );
167
168	req->func( req );
169	http_free( req );
170	return FALSE;
171	}
172
173	static gboolean http_ssl_connected( gpointer data, int returncode, void *source, b_input_condition cond )
174	{
175	struct http_request *req = data;
176
177	if( source == NULL )
178	{
179	if( returncode != 0 )
180	{
181	char *err = ssl_verify_strerror( returncode );
182	req->status_string = g_strdup_printf(
183	"Certificate verification problem 0x%x: %s",
184	returncode, err ? err : "Unknown" );
185	g_free( err );
186	}
187	return http_connected( data, -1, cond );
188	}
189
190	req->fd = ssl_getfd( source );
191
192	return http_connected( data, req->fd, cond );
193	}
194
195	static gboolean http_incoming_data( gpointer data, int source, b_input_condition cond )
196	{
197	struct http_request *req = data;
198	int evil_server = 0;
199	char buffer[2048];
200	char end1, end2, *s;
201	size_t content_length;
202	int st;
203
204	if( req->inpa > 0 )
205	b_event_remove( req->inpa );
206
207	if( req->ssl )
208	{
209	st = ssl_read( req->ssl, buffer, sizeof( buffer ) );
210	if( st < 0 )
211	{
212	if( ssl_errno != SSL_AGAIN )
213	{
214	/* goto cleanup; */
215
216	/* YAY! We have to deal with crappy Microsoft
217	servers that LOVE to send invalid TLS
218	packets that abort connections! \o/ */
219
220	goto got_reply;
221	}
222	}
223	else if( st == 0 )
224	{
225	goto got_reply;
226	}
227	}
228	else
229	{
230	st = read( req->fd, buffer, sizeof( buffer ) );
231	if( st < 0 )
232	{
233	if( !sockerr_again() )
234	{
235	req->status_string = g_strdup( strerror( errno ) );
236	goto cleanup;
237	}
238	}
239	else if( st == 0 )
240	{
241	goto got_reply;
242	}
243	}
244
245	if( st > 0 )
246	{
247	req->reply_headers = g_realloc( req->reply_headers, req->bytes_read + st + 1 );
248	memcpy( req->reply_headers + req->bytes_read, buffer, st );
249	req->bytes_read += st;
250	}
251
252	/* There will be more! */
253	req->inpa = b_input_add( req->fd,
254	req->ssl ? ssl_getdirection( req->ssl ) : B_EV_IO_READ,
255	http_incoming_data, req );
256
257	if( ssl_pending( req->ssl ) )
258	return http_incoming_data( data, source, cond );
259	else
260	return FALSE;
261
262	got_reply:
263	/* Maybe if the webserver is overloaded, or when there's bad SSL
264	support... */
265	if( req->bytes_read == 0 )
266	{
267	req->status_string = g_strdup( "Empty HTTP reply" );
268	goto cleanup;
269	}
270
271	/* Zero termination is very convenient. */
272	req->reply_headers[req->bytes_read] = 0;
273
274	/* Find the separation between headers and body, and keep stupid
275	webservers in mind. */
276	end1 = strstr( req->reply_headers, "\r\n\r\n" );
277	end2 = strstr( req->reply_headers, "\n\n" );
278
279	if( end2 && end2 < end1 )
280	{
281	end1 = end2 + 1;
282	evil_server = 1;
283	}
284	else if( end1 )
285	{
286	end1 += 2;
287	}
288	else
289	{
290	req->status_string = g_strdup( "Malformed HTTP reply" );
291	goto cleanup;
292	}
293
294	*end1 = 0;
295
296	if( getenv( "BITLBEE_DEBUG" ) )
297	printf( "HTTP response headers:\n%s\n", req->reply_headers );
298
299	if( evil_server )
300	req->reply_body = end1 + 1;
301	else
302	req->reply_body = end1 + 2;
303
304	req->body_size = req->reply_headers + req->bytes_read - req->reply_body;
305
306	if( ( end1 = strchr( req->reply_headers, ' ' ) ) != NULL )
307	{
308	if( sscanf( end1 + 1, "%d", &req->status_code ) != 1 )
309	{
310	req->status_string = g_strdup( "Can't parse status code" );
311	req->status_code = -1;
312	}
313	else
314	{
315	char *eol;
316
317	if( evil_server )
318	eol = strchr( end1, '\n' );
319	else
320	eol = strchr( end1, '\r' );
321
322	req->status_string = g_strndup( end1 + 1, eol - end1 - 1 );
323
324	/* Just to be sure... */
325	if( ( eol = strchr( req->status_string, '\r' ) ) )
326	*eol = 0;
327	if( ( eol = strchr( req->status_string, '\n' ) ) )
328	*eol = 0;
329	}
330	}
331	else
332	{
333	req->status_string = g_strdup( "Can't locate status code" );
334	req->status_code = -1;
335	}
336
337	if( ( ( req->status_code >= 301 && req->status_code <= 303 ) \|\|
338	req->status_code == 307 ) && req->redir_ttl-- > 0 )
339	{
340	char loc, new_request, *new_host;
341	int error = 0, new_port, new_proto;
342
343	/* We might fill it again, so let's not leak any memory. */
344	g_free( req->status_string );
345	req->status_string = NULL;
346
347	loc = strstr( req->reply_headers, "\nLocation: " );
348	if( loc == NULL ) /* We can't handle this redirect... */
349	{
350	req->status_string = g_strdup( "Can't locate Location: header" );
351	goto cleanup;
352	}
353
354	loc += 11;
355	while( *loc == ' ' )
356	loc ++;
357
358	/* TODO/FIXME: Possibly have to handle relative redirections,
359	and rewrite Host: headers. Not necessary for now, it's
360	enough for passport authentication like this. */
361
362	if( *loc == '/' )
363	{
364	/* Just a different pathname... */
365
366	/* Since we don't cache the servername, and since we
367	don't need this yet anyway, I won't implement it. */
368
369	req->status_string = g_strdup( "Can't handle recursive redirects" );
370
371	goto cleanup;
372	}
373	else
374	{
375	/* A whole URL */
376	url_t *url;
377	char *s;
378	const char *new_method;
379
380	s = strstr( loc, "\r\n" );
381	if( s == NULL )
382	goto cleanup;
383
384	url = g_new0( url_t, 1 );
385	*s = 0;
386
387	if( !url_set( url, loc ) )
388	{
389	req->status_string = g_strdup( "Malformed redirect URL" );
390	g_free( url );
391	goto cleanup;
392	}
393
394	/* Find all headers and, if necessary, the POST request contents.
395	Skip the old Host: header though. This crappy code here means
396	anything using this http_client MUST put the Host: header at
397	the top. */
398	if( !( ( s = strstr( req->request, "\r\nHost: " ) ) &&
399	( s = strstr( s + strlen( "\r\nHost: " ), "\r\n" ) ) ) )
400	{
401	req->status_string = g_strdup( "Error while rebuilding request string" );
402	g_free( url );
403	goto cleanup;
404	}
405
406	/* More or less HTTP/1.0 compliant, from my reading of RFC 2616.
407	Always perform a GET request unless we received a 301. 303 was
408	meant for this but it's HTTP/1.1-only and we're specifically
409	speaking HTTP/1.0. ...
410
411	Well except someone at identi.ca's didn't bother reading any
412	RFCs and just return HTTP/1.1-specific status codes to HTTP/1.0
413	requests. Fuckers. So here we are, handle 301..303,307. */
414	if( strncmp( req->request, "GET", 3 ) == 0 )
415	/* GETs never become POSTs. */
416	new_method = "GET";
417	else if( req->status_code == 302 \|\| req->status_code == 303 )
418	/* 302 de-facto becomes GET, 303 as specified by RFC 2616#10.3.3 */
419	new_method = "GET";
420	else
421	/* 301 de-facto should stay POST, 307 specifally RFC 2616#10.3.8 */
422	new_method = "POST";
423
424	/* Okay, this isn't fun! We have to rebuild the request... :-( */
425	new_request = g_strdup_printf( "%s %s HTTP/1.0\r\nHost: %s%s",
426	new_method, url->file, url->host, s );
427
428	new_host = g_strdup( url->host );
429	new_port = url->port;
430	new_proto = url->proto;
431
432	/* If we went from POST to GET, truncate the request content. */
433	if( new_request[0] != req->request[0] && new_request[0] == 'G' &&
434	( s = strstr( new_request, "\r\n\r\n" ) ) )
435	s[4] = '\0';
436
437	g_free( url );
438	}
439
440	if( req->ssl )
441	ssl_disconnect( req->ssl );
442	else
443	closesocket( req->fd );
444
445	req->fd = -1;
446	req->ssl = NULL;
447
448	if( getenv( "BITLBEE_DEBUG" ) )
449	printf( "New headers for redirected HTTP request:\n%s\n", new_request );
450
451	if( new_proto == PROTO_HTTPS )
452	{
453	req->ssl = ssl_connect( new_host, new_port, TRUE, http_ssl_connected, req );
454	if( req->ssl == NULL )
455	error = 1;
456	}
457	else
458	{
459	req->fd = proxy_connect( new_host, new_port, http_connected, req );
460	if( req->fd < 0 )
461	error = 1;
462	}
463	g_free( new_host );
464
465	if( error )
466	{
467	req->status_string = g_strdup( "Connection problem during redirect" );
468	g_free( new_request );
469	goto cleanup;
470	}
471
472	g_free( req->request );
473	g_free( req->reply_headers );
474	req->request = new_request;
475	req->request_length = strlen( new_request );
476	req->bytes_read = req->bytes_written = req->inpa = 0;
477	req->reply_headers = req->reply_body = NULL;
478
479	return FALSE;
480	}
481
482	/* Assume that a closed connection means we're finished, this indeed
483	breaks with keep-alive connections and faulty connections. */
484	/* req->finished = 1; */
485
486	cleanup:
487	if( req->ssl )
488	ssl_disconnect( req->ssl );
489	else
490	closesocket( req->fd );
491
492	if( ( s = get_rfc822_header( req->reply_headers, "Content-Length", 0 ) ) &&
493	sscanf( s, "%zd", &content_length ) == 1 )
494	{
495	if( content_length < req->body_size )
496	{
497	req->status_code = -1;
498	g_free( req->status_string );
499	req->status_string = g_strdup( "Response truncated" );
500	}
501	}
502	g_free( s );
503
504	if( getenv( "BITLBEE_DEBUG" ) && req )
505	printf( "Finishing HTTP request with status: %s\n",
506	req->status_string ? req->status_string : "NULL" );
507
508	req->func( req );
509	http_free( req );
510	return FALSE;
511	}
512
513	static void http_free( struct http_request *req )
514	{
515	g_free( req->request );
516	g_free( req->reply_headers );
517	g_free( req->status_string );
518	g_free( req );
519	}
520

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

Download in other formats: