]> git.ipfire.org Git - thirdparty/git.git/blame - http.c
remote-http(s): support SOCKS proxies
[thirdparty/git.git] / http.c
CommitLineData
1c4b6604 1#include "git-compat-util.h"
29508e1e 2#include "http.h"
2264dfa5 3#include "pack.h"
de1a2fdd 4#include "sideband.h"
fe72d420 5#include "run-command.h"
f39f72d8 6#include "url.h"
6a56993b 7#include "urlmatch.h"
148bb6a7 8#include "credential.h"
745c7c8e 9#include "version.h"
047ec602 10#include "pkt-line.h"
93f7d910 11#include "gettext.h"
f4113cac 12#include "transport.h"
29508e1e 13
4251ccbd 14int active_requests;
e9176745 15int http_is_verbose;
de1a2fdd 16size_t http_post_buffer = 16 * LARGE_PACKET_MAX;
29508e1e 17
b8ac9230
MS
18#if LIBCURL_VERSION_NUM >= 0x070a06
19#define LIBCURL_CAN_HANDLE_AUTH_ANY
20#endif
21
ad75ebe5
TRC
22static int min_curl_sessions = 1;
23static int curl_session_count;
29508e1e 24#ifdef USE_CURL_MULTI
cc3530e8
MH
25static int max_requests = -1;
26static CURLM *curlm;
29508e1e
NH
27#endif
28#ifndef NO_CURL_EASY_DUPHANDLE
cc3530e8 29static CURL *curl_default;
29508e1e 30#endif
5424bc55
TRC
31
32#define PREV_BUF_SIZE 4096
33#define RANGE_HEADER_SIZE 30
34
29508e1e
NH
35char curl_errorstr[CURL_ERROR_SIZE];
36
cc3530e8 37static int curl_ssl_verify = -1;
4bc444eb 38static int curl_ssl_try;
4251ccbd 39static const char *ssl_cert;
ef52aafa 40#if LIBCURL_VERSION_NUM >= 0x070903
4251ccbd 41static const char *ssl_key;
29508e1e
NH
42#endif
43#if LIBCURL_VERSION_NUM >= 0x070908
4251ccbd 44static const char *ssl_capath;
29508e1e 45#endif
4251ccbd 46static const char *ssl_cainfo;
cc3530e8
MH
47static long curl_low_speed_limit = -1;
48static long curl_low_speed_time = -1;
4251ccbd
JH
49static int curl_ftp_no_epsv;
50static const char *curl_http_proxy;
bcfb95dd 51static const char *curl_cookie_file;
912b2acf 52static int curl_save_cookies;
2501aff8 53struct credential http_auth = CREDENTIAL_INIT;
a4ddbc33 54static int http_proactive_auth;
b1d1058c 55static const char *user_agent;
29508e1e 56
30dd9163
ML
57#if LIBCURL_VERSION_NUM >= 0x071700
58/* Use CURLOPT_KEYPASSWD as is */
59#elif LIBCURL_VERSION_NUM >= 0x070903
60#define CURLOPT_KEYPASSWD CURLOPT_SSLKEYPASSWD
61#else
62#define CURLOPT_KEYPASSWD CURLOPT_SSLCERTPASSWD
63#endif
64
148bb6a7 65static struct credential cert_auth = CREDENTIAL_INIT;
30dd9163 66static int ssl_cert_password_required;
4dbe6646 67#ifdef LIBCURL_CAN_HANDLE_AUTH_ANY
68static unsigned long http_auth_methods = CURLAUTH_ANY;
69#endif
30dd9163 70
cc3530e8 71static struct curl_slist *pragma_header;
5424bc55 72static struct curl_slist *no_pragma_header;
e9176745 73
4251ccbd 74static struct active_request_slot *active_queue_head;
29508e1e 75
f18604bb
YE
76static char *cached_accept_language;
77
a04ff3ec 78size_t fread_buffer(char *ptr, size_t eltsize, size_t nmemb, void *buffer_)
29508e1e
NH
79{
80 size_t size = eltsize * nmemb;
f444e528
JH
81 struct buffer *buffer = buffer_;
82
028c2976
MH
83 if (size > buffer->buf.len - buffer->posn)
84 size = buffer->buf.len - buffer->posn;
85 memcpy(ptr, buffer->buf.buf + buffer->posn, size);
29508e1e 86 buffer->posn += size;
028c2976 87
29508e1e
NH
88 return size;
89}
90
3944ba0c
MS
91#ifndef NO_CURL_IOCTL
92curlioerr ioctl_buffer(CURL *handle, int cmd, void *clientp)
93{
94 struct buffer *buffer = clientp;
95
96 switch (cmd) {
97 case CURLIOCMD_NOP:
98 return CURLIOE_OK;
99
100 case CURLIOCMD_RESTARTREAD:
101 buffer->posn = 0;
102 return CURLIOE_OK;
103
104 default:
105 return CURLIOE_UNKNOWNCMD;
106 }
107}
108#endif
109
a04ff3ec 110size_t fwrite_buffer(char *ptr, size_t eltsize, size_t nmemb, void *buffer_)
29508e1e
NH
111{
112 size_t size = eltsize * nmemb;
f444e528
JH
113 struct strbuf *buffer = buffer_;
114
028c2976 115 strbuf_add(buffer, ptr, size);
29508e1e
NH
116 return size;
117}
118
a04ff3ec 119size_t fwrite_null(char *ptr, size_t eltsize, size_t nmemb, void *strbuf)
29508e1e 120{
29508e1e
NH
121 return eltsize * nmemb;
122}
123
b90a3d7b
JH
124static void closedown_active_slot(struct active_request_slot *slot)
125{
126 active_requests--;
127 slot->in_use = 0;
128}
129
130static void finish_active_slot(struct active_request_slot *slot)
131{
132 closedown_active_slot(slot);
133 curl_easy_getinfo(slot->curl, CURLINFO_HTTP_CODE, &slot->http_code);
134
135 if (slot->finished != NULL)
136 (*slot->finished) = 1;
137
138 /* Store slot results so they can be read after the slot is reused */
139 if (slot->results != NULL) {
140 slot->results->curl_result = slot->curl_result;
141 slot->results->http_code = slot->http_code;
142#if LIBCURL_VERSION_NUM >= 0x070a08
143 curl_easy_getinfo(slot->curl, CURLINFO_HTTPAUTH_AVAIL,
144 &slot->results->auth_avail);
145#else
146 slot->results->auth_avail = 0;
147#endif
148 }
149
150 /* Run callback if appropriate */
151 if (slot->callback_func != NULL)
152 slot->callback_func(slot->callback_data);
153}
154
29508e1e
NH
155#ifdef USE_CURL_MULTI
156static void process_curl_messages(void)
157{
158 int num_messages;
159 struct active_request_slot *slot;
160 CURLMsg *curl_message = curl_multi_info_read(curlm, &num_messages);
161
162 while (curl_message != NULL) {
163 if (curl_message->msg == CURLMSG_DONE) {
164 int curl_result = curl_message->data.result;
165 slot = active_queue_head;
166 while (slot != NULL &&
167 slot->curl != curl_message->easy_handle)
168 slot = slot->next;
169 if (slot != NULL) {
170 curl_multi_remove_handle(curlm, slot->curl);
171 slot->curl_result = curl_result;
172 finish_active_slot(slot);
173 } else {
174 fprintf(stderr, "Received DONE message for unknown request!\n");
175 }
176 } else {
177 fprintf(stderr, "Unknown CURL message received: %d\n",
178 (int)curl_message->msg);
179 }
180 curl_message = curl_multi_info_read(curlm, &num_messages);
181 }
182}
183#endif
184
ef90d6d4 185static int http_options(const char *var, const char *value, void *cb)
29508e1e
NH
186{
187 if (!strcmp("http.sslverify", var)) {
7059cd99 188 curl_ssl_verify = git_config_bool(var, value);
29508e1e
NH
189 return 0;
190 }
7059cd99
JH
191 if (!strcmp("http.sslcert", var))
192 return git_config_string(&ssl_cert, var, value);
ef52aafa 193#if LIBCURL_VERSION_NUM >= 0x070903
7059cd99
JH
194 if (!strcmp("http.sslkey", var))
195 return git_config_string(&ssl_key, var, value);
29508e1e
NH
196#endif
197#if LIBCURL_VERSION_NUM >= 0x070908
7059cd99
JH
198 if (!strcmp("http.sslcapath", var))
199 return git_config_string(&ssl_capath, var, value);
29508e1e 200#endif
7059cd99
JH
201 if (!strcmp("http.sslcainfo", var))
202 return git_config_string(&ssl_cainfo, var, value);
754ae192 203 if (!strcmp("http.sslcertpasswordprotected", var)) {
3f4ccd2b 204 ssl_cert_password_required = git_config_bool(var, value);
754ae192
ML
205 return 0;
206 }
4bc444eb
MV
207 if (!strcmp("http.ssltry", var)) {
208 curl_ssl_try = git_config_bool(var, value);
209 return 0;
210 }
ad75ebe5
TRC
211 if (!strcmp("http.minsessions", var)) {
212 min_curl_sessions = git_config_int(var, value);
213#ifndef USE_CURL_MULTI
214 if (min_curl_sessions > 1)
215 min_curl_sessions = 1;
216#endif
217 return 0;
218 }
a6080a0a 219#ifdef USE_CURL_MULTI
29508e1e 220 if (!strcmp("http.maxrequests", var)) {
7059cd99 221 max_requests = git_config_int(var, value);
29508e1e
NH
222 return 0;
223 }
224#endif
29508e1e 225 if (!strcmp("http.lowspeedlimit", var)) {
7059cd99 226 curl_low_speed_limit = (long)git_config_int(var, value);
29508e1e
NH
227 return 0;
228 }
229 if (!strcmp("http.lowspeedtime", var)) {
7059cd99 230 curl_low_speed_time = (long)git_config_int(var, value);
29508e1e
NH
231 return 0;
232 }
233
3ea099d4
SK
234 if (!strcmp("http.noepsv", var)) {
235 curl_ftp_no_epsv = git_config_bool(var, value);
236 return 0;
237 }
7059cd99
JH
238 if (!strcmp("http.proxy", var))
239 return git_config_string(&curl_http_proxy, var, value);
3ea099d4 240
bcfb95dd
DB
241 if (!strcmp("http.cookiefile", var))
242 return git_config_string(&curl_cookie_file, var, value);
912b2acf
DB
243 if (!strcmp("http.savecookies", var)) {
244 curl_save_cookies = git_config_bool(var, value);
245 return 0;
246 }
bcfb95dd 247
de1a2fdd
SP
248 if (!strcmp("http.postbuffer", var)) {
249 http_post_buffer = git_config_int(var, value);
250 if (http_post_buffer < LARGE_PACKET_MAX)
251 http_post_buffer = LARGE_PACKET_MAX;
252 return 0;
253 }
254
b1d1058c
SO
255 if (!strcmp("http.useragent", var))
256 return git_config_string(&user_agent, var, value);
257
29508e1e 258 /* Fall back on the default ones */
ef90d6d4 259 return git_default_config(var, value, cb);
29508e1e
NH
260}
261
c33976cb
JH
262static void init_curl_http_auth(CURL *result)
263{
6f4c347c
JK
264 if (!http_auth.username)
265 return;
266
267 credential_fill(&http_auth);
268
269#if LIBCURL_VERSION_NUM >= 0x071301
270 curl_easy_setopt(result, CURLOPT_USERNAME, http_auth.username);
271 curl_easy_setopt(result, CURLOPT_PASSWORD, http_auth.password);
272#else
273 {
aa0834a0 274 static struct strbuf up = STRBUF_INIT;
a94cf2cb
BC
275 /*
276 * Note that we assume we only ever have a single set of
277 * credentials in a given program run, so we do not have
278 * to worry about updating this buffer, only setting its
279 * initial value.
280 */
281 if (!up.len)
282 strbuf_addf(&up, "%s:%s",
283 http_auth.username, http_auth.password);
aa0834a0 284 curl_easy_setopt(result, CURLOPT_USERPWD, up.buf);
c33976cb 285 }
6f4c347c 286#endif
c33976cb
JH
287}
288
30dd9163
ML
289static int has_cert_password(void)
290{
30dd9163
ML
291 if (ssl_cert == NULL || ssl_cert_password_required != 1)
292 return 0;
148bb6a7
JK
293 if (!cert_auth.password) {
294 cert_auth.protocol = xstrdup("cert");
75e9a405 295 cert_auth.username = xstrdup("");
148bb6a7
JK
296 cert_auth.path = xstrdup(ssl_cert);
297 credential_fill(&cert_auth);
298 }
299 return 1;
30dd9163
ML
300}
301
47ce1153
JK
302#if LIBCURL_VERSION_NUM >= 0x071900
303static void set_curl_keepalive(CURL *c)
304{
305 curl_easy_setopt(c, CURLOPT_TCP_KEEPALIVE, 1);
306}
307
308#elif LIBCURL_VERSION_NUM >= 0x071000
a15d069a
EW
309static int sockopt_callback(void *client, curl_socket_t fd, curlsocktype type)
310{
311 int ka = 1;
312 int rc;
313 socklen_t len = (socklen_t)sizeof(ka);
314
315 if (type != CURLSOCKTYPE_IPCXN)
316 return 0;
317
318 rc = setsockopt(fd, SOL_SOCKET, SO_KEEPALIVE, (void *)&ka, len);
319 if (rc < 0)
320 warning("unable to set SO_KEEPALIVE on socket %s",
321 strerror(errno));
322
323 return 0; /* CURL_SOCKOPT_OK only exists since curl 7.21.5 */
324}
325
47ce1153
JK
326static void set_curl_keepalive(CURL *c)
327{
328 curl_easy_setopt(c, CURLOPT_SOCKOPTFUNCTION, sockopt_callback);
329}
330
331#else
332static void set_curl_keepalive(CURL *c)
333{
334 /* not supported on older curl versions */
335}
336#endif
337
4251ccbd 338static CURL *get_curl_handle(void)
11979b98 339{
4251ccbd 340 CURL *result = curl_easy_init();
f4113cac 341 long allowed_protocols = 0;
11979b98 342
faa3807c
BR
343 if (!result)
344 die("curl_easy_init failed");
345
a5ccc597
JH
346 if (!curl_ssl_verify) {
347 curl_easy_setopt(result, CURLOPT_SSL_VERIFYPEER, 0);
348 curl_easy_setopt(result, CURLOPT_SSL_VERIFYHOST, 0);
349 } else {
350 /* Verify authenticity of the peer's certificate */
351 curl_easy_setopt(result, CURLOPT_SSL_VERIFYPEER, 1);
352 /* The name in the cert must match whom we tried to connect */
353 curl_easy_setopt(result, CURLOPT_SSL_VERIFYHOST, 2);
354 }
355
11979b98
JH
356#if LIBCURL_VERSION_NUM >= 0x070907
357 curl_easy_setopt(result, CURLOPT_NETRC, CURL_NETRC_OPTIONAL);
358#endif
b8ac9230 359#ifdef LIBCURL_CAN_HANDLE_AUTH_ANY
525ecd26 360 curl_easy_setopt(result, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
b8ac9230 361#endif
11979b98 362
a4ddbc33
JK
363 if (http_proactive_auth)
364 init_curl_http_auth(result);
365
11979b98
JH
366 if (ssl_cert != NULL)
367 curl_easy_setopt(result, CURLOPT_SSLCERT, ssl_cert);
30dd9163 368 if (has_cert_password())
148bb6a7 369 curl_easy_setopt(result, CURLOPT_KEYPASSWD, cert_auth.password);
ef52aafa 370#if LIBCURL_VERSION_NUM >= 0x070903
11979b98
JH
371 if (ssl_key != NULL)
372 curl_easy_setopt(result, CURLOPT_SSLKEY, ssl_key);
373#endif
374#if LIBCURL_VERSION_NUM >= 0x070908
375 if (ssl_capath != NULL)
376 curl_easy_setopt(result, CURLOPT_CAPATH, ssl_capath);
377#endif
378 if (ssl_cainfo != NULL)
379 curl_easy_setopt(result, CURLOPT_CAINFO, ssl_cainfo);
11979b98
JH
380
381 if (curl_low_speed_limit > 0 && curl_low_speed_time > 0) {
382 curl_easy_setopt(result, CURLOPT_LOW_SPEED_LIMIT,
383 curl_low_speed_limit);
384 curl_easy_setopt(result, CURLOPT_LOW_SPEED_TIME,
385 curl_low_speed_time);
386 }
387
388 curl_easy_setopt(result, CURLOPT_FOLLOWLOCATION, 1);
b2581164 389 curl_easy_setopt(result, CURLOPT_MAXREDIRS, 20);
311e2ea0
TRC
390#if LIBCURL_VERSION_NUM >= 0x071301
391 curl_easy_setopt(result, CURLOPT_POSTREDIR, CURL_REDIR_POST_ALL);
392#elif LIBCURL_VERSION_NUM >= 0x071101
393 curl_easy_setopt(result, CURLOPT_POST301, 1);
394#endif
f4113cac
BB
395#if LIBCURL_VERSION_NUM >= 0x071304
396 if (is_transport_allowed("http"))
397 allowed_protocols |= CURLPROTO_HTTP;
398 if (is_transport_allowed("https"))
399 allowed_protocols |= CURLPROTO_HTTPS;
400 if (is_transport_allowed("ftp"))
401 allowed_protocols |= CURLPROTO_FTP;
402 if (is_transport_allowed("ftps"))
403 allowed_protocols |= CURLPROTO_FTPS;
404 curl_easy_setopt(result, CURLOPT_REDIR_PROTOCOLS, allowed_protocols);
405#else
406 if (transport_restrict_protocols())
407 warning("protocol restrictions not applied to curl redirects because\n"
408 "your curl version is too old (>= 7.19.4)");
409#endif
11979b98 410
7982d74e
MW
411 if (getenv("GIT_CURL_VERBOSE"))
412 curl_easy_setopt(result, CURLOPT_VERBOSE, 1);
413
b1d1058c 414 curl_easy_setopt(result, CURLOPT_USERAGENT,
745c7c8e 415 user_agent ? user_agent : git_user_agent());
20fc9bc5 416
3ea099d4
SK
417 if (curl_ftp_no_epsv)
418 curl_easy_setopt(result, CURLOPT_FTP_USE_EPSV, 0);
419
4bc444eb
MV
420#ifdef CURLOPT_USE_SSL
421 if (curl_ssl_try)
422 curl_easy_setopt(result, CURLOPT_USE_SSL, CURLUSESSL_TRY);
423#endif
424
dd613997 425 if (curl_http_proxy) {
9c5665aa 426 curl_easy_setopt(result, CURLOPT_PROXY, curl_http_proxy);
6d7afe07
PT
427#if LIBCURL_VERSION_NUM >= 0x071800
428 if (starts_with(curl_http_proxy, "socks5"))
429 curl_easy_setopt(result,
430 CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5);
431 else if (starts_with(curl_http_proxy, "socks4a"))
432 curl_easy_setopt(result,
433 CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4A);
434 else if (starts_with(curl_http_proxy, "socks"))
435 curl_easy_setopt(result,
436 CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4);
437#endif
5841520b 438 }
1c2dbf20 439#if LIBCURL_VERSION_NUM >= 0x070a07
5841520b 440 curl_easy_setopt(result, CURLOPT_PROXYAUTH, CURLAUTH_ANY);
1c2dbf20 441#endif
9c5665aa 442
47ce1153 443 set_curl_keepalive(result);
a15d069a 444
11979b98
JH
445 return result;
446}
447
7059cd99
JH
448static void set_from_env(const char **var, const char *envname)
449{
450 const char *val = getenv(envname);
451 if (val)
452 *var = val;
453}
454
a4ddbc33 455void http_init(struct remote *remote, const char *url, int proactive_auth)
29508e1e
NH
456{
457 char *low_speed_limit;
458 char *low_speed_time;
6a56993b
KM
459 char *normalized_url;
460 struct urlmatch_config config = { STRING_LIST_INIT_DUP };
461
462 config.section = "http";
463 config.key = NULL;
464 config.collect_fn = http_options;
465 config.cascade_fn = git_default_config;
466 config.cb = NULL;
29508e1e 467
e9176745 468 http_is_verbose = 0;
6a56993b 469 normalized_url = url_normalize(url, &config.url);
e9176745 470
6a56993b
KM
471 git_config(urlmatch_config_entry, &config);
472 free(normalized_url);
7059cd99 473
faa3807c
BR
474 if (curl_global_init(CURL_GLOBAL_ALL) != CURLE_OK)
475 die("curl_global_init failed");
29508e1e 476
a4ddbc33
JK
477 http_proactive_auth = proactive_auth;
478
9fc6440d
MH
479 if (remote && remote->http_proxy)
480 curl_http_proxy = xstrdup(remote->http_proxy);
481
29508e1e 482 pragma_header = curl_slist_append(pragma_header, "Pragma: no-cache");
e9176745 483 no_pragma_header = curl_slist_append(no_pragma_header, "Pragma:");
29508e1e
NH
484
485#ifdef USE_CURL_MULTI
486 {
487 char *http_max_requests = getenv("GIT_HTTP_MAX_REQUESTS");
488 if (http_max_requests != NULL)
489 max_requests = atoi(http_max_requests);
490 }
491
492 curlm = curl_multi_init();
8837eb47
JK
493 if (!curlm)
494 die("curl_multi_init failed");
29508e1e
NH
495#endif
496
497 if (getenv("GIT_SSL_NO_VERIFY"))
498 curl_ssl_verify = 0;
499
7059cd99 500 set_from_env(&ssl_cert, "GIT_SSL_CERT");
ef52aafa 501#if LIBCURL_VERSION_NUM >= 0x070903
7059cd99 502 set_from_env(&ssl_key, "GIT_SSL_KEY");
29508e1e
NH
503#endif
504#if LIBCURL_VERSION_NUM >= 0x070908
7059cd99 505 set_from_env(&ssl_capath, "GIT_SSL_CAPATH");
29508e1e 506#endif
7059cd99 507 set_from_env(&ssl_cainfo, "GIT_SSL_CAINFO");
29508e1e 508
b1d1058c
SO
509 set_from_env(&user_agent, "GIT_HTTP_USER_AGENT");
510
29508e1e
NH
511 low_speed_limit = getenv("GIT_HTTP_LOW_SPEED_LIMIT");
512 if (low_speed_limit != NULL)
513 curl_low_speed_limit = strtol(low_speed_limit, NULL, 10);
514 low_speed_time = getenv("GIT_HTTP_LOW_SPEED_TIME");
515 if (low_speed_time != NULL)
516 curl_low_speed_time = strtol(low_speed_time, NULL, 10);
517
29508e1e
NH
518 if (curl_ssl_verify == -1)
519 curl_ssl_verify = 1;
520
ad75ebe5 521 curl_session_count = 0;
29508e1e
NH
522#ifdef USE_CURL_MULTI
523 if (max_requests < 1)
524 max_requests = DEFAULT_MAX_REQUESTS;
525#endif
526
3ea099d4
SK
527 if (getenv("GIT_CURL_FTP_NO_EPSV"))
528 curl_ftp_no_epsv = 1;
529
deba4937 530 if (url) {
148bb6a7 531 credential_from_url(&http_auth, url);
754ae192
ML
532 if (!ssl_cert_password_required &&
533 getenv("GIT_SSL_CERT_PASSWORD_PROTECTED") &&
59556548 534 starts_with(url, "https://"))
30dd9163
ML
535 ssl_cert_password_required = 1;
536 }
c33976cb 537
29508e1e
NH
538#ifndef NO_CURL_EASY_DUPHANDLE
539 curl_default = get_curl_handle();
540#endif
541}
542
543void http_cleanup(void)
544{
545 struct active_request_slot *slot = active_queue_head;
29508e1e
NH
546
547 while (slot != NULL) {
3278cd0a 548 struct active_request_slot *next = slot->next;
f23d1f76 549 if (slot->curl != NULL) {
29508e1e 550#ifdef USE_CURL_MULTI
f23d1f76 551 curl_multi_remove_handle(curlm, slot->curl);
29508e1e 552#endif
29508e1e 553 curl_easy_cleanup(slot->curl);
f23d1f76 554 }
3278cd0a
SP
555 free(slot);
556 slot = next;
29508e1e 557 }
3278cd0a 558 active_queue_head = NULL;
29508e1e
NH
559
560#ifndef NO_CURL_EASY_DUPHANDLE
561 curl_easy_cleanup(curl_default);
562#endif
563
564#ifdef USE_CURL_MULTI
565 curl_multi_cleanup(curlm);
566#endif
567 curl_global_cleanup();
b3ca4e4e
NH
568
569 curl_slist_free_all(pragma_header);
3278cd0a 570 pragma_header = NULL;
9fc6440d 571
e9176745
TRC
572 curl_slist_free_all(no_pragma_header);
573 no_pragma_header = NULL;
574
9fc6440d 575 if (curl_http_proxy) {
e4a80ecf 576 free((void *)curl_http_proxy);
9fc6440d
MH
577 curl_http_proxy = NULL;
578 }
30dd9163 579
148bb6a7
JK
580 if (cert_auth.password != NULL) {
581 memset(cert_auth.password, 0, strlen(cert_auth.password));
582 free(cert_auth.password);
583 cert_auth.password = NULL;
30dd9163
ML
584 }
585 ssl_cert_password_required = 0;
f18604bb
YE
586
587 free(cached_accept_language);
588 cached_accept_language = NULL;
29508e1e
NH
589}
590
29508e1e
NH
591struct active_request_slot *get_active_slot(void)
592{
593 struct active_request_slot *slot = active_queue_head;
594 struct active_request_slot *newslot;
595
596#ifdef USE_CURL_MULTI
597 int num_transfers;
598
599 /* Wait for a slot to open up if the queue is full */
600 while (active_requests >= max_requests) {
601 curl_multi_perform(curlm, &num_transfers);
4251ccbd 602 if (num_transfers < active_requests)
29508e1e 603 process_curl_messages();
29508e1e
NH
604 }
605#endif
606
4251ccbd 607 while (slot != NULL && slot->in_use)
29508e1e 608 slot = slot->next;
4251ccbd 609
29508e1e
NH
610 if (slot == NULL) {
611 newslot = xmalloc(sizeof(*newslot));
612 newslot->curl = NULL;
613 newslot->in_use = 0;
614 newslot->next = NULL;
615
616 slot = active_queue_head;
617 if (slot == NULL) {
618 active_queue_head = newslot;
619 } else {
4251ccbd 620 while (slot->next != NULL)
29508e1e 621 slot = slot->next;
29508e1e
NH
622 slot->next = newslot;
623 }
624 slot = newslot;
625 }
626
627 if (slot->curl == NULL) {
628#ifdef NO_CURL_EASY_DUPHANDLE
629 slot->curl = get_curl_handle();
630#else
631 slot->curl = curl_easy_duphandle(curl_default);
632#endif
ad75ebe5 633 curl_session_count++;
29508e1e
NH
634 }
635
636 active_requests++;
637 slot->in_use = 1;
c8568e13 638 slot->results = NULL;
baa7b67d 639 slot->finished = NULL;
29508e1e
NH
640 slot->callback_data = NULL;
641 slot->callback_func = NULL;
bcfb95dd 642 curl_easy_setopt(slot->curl, CURLOPT_COOKIEFILE, curl_cookie_file);
912b2acf
DB
643 if (curl_save_cookies)
644 curl_easy_setopt(slot->curl, CURLOPT_COOKIEJAR, curl_cookie_file);
29508e1e 645 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, pragma_header);
29508e1e 646 curl_easy_setopt(slot->curl, CURLOPT_ERRORBUFFER, curl_errorstr);
9094950d
NH
647 curl_easy_setopt(slot->curl, CURLOPT_CUSTOMREQUEST, NULL);
648 curl_easy_setopt(slot->curl, CURLOPT_READFUNCTION, NULL);
649 curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, NULL);
1e41827d 650 curl_easy_setopt(slot->curl, CURLOPT_POSTFIELDS, NULL);
9094950d
NH
651 curl_easy_setopt(slot->curl, CURLOPT_UPLOAD, 0);
652 curl_easy_setopt(slot->curl, CURLOPT_HTTPGET, 1);
b793acf1 653 curl_easy_setopt(slot->curl, CURLOPT_FAILONERROR, 1);
4dbe6646 654#ifdef LIBCURL_CAN_HANDLE_AUTH_ANY
655 curl_easy_setopt(slot->curl, CURLOPT_HTTPAUTH, http_auth_methods);
656#endif
dfa1725a
JK
657 if (http_auth.password)
658 init_curl_http_auth(slot->curl);
29508e1e
NH
659
660 return slot;
661}
662
663int start_active_slot(struct active_request_slot *slot)
664{
665#ifdef USE_CURL_MULTI
666 CURLMcode curlm_result = curl_multi_add_handle(curlm, slot->curl);
45c17412 667 int num_transfers;
29508e1e
NH
668
669 if (curlm_result != CURLM_OK &&
670 curlm_result != CURLM_CALL_MULTI_PERFORM) {
671 active_requests--;
672 slot->in_use = 0;
673 return 0;
674 }
45c17412
DB
675
676 /*
677 * We know there must be something to do, since we just added
678 * something.
679 */
680 curl_multi_perform(curlm, &num_transfers);
29508e1e
NH
681#endif
682 return 1;
683}
684
685#ifdef USE_CURL_MULTI
fc57b6aa
DB
686struct fill_chain {
687 void *data;
688 int (*fill)(void *);
689 struct fill_chain *next;
690};
691
4251ccbd 692static struct fill_chain *fill_cfg;
fc57b6aa
DB
693
694void add_fill_function(void *data, int (*fill)(void *))
695{
e8eec71d 696 struct fill_chain *new = xmalloc(sizeof(*new));
fc57b6aa
DB
697 struct fill_chain **linkp = &fill_cfg;
698 new->data = data;
699 new->fill = fill;
700 new->next = NULL;
701 while (*linkp)
702 linkp = &(*linkp)->next;
703 *linkp = new;
704}
705
45c17412
DB
706void fill_active_slots(void)
707{
708 struct active_request_slot *slot = active_queue_head;
709
fc57b6aa
DB
710 while (active_requests < max_requests) {
711 struct fill_chain *fill;
712 for (fill = fill_cfg; fill; fill = fill->next)
713 if (fill->fill(fill->data))
714 break;
715
716 if (!fill)
45c17412 717 break;
fc57b6aa 718 }
45c17412
DB
719
720 while (slot != NULL) {
ad75ebe5
TRC
721 if (!slot->in_use && slot->curl != NULL
722 && curl_session_count > min_curl_sessions) {
45c17412
DB
723 curl_easy_cleanup(slot->curl);
724 slot->curl = NULL;
ad75ebe5 725 curl_session_count--;
45c17412
DB
726 }
727 slot = slot->next;
728 }
729}
730
29508e1e
NH
731void step_active_slots(void)
732{
733 int num_transfers;
734 CURLMcode curlm_result;
735
736 do {
737 curlm_result = curl_multi_perform(curlm, &num_transfers);
738 } while (curlm_result == CURLM_CALL_MULTI_PERFORM);
739 if (num_transfers < active_requests) {
740 process_curl_messages();
741 fill_active_slots();
742 }
743}
744#endif
745
746void run_active_slot(struct active_request_slot *slot)
747{
748#ifdef USE_CURL_MULTI
29508e1e
NH
749 fd_set readfds;
750 fd_set writefds;
751 fd_set excfds;
752 int max_fd;
753 struct timeval select_timeout;
baa7b67d 754 int finished = 0;
29508e1e 755
baa7b67d
NH
756 slot->finished = &finished;
757 while (!finished) {
29508e1e
NH
758 step_active_slots();
759
df26c471 760 if (slot->in_use) {
eb56c821
MF
761#if LIBCURL_VERSION_NUM >= 0x070f04
762 long curl_timeout;
763 curl_multi_timeout(curlm, &curl_timeout);
764 if (curl_timeout == 0) {
765 continue;
766 } else if (curl_timeout == -1) {
767 select_timeout.tv_sec = 0;
768 select_timeout.tv_usec = 50000;
769 } else {
770 select_timeout.tv_sec = curl_timeout / 1000;
771 select_timeout.tv_usec = (curl_timeout % 1000) * 1000;
772 }
773#else
774 select_timeout.tv_sec = 0;
775 select_timeout.tv_usec = 50000;
776#endif
29508e1e 777
6f9dd67f 778 max_fd = -1;
29508e1e
NH
779 FD_ZERO(&readfds);
780 FD_ZERO(&writefds);
781 FD_ZERO(&excfds);
6f9dd67f 782 curl_multi_fdset(curlm, &readfds, &writefds, &excfds, &max_fd);
eb56c821 783
7202b81f
SZ
784 /*
785 * It can happen that curl_multi_timeout returns a pathologically
786 * long timeout when curl_multi_fdset returns no file descriptors
787 * to read. See commit message for more details.
788 */
789 if (max_fd < 0 &&
790 (select_timeout.tv_sec > 0 ||
791 select_timeout.tv_usec > 50000)) {
792 select_timeout.tv_sec = 0;
793 select_timeout.tv_usec = 50000;
794 }
795
6f9dd67f 796 select(max_fd+1, &readfds, &writefds, &excfds, &select_timeout);
29508e1e
NH
797 }
798 }
799#else
800 while (slot->in_use) {
801 slot->curl_result = curl_easy_perform(slot->curl);
802 finish_active_slot(slot);
803 }
804#endif
805}
806
83e41e2e 807static void release_active_slot(struct active_request_slot *slot)
53f31389
MW
808{
809 closedown_active_slot(slot);
ad75ebe5 810 if (slot->curl && curl_session_count > min_curl_sessions) {
b3ca4e4e 811#ifdef USE_CURL_MULTI
53f31389 812 curl_multi_remove_handle(curlm, slot->curl);
b3ca4e4e 813#endif
53f31389
MW
814 curl_easy_cleanup(slot->curl);
815 slot->curl = NULL;
ad75ebe5 816 curl_session_count--;
53f31389 817 }
b3ca4e4e 818#ifdef USE_CURL_MULTI
53f31389 819 fill_active_slots();
b3ca4e4e 820#endif
53f31389
MW
821}
822
29508e1e
NH
823void finish_all_active_slots(void)
824{
825 struct active_request_slot *slot = active_queue_head;
826
827 while (slot != NULL)
828 if (slot->in_use) {
829 run_active_slot(slot);
830 slot = active_queue_head;
831 } else {
832 slot = slot->next;
833 }
834}
d7e92806 835
5ace994f 836/* Helpers for modifying and creating URLs */
d7e92806
MH
837static inline int needs_quote(int ch)
838{
839 if (((ch >= 'A') && (ch <= 'Z'))
840 || ((ch >= 'a') && (ch <= 'z'))
841 || ((ch >= '0') && (ch <= '9'))
842 || (ch == '/')
843 || (ch == '-')
844 || (ch == '.'))
845 return 0;
846 return 1;
847}
848
d7e92806
MH
849static char *quote_ref_url(const char *base, const char *ref)
850{
113106e0 851 struct strbuf buf = STRBUF_INIT;
d7e92806 852 const char *cp;
113106e0 853 int ch;
d7e92806 854
5ace994f 855 end_url_with_slash(&buf, base);
113106e0
TRC
856
857 for (cp = ref; (ch = *cp) != 0; cp++)
d7e92806 858 if (needs_quote(ch))
113106e0 859 strbuf_addf(&buf, "%%%02x", ch);
d7e92806 860 else
113106e0 861 strbuf_addch(&buf, *cp);
d7e92806 862
113106e0 863 return strbuf_detach(&buf, NULL);
d7e92806
MH
864}
865
5424bc55
TRC
866void append_remote_object_url(struct strbuf *buf, const char *url,
867 const char *hex,
868 int only_two_digit_prefix)
869{
800324c3
TRC
870 end_url_with_slash(buf, url);
871
872 strbuf_addf(buf, "objects/%.*s/", 2, hex);
5424bc55
TRC
873 if (!only_two_digit_prefix)
874 strbuf_addf(buf, "%s", hex+2);
875}
876
877char *get_remote_object_url(const char *url, const char *hex,
878 int only_two_digit_prefix)
879{
880 struct strbuf buf = STRBUF_INIT;
881 append_remote_object_url(&buf, url, hex, only_two_digit_prefix);
882 return strbuf_detach(&buf, NULL);
883}
884
b90a3d7b 885static int handle_curl_result(struct slot_results *results)
88097030 886{
6d052d78
JK
887 /*
888 * If we see a failing http code with CURLE_OK, we have turned off
889 * FAILONERROR (to keep the server's custom error response), and should
890 * translate the code into failure here.
891 */
892 if (results->curl_result == CURLE_OK &&
893 results->http_code >= 400) {
894 results->curl_result = CURLE_HTTP_RETURNED_ERROR;
895 /*
896 * Normally curl will already have put the "reason phrase"
897 * from the server into curl_errorstr; unfortunately without
898 * FAILONERROR it is lost, so we can give only the numeric
899 * status code.
900 */
901 snprintf(curl_errorstr, sizeof(curl_errorstr),
902 "The requested URL returned error: %ld",
903 results->http_code);
904 }
905
88097030
JK
906 if (results->curl_result == CURLE_OK) {
907 credential_approve(&http_auth);
908 return HTTP_OK;
909 } else if (missing_target(results))
910 return HTTP_MISSING_TARGET;
911 else if (results->http_code == 401) {
912 if (http_auth.username && http_auth.password) {
913 credential_reject(&http_auth);
914 return HTTP_NOAUTH;
915 } else {
4dbe6646 916#ifdef LIBCURL_CAN_HANDLE_AUTH_ANY
917 http_auth_methods &= ~CURLAUTH_GSSNEGOTIATE;
918#endif
88097030
JK
919 return HTTP_REAUTH;
920 }
921 } else {
3503e9ab 922#if LIBCURL_VERSION_NUM >= 0x070c00
88097030
JK
923 if (!curl_errorstr[0])
924 strlcpy(curl_errorstr,
925 curl_easy_strerror(results->curl_result),
926 sizeof(curl_errorstr));
3503e9ab 927#endif
88097030
JK
928 return HTTP_ERROR;
929 }
930}
931
beed336c
JK
932int run_one_slot(struct active_request_slot *slot,
933 struct slot_results *results)
934{
935 slot->results = results;
936 if (!start_active_slot(slot)) {
937 snprintf(curl_errorstr, sizeof(curl_errorstr),
938 "failed to start HTTP request");
939 return HTTP_START_FAILED;
940 }
941
942 run_active_slot(slot);
943 return handle_curl_result(results);
944}
945
132b70a2
JK
946static CURLcode curlinfo_strbuf(CURL *curl, CURLINFO info, struct strbuf *buf)
947{
948 char *ptr;
949 CURLcode ret;
950
951 strbuf_reset(buf);
952 ret = curl_easy_getinfo(curl, info, &ptr);
953 if (!ret && ptr)
954 strbuf_addstr(buf, ptr);
955 return ret;
956}
957
e3131626
JK
958/*
959 * Check for and extract a content-type parameter. "raw"
960 * should be positioned at the start of the potential
961 * parameter, with any whitespace already removed.
962 *
963 * "name" is the name of the parameter. The value is appended
964 * to "out".
965 */
966static int extract_param(const char *raw, const char *name,
967 struct strbuf *out)
968{
969 size_t len = strlen(name);
970
971 if (strncasecmp(raw, name, len))
972 return -1;
973 raw += len;
974
975 if (*raw != '=')
976 return -1;
977 raw++;
978
f34a655d 979 while (*raw && !isspace(*raw) && *raw != ';')
e3131626
JK
980 strbuf_addch(out, *raw++);
981 return 0;
982}
983
bf197fd7
JK
984/*
985 * Extract a normalized version of the content type, with any
986 * spaces suppressed, all letters lowercased, and no trailing ";"
987 * or parameters.
988 *
989 * Note that we will silently remove even invalid whitespace. For
990 * example, "text / plain" is specifically forbidden by RFC 2616,
991 * but "text/plain" is the only reasonable output, and this keeps
992 * our code simple.
993 *
e3131626
JK
994 * If the "charset" argument is not NULL, store the value of any
995 * charset parameter there.
996 *
bf197fd7 997 * Example:
e3131626 998 * "TEXT/PLAIN; charset=utf-8" -> "text/plain", "utf-8"
bf197fd7
JK
999 * "text / plain" -> "text/plain"
1000 */
e3131626
JK
1001static void extract_content_type(struct strbuf *raw, struct strbuf *type,
1002 struct strbuf *charset)
bf197fd7
JK
1003{
1004 const char *p;
1005
1006 strbuf_reset(type);
1007 strbuf_grow(type, raw->len);
1008 for (p = raw->buf; *p; p++) {
1009 if (isspace(*p))
1010 continue;
e3131626
JK
1011 if (*p == ';') {
1012 p++;
bf197fd7 1013 break;
e3131626 1014 }
bf197fd7
JK
1015 strbuf_addch(type, tolower(*p));
1016 }
e3131626
JK
1017
1018 if (!charset)
1019 return;
1020
1021 strbuf_reset(charset);
1022 while (*p) {
f34a655d 1023 while (isspace(*p) || *p == ';')
e3131626
JK
1024 p++;
1025 if (!extract_param(p, "charset", charset))
1026 return;
1027 while (*p && !isspace(*p))
1028 p++;
1029 }
c553fd1c
JK
1030
1031 if (!charset->len && starts_with(type->buf, "text/"))
1032 strbuf_addstr(charset, "ISO-8859-1");
bf197fd7
JK
1033}
1034
f18604bb
YE
1035static void write_accept_language(struct strbuf *buf)
1036{
1037 /*
1038 * MAX_DECIMAL_PLACES must not be larger than 3. If it is larger than
1039 * that, q-value will be smaller than 0.001, the minimum q-value the
1040 * HTTP specification allows. See
1041 * http://tools.ietf.org/html/rfc7231#section-5.3.1 for q-value.
1042 */
1043 const int MAX_DECIMAL_PLACES = 3;
1044 const int MAX_LANGUAGE_TAGS = 1000;
1045 const int MAX_ACCEPT_LANGUAGE_HEADER_SIZE = 4000;
1046 char **language_tags = NULL;
1047 int num_langs = 0;
1048 const char *s = get_preferred_languages();
1049 int i;
1050 struct strbuf tag = STRBUF_INIT;
1051
1052 /* Don't add Accept-Language header if no language is preferred. */
1053 if (!s)
1054 return;
1055
1056 /*
1057 * Split the colon-separated string of preferred languages into
1058 * language_tags array.
1059 */
1060 do {
1061 /* collect language tag */
1062 for (; *s && (isalnum(*s) || *s == '_'); s++)
1063 strbuf_addch(&tag, *s == '_' ? '-' : *s);
1064
1065 /* skip .codeset, @modifier and any other unnecessary parts */
1066 while (*s && *s != ':')
1067 s++;
1068
1069 if (tag.len) {
1070 num_langs++;
1071 REALLOC_ARRAY(language_tags, num_langs);
1072 language_tags[num_langs - 1] = strbuf_detach(&tag, NULL);
1073 if (num_langs >= MAX_LANGUAGE_TAGS - 1) /* -1 for '*' */
1074 break;
1075 }
1076 } while (*s++);
1077
1078 /* write Accept-Language header into buf */
1079 if (num_langs) {
1080 int last_buf_len = 0;
1081 int max_q;
1082 int decimal_places;
1083 char q_format[32];
1084
1085 /* add '*' */
1086 REALLOC_ARRAY(language_tags, num_langs + 1);
1087 language_tags[num_langs++] = "*"; /* it's OK; this won't be freed */
1088
1089 /* compute decimal_places */
1090 for (max_q = 1, decimal_places = 0;
1091 max_q < num_langs && decimal_places <= MAX_DECIMAL_PLACES;
1092 decimal_places++, max_q *= 10)
1093 ;
1094
1095 sprintf(q_format, ";q=0.%%0%dd", decimal_places);
1096
1097 strbuf_addstr(buf, "Accept-Language: ");
1098
1099 for (i = 0; i < num_langs; i++) {
1100 if (i > 0)
1101 strbuf_addstr(buf, ", ");
1102
1103 strbuf_addstr(buf, language_tags[i]);
1104
1105 if (i > 0)
1106 strbuf_addf(buf, q_format, max_q - i);
1107
1108 if (buf->len > MAX_ACCEPT_LANGUAGE_HEADER_SIZE) {
1109 strbuf_remove(buf, last_buf_len, buf->len - last_buf_len);
1110 break;
1111 }
1112
1113 last_buf_len = buf->len;
1114 }
1115 }
1116
1117 /* free language tags -- last one is a static '*' */
1118 for (i = 0; i < num_langs - 1; i++)
1119 free(language_tags[i]);
1120 free(language_tags);
1121}
1122
1123/*
1124 * Get an Accept-Language header which indicates user's preferred languages.
1125 *
1126 * Examples:
1127 * LANGUAGE= -> ""
1128 * LANGUAGE=ko:en -> "Accept-Language: ko, en; q=0.9, *; q=0.1"
1129 * LANGUAGE=ko_KR.UTF-8:sr@latin -> "Accept-Language: ko-KR, sr; q=0.9, *; q=0.1"
1130 * LANGUAGE=ko LANG=en_US.UTF-8 -> "Accept-Language: ko, *; q=0.1"
1131 * LANGUAGE= LANG=en_US.UTF-8 -> "Accept-Language: en-US, *; q=0.1"
1132 * LANGUAGE= LANG=C -> ""
1133 */
1134static const char *get_accept_language(void)
1135{
1136 if (!cached_accept_language) {
1137 struct strbuf buf = STRBUF_INIT;
1138 write_accept_language(&buf);
1139 if (buf.len > 0)
1140 cached_accept_language = strbuf_detach(&buf, NULL);
1141 }
1142
1143 return cached_accept_language;
1144}
1145
e929cd20
MH
1146/* http_request() targets */
1147#define HTTP_REQUEST_STRBUF 0
1148#define HTTP_REQUEST_FILE 1
1149
1bbcc224
JK
1150static int http_request(const char *url,
1151 void *result, int target,
1152 const struct http_get_options *options)
e929cd20
MH
1153{
1154 struct active_request_slot *slot;
1155 struct slot_results results;
1156 struct curl_slist *headers = NULL;
1157 struct strbuf buf = STRBUF_INIT;
f18604bb 1158 const char *accept_language;
e929cd20
MH
1159 int ret;
1160
1161 slot = get_active_slot();
e929cd20
MH
1162 curl_easy_setopt(slot->curl, CURLOPT_HTTPGET, 1);
1163
1164 if (result == NULL) {
1165 curl_easy_setopt(slot->curl, CURLOPT_NOBODY, 1);
1166 } else {
1167 curl_easy_setopt(slot->curl, CURLOPT_NOBODY, 0);
1168 curl_easy_setopt(slot->curl, CURLOPT_FILE, result);
1169
1170 if (target == HTTP_REQUEST_FILE) {
1171 long posn = ftell(result);
1172 curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
1173 fwrite);
1174 if (posn > 0) {
1175 strbuf_addf(&buf, "Range: bytes=%ld-", posn);
1176 headers = curl_slist_append(headers, buf.buf);
1177 strbuf_reset(&buf);
1178 }
e929cd20
MH
1179 } else
1180 curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION,
1181 fwrite_buffer);
1182 }
1183
f18604bb
YE
1184 accept_language = get_accept_language();
1185
1186 if (accept_language)
1187 headers = curl_slist_append(headers, accept_language);
1188
e929cd20 1189 strbuf_addstr(&buf, "Pragma:");
1bbcc224 1190 if (options && options->no_cache)
e929cd20 1191 strbuf_addstr(&buf, " no-cache");
1bbcc224 1192 if (options && options->keep_error)
6d052d78 1193 curl_easy_setopt(slot->curl, CURLOPT_FAILONERROR, 0);
e929cd20
MH
1194
1195 headers = curl_slist_append(headers, buf.buf);
1196
1197 curl_easy_setopt(slot->curl, CURLOPT_URL, url);
1198 curl_easy_setopt(slot->curl, CURLOPT_HTTPHEADER, headers);
aa90b969 1199 curl_easy_setopt(slot->curl, CURLOPT_ENCODING, "gzip");
e929cd20 1200
beed336c 1201 ret = run_one_slot(slot, &results);
e929cd20 1202
bf197fd7
JK
1203 if (options && options->content_type) {
1204 struct strbuf raw = STRBUF_INIT;
1205 curlinfo_strbuf(slot->curl, CURLINFO_CONTENT_TYPE, &raw);
e3131626
JK
1206 extract_content_type(&raw, options->content_type,
1207 options->charset);
bf197fd7
JK
1208 strbuf_release(&raw);
1209 }
4656bf47 1210
78868962
JK
1211 if (options && options->effective_url)
1212 curlinfo_strbuf(slot->curl, CURLINFO_EFFECTIVE_URL,
1213 options->effective_url);
4656bf47 1214
e929cd20
MH
1215 curl_slist_free_all(headers);
1216 strbuf_release(&buf);
1217
1218 return ret;
1219}
1220
c93c92f3
JK
1221/*
1222 * Update the "base" url to a more appropriate value, as deduced by
1223 * redirects seen when requesting a URL starting with "url".
1224 *
1225 * The "asked" parameter is a URL that we asked curl to access, and must begin
1226 * with "base".
1227 *
1228 * The "got" parameter is the URL that curl reported to us as where we ended
1229 * up.
1230 *
1231 * Returns 1 if we updated the base url, 0 otherwise.
1232 *
1233 * Our basic strategy is to compare "base" and "asked" to find the bits
1234 * specific to our request. We then strip those bits off of "got" to yield the
1235 * new base. So for example, if our base is "http://example.com/foo.git",
1236 * and we ask for "http://example.com/foo.git/info/refs", we might end up
1237 * with "https://other.example.com/foo.git/info/refs". We would want the
1238 * new URL to become "https://other.example.com/foo.git".
1239 *
1240 * Note that this assumes a sane redirect scheme. It's entirely possible
1241 * in the example above to end up at a URL that does not even end in
1242 * "info/refs". In such a case we simply punt, as there is not much we can
1243 * do (and such a scheme is unlikely to represent a real git repository,
1244 * which means we are likely about to abort anyway).
1245 */
1246static int update_url_from_redirect(struct strbuf *base,
1247 const char *asked,
1248 const struct strbuf *got)
1249{
1250 const char *tail;
1251 size_t tail_len;
1252
1253 if (!strcmp(asked, got->buf))
1254 return 0;
1255
de8118e1 1256 if (!skip_prefix(asked, base->buf, &tail))
c93c92f3
JK
1257 die("BUG: update_url_from_redirect: %s is not a superset of %s",
1258 asked, base->buf);
1259
c93c92f3
JK
1260 tail_len = strlen(tail);
1261
1262 if (got->len < tail_len ||
1263 strcmp(tail, got->buf + got->len - tail_len))
1264 return 0; /* insane redirect scheme */
1265
1266 strbuf_reset(base);
1267 strbuf_add(base, got->buf, got->len - tail_len);
1268 return 1;
1269}
1270
4656bf47 1271static int http_request_reauth(const char *url,
4656bf47 1272 void *result, int target,
1bbcc224 1273 struct http_get_options *options)
8d677edc 1274{
1bbcc224 1275 int ret = http_request(url, result, target, options);
c93c92f3
JK
1276
1277 if (options && options->effective_url && options->base_url) {
1278 if (update_url_from_redirect(options->base_url,
1279 url, options->effective_url)) {
1280 credential_from_url(&http_auth, options->base_url->buf);
1281 url = options->effective_url->buf;
1282 }
1283 }
1284
8d677edc
JK
1285 if (ret != HTTP_REAUTH)
1286 return ret;
6d052d78
JK
1287
1288 /*
1289 * If we are using KEEP_ERROR, the previous request may have
1290 * put cruft into our output stream; we should clear it out before
1291 * making our next request. We only know how to do this for
1292 * the strbuf case, but that is enough to satisfy current callers.
1293 */
1bbcc224 1294 if (options && options->keep_error) {
6d052d78
JK
1295 switch (target) {
1296 case HTTP_REQUEST_STRBUF:
1297 strbuf_reset(result);
1298 break;
1299 default:
1300 die("BUG: HTTP_KEEP_ERROR is only supported with strbufs");
1301 }
1302 }
2501aff8
JK
1303
1304 credential_fill(&http_auth);
1305
1bbcc224 1306 return http_request(url, result, target, options);
8d677edc
JK
1307}
1308
4656bf47 1309int http_get_strbuf(const char *url,
1bbcc224
JK
1310 struct strbuf *result,
1311 struct http_get_options *options)
e929cd20 1312{
1bbcc224 1313 return http_request_reauth(url, result, HTTP_REQUEST_STRBUF, options);
e929cd20
MH
1314}
1315
83e41e2e 1316/*
a7793a74 1317 * Downloads a URL and stores the result in the given file.
83e41e2e
JH
1318 *
1319 * If a previous interrupted download is detected (i.e. a previous temporary
1320 * file is still around) the download is resumed.
1321 */
1bbcc224
JK
1322static int http_get_file(const char *url, const char *filename,
1323 struct http_get_options *options)
e929cd20
MH
1324{
1325 int ret;
1326 struct strbuf tmpfile = STRBUF_INIT;
1327 FILE *result;
1328
1329 strbuf_addf(&tmpfile, "%s.temp", filename);
1330 result = fopen(tmpfile.buf, "a");
3d1fb769 1331 if (!result) {
e929cd20
MH
1332 error("Unable to open local file %s", tmpfile.buf);
1333 ret = HTTP_ERROR;
1334 goto cleanup;
1335 }
1336
1bbcc224 1337 ret = http_request_reauth(url, result, HTTP_REQUEST_FILE, options);
e929cd20
MH
1338 fclose(result);
1339
3d1fb769 1340 if (ret == HTTP_OK && move_temp_to_file(tmpfile.buf, filename))
e929cd20
MH
1341 ret = HTTP_ERROR;
1342cleanup:
1343 strbuf_release(&tmpfile);
1344 return ret;
1345}
1346
c13b2633 1347int http_fetch_ref(const char *base, struct ref *ref)
d7e92806 1348{
1bbcc224 1349 struct http_get_options options = {0};
d7e92806
MH
1350 char *url;
1351 struct strbuf buffer = STRBUF_INIT;
0d5896e1 1352 int ret = -1;
d7e92806 1353
1bbcc224
JK
1354 options.no_cache = 1;
1355
c13b2633 1356 url = quote_ref_url(base, ref->name);
1bbcc224 1357 if (http_get_strbuf(url, &buffer, &options) == HTTP_OK) {
0d5896e1
MH
1358 strbuf_rtrim(&buffer);
1359 if (buffer.len == 40)
1360 ret = get_sha1_hex(buffer.buf, ref->old_sha1);
59556548 1361 else if (starts_with(buffer.buf, "ref: ")) {
0d5896e1
MH
1362 ref->symref = xstrdup(buffer.buf + 5);
1363 ret = 0;
d7e92806 1364 }
d7e92806
MH
1365 }
1366
1367 strbuf_release(&buffer);
1368 free(url);
1369 return ret;
1370}
b8caac2b
TRC
1371
1372/* Helpers for fetching packs */
750ef425 1373static char *fetch_pack_index(unsigned char *sha1, const char *base_url)
b8caac2b 1374{
750ef425 1375 char *url, *tmp;
b8caac2b 1376 struct strbuf buf = STRBUF_INIT;
b8caac2b 1377
b8caac2b 1378 if (http_is_verbose)
162eb5f8 1379 fprintf(stderr, "Getting index for pack %s\n", sha1_to_hex(sha1));
b8caac2b
TRC
1380
1381 end_url_with_slash(&buf, base_url);
162eb5f8 1382 strbuf_addf(&buf, "objects/pack/pack-%s.idx", sha1_to_hex(sha1));
b8caac2b
TRC
1383 url = strbuf_detach(&buf, NULL);
1384
750ef425
SP
1385 strbuf_addf(&buf, "%s.temp", sha1_pack_index_name(sha1));
1386 tmp = strbuf_detach(&buf, NULL);
1387
70900eda 1388 if (http_get_file(url, tmp, NULL) != HTTP_OK) {
82247e9b 1389 error("Unable to get pack index %s", url);
750ef425
SP
1390 free(tmp);
1391 tmp = NULL;
1392 }
b8caac2b 1393
b8caac2b 1394 free(url);
750ef425 1395 return tmp;
b8caac2b
TRC
1396}
1397
1398static int fetch_and_setup_pack_index(struct packed_git **packs_head,
1399 unsigned char *sha1, const char *base_url)
1400{
1401 struct packed_git *new_pack;
750ef425
SP
1402 char *tmp_idx = NULL;
1403 int ret;
b8caac2b 1404
750ef425 1405 if (has_pack_index(sha1)) {
8b9c2dd4 1406 new_pack = parse_pack_index(sha1, sha1_pack_index_name(sha1));
750ef425
SP
1407 if (!new_pack)
1408 return -1; /* parse_pack_index() already issued error message */
1409 goto add_pack;
1410 }
1411
1412 tmp_idx = fetch_pack_index(sha1, base_url);
1413 if (!tmp_idx)
b8caac2b
TRC
1414 return -1;
1415
750ef425
SP
1416 new_pack = parse_pack_index(sha1, tmp_idx);
1417 if (!new_pack) {
1418 unlink(tmp_idx);
1419 free(tmp_idx);
1420
b8caac2b 1421 return -1; /* parse_pack_index() already issued error message */
750ef425
SP
1422 }
1423
1424 ret = verify_pack_index(new_pack);
1425 if (!ret) {
1426 close_pack_index(new_pack);
1427 ret = move_temp_to_file(tmp_idx, sha1_pack_index_name(sha1));
1428 }
1429 free(tmp_idx);
1430 if (ret)
1431 return -1;
1432
1433add_pack:
b8caac2b
TRC
1434 new_pack->next = *packs_head;
1435 *packs_head = new_pack;
1436 return 0;
1437}
1438
1439int http_get_info_packs(const char *base_url, struct packed_git **packs_head)
1440{
1bbcc224 1441 struct http_get_options options = {0};
b8caac2b
TRC
1442 int ret = 0, i = 0;
1443 char *url, *data;
1444 struct strbuf buf = STRBUF_INIT;
1445 unsigned char sha1[20];
1446
1447 end_url_with_slash(&buf, base_url);
1448 strbuf_addstr(&buf, "objects/info/packs");
1449 url = strbuf_detach(&buf, NULL);
1450
1bbcc224
JK
1451 options.no_cache = 1;
1452 ret = http_get_strbuf(url, &buf, &options);
b8caac2b
TRC
1453 if (ret != HTTP_OK)
1454 goto cleanup;
1455
1456 data = buf.buf;
1457 while (i < buf.len) {
1458 switch (data[i]) {
1459 case 'P':
1460 i++;
1461 if (i + 52 <= buf.len &&
59556548
CC
1462 starts_with(data + i, " pack-") &&
1463 starts_with(data + i + 46, ".pack\n")) {
b8caac2b
TRC
1464 get_sha1_hex(data + i + 6, sha1);
1465 fetch_and_setup_pack_index(packs_head, sha1,
1466 base_url);
1467 i += 51;
1468 break;
1469 }
1470 default:
1471 while (i < buf.len && data[i] != '\n')
1472 i++;
1473 }
1474 i++;
1475 }
1476
1477cleanup:
1478 free(url);
1479 return ret;
1480}
2264dfa5
TRC
1481
1482void release_http_pack_request(struct http_pack_request *preq)
1483{
1484 if (preq->packfile != NULL) {
1485 fclose(preq->packfile);
1486 preq->packfile = NULL;
2264dfa5
TRC
1487 }
1488 if (preq->range_header != NULL) {
1489 curl_slist_free_all(preq->range_header);
1490 preq->range_header = NULL;
1491 }
1492 preq->slot = NULL;
1493 free(preq->url);
826aed50 1494 free(preq);
2264dfa5
TRC
1495}
1496
1497int finish_http_pack_request(struct http_pack_request *preq)
1498{
2264dfa5 1499 struct packed_git **lst;
021ab6f0 1500 struct packed_git *p = preq->target;
fe72d420 1501 char *tmp_idx;
d3180279 1502 struct child_process ip = CHILD_PROCESS_INIT;
fe72d420 1503 const char *ip_argv[8];
2264dfa5 1504
fe72d420 1505 close_pack_index(p);
2264dfa5 1506
3065274c
SP
1507 fclose(preq->packfile);
1508 preq->packfile = NULL;
2264dfa5
TRC
1509
1510 lst = preq->lst;
021ab6f0 1511 while (*lst != p)
2264dfa5
TRC
1512 lst = &((*lst)->next);
1513 *lst = (*lst)->next;
1514
fe72d420
SP
1515 tmp_idx = xstrdup(preq->tmpfile);
1516 strcpy(tmp_idx + strlen(tmp_idx) - strlen(".pack.temp"),
1517 ".idx.temp");
1518
1519 ip_argv[0] = "index-pack";
1520 ip_argv[1] = "-o";
1521 ip_argv[2] = tmp_idx;
1522 ip_argv[3] = preq->tmpfile;
1523 ip_argv[4] = NULL;
1524
fe72d420
SP
1525 ip.argv = ip_argv;
1526 ip.git_cmd = 1;
1527 ip.no_stdin = 1;
1528 ip.no_stdout = 1;
1529
1530 if (run_command(&ip)) {
1531 unlink(preq->tmpfile);
1532 unlink(tmp_idx);
1533 free(tmp_idx);
2264dfa5 1534 return -1;
fe72d420
SP
1535 }
1536
1537 unlink(sha1_pack_index_name(p->sha1));
2264dfa5 1538
fe72d420
SP
1539 if (move_temp_to_file(preq->tmpfile, sha1_pack_name(p->sha1))
1540 || move_temp_to_file(tmp_idx, sha1_pack_index_name(p->sha1))) {
1541 free(tmp_idx);
2264dfa5 1542 return -1;
fe72d420 1543 }
2264dfa5 1544
fe72d420
SP
1545 install_packed_git(p);
1546 free(tmp_idx);
2264dfa5
TRC
1547 return 0;
1548}
1549
1550struct http_pack_request *new_http_pack_request(
1551 struct packed_git *target, const char *base_url)
1552{
2264dfa5
TRC
1553 long prev_posn = 0;
1554 char range[RANGE_HEADER_SIZE];
1555 struct strbuf buf = STRBUF_INIT;
1556 struct http_pack_request *preq;
1557
ec99c9a8 1558 preq = xcalloc(1, sizeof(*preq));
2264dfa5 1559 preq->target = target;
2264dfa5
TRC
1560
1561 end_url_with_slash(&buf, base_url);
1562 strbuf_addf(&buf, "objects/pack/pack-%s.pack",
1563 sha1_to_hex(target->sha1));
bb99190e 1564 preq->url = strbuf_detach(&buf, NULL);
2264dfa5 1565
90d05713
TRC
1566 snprintf(preq->tmpfile, sizeof(preq->tmpfile), "%s.temp",
1567 sha1_pack_name(target->sha1));
2264dfa5
TRC
1568 preq->packfile = fopen(preq->tmpfile, "a");
1569 if (!preq->packfile) {
1570 error("Unable to open local file %s for pack",
1571 preq->tmpfile);
1572 goto abort;
1573 }
1574
1575 preq->slot = get_active_slot();
2264dfa5
TRC
1576 curl_easy_setopt(preq->slot->curl, CURLOPT_FILE, preq->packfile);
1577 curl_easy_setopt(preq->slot->curl, CURLOPT_WRITEFUNCTION, fwrite);
bb99190e 1578 curl_easy_setopt(preq->slot->curl, CURLOPT_URL, preq->url);
2264dfa5
TRC
1579 curl_easy_setopt(preq->slot->curl, CURLOPT_HTTPHEADER,
1580 no_pragma_header);
1581
1582 /*
1583 * If there is data present from a previous transfer attempt,
1584 * resume where it left off
1585 */
1586 prev_posn = ftell(preq->packfile);
1587 if (prev_posn>0) {
1588 if (http_is_verbose)
1589 fprintf(stderr,
1590 "Resuming fetch of pack %s at byte %ld\n",
1591 sha1_to_hex(target->sha1), prev_posn);
1592 sprintf(range, "Range: bytes=%ld-", prev_posn);
1593 preq->range_header = curl_slist_append(NULL, range);
1594 curl_easy_setopt(preq->slot->curl, CURLOPT_HTTPHEADER,
1595 preq->range_header);
1596 }
1597
1598 return preq;
1599
1600abort:
bb99190e 1601 free(preq->url);
5ae9ebfd 1602 free(preq);
2264dfa5
TRC
1603 return NULL;
1604}
5424bc55
TRC
1605
1606/* Helpers for fetching objects (loose) */
a04ff3ec 1607static size_t fwrite_sha1_file(char *ptr, size_t eltsize, size_t nmemb,
5424bc55
TRC
1608 void *data)
1609{
1610 unsigned char expn[4096];
1611 size_t size = eltsize * nmemb;
1612 int posn = 0;
1613 struct http_object_request *freq =
1614 (struct http_object_request *)data;
1615 do {
1616 ssize_t retval = xwrite(freq->localfile,
1617 (char *) ptr + posn, size - posn);
1618 if (retval < 0)
1619 return posn;
1620 posn += retval;
1621 } while (posn < size);
1622
1623 freq->stream.avail_in = size;
a04ff3ec 1624 freq->stream.next_in = (void *)ptr;
5424bc55
TRC
1625 do {
1626 freq->stream.next_out = expn;
1627 freq->stream.avail_out = sizeof(expn);
1628 freq->zret = git_inflate(&freq->stream, Z_SYNC_FLUSH);
1629 git_SHA1_Update(&freq->c, expn,
1630 sizeof(expn) - freq->stream.avail_out);
1631 } while (freq->stream.avail_in && freq->zret == Z_OK);
5424bc55
TRC
1632 return size;
1633}
1634
1635struct http_object_request *new_http_object_request(const char *base_url,
1636 unsigned char *sha1)
1637{
1638 char *hex = sha1_to_hex(sha1);
30d6c6ea 1639 const char *filename;
5424bc55 1640 char prevfile[PATH_MAX];
5424bc55 1641 int prevlocal;
a04ff3ec 1642 char prev_buf[PREV_BUF_SIZE];
5424bc55
TRC
1643 ssize_t prev_read = 0;
1644 long prev_posn = 0;
1645 char range[RANGE_HEADER_SIZE];
1646 struct curl_slist *range_header = NULL;
1647 struct http_object_request *freq;
1648
ec99c9a8 1649 freq = xcalloc(1, sizeof(*freq));
5424bc55
TRC
1650 hashcpy(freq->sha1, sha1);
1651 freq->localfile = -1;
1652
1653 filename = sha1_file_name(sha1);
5424bc55
TRC
1654 snprintf(freq->tmpfile, sizeof(freq->tmpfile),
1655 "%s.temp", filename);
1656
1657 snprintf(prevfile, sizeof(prevfile), "%s.prev", filename);
1658 unlink_or_warn(prevfile);
1659 rename(freq->tmpfile, prevfile);
1660 unlink_or_warn(freq->tmpfile);
1661
1662 if (freq->localfile != -1)
1663 error("fd leakage in start: %d", freq->localfile);
1664 freq->localfile = open(freq->tmpfile,
1665 O_WRONLY | O_CREAT | O_EXCL, 0666);
1666 /*
1667 * This could have failed due to the "lazy directory creation";
1668 * try to mkdir the last path component.
1669 */
1670 if (freq->localfile < 0 && errno == ENOENT) {
1671 char *dir = strrchr(freq->tmpfile, '/');
1672 if (dir) {
1673 *dir = 0;
1674 mkdir(freq->tmpfile, 0777);
1675 *dir = '/';
1676 }
1677 freq->localfile = open(freq->tmpfile,
1678 O_WRONLY | O_CREAT | O_EXCL, 0666);
1679 }
1680
1681 if (freq->localfile < 0) {
0da8b2e7
SP
1682 error("Couldn't create temporary file %s: %s",
1683 freq->tmpfile, strerror(errno));
5424bc55
TRC
1684 goto abort;
1685 }
1686
5424bc55
TRC
1687 git_inflate_init(&freq->stream);
1688
1689 git_SHA1_Init(&freq->c);
1690
bb99190e 1691 freq->url = get_remote_object_url(base_url, hex, 0);
5424bc55
TRC
1692
1693 /*
1694 * If a previous temp file is present, process what was already
1695 * fetched.
1696 */
1697 prevlocal = open(prevfile, O_RDONLY);
1698 if (prevlocal != -1) {
1699 do {
1700 prev_read = xread(prevlocal, prev_buf, PREV_BUF_SIZE);
1701 if (prev_read>0) {
1702 if (fwrite_sha1_file(prev_buf,
1703 1,
1704 prev_read,
1705 freq) == prev_read) {
1706 prev_posn += prev_read;
1707 } else {
1708 prev_read = -1;
1709 }
1710 }
1711 } while (prev_read > 0);
1712 close(prevlocal);
1713 }
1714 unlink_or_warn(prevfile);
1715
1716 /*
1717 * Reset inflate/SHA1 if there was an error reading the previous temp
1718 * file; also rewind to the beginning of the local file.
1719 */
1720 if (prev_read == -1) {
1721 memset(&freq->stream, 0, sizeof(freq->stream));
1722 git_inflate_init(&freq->stream);
1723 git_SHA1_Init(&freq->c);
1724 if (prev_posn>0) {
1725 prev_posn = 0;
1726 lseek(freq->localfile, 0, SEEK_SET);
0c4f21e4 1727 if (ftruncate(freq->localfile, 0) < 0) {
0da8b2e7
SP
1728 error("Couldn't truncate temporary file %s: %s",
1729 freq->tmpfile, strerror(errno));
0c4f21e4
JL
1730 goto abort;
1731 }
5424bc55
TRC
1732 }
1733 }
1734
1735 freq->slot = get_active_slot();
1736
1737 curl_easy_setopt(freq->slot->curl, CURLOPT_FILE, freq);
1738 curl_easy_setopt(freq->slot->curl, CURLOPT_WRITEFUNCTION, fwrite_sha1_file);
1739 curl_easy_setopt(freq->slot->curl, CURLOPT_ERRORBUFFER, freq->errorstr);
bb99190e 1740 curl_easy_setopt(freq->slot->curl, CURLOPT_URL, freq->url);
5424bc55
TRC
1741 curl_easy_setopt(freq->slot->curl, CURLOPT_HTTPHEADER, no_pragma_header);
1742
1743 /*
1744 * If we have successfully processed data from a previous fetch
1745 * attempt, only fetch the data we don't already have.
1746 */
1747 if (prev_posn>0) {
1748 if (http_is_verbose)
1749 fprintf(stderr,
1750 "Resuming fetch of object %s at byte %ld\n",
1751 hex, prev_posn);
1752 sprintf(range, "Range: bytes=%ld-", prev_posn);
1753 range_header = curl_slist_append(range_header, range);
1754 curl_easy_setopt(freq->slot->curl,
1755 CURLOPT_HTTPHEADER, range_header);
1756 }
1757
1758 return freq;
1759
5424bc55 1760abort:
bb99190e 1761 free(freq->url);
5424bc55
TRC
1762 free(freq);
1763 return NULL;
1764}
1765
1766void process_http_object_request(struct http_object_request *freq)
1767{
1768 if (freq->slot == NULL)
1769 return;
1770 freq->curl_result = freq->slot->curl_result;
1771 freq->http_code = freq->slot->http_code;
1772 freq->slot = NULL;
1773}
1774
1775int finish_http_object_request(struct http_object_request *freq)
1776{
1777 struct stat st;
1778
1779 close(freq->localfile);
1780 freq->localfile = -1;
1781
1782 process_http_object_request(freq);
1783
1784 if (freq->http_code == 416) {
bd757c18 1785 warning("requested range invalid; we may already have all the data.");
5424bc55
TRC
1786 } else if (freq->curl_result != CURLE_OK) {
1787 if (stat(freq->tmpfile, &st) == 0)
1788 if (st.st_size == 0)
1789 unlink_or_warn(freq->tmpfile);
1790 return -1;
1791 }
1792
1793 git_inflate_end(&freq->stream);
1794 git_SHA1_Final(freq->real_sha1, &freq->c);
1795 if (freq->zret != Z_STREAM_END) {
1796 unlink_or_warn(freq->tmpfile);
1797 return -1;
1798 }
1799 if (hashcmp(freq->sha1, freq->real_sha1)) {
1800 unlink_or_warn(freq->tmpfile);
1801 return -1;
1802 }
1803 freq->rename =
0da8b2e7 1804 move_temp_to_file(freq->tmpfile, sha1_file_name(freq->sha1));
5424bc55
TRC
1805
1806 return freq->rename;
1807}
1808
1809void abort_http_object_request(struct http_object_request *freq)
1810{
1811 unlink_or_warn(freq->tmpfile);
1812
1813 release_http_object_request(freq);
1814}
1815
1816void release_http_object_request(struct http_object_request *freq)
1817{
1818 if (freq->localfile != -1) {
1819 close(freq->localfile);
1820 freq->localfile = -1;
1821 }
1822 if (freq->url != NULL) {
1823 free(freq->url);
1824 freq->url = NULL;
1825 }
4b9fa0e3
TRC
1826 if (freq->slot != NULL) {
1827 freq->slot->callback_func = NULL;
1828 freq->slot->callback_data = NULL;
1829 release_active_slot(freq->slot);
1830 freq->slot = NULL;
1831 }
5424bc55 1832}