Merge branch 'jn/vcs-svn-cleanup'
[git/git.git] / vcs-svn / svndump.c
1 /*
2 * Parse and rearrange a svnadmin dump.
3 * Create the dump with:
4 * svnadmin dump --incremental -r<startrev>:<endrev> <repository> >outfile
5 *
6 * Licensed under a two-clause BSD-style license.
7 * See LICENSE for details.
8 */
9
10 #include "cache.h"
11 #include "fast_export.h"
12 #include "line_buffer.h"
13 #include "strbuf.h"
14 #include "svndump.h"
15
16 /*
17 * Compare start of string to literal of equal length;
18 * must be guarded by length test.
19 */
20 #define constcmp(s, ref) memcmp(s, ref, sizeof(ref) - 1)
21
22 #define REPORT_FILENO 3
23
24 #define NODEACT_REPLACE 4
25 #define NODEACT_DELETE 3
26 #define NODEACT_ADD 2
27 #define NODEACT_CHANGE 1
28 #define NODEACT_UNKNOWN 0
29
30 /* States: */
31 #define DUMP_CTX 0 /* dump metadata */
32 #define REV_CTX 1 /* revision metadata */
33 #define NODE_CTX 2 /* node metadata */
34 #define INTERNODE_CTX 3 /* between nodes */
35
36 #define DATE_RFC2822_LEN 31
37
38 static struct line_buffer input = LINE_BUFFER_INIT;
39
40 static struct {
41 uint32_t action, srcRev, type;
42 off_t prop_length, text_length;
43 struct strbuf src, dst;
44 uint32_t text_delta, prop_delta;
45 } node_ctx;
46
47 static struct {
48 uint32_t revision;
49 timestamp_t timestamp;
50 struct strbuf log, author, note;
51 } rev_ctx;
52
53 static struct {
54 uint32_t version;
55 struct strbuf uuid, url;
56 } dump_ctx;
57
58 static void reset_node_ctx(char *fname)
59 {
60 node_ctx.type = 0;
61 node_ctx.action = NODEACT_UNKNOWN;
62 node_ctx.prop_length = -1;
63 node_ctx.text_length = -1;
64 strbuf_reset(&node_ctx.src);
65 node_ctx.srcRev = 0;
66 strbuf_reset(&node_ctx.dst);
67 if (fname)
68 strbuf_addstr(&node_ctx.dst, fname);
69 node_ctx.text_delta = 0;
70 node_ctx.prop_delta = 0;
71 }
72
73 static void reset_rev_ctx(uint32_t revision)
74 {
75 rev_ctx.revision = revision;
76 rev_ctx.timestamp = 0;
77 strbuf_reset(&rev_ctx.log);
78 strbuf_reset(&rev_ctx.author);
79 strbuf_reset(&rev_ctx.note);
80 }
81
82 static void reset_dump_ctx(const char *url)
83 {
84 strbuf_reset(&dump_ctx.url);
85 if (url)
86 strbuf_addstr(&dump_ctx.url, url);
87 dump_ctx.version = 1;
88 strbuf_reset(&dump_ctx.uuid);
89 }
90
91 static void handle_property(const struct strbuf *key_buf,
92 struct strbuf *val,
93 uint32_t *type_set)
94 {
95 const char *key = key_buf->buf;
96 size_t keylen = key_buf->len;
97
98 switch (keylen + 1) {
99 case sizeof("svn:log"):
100 if (constcmp(key, "svn:log"))
101 break;
102 if (!val)
103 die("invalid dump: unsets svn:log");
104 strbuf_swap(&rev_ctx.log, val);
105 break;
106 case sizeof("svn:author"):
107 if (constcmp(key, "svn:author"))
108 break;
109 if (!val)
110 strbuf_reset(&rev_ctx.author);
111 else
112 strbuf_swap(&rev_ctx.author, val);
113 break;
114 case sizeof("svn:date"):
115 if (constcmp(key, "svn:date"))
116 break;
117 if (!val)
118 die("invalid dump: unsets svn:date");
119 if (parse_date_basic(val->buf, &rev_ctx.timestamp, NULL))
120 warning("invalid timestamp: %s", val->buf);
121 break;
122 case sizeof("svn:executable"):
123 case sizeof("svn:special"):
124 if (keylen == strlen("svn:executable") &&
125 constcmp(key, "svn:executable"))
126 break;
127 if (keylen == strlen("svn:special") &&
128 constcmp(key, "svn:special"))
129 break;
130 if (*type_set) {
131 if (!val)
132 return;
133 die("invalid dump: sets type twice");
134 }
135 if (!val) {
136 node_ctx.type = S_IFREG | 0644;
137 return;
138 }
139 *type_set = 1;
140 node_ctx.type = keylen == strlen("svn:executable") ?
141 (S_IFREG | 0755) :
142 S_IFLNK;
143 }
144 }
145
146 static void die_short_read(void)
147 {
148 if (buffer_ferror(&input))
149 die_errno("error reading dump file");
150 die("invalid dump: unexpected end of file");
151 }
152
153 static void read_props(void)
154 {
155 static struct strbuf key = STRBUF_INIT;
156 static struct strbuf val = STRBUF_INIT;
157 const char *t;
158 /*
159 * NEEDSWORK: to support simple mode changes like
160 * K 11
161 * svn:special
162 * V 1
163 * *
164 * D 14
165 * svn:executable
166 * we keep track of whether a mode has been set and reset to
167 * plain file only if not. We should be keeping track of the
168 * symlink and executable bits separately instead.
169 */
170 uint32_t type_set = 0;
171 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
172 uint32_t len;
173 const char type = t[0];
174 int ch;
175
176 if (!type || t[1] != ' ')
177 die("invalid property line: %s", t);
178 len = atoi(&t[2]);
179 strbuf_reset(&val);
180 buffer_read_binary(&input, &val, len);
181 if (val.len < len)
182 die_short_read();
183
184 /* Discard trailing newline. */
185 ch = buffer_read_char(&input);
186 if (ch == EOF)
187 die_short_read();
188 if (ch != '\n')
189 die("invalid dump: expected newline after %s", val.buf);
190
191 switch (type) {
192 case 'K':
193 strbuf_swap(&key, &val);
194 continue;
195 case 'D':
196 handle_property(&val, NULL, &type_set);
197 continue;
198 case 'V':
199 handle_property(&key, &val, &type_set);
200 strbuf_reset(&key);
201 continue;
202 default:
203 die("invalid property line: %s", t);
204 }
205 }
206 }
207
208 static void handle_node(void)
209 {
210 const uint32_t type = node_ctx.type;
211 const int have_props = node_ctx.prop_length != -1;
212 const int have_text = node_ctx.text_length != -1;
213 /*
214 * Old text for this node:
215 * NULL - directory or bug
216 * empty_blob - empty
217 * "<dataref>" - data retrievable from fast-import
218 */
219 static const char *const empty_blob = "::empty::";
220 const char *old_data = NULL;
221 uint32_t old_mode = S_IFREG | 0644;
222
223 if (node_ctx.action == NODEACT_DELETE) {
224 if (have_text || have_props || node_ctx.srcRev)
225 die("invalid dump: deletion node has "
226 "copyfrom info, text, or properties");
227 fast_export_delete(node_ctx.dst.buf);
228 return;
229 }
230 if (node_ctx.action == NODEACT_REPLACE) {
231 fast_export_delete(node_ctx.dst.buf);
232 node_ctx.action = NODEACT_ADD;
233 }
234 if (node_ctx.srcRev) {
235 fast_export_copy(node_ctx.srcRev, node_ctx.src.buf, node_ctx.dst.buf);
236 if (node_ctx.action == NODEACT_ADD)
237 node_ctx.action = NODEACT_CHANGE;
238 }
239 if (have_text && type == S_IFDIR)
240 die("invalid dump: directories cannot have text attached");
241
242 /*
243 * Find old content (old_data) and decide on the new mode.
244 */
245 if (node_ctx.action == NODEACT_CHANGE && !*node_ctx.dst.buf) {
246 if (type != S_IFDIR)
247 die("invalid dump: root of tree is not a regular file");
248 old_data = NULL;
249 } else if (node_ctx.action == NODEACT_CHANGE) {
250 uint32_t mode;
251 old_data = fast_export_read_path(node_ctx.dst.buf, &mode);
252 if (mode == S_IFDIR && type != S_IFDIR)
253 die("invalid dump: cannot modify a directory into a file");
254 if (mode != S_IFDIR && type == S_IFDIR)
255 die("invalid dump: cannot modify a file into a directory");
256 node_ctx.type = mode;
257 old_mode = mode;
258 } else if (node_ctx.action == NODEACT_ADD) {
259 if (type == S_IFDIR)
260 old_data = NULL;
261 else if (have_text)
262 old_data = empty_blob;
263 else
264 die("invalid dump: adds node without text");
265 } else {
266 die("invalid dump: Node-path block lacks Node-action");
267 }
268
269 /*
270 * Adjust mode to reflect properties.
271 */
272 if (have_props) {
273 if (!node_ctx.prop_delta)
274 node_ctx.type = type;
275 if (node_ctx.prop_length)
276 read_props();
277 }
278
279 /*
280 * Save the result.
281 */
282 if (type == S_IFDIR) /* directories are not tracked. */
283 return;
284 assert(old_data);
285 if (old_data == empty_blob)
286 /* For the fast_export_* functions, NULL means empty. */
287 old_data = NULL;
288 if (!have_text) {
289 fast_export_modify(node_ctx.dst.buf, node_ctx.type, old_data);
290 return;
291 }
292 if (!node_ctx.text_delta) {
293 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
294 fast_export_data(node_ctx.type, node_ctx.text_length, &input);
295 return;
296 }
297 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
298 fast_export_blob_delta(node_ctx.type, old_mode, old_data,
299 node_ctx.text_length, &input);
300 }
301
302 static void begin_revision(const char *remote_ref)
303 {
304 if (!rev_ctx.revision) /* revision 0 gets no git commit. */
305 return;
306 fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
307 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
308 rev_ctx.timestamp, remote_ref);
309 }
310
311 static void end_revision(const char *note_ref)
312 {
313 struct strbuf mark = STRBUF_INIT;
314 if (rev_ctx.revision) {
315 fast_export_end_commit(rev_ctx.revision);
316 fast_export_begin_note(rev_ctx.revision, "remote-svn",
317 "Note created by remote-svn.", rev_ctx.timestamp, note_ref);
318 strbuf_addf(&mark, ":%"PRIu32, rev_ctx.revision);
319 fast_export_note(mark.buf, "inline");
320 fast_export_buf_to_data(&rev_ctx.note);
321 }
322 }
323
324 void svndump_read(const char *url, const char *local_ref, const char *notes_ref)
325 {
326 char *val;
327 char *t;
328 uint32_t active_ctx = DUMP_CTX;
329 uint32_t len;
330
331 reset_dump_ctx(url);
332 while ((t = buffer_read_line(&input))) {
333 val = strchr(t, ':');
334 if (!val)
335 continue;
336 val++;
337 if (*val != ' ')
338 continue;
339 val++;
340
341 /* strlen(key) + 1 */
342 switch (val - t - 1) {
343 case sizeof("SVN-fs-dump-format-version"):
344 if (constcmp(t, "SVN-fs-dump-format-version"))
345 continue;
346 dump_ctx.version = atoi(val);
347 if (dump_ctx.version > 3)
348 die("expected svn dump format version <= 3, found %"PRIu32,
349 dump_ctx.version);
350 break;
351 case sizeof("UUID"):
352 if (constcmp(t, "UUID"))
353 continue;
354 strbuf_reset(&dump_ctx.uuid);
355 strbuf_addstr(&dump_ctx.uuid, val);
356 break;
357 case sizeof("Revision-number"):
358 if (constcmp(t, "Revision-number"))
359 continue;
360 if (active_ctx == NODE_CTX)
361 handle_node();
362 if (active_ctx == REV_CTX)
363 begin_revision(local_ref);
364 if (active_ctx != DUMP_CTX)
365 end_revision(notes_ref);
366 active_ctx = REV_CTX;
367 reset_rev_ctx(atoi(val));
368 strbuf_addf(&rev_ctx.note, "%s\n", t);
369 break;
370 case sizeof("Node-path"):
371 if (constcmp(t, "Node-"))
372 continue;
373 if (!constcmp(t + strlen("Node-"), "path")) {
374 if (active_ctx == NODE_CTX)
375 handle_node();
376 if (active_ctx == REV_CTX)
377 begin_revision(local_ref);
378 active_ctx = NODE_CTX;
379 reset_node_ctx(val);
380 strbuf_addf(&rev_ctx.note, "%s\n", t);
381 break;
382 }
383 if (constcmp(t + strlen("Node-"), "kind"))
384 continue;
385 strbuf_addf(&rev_ctx.note, "%s\n", t);
386 if (!strcmp(val, "dir"))
387 node_ctx.type = S_IFDIR;
388 else if (!strcmp(val, "file"))
389 node_ctx.type = S_IFREG | 0644;
390 else
391 fprintf(stderr, "Unknown node-kind: %s\n", val);
392 break;
393 case sizeof("Node-action"):
394 if (constcmp(t, "Node-action"))
395 continue;
396 strbuf_addf(&rev_ctx.note, "%s\n", t);
397 if (!strcmp(val, "delete")) {
398 node_ctx.action = NODEACT_DELETE;
399 } else if (!strcmp(val, "add")) {
400 node_ctx.action = NODEACT_ADD;
401 } else if (!strcmp(val, "change")) {
402 node_ctx.action = NODEACT_CHANGE;
403 } else if (!strcmp(val, "replace")) {
404 node_ctx.action = NODEACT_REPLACE;
405 } else {
406 fprintf(stderr, "Unknown node-action: %s\n", val);
407 node_ctx.action = NODEACT_UNKNOWN;
408 }
409 break;
410 case sizeof("Node-copyfrom-path"):
411 if (constcmp(t, "Node-copyfrom-path"))
412 continue;
413 strbuf_reset(&node_ctx.src);
414 strbuf_addstr(&node_ctx.src, val);
415 strbuf_addf(&rev_ctx.note, "%s\n", t);
416 break;
417 case sizeof("Node-copyfrom-rev"):
418 if (constcmp(t, "Node-copyfrom-rev"))
419 continue;
420 node_ctx.srcRev = atoi(val);
421 strbuf_addf(&rev_ctx.note, "%s\n", t);
422 break;
423 case sizeof("Text-content-length"):
424 if (constcmp(t, "Text") && constcmp(t, "Prop"))
425 continue;
426 if (constcmp(t + 4, "-content-length"))
427 continue;
428 {
429 char *end;
430 uintmax_t len;
431
432 len = strtoumax(val, &end, 10);
433 if (!isdigit(*val) || *end)
434 die("invalid dump: non-numeric length %s", val);
435 if (len > maximum_signed_value_of_type(off_t))
436 die("unrepresentable length in dump: %s", val);
437
438 if (*t == 'T')
439 node_ctx.text_length = (off_t) len;
440 else
441 node_ctx.prop_length = (off_t) len;
442 break;
443 }
444 case sizeof("Text-delta"):
445 if (!constcmp(t, "Text-delta")) {
446 node_ctx.text_delta = !strcmp(val, "true");
447 break;
448 }
449 if (constcmp(t, "Prop-delta"))
450 continue;
451 node_ctx.prop_delta = !strcmp(val, "true");
452 break;
453 case sizeof("Content-length"):
454 if (constcmp(t, "Content-length"))
455 continue;
456 len = atoi(val);
457 t = buffer_read_line(&input);
458 if (!t)
459 die_short_read();
460 if (*t)
461 die("invalid dump: expected blank line after content length header");
462 if (active_ctx == REV_CTX) {
463 read_props();
464 } else if (active_ctx == NODE_CTX) {
465 handle_node();
466 active_ctx = INTERNODE_CTX;
467 } else {
468 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
469 if (buffer_skip_bytes(&input, len) != len)
470 die_short_read();
471 }
472 }
473 }
474 if (buffer_ferror(&input))
475 die_short_read();
476 if (active_ctx == NODE_CTX)
477 handle_node();
478 if (active_ctx == REV_CTX)
479 begin_revision(local_ref);
480 if (active_ctx != DUMP_CTX)
481 end_revision(notes_ref);
482 }
483
484 static void init(int report_fd)
485 {
486 fast_export_init(report_fd);
487 strbuf_init(&dump_ctx.uuid, 4096);
488 strbuf_init(&dump_ctx.url, 4096);
489 strbuf_init(&rev_ctx.log, 4096);
490 strbuf_init(&rev_ctx.author, 4096);
491 strbuf_init(&rev_ctx.note, 4096);
492 strbuf_init(&node_ctx.src, 4096);
493 strbuf_init(&node_ctx.dst, 4096);
494 reset_dump_ctx(NULL);
495 reset_rev_ctx(0);
496 reset_node_ctx(NULL);
497 return;
498 }
499
500 int svndump_init(const char *filename)
501 {
502 if (buffer_init(&input, filename))
503 return error_errno("cannot open %s", filename ? filename : "NULL");
504 init(REPORT_FILENO);
505 return 0;
506 }
507
508 int svndump_init_fd(int in_fd, int back_fd)
509 {
510 if(buffer_fdinit(&input, xdup(in_fd)))
511 return error_errno("cannot open fd %d", in_fd);
512 init(xdup(back_fd));
513 return 0;
514 }
515
516 void svndump_deinit(void)
517 {
518 fast_export_deinit();
519 reset_dump_ctx(NULL);
520 reset_rev_ctx(0);
521 reset_node_ctx(NULL);
522 strbuf_release(&rev_ctx.log);
523 strbuf_release(&rev_ctx.author);
524 strbuf_release(&rev_ctx.note);
525 strbuf_release(&node_ctx.src);
526 strbuf_release(&node_ctx.dst);
527 if (buffer_deinit(&input))
528 fprintf(stderr, "Input error\n");
529 if (ferror(stdout))
530 fprintf(stderr, "Output error\n");
531 }
532
533 void svndump_reset(void)
534 {
535 strbuf_release(&dump_ctx.uuid);
536 strbuf_release(&dump_ctx.url);
537 strbuf_release(&rev_ctx.log);
538 strbuf_release(&rev_ctx.author);
539 }