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#include "cache.h"
11#include "repo_tree.h"
12#include "fast_export.h"
13#include "line_buffer.h"
14#include "strbuf.h"
15#include "svndump.h"
16/*
18 * Compare start of string to literal of equal length;
19 * must be guarded by length test.
20 */
21#define constcmp(s, ref) memcmp(s, ref, sizeof(ref) - 1)
22#define REPORT_FILENO 3
24#define NODEACT_REPLACE 4
26#define NODEACT_DELETE 3
27#define NODEACT_ADD 2
28#define NODEACT_CHANGE 1
29#define NODEACT_UNKNOWN 0
30/* States: */
32#define DUMP_CTX 0 /* dump metadata */
33#define REV_CTX 1 /* revision metadata */
34#define NODE_CTX 2 /* node metadata */
35#define INTERNODE_CTX 3 /* between nodes */
36#define DATE_RFC2822_LEN 31
38static struct line_buffer input = LINE_BUFFER_INIT;
40static struct {
42 uint32_t action, srcRev, type;
43 off_t prop_length, text_length;
44 struct strbuf src, dst;
45 uint32_t text_delta, prop_delta;
46} node_ctx;
47static struct {
49 uint32_t revision;
50 unsigned long timestamp;
51 struct strbuf log, author;
52} rev_ctx;
53static struct {
55 uint32_t version;
56 struct strbuf uuid, url;
57} dump_ctx;
58static void reset_node_ctx(char *fname)
60{
61 node_ctx.type = 0;
62 node_ctx.action = NODEACT_UNKNOWN;
63 node_ctx.prop_length = -1;
64 node_ctx.text_length = -1;
65 strbuf_reset(&node_ctx.src);
66 node_ctx.srcRev = 0;
67 strbuf_reset(&node_ctx.dst);
68 if (fname)
69 strbuf_addstr(&node_ctx.dst, fname);
70 node_ctx.text_delta = 0;
71 node_ctx.prop_delta = 0;
72}
73static void reset_rev_ctx(uint32_t revision)
75{
76 rev_ctx.revision = revision;
77 rev_ctx.timestamp = 0;
78 strbuf_reset(&rev_ctx.log);
79 strbuf_reset(&rev_ctx.author);
80}
81static 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}
90static 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 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 = REPO_MODE_BLB;
137 return;
138 }
139 *type_set = 1;
140 node_ctx.type = keylen == strlen("svn:executable") ?
141 REPO_MODE_EXE :
142 REPO_MODE_LNK;
143 }
144}
145static 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}
152static 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 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 /* 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 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}
207static 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 = REPO_MODE_BLB;
222 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 repo_delete(node_ctx.dst.buf);
228 return;
229 }
230 if (node_ctx.action == NODEACT_REPLACE) {
231 repo_delete(node_ctx.dst.buf);
232 node_ctx.action = NODEACT_ADD;
233 }
234 if (node_ctx.srcRev) {
235 repo_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 == REPO_MODE_DIR)
240 die("invalid dump: directories cannot have text attached");
241 /*
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 != REPO_MODE_DIR)
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 = repo_read_path(node_ctx.dst.buf, &mode);
252 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
253 die("invalid dump: cannot modify a directory into a file");
254 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
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 == REPO_MODE_DIR)
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 /*
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 /*
280 * Save the result.
281 */
282 if (type == REPO_MODE_DIR) /* 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}
301static void begin_revision(void)
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);
309}
310static void end_revision(void)
312{
313 if (rev_ctx.revision)
314 fast_export_end_commit(rev_ctx.revision);
315}
316void svndump_read(const char *url)
318{
319 char *val;
320 char *t;
321 uint32_t active_ctx = DUMP_CTX;
322 uint32_t len;
323 reset_dump_ctx(url);
325 while ((t = buffer_read_line(&input))) {
326 val = strchr(t, ':');
327 if (!val)
328 continue;
329 val++;
330 if (*val != ' ')
331 continue;
332 val++;
333 /* strlen(key) + 1 */
335 switch (val - t - 1) {
336 case sizeof("SVN-fs-dump-format-version"):
337 if (constcmp(t, "SVN-fs-dump-format-version"))
338 continue;
339 dump_ctx.version = atoi(val);
340 if (dump_ctx.version > 3)
341 die("expected svn dump format version <= 3, found %"PRIu32,
342 dump_ctx.version);
343 break;
344 case sizeof("UUID"):
345 if (constcmp(t, "UUID"))
346 continue;
347 strbuf_reset(&dump_ctx.uuid);
348 strbuf_addstr(&dump_ctx.uuid, val);
349 break;
350 case sizeof("Revision-number"):
351 if (constcmp(t, "Revision-number"))
352 continue;
353 if (active_ctx == NODE_CTX)
354 handle_node();
355 if (active_ctx == REV_CTX)
356 begin_revision();
357 if (active_ctx != DUMP_CTX)
358 end_revision();
359 active_ctx = REV_CTX;
360 reset_rev_ctx(atoi(val));
361 break;
362 case sizeof("Node-path"):
363 if (constcmp(t, "Node-"))
364 continue;
365 if (!constcmp(t + strlen("Node-"), "path")) {
366 if (active_ctx == NODE_CTX)
367 handle_node();
368 if (active_ctx == REV_CTX)
369 begin_revision();
370 active_ctx = NODE_CTX;
371 reset_node_ctx(val);
372 break;
373 }
374 if (constcmp(t + strlen("Node-"), "kind"))
375 continue;
376 if (!strcmp(val, "dir"))
377 node_ctx.type = REPO_MODE_DIR;
378 else if (!strcmp(val, "file"))
379 node_ctx.type = REPO_MODE_BLB;
380 else
381 fprintf(stderr, "Unknown node-kind: %s\n", val);
382 break;
383 case sizeof("Node-action"):
384 if (constcmp(t, "Node-action"))
385 continue;
386 if (!strcmp(val, "delete")) {
387 node_ctx.action = NODEACT_DELETE;
388 } else if (!strcmp(val, "add")) {
389 node_ctx.action = NODEACT_ADD;
390 } else if (!strcmp(val, "change")) {
391 node_ctx.action = NODEACT_CHANGE;
392 } else if (!strcmp(val, "replace")) {
393 node_ctx.action = NODEACT_REPLACE;
394 } else {
395 fprintf(stderr, "Unknown node-action: %s\n", val);
396 node_ctx.action = NODEACT_UNKNOWN;
397 }
398 break;
399 case sizeof("Node-copyfrom-path"):
400 if (constcmp(t, "Node-copyfrom-path"))
401 continue;
402 strbuf_reset(&node_ctx.src);
403 strbuf_addstr(&node_ctx.src, val);
404 break;
405 case sizeof("Node-copyfrom-rev"):
406 if (constcmp(t, "Node-copyfrom-rev"))
407 continue;
408 node_ctx.srcRev = atoi(val);
409 break;
410 case sizeof("Text-content-length"):
411 if (constcmp(t, "Text") && constcmp(t, "Prop"))
412 continue;
413 if (constcmp(t + 4, "-content-length"))
414 continue;
415 {
416 char *end;
417 uintmax_t len;
418 len = strtoumax(val, &end, 10);
420 if (!isdigit(*val) || *end)
421 die("invalid dump: non-numeric length %s", val);
422 if (len > maximum_signed_value_of_type(off_t))
423 die("unrepresentable length in dump: %s", val);
424 if (*t == 'T')
426 node_ctx.text_length = (off_t) len;
427 else
428 node_ctx.prop_length = (off_t) len;
429 break;
430 }
431 case sizeof("Text-delta"):
432 if (!constcmp(t, "Text-delta")) {
433 node_ctx.text_delta = !strcmp(val, "true");
434 break;
435 }
436 if (constcmp(t, "Prop-delta"))
437 continue;
438 node_ctx.prop_delta = !strcmp(val, "true");
439 break;
440 case sizeof("Content-length"):
441 if (constcmp(t, "Content-length"))
442 continue;
443 len = atoi(val);
444 t = buffer_read_line(&input);
445 if (!t)
446 die_short_read();
447 if (*t)
448 die("invalid dump: expected blank line after content length header");
449 if (active_ctx == REV_CTX) {
450 read_props();
451 } else if (active_ctx == NODE_CTX) {
452 handle_node();
453 active_ctx = INTERNODE_CTX;
454 } else {
455 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
456 if (buffer_skip_bytes(&input, len) != len)
457 die_short_read();
458 }
459 }
460 }
461 if (buffer_ferror(&input))
462 die_short_read();
463 if (active_ctx == NODE_CTX)
464 handle_node();
465 if (active_ctx == REV_CTX)
466 begin_revision();
467 if (active_ctx != DUMP_CTX)
468 end_revision();
469}
470static void init(int report_fd)
472{
473 fast_export_init(report_fd);
474 strbuf_init(&dump_ctx.uuid, 4096);
475 strbuf_init(&dump_ctx.url, 4096);
476 strbuf_init(&rev_ctx.log, 4096);
477 strbuf_init(&rev_ctx.author, 4096);
478 strbuf_init(&node_ctx.src, 4096);
479 strbuf_init(&node_ctx.dst, 4096);
480 reset_dump_ctx(NULL);
481 reset_rev_ctx(0);
482 reset_node_ctx(NULL);
483 return;
484}
485int svndump_init(const char *filename)
487{
488 if (buffer_init(&input, filename))
489 return error("cannot open %s: %s", filename ? filename : "NULL", strerror(errno));
490 init(REPORT_FILENO);
491 return 0;
492}
493int svndump_init_fd(int in_fd, int back_fd)
495{
496 if(buffer_fdinit(&input, xdup(in_fd)))
497 return error("cannot open fd %d: %s", in_fd, strerror(errno));
498 init(xdup(back_fd));
499 return 0;
500}
501void svndump_deinit(void)
503{
504 fast_export_deinit();
505 reset_dump_ctx(NULL);
506 reset_rev_ctx(0);
507 reset_node_ctx(NULL);
508 strbuf_release(&rev_ctx.log);
509 strbuf_release(&node_ctx.src);
510 strbuf_release(&node_ctx.dst);
511 if (buffer_deinit(&input))
512 fprintf(stderr, "Input error\n");
513 if (ferror(stdout))
514 fprintf(stderr, "Output error\n");
515}
516void svndump_reset(void)
518{
519 strbuf_release(&dump_ctx.uuid);
520 strbuf_release(&dump_ctx.url);
521 strbuf_release(&rev_ctx.log);
522 strbuf_release(&rev_ctx.author);
523}