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#define REPORT_FILENO 3
17/*
19 * Compare start of string to literal of equal length;
20 * must be guarded by length test.
21 */
22#define constcmp(s, ref) memcmp(s, ref, sizeof(ref) - 1)
23#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/* 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#define LENGTH_UNKNOWN (~0)
37#define DATE_RFC2822_LEN 31
38static struct line_buffer input = LINE_BUFFER_INIT;
40static struct {
42 uint32_t action, propLength, textLength, srcRev, type;
43 struct strbuf src, dst;
44 uint32_t text_delta, prop_delta;
45} node_ctx;
46static struct {
48 uint32_t revision;
49 unsigned long timestamp;
50 struct strbuf log, author;
51} rev_ctx;
52static struct {
54 uint32_t version;
55 struct strbuf uuid, url;
56} dump_ctx;
57static void reset_node_ctx(char *fname)
59{
60 node_ctx.type = 0;
61 node_ctx.action = NODEACT_UNKNOWN;
62 node_ctx.propLength = LENGTH_UNKNOWN;
63 node_ctx.textLength = LENGTH_UNKNOWN;
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}
72static 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}
80static void reset_dump_ctx(const char *url)
82{
83 strbuf_reset(&dump_ctx.url);
84 if (url)
85 strbuf_addstr(&dump_ctx.url, url);
86 dump_ctx.version = 1;
87 strbuf_reset(&dump_ctx.uuid);
88}
89static void handle_property(const struct strbuf *key_buf,
91 const char *val, uint32_t len,
92 uint32_t *type_set)
93{
94 const char *key = key_buf->buf;
95 size_t keylen = key_buf->len;
96 switch (keylen + 1) {
98 case sizeof("svn:log"):
99 if (constcmp(key, "svn:log"))
100 break;
101 if (!val)
102 die("invalid dump: unsets svn:log");
103 strbuf_reset(&rev_ctx.log);
104 strbuf_add(&rev_ctx.log, val, len);
105 break;
106 case sizeof("svn:author"):
107 if (constcmp(key, "svn:author"))
108 break;
109 strbuf_reset(&rev_ctx.author);
110 if (val)
111 strbuf_add(&rev_ctx.author, val, len);
112 break;
113 case sizeof("svn:date"):
114 if (constcmp(key, "svn:date"))
115 break;
116 if (!val)
117 die("invalid dump: unsets svn:date");
118 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
119 warning("invalid timestamp: %s", val);
120 break;
121 case sizeof("svn:executable"):
122 case sizeof("svn:special"):
123 if (keylen == strlen("svn:executable") &&
124 constcmp(key, "svn:executable"))
125 break;
126 if (keylen == strlen("svn:special") &&
127 constcmp(key, "svn:special"))
128 break;
129 if (*type_set) {
130 if (!val)
131 return;
132 die("invalid dump: sets type twice");
133 }
134 if (!val) {
135 node_ctx.type = REPO_MODE_BLB;
136 return;
137 }
138 *type_set = 1;
139 node_ctx.type = keylen == strlen("svn:executable") ?
140 REPO_MODE_EXE :
141 REPO_MODE_LNK;
142 }
143}
144static void die_short_read(void)
146{
147 if (buffer_ferror(&input))
148 die_errno("error reading dump file");
149 die("invalid dump: unexpected end of file");
150}
151static void read_props(void)
153{
154 static struct strbuf key = STRBUF_INIT;
155 const char *t;
156 /*
157 * NEEDSWORK: to support simple mode changes like
158 * K 11
159 * svn:special
160 * V 1
161 * *
162 * D 14
163 * svn:executable
164 * we keep track of whether a mode has been set and reset to
165 * plain file only if not. We should be keeping track of the
166 * symlink and executable bits separately instead.
167 */
168 uint32_t type_set = 0;
169 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
170 uint32_t len;
171 const char *val;
172 const char type = t[0];
173 int ch;
174 if (!type || t[1] != ' ')
176 die("invalid property line: %s\n", t);
177 len = atoi(&t[2]);
178 val = buffer_read_string(&input, len);
179 if (!val || strlen(val) != len)
180 die_short_read();
181 /* Discard trailing newline. */
183 ch = buffer_read_char(&input);
184 if (ch == EOF)
185 die_short_read();
186 if (ch != '\n')
187 die("invalid dump: expected newline after %s", val);
188 switch (type) {
190 case 'K':
191 case 'D':
192 strbuf_reset(&key);
193 if (val)
194 strbuf_add(&key, val, len);
195 if (type == 'K')
196 continue;
197 assert(type == 'D');
198 val = NULL;
199 len = 0;
200 /* fall through */
201 case 'V':
202 handle_property(&key, val, len, &type_set);
203 strbuf_reset(&key);
204 continue;
205 default:
206 die("invalid property line: %s\n", t);
207 }
208 }
209}
210static void handle_node(void)
212{
213 const uint32_t type = node_ctx.type;
214 const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
215 const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
216 /*
217 * Old text for this node:
218 * NULL - directory or bug
219 * empty_blob - empty
220 * "<dataref>" - data retrievable from fast-import
221 */
222 static const char *const empty_blob = "::empty::";
223 const char *old_data = NULL;
224 if (node_ctx.text_delta)
226 die("text deltas not supported");
227 if (node_ctx.action == NODEACT_DELETE) {
229 if (have_text || have_props || node_ctx.srcRev)
230 die("invalid dump: deletion node has "
231 "copyfrom info, text, or properties");
232 return repo_delete(node_ctx.dst.buf);
233 }
234 if (node_ctx.action == NODEACT_REPLACE) {
235 repo_delete(node_ctx.dst.buf);
236 node_ctx.action = NODEACT_ADD;
237 }
238 if (node_ctx.srcRev) {
239 repo_copy(node_ctx.srcRev, node_ctx.src.buf, node_ctx.dst.buf);
240 if (node_ctx.action == NODEACT_ADD)
241 node_ctx.action = NODEACT_CHANGE;
242 }
243 if (have_text && type == REPO_MODE_DIR)
244 die("invalid dump: directories cannot have text attached");
245 /*
247 * Find old content (old_data) and decide on the new mode.
248 */
249 if (node_ctx.action == NODEACT_CHANGE && !*node_ctx.dst.buf) {
250 if (type != REPO_MODE_DIR)
251 die("invalid dump: root of tree is not a regular file");
252 old_data = NULL;
253 } else if (node_ctx.action == NODEACT_CHANGE) {
254 uint32_t mode;
255 old_data = repo_read_path(node_ctx.dst.buf);
256 mode = repo_read_mode(node_ctx.dst.buf);
257 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
258 die("invalid dump: cannot modify a directory into a file");
259 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
260 die("invalid dump: cannot modify a file into a directory");
261 node_ctx.type = mode;
262 } else if (node_ctx.action == NODEACT_ADD) {
263 if (type == REPO_MODE_DIR)
264 old_data = NULL;
265 else if (have_text)
266 old_data = empty_blob;
267 else
268 die("invalid dump: adds node without text");
269 } else {
270 die("invalid dump: Node-path block lacks Node-action");
271 }
272 /*
274 * Adjust mode to reflect properties.
275 */
276 if (have_props) {
277 if (!node_ctx.prop_delta)
278 node_ctx.type = type;
279 if (node_ctx.propLength)
280 read_props();
281 }
282 /*
284 * Save the result.
285 */
286 if (type == REPO_MODE_DIR) /* directories are not tracked. */
287 return;
288 assert(old_data);
289 if (old_data == empty_blob)
290 /* For the fast_export_* functions, NULL means empty. */
291 old_data = NULL;
292 if (!have_text) {
293 fast_export_modify(node_ctx.dst.buf, node_ctx.type, old_data);
294 return;
295 }
296 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
297 fast_export_data(node_ctx.type, node_ctx.textLength, &input);
298}
299static void begin_revision(void)
301{
302 if (!rev_ctx.revision) /* revision 0 gets no git commit. */
303 return;
304 fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
305 rev_ctx.log.buf, dump_ctx.uuid.buf, dump_ctx.url.buf,
306 rev_ctx.timestamp);
307}
308static void end_revision(void)
310{
311 if (rev_ctx.revision)
312 fast_export_end_commit(rev_ctx.revision);
313}
314void svndump_read(const char *url)
316{
317 char *val;
318 char *t;
319 uint32_t active_ctx = DUMP_CTX;
320 uint32_t len;
321 reset_dump_ctx(url);
323 while ((t = buffer_read_line(&input))) {
324 val = strstr(t, ": ");
325 if (!val)
326 continue;
327 val += 2;
328 /* strlen(key) + 1 */
330 switch (val - t - 1) {
331 case sizeof("SVN-fs-dump-format-version"):
332 if (constcmp(t, "SVN-fs-dump-format-version"))
333 continue;
334 dump_ctx.version = atoi(val);
335 if (dump_ctx.version > 3)
336 die("expected svn dump format version <= 3, found %"PRIu32,
337 dump_ctx.version);
338 break;
339 case sizeof("UUID"):
340 if (constcmp(t, "UUID"))
341 continue;
342 strbuf_reset(&dump_ctx.uuid);
343 strbuf_addstr(&dump_ctx.uuid, val);
344 break;
345 case sizeof("Revision-number"):
346 if (constcmp(t, "Revision-number"))
347 continue;
348 if (active_ctx == NODE_CTX)
349 handle_node();
350 if (active_ctx == REV_CTX)
351 begin_revision();
352 if (active_ctx != DUMP_CTX)
353 end_revision();
354 active_ctx = REV_CTX;
355 reset_rev_ctx(atoi(val));
356 break;
357 case sizeof("Node-path"):
358 if (prefixcmp(t, "Node-"))
359 continue;
360 if (!constcmp(t + strlen("Node-"), "path")) {
361 if (active_ctx == NODE_CTX)
362 handle_node();
363 if (active_ctx == REV_CTX)
364 begin_revision();
365 active_ctx = NODE_CTX;
366 reset_node_ctx(val);
367 break;
368 }
369 if (constcmp(t + strlen("Node-"), "kind"))
370 continue;
371 if (!strcmp(val, "dir"))
372 node_ctx.type = REPO_MODE_DIR;
373 else if (!strcmp(val, "file"))
374 node_ctx.type = REPO_MODE_BLB;
375 else
376 fprintf(stderr, "Unknown node-kind: %s\n", val);
377 break;
378 case sizeof("Node-action"):
379 if (constcmp(t, "Node-action"))
380 continue;
381 if (!strcmp(val, "delete")) {
382 node_ctx.action = NODEACT_DELETE;
383 } else if (!strcmp(val, "add")) {
384 node_ctx.action = NODEACT_ADD;
385 } else if (!strcmp(val, "change")) {
386 node_ctx.action = NODEACT_CHANGE;
387 } else if (!strcmp(val, "replace")) {
388 node_ctx.action = NODEACT_REPLACE;
389 } else {
390 fprintf(stderr, "Unknown node-action: %s\n", val);
391 node_ctx.action = NODEACT_UNKNOWN;
392 }
393 break;
394 case sizeof("Node-copyfrom-path"):
395 if (constcmp(t, "Node-copyfrom-path"))
396 continue;
397 strbuf_reset(&node_ctx.src);
398 strbuf_addstr(&node_ctx.src, val);
399 break;
400 case sizeof("Node-copyfrom-rev"):
401 if (constcmp(t, "Node-copyfrom-rev"))
402 continue;
403 node_ctx.srcRev = atoi(val);
404 break;
405 case sizeof("Text-content-length"):
406 if (!constcmp(t, "Text-content-length")) {
407 node_ctx.textLength = atoi(val);
408 break;
409 }
410 if (constcmp(t, "Prop-content-length"))
411 continue;
412 node_ctx.propLength = atoi(val);
413 break;
414 case sizeof("Text-delta"):
415 if (!constcmp(t, "Text-delta")) {
416 node_ctx.text_delta = !strcmp(val, "true");
417 break;
418 }
419 if (constcmp(t, "Prop-delta"))
420 continue;
421 node_ctx.prop_delta = !strcmp(val, "true");
422 break;
423 case sizeof("Content-length"):
424 if (constcmp(t, "Content-length"))
425 continue;
426 len = atoi(val);
427 t = buffer_read_line(&input);
428 if (!t)
429 die_short_read();
430 if (*t)
431 die("invalid dump: expected blank line after content length header");
432 if (active_ctx == REV_CTX) {
433 read_props();
434 } else if (active_ctx == NODE_CTX) {
435 handle_node();
436 active_ctx = INTERNODE_CTX;
437 } else {
438 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
439 if (buffer_skip_bytes(&input, len) != len)
440 die_short_read();
441 }
442 }
443 }
444 if (buffer_ferror(&input))
445 die_short_read();
446 if (active_ctx == NODE_CTX)
447 handle_node();
448 if (active_ctx == REV_CTX)
449 begin_revision();
450 if (active_ctx != DUMP_CTX)
451 end_revision();
452}
453int svndump_init(const char *filename)
455{
456 if (buffer_init(&input, filename))
457 return error("cannot open %s: %s", filename, strerror(errno));
458 fast_export_init(REPORT_FILENO);
459 strbuf_init(&dump_ctx.uuid, 4096);
460 strbuf_init(&dump_ctx.url, 4096);
461 strbuf_init(&rev_ctx.log, 4096);
462 strbuf_init(&rev_ctx.author, 4096);
463 strbuf_init(&node_ctx.src, 4096);
464 strbuf_init(&node_ctx.dst, 4096);
465 reset_dump_ctx(NULL);
466 reset_rev_ctx(0);
467 reset_node_ctx(NULL);
468 return 0;
469}
470void svndump_deinit(void)
472{
473 fast_export_deinit();
474 reset_dump_ctx(NULL);
475 reset_rev_ctx(0);
476 reset_node_ctx(NULL);
477 strbuf_release(&rev_ctx.log);
478 strbuf_release(&node_ctx.src);
479 strbuf_release(&node_ctx.dst);
480 if (buffer_deinit(&input))
481 fprintf(stderr, "Input error\n");
482 if (ferror(stdout))
483 fprintf(stderr, "Output error\n");
484}
485void svndump_reset(void)
487{
488 fast_export_reset();
489 buffer_reset(&input);
490 strbuf_release(&dump_ctx.uuid);
491 strbuf_release(&dump_ctx.url);
492 strbuf_release(&rev_ctx.log);
493 strbuf_release(&rev_ctx.author);
494}