http-push: use strbuf instead of fwrite_buffer
[git/git.git] / http-walker.c
CommitLineData
6eb7ed54
DB
1#include "cache.h"
2#include "commit.h"
30ae764b 3#include "walker.h"
29508e1e 4#include "http.h"
7baa3e86 5
9cba13ca 6struct alt_base {
2afea3bc 7 char *base;
b3661567
DB
8 int got_indices;
9 struct packed_git *packs;
10 struct alt_base *next;
11};
12
e388ab74 13enum object_request_state {
1d389ab6
NH
14 WAITING,
15 ABORTED,
16 ACTIVE,
4b05548f 17 COMPLETE
1d389ab6 18};
6eb7ed54 19
9cba13ca 20struct object_request {
30ae764b 21 struct walker *walker;
1d389ab6
NH
22 unsigned char sha1[20];
23 struct alt_base *repo;
e388ab74 24 enum object_request_state state;
5424bc55 25 struct http_object_request *req;
e388ab74 26 struct object_request *next;
1d389ab6
NH
27};
28
e388ab74 29struct alternates_request {
30ae764b 30 struct walker *walker;
8e29f6a0 31 const char *base;
acc075a8 32 char *url;
028c2976 33 struct strbuf *buffer;
acc075a8
NH
34 struct active_request_slot *slot;
35 int http_specific;
36};
37
30ae764b
DB
38struct walker_data {
39 const char *url;
40 int got_alternates;
41 struct alt_base *alt;
30ae764b
DB
42};
43
96f1e58f 44static struct object_request *object_queue_head;
bc8f2652 45
30ae764b 46static void fetch_alternates(struct walker *walker, const char *base);
1d389ab6 47
29508e1e 48static void process_object_response(void *callback_data);
1d389ab6 49
30ae764b
DB
50static void start_object_request(struct walker *walker,
51 struct object_request *obj_req)
1d389ab6 52{
1d389ab6 53 struct active_request_slot *slot;
5424bc55 54 struct http_object_request *req;
1d389ab6 55
5424bc55
TRC
56 req = new_http_object_request(obj_req->repo->base, obj_req->sha1);
57 if (req == NULL) {
e388ab74 58 obj_req->state = ABORTED;
1d389ab6
NH
59 return;
60 }
5424bc55 61 obj_req->req = req;
1d389ab6 62
5424bc55 63 slot = req->slot;
29508e1e 64 slot->callback_func = process_object_response;
e388ab74 65 slot->callback_data = obj_req;
1d389ab6 66
a7a8d378 67 /* Try to get the request started, abort the request on error */
e388ab74 68 obj_req->state = ACTIVE;
1d389ab6 69 if (!start_active_slot(slot)) {
e388ab74 70 obj_req->state = ABORTED;
5424bc55 71 release_http_object_request(req);
e388ab74 72 return;
1d389ab6 73 }
1d389ab6
NH
74}
75
e388ab74 76static void finish_object_request(struct object_request *obj_req)
1d389ab6 77{
5424bc55 78 if (finish_http_object_request(obj_req->req))
1d389ab6 79 return;
1d389ab6 80
5424bc55 81 if (obj_req->req->rename == 0)
30ae764b 82 walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
1d389ab6
NH
83}
84
29508e1e
NH
85static void process_object_response(void *callback_data)
86{
e388ab74
NH
87 struct object_request *obj_req =
88 (struct object_request *)callback_data;
30ae764b
DB
89 struct walker *walker = obj_req->walker;
90 struct walker_data *data = walker->data;
91 struct alt_base *alt = data->alt;
29508e1e 92
5424bc55 93 process_http_object_request(obj_req->req);
e388ab74 94 obj_req->state = COMPLETE;
29508e1e
NH
95
96 /* Use alternates if necessary */
5424bc55 97 if (missing_target(obj_req->req)) {
30ae764b 98 fetch_alternates(walker, alt->base);
e388ab74
NH
99 if (obj_req->repo->next != NULL) {
100 obj_req->repo =
101 obj_req->repo->next;
5424bc55 102 release_http_object_request(obj_req->req);
30ae764b 103 start_object_request(walker, obj_req);
29508e1e
NH
104 return;
105 }
106 }
107
e388ab74 108 finish_object_request(obj_req);
29508e1e
NH
109}
110
e388ab74 111static void release_object_request(struct object_request *obj_req)
1d389ab6 112{
e388ab74 113 struct object_request *entry = object_queue_head;
1d389ab6 114
5424bc55
TRC
115 if (obj_req->req !=NULL && obj_req->req->localfile != -1)
116 error("fd leakage in release: %d", obj_req->req->localfile);
e388ab74
NH
117 if (obj_req == object_queue_head) {
118 object_queue_head = obj_req->next;
1d389ab6 119 } else {
e388ab74 120 while (entry->next != NULL && entry->next != obj_req)
1d389ab6 121 entry = entry->next;
e388ab74 122 if (entry->next == obj_req)
1d389ab6
NH
123 entry->next = entry->next->next;
124 }
125
e388ab74 126 free(obj_req);
1d389ab6
NH
127}
128
a7a8d378 129#ifdef USE_CURL_MULTI
30ae764b 130static int fill_active_slot(struct walker *walker)
1d389ab6 131{
45c17412 132 struct object_request *obj_req;
1d389ab6 133
45c17412 134 for (obj_req = object_queue_head; obj_req; obj_req = obj_req->next) {
e388ab74
NH
135 if (obj_req->state == WAITING) {
136 if (has_sha1_file(obj_req->sha1))
09db444f 137 obj_req->state = COMPLETE;
45c17412 138 else {
30ae764b 139 start_object_request(walker, obj_req);
45c17412
DB
140 return 1;
141 }
f1a906a3 142 }
8fcf7f9a 143 }
45c17412 144 return 0;
1d389ab6 145}
a7a8d378 146#endif
1d389ab6 147
30ae764b 148static void prefetch(struct walker *walker, unsigned char *sha1)
1d389ab6 149{
e388ab74
NH
150 struct object_request *newreq;
151 struct object_request *tail;
30ae764b 152 struct walker_data *data = walker->data;
1d389ab6
NH
153
154 newreq = xmalloc(sizeof(*newreq));
30ae764b 155 newreq->walker = walker;
e702496e 156 hashcpy(newreq->sha1, sha1);
30ae764b 157 newreq->repo = data->alt;
1d389ab6 158 newreq->state = WAITING;
5424bc55 159 newreq->req = NULL;
1d389ab6
NH
160 newreq->next = NULL;
161
e9176745
TRC
162 http_is_verbose = walker->get_verbosely;
163
e388ab74
NH
164 if (object_queue_head == NULL) {
165 object_queue_head = newreq;
1d389ab6 166 } else {
e388ab74 167 tail = object_queue_head;
4c42aa1a 168 while (tail->next != NULL)
1d389ab6 169 tail = tail->next;
1d389ab6
NH
170 tail->next = newreq;
171 }
29508e1e 172
a7a8d378 173#ifdef USE_CURL_MULTI
29508e1e
NH
174 fill_active_slots();
175 step_active_slots();
a7a8d378 176#endif
1d389ab6
NH
177}
178
e388ab74 179static void process_alternates_response(void *callback_data)
b3661567 180{
e388ab74
NH
181 struct alternates_request *alt_req =
182 (struct alternates_request *)callback_data;
30ae764b
DB
183 struct walker *walker = alt_req->walker;
184 struct walker_data *cdata = walker->data;
acc075a8 185 struct active_request_slot *slot = alt_req->slot;
30ae764b 186 struct alt_base *tail = cdata->alt;
8e29f6a0 187 const char *base = alt_req->base;
a04ff3ec 188 const char null_byte = '\0';
acc075a8
NH
189 char *data;
190 int i = 0;
1d389ab6 191
acc075a8
NH
192 if (alt_req->http_specific) {
193 if (slot->curl_result != CURLE_OK ||
028c2976 194 !alt_req->buffer->len) {
acc075a8
NH
195
196 /* Try reusing the slot to get non-http alternates */
197 alt_req->http_specific = 0;
198 sprintf(alt_req->url, "%s/objects/info/alternates",
199 base);
200 curl_easy_setopt(slot->curl, CURLOPT_URL,
201 alt_req->url);
202 active_requests++;
203 slot->in_use = 1;
c9826473
NH
204 if (slot->finished != NULL)
205 (*slot->finished) = 0;
a3f583cb 206 if (!start_active_slot(slot)) {
30ae764b 207 cdata->got_alternates = -1;
29508e1e 208 slot->in_use = 0;
c9826473
NH
209 if (slot->finished != NULL)
210 (*slot->finished) = 1;
1d389ab6 211 }
a3f583cb 212 return;
b3661567 213 }
acc075a8 214 } else if (slot->curl_result != CURLE_OK) {
be4a015b 215 if (!missing_target(slot)) {
30ae764b 216 cdata->got_alternates = -1;
acc075a8
NH
217 return;
218 }
b3661567
DB
219 }
220
a04ff3ec 221 fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
028c2976
MH
222 alt_req->buffer->len--;
223 data = alt_req->buffer->buf;
1b0c1e67 224
028c2976 225 while (i < alt_req->buffer->len) {
b3661567 226 int posn = i;
028c2976 227 while (posn < alt_req->buffer->len && data[posn] != '\n')
b3661567
DB
228 posn++;
229 if (data[posn] == '\n') {
1b0c1e67
DB
230 int okay = 0;
231 int serverlen = 0;
232 struct alt_base *newalt;
b3661567 233 if (data[i] == '/') {
4c42aa1a
TRC
234 /*
235 * This counts
5df1e0d0
JH
236 * http://git.host/pub/scm/linux.git/
237 * -----------here^
238 * so memcpy(dst, base, serverlen) will
239 * copy up to "...git.host".
240 */
241 const char *colon_ss = strstr(base,"://");
242 if (colon_ss) {
243 serverlen = (strchr(colon_ss + 3, '/')
244 - base);
245 okay = 1;
246 }
1b0c1e67 247 } else if (!memcmp(data + i, "../", 3)) {
4c42aa1a
TRC
248 /*
249 * Relative URL; chop the corresponding
5df1e0d0
JH
250 * number of subpath from base (and ../
251 * from data), and concatenate the result.
252 *
253 * The code first drops ../ from data, and
254 * then drops one ../ from data and one path
255 * from base. IOW, one extra ../ is dropped
256 * from data than path is dropped from base.
257 *
258 * This is not wrong. The alternate in
259 * http://git.host/pub/scm/linux.git/
260 * to borrow from
261 * http://git.host/pub/scm/linus.git/
262 * is ../../linus.git/objects/. You need
263 * two ../../ to borrow from your direct
264 * neighbour.
265 */
1b0c1e67
DB
266 i += 3;
267 serverlen = strlen(base);
8fcf7f9a 268 while (i + 2 < posn &&
1b0c1e67
DB
269 !memcmp(data + i, "../", 3)) {
270 do {
271 serverlen--;
272 } while (serverlen &&
273 base[serverlen - 1] != '/');
274 i += 3;
275 }
a9486b02 276 /* If the server got removed, give up. */
8fcf7f9a 277 okay = strchr(base, ':') - base + 3 <
4c42aa1a 278 serverlen;
acc075a8 279 } else if (alt_req->http_specific) {
1b0c1e67
DB
280 char *colon = strchr(data + i, ':');
281 char *slash = strchr(data + i, '/');
282 if (colon && slash && colon < data + posn &&
283 slash < data + posn && colon < slash) {
284 okay = 1;
285 }
286 }
5df1e0d0 287 /* skip "objects\n" at end */
1b0c1e67 288 if (okay) {
59b8263a
RS
289 struct strbuf target = STRBUF_INIT;
290 strbuf_add(&target, base, serverlen);
291 strbuf_add(&target, data + i, posn - i - 7);
30ae764b 292 if (walker->get_verbosely)
59b8263a
RS
293 fprintf(stderr, "Also look at %s\n",
294 target.buf);
b3661567 295 newalt = xmalloc(sizeof(*newalt));
1d389ab6 296 newalt->next = NULL;
59b8263a 297 newalt->base = strbuf_detach(&target, NULL);
b3661567
DB
298 newalt->got_indices = 0;
299 newalt->packs = NULL;
8d9fbe57 300
1d389ab6
NH
301 while (tail->next != NULL)
302 tail = tail->next;
303 tail->next = newalt;
b3661567
DB
304 }
305 }
306 i = posn + 1;
307 }
bc8f2652 308
30ae764b 309 cdata->got_alternates = 1;
acc075a8
NH
310}
311
30ae764b 312static void fetch_alternates(struct walker *walker, const char *base)
acc075a8 313{
028c2976 314 struct strbuf buffer = STRBUF_INIT;
acc075a8 315 char *url;
acc075a8 316 struct active_request_slot *slot;
cb754fdf 317 struct alternates_request alt_req;
30ae764b 318 struct walker_data *cdata = walker->data;
acc075a8 319
4c42aa1a
TRC
320 /*
321 * If another request has already started fetching alternates,
322 * wait for them to arrive and return to processing this request's
323 * curl message
324 */
29508e1e 325#ifdef USE_CURL_MULTI
30ae764b 326 while (cdata->got_alternates == 0) {
29508e1e 327 step_active_slots();
acc075a8 328 }
29508e1e 329#endif
acc075a8
NH
330
331 /* Nothing to do if they've already been fetched */
30ae764b 332 if (cdata->got_alternates == 1)
acc075a8
NH
333 return;
334
335 /* Start the fetch */
30ae764b 336 cdata->got_alternates = 0;
acc075a8 337
30ae764b 338 if (walker->get_verbosely)
acc075a8 339 fprintf(stderr, "Getting alternates list for %s\n", base);
8fcf7f9a 340
28310186 341 url = xstrfmt("%s/objects/info/http-alternates", base);
acc075a8 342
4c42aa1a
TRC
343 /*
344 * Use a callback to process the result, since another request
345 * may fail and need to have alternates loaded before continuing
346 */
acc075a8 347 slot = get_active_slot();
e388ab74 348 slot->callback_func = process_alternates_response;
30ae764b 349 alt_req.walker = walker;
acc075a8
NH
350 slot->callback_data = &alt_req;
351
352 curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
29508e1e 353 curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
acc075a8
NH
354 curl_easy_setopt(slot->curl, CURLOPT_URL, url);
355
356 alt_req.base = base;
357 alt_req.url = url;
358 alt_req.buffer = &buffer;
359 alt_req.http_specific = 1;
360 alt_req.slot = slot;
361
362 if (start_active_slot(slot))
363 run_active_slot(slot);
364 else
30ae764b 365 cdata->got_alternates = -1;
acc075a8 366
028c2976 367 strbuf_release(&buffer);
acc075a8 368 free(url);
b3661567
DB
369}
370
30ae764b 371static int fetch_indices(struct walker *walker, struct alt_base *repo)
182005b9 372{
b8caac2b 373 int ret;
1d389ab6 374
b3661567 375 if (repo->got_indices)
182005b9
DB
376 return 0;
377
30ae764b 378 if (walker->get_verbosely)
6fd72e39 379 fprintf(stderr, "Getting pack list for %s\n", repo->base);
8fcf7f9a 380
b8caac2b
TRC
381 switch (http_get_info_packs(repo->base, &repo->packs)) {
382 case HTTP_OK:
383 case HTTP_MISSING_TARGET:
384 repo->got_indices = 1;
385 ret = 0;
386 break;
387 default:
5e3a7691 388 repo->got_indices = 0;
b8caac2b 389 ret = -1;
b3661567 390 }
182005b9 391
3a462bc9 392 return ret;
182005b9
DB
393}
394
07c19e72 395static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
182005b9 396{
182005b9 397 struct packed_git *target;
49a0f240 398 int ret;
cb754fdf 399 struct slot_results results;
2264dfa5 400 struct http_pack_request *preq;
182005b9 401
30ae764b 402 if (fetch_indices(walker, repo))
182005b9 403 return -1;
b3661567 404 target = find_sha1_pack(sha1, repo->packs);
182005b9 405 if (!target)
b3661567 406 return -1;
182005b9 407
30ae764b 408 if (walker->get_verbosely) {
182005b9
DB
409 fprintf(stderr, "Getting pack %s\n",
410 sha1_to_hex(target->sha1));
411 fprintf(stderr, " which contains %s\n",
412 sha1_to_hex(sha1));
413 }
414
2264dfa5
TRC
415 preq = new_http_pack_request(target, repo->base);
416 if (preq == NULL)
417 goto abort;
418 preq->lst = &repo->packs;
419 preq->slot->results = &results;
182005b9 420
2264dfa5
TRC
421 if (start_active_slot(preq->slot)) {
422 run_active_slot(preq->slot);
c8568e13 423 if (results.curl_result != CURLE_OK) {
2264dfa5
TRC
424 error("Unable to get pack file %s\n%s", preq->url,
425 curl_errorstr);
426 goto abort;
1d389ab6
NH
427 }
428 } else {
2264dfa5
TRC
429 error("Unable to start request");
430 goto abort;
182005b9
DB
431 }
432
2264dfa5
TRC
433 ret = finish_http_pack_request(preq);
434 release_http_pack_request(preq);
49a0f240 435 if (ret)
b721e01f 436 return ret;
49a0f240 437
182005b9 438 return 0;
2264dfa5
TRC
439
440abort:
441 return -1;
182005b9
DB
442}
443
53f31389
MW
444static void abort_object_request(struct object_request *obj_req)
445{
53f31389
MW
446 release_object_request(obj_req);
447}
448
30ae764b 449static int fetch_object(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
6eb7ed54
DB
450{
451 char *hex = sha1_to_hex(sha1);
29508e1e 452 int ret = 0;
e388ab74 453 struct object_request *obj_req = object_queue_head;
5424bc55 454 struct http_object_request *req;
1d389ab6 455
a89fccd2 456 while (obj_req != NULL && hashcmp(obj_req->sha1, sha1))
e388ab74
NH
457 obj_req = obj_req->next;
458 if (obj_req == NULL)
1d389ab6
NH
459 return error("Couldn't find request for %s in the queue", hex);
460
e388ab74 461 if (has_sha1_file(obj_req->sha1)) {
5424bc55
TRC
462 if (obj_req->req != NULL)
463 abort_http_object_request(obj_req->req);
53f31389 464 abort_object_request(obj_req);
11f0dafe
NH
465 return 0;
466 }
467
a7a8d378 468#ifdef USE_CURL_MULTI
4c42aa1a 469 while (obj_req->state == WAITING)
29508e1e 470 step_active_slots();
a7a8d378 471#else
30ae764b 472 start_object_request(walker, obj_req);
a7a8d378 473#endif
6eb7ed54 474
5424bc55
TRC
475 /*
476 * obj_req->req might change when fetching alternates in the callback
477 * process_object_response; therefore, the "shortcut" variable, req,
478 * is used only after we're done with slots.
479 */
4c42aa1a 480 while (obj_req->state == ACTIVE)
5424bc55
TRC
481 run_active_slot(obj_req->req->slot);
482
483 req = obj_req->req;
4c42aa1a 484
5424bc55
TRC
485 if (req->localfile != -1) {
486 close(req->localfile);
487 req->localfile = -1;
313c4714 488 }
6eb7ed54 489
e388ab74 490 if (obj_req->state == ABORTED) {
29508e1e 491 ret = error("Request for %s aborted", hex);
5424bc55
TRC
492 } else if (req->curl_result != CURLE_OK &&
493 req->http_code != 416) {
494 if (missing_target(req))
e2029eb9
PB
495 ret = -1; /* Be silent, it is probably in a pack. */
496 else
497 ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
5424bc55
TRC
498 req->errorstr, req->curl_result,
499 req->http_code, hex);
500 } else if (req->zret != Z_STREAM_END) {
30ae764b 501 walker->corrupt_object_found++;
5424bc55
TRC
502 ret = error("File %s (%s) corrupt", hex, req->url);
503 } else if (hashcmp(obj_req->sha1, req->real_sha1)) {
bd2afde8 504 ret = error("File %s has bad hash", hex);
5424bc55 505 } else if (req->rename < 0) {
7b934ec0 506 ret = error("unable to write sha1 filename %s",
0da8b2e7 507 sha1_file_name(req->sha1));
6eb7ed54 508 }
49a0f240 509
5424bc55 510 release_http_object_request(req);
e388ab74 511 release_object_request(obj_req);
29508e1e 512 return ret;
6eb7ed54
DB
513}
514
30ae764b 515static int fetch(struct walker *walker, unsigned char *sha1)
b3661567 516{
30ae764b
DB
517 struct walker_data *data = walker->data;
518 struct alt_base *altbase = data->alt;
1d389ab6 519
30ae764b 520 if (!fetch_object(walker, altbase, sha1))
1d389ab6 521 return 0;
b3661567 522 while (altbase) {
07c19e72 523 if (!http_fetch_pack(walker, altbase, sha1))
b3661567 524 return 0;
30ae764b 525 fetch_alternates(walker, data->alt->base);
b3661567
DB
526 altbase = altbase->next;
527 }
bd2afde8 528 return error("Unable to find %s under %s", sha1_to_hex(sha1),
30ae764b 529 data->alt->base);
b3661567
DB
530}
531
c13b2633 532static int fetch_ref(struct walker *walker, struct ref *ref)
cd541a68 533{
30ae764b 534 struct walker_data *data = walker->data;
c13b2633 535 return http_fetch_ref(data->alt->base, ref);
cd541a68
DB
536}
537
30ae764b
DB
538static void cleanup(struct walker *walker)
539{
09ae9aca
TRC
540 struct walker_data *data = walker->data;
541 struct alt_base *alt, *alt_next;
542
543 if (data) {
544 alt = data->alt;
545 while (alt) {
546 alt_next = alt->next;
547
548 free(alt->base);
549 free(alt);
550
551 alt = alt_next;
552 }
553 free(data);
554 walker->data = NULL;
555 }
30ae764b
DB
556}
557
888692b7 558struct walker *get_http_walker(const char *url)
6eb7ed54 559{
9c880b3e 560 char *s;
30ae764b
DB
561 struct walker_data *data = xmalloc(sizeof(struct walker_data));
562 struct walker *walker = xmalloc(sizeof(struct walker));
6eb7ed54 563
30ae764b 564 data->alt = xmalloc(sizeof(*data->alt));
95244ae3 565 data->alt->base = xstrdup(url);
30ae764b 566 for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
9c880b3e 567 *s = 0;
6eb7ed54 568
30ae764b
DB
569 data->alt->got_indices = 0;
570 data->alt->packs = NULL;
571 data->alt->next = NULL;
572 data->got_alternates = -1;
fc57b6aa 573
30ae764b
DB
574 walker->corrupt_object_found = 0;
575 walker->fetch = fetch;
576 walker->fetch_ref = fetch_ref;
577 walker->prefetch = prefetch;
578 walker->cleanup = cleanup;
579 walker->data = data;
6eb7ed54 580
30ae764b
DB
581#ifdef USE_CURL_MULTI
582 add_fill_function(walker, (int (*)(void *)) fill_active_slot);
583#endif
8e29f6a0 584
30ae764b 585 return walker;
6eb7ed54 586}