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/*
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#define REPORT_FILENO 3
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 struct strbuf *val,
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_swap(&rev_ctx.log, val);
104 break;
105 case sizeof("svn:author"):
106 if (constcmp(key, "svn:author"))
107 break;
108 if (!val)
109 strbuf_reset(&rev_ctx.author);
110 else
111 strbuf_swap(&rev_ctx.author, val);
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->buf, &rev_ctx.timestamp, NULL))
119 warning("invalid timestamp: %s", val->buf);
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 static struct strbuf val = STRBUF_INIT;
156 const char *t;
157 /*
158 * NEEDSWORK: to support simple mode changes like
159 * K 11
160 * svn:special
161 * V 1
162 * *
163 * D 14
164 * svn:executable
165 * we keep track of whether a mode has been set and reset to
166 * plain file only if not. We should be keeping track of the
167 * symlink and executable bits separately instead.
168 */
169 uint32_t type_set = 0;
170 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
171 uint32_t len;
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 strbuf_reset(&val);
179 buffer_read_binary(&input, &val, len);
180 if (val.len < len)
181 die_short_read();
182 /* Discard trailing newline. */
184 ch = buffer_read_char(&input);
185 if (ch == EOF)
186 die_short_read();
187 if (ch != '\n')
188 die("invalid dump: expected newline after %s", val.buf);
189 switch (type) {
191 case 'K':
192 strbuf_swap(&key, &val);
193 continue;
194 case 'D':
195 handle_property(&val, NULL, &type_set);
196 continue;
197 case 'V':
198 handle_property(&key, &val, &type_set);
199 strbuf_reset(&key);
200 continue;
201 default:
202 die("invalid property line: %s\n", t);
203 }
204 }
205}
206static void handle_node(void)
208{
209 const uint32_t type = node_ctx.type;
210 const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
211 const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
212 /*
213 * Old text for this node:
214 * NULL - directory or bug
215 * empty_blob - empty
216 * "<dataref>" - data retrievable from fast-import
217 */
218 static const char *const empty_blob = "::empty::";
219 const char *old_data = NULL;
220 uint32_t old_mode = REPO_MODE_BLB;
221 if (node_ctx.action == NODEACT_DELETE) {
223 if (have_text || have_props || node_ctx.srcRev)
224 die("invalid dump: deletion node has "
225 "copyfrom info, text, or properties");
226 repo_delete(node_ctx.dst.buf);
227 return;
228 }
229 if (node_ctx.action == NODEACT_REPLACE) {
230 repo_delete(node_ctx.dst.buf);
231 node_ctx.action = NODEACT_ADD;
232 }
233 if (node_ctx.srcRev) {
234 repo_copy(node_ctx.srcRev, node_ctx.src.buf, node_ctx.dst.buf);
235 if (node_ctx.action == NODEACT_ADD)
236 node_ctx.action = NODEACT_CHANGE;
237 }
238 if (have_text && type == REPO_MODE_DIR)
239 die("invalid dump: directories cannot have text attached");
240 /*
242 * Find old content (old_data) and decide on the new mode.
243 */
244 if (node_ctx.action == NODEACT_CHANGE && !*node_ctx.dst.buf) {
245 if (type != REPO_MODE_DIR)
246 die("invalid dump: root of tree is not a regular file");
247 old_data = NULL;
248 } else if (node_ctx.action == NODEACT_CHANGE) {
249 uint32_t mode;
250 old_data = repo_read_path(node_ctx.dst.buf, &mode);
251 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
252 die("invalid dump: cannot modify a directory into a file");
253 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
254 die("invalid dump: cannot modify a file into a directory");
255 node_ctx.type = mode;
256 old_mode = mode;
257 } else if (node_ctx.action == NODEACT_ADD) {
258 if (type == REPO_MODE_DIR)
259 old_data = NULL;
260 else if (have_text)
261 old_data = empty_blob;
262 else
263 die("invalid dump: adds node without text");
264 } else {
265 die("invalid dump: Node-path block lacks Node-action");
266 }
267 /*
269 * Adjust mode to reflect properties.
270 */
271 if (have_props) {
272 if (!node_ctx.prop_delta)
273 node_ctx.type = type;
274 if (node_ctx.propLength)
275 read_props();
276 }
277 /*
279 * Save the result.
280 */
281 if (type == REPO_MODE_DIR) /* directories are not tracked. */
282 return;
283 assert(old_data);
284 if (old_data == empty_blob)
285 /* For the fast_export_* functions, NULL means empty. */
286 old_data = NULL;
287 if (!have_text) {
288 fast_export_modify(node_ctx.dst.buf, node_ctx.type, old_data);
289 return;
290 }
291 if (!node_ctx.text_delta) {
292 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
293 fast_export_data(node_ctx.type, node_ctx.textLength, &input);
294 return;
295 }
296 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
297 fast_export_blob_delta(node_ctx.type, old_mode, old_data,
298 node_ctx.textLength, &input);
299}
300static void begin_revision(void)
302{
303 if (!rev_ctx.revision) /* revision 0 gets no git commit. */
304 return;
305 fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
306 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
307 rev_ctx.timestamp);
308}
309static void end_revision(void)
311{
312 if (rev_ctx.revision)
313 fast_export_end_commit(rev_ctx.revision);
314}
315void svndump_read(const char *url)
317{
318 char *val;
319 char *t;
320 uint32_t active_ctx = DUMP_CTX;
321 uint32_t len;
322 reset_dump_ctx(url);
324 while ((t = buffer_read_line(&input))) {
325 val = strchr(t, ':');
326 if (!val)
327 continue;
328 val++;
329 if (*val != ' ')
330 continue;
331 val++;
332 /* strlen(key) + 1 */
334 switch (val - t - 1) {
335 case sizeof("SVN-fs-dump-format-version"):
336 if (constcmp(t, "SVN-fs-dump-format-version"))
337 continue;
338 dump_ctx.version = atoi(val);
339 if (dump_ctx.version > 3)
340 die("expected svn dump format version <= 3, found %"PRIu32,
341 dump_ctx.version);
342 break;
343 case sizeof("UUID"):
344 if (constcmp(t, "UUID"))
345 continue;
346 strbuf_reset(&dump_ctx.uuid);
347 strbuf_addstr(&dump_ctx.uuid, val);
348 break;
349 case sizeof("Revision-number"):
350 if (constcmp(t, "Revision-number"))
351 continue;
352 if (active_ctx == NODE_CTX)
353 handle_node();
354 if (active_ctx == REV_CTX)
355 begin_revision();
356 if (active_ctx != DUMP_CTX)
357 end_revision();
358 active_ctx = REV_CTX;
359 reset_rev_ctx(atoi(val));
360 break;
361 case sizeof("Node-path"):
362 if (prefixcmp(t, "Node-"))
363 continue;
364 if (!constcmp(t + strlen("Node-"), "path")) {
365 if (active_ctx == NODE_CTX)
366 handle_node();
367 if (active_ctx == REV_CTX)
368 begin_revision();
369 active_ctx = NODE_CTX;
370 reset_node_ctx(val);
371 break;
372 }
373 if (constcmp(t + strlen("Node-"), "kind"))
374 continue;
375 if (!strcmp(val, "dir"))
376 node_ctx.type = REPO_MODE_DIR;
377 else if (!strcmp(val, "file"))
378 node_ctx.type = REPO_MODE_BLB;
379 else
380 fprintf(stderr, "Unknown node-kind: %s\n", val);
381 break;
382 case sizeof("Node-action"):
383 if (constcmp(t, "Node-action"))
384 continue;
385 if (!strcmp(val, "delete")) {
386 node_ctx.action = NODEACT_DELETE;
387 } else if (!strcmp(val, "add")) {
388 node_ctx.action = NODEACT_ADD;
389 } else if (!strcmp(val, "change")) {
390 node_ctx.action = NODEACT_CHANGE;
391 } else if (!strcmp(val, "replace")) {
392 node_ctx.action = NODEACT_REPLACE;
393 } else {
394 fprintf(stderr, "Unknown node-action: %s\n", val);
395 node_ctx.action = NODEACT_UNKNOWN;
396 }
397 break;
398 case sizeof("Node-copyfrom-path"):
399 if (constcmp(t, "Node-copyfrom-path"))
400 continue;
401 strbuf_reset(&node_ctx.src);
402 strbuf_addstr(&node_ctx.src, val);
403 break;
404 case sizeof("Node-copyfrom-rev"):
405 if (constcmp(t, "Node-copyfrom-rev"))
406 continue;
407 node_ctx.srcRev = atoi(val);
408 break;
409 case sizeof("Text-content-length"):
410 if (!constcmp(t, "Text-content-length")) {
411 node_ctx.textLength = atoi(val);
412 break;
413 }
414 if (constcmp(t, "Prop-content-length"))
415 continue;
416 node_ctx.propLength = atoi(val);
417 break;
418 case sizeof("Text-delta"):
419 if (!constcmp(t, "Text-delta")) {
420 node_ctx.text_delta = !strcmp(val, "true");
421 break;
422 }
423 if (constcmp(t, "Prop-delta"))
424 continue;
425 node_ctx.prop_delta = !strcmp(val, "true");
426 break;
427 case sizeof("Content-length"):
428 if (constcmp(t, "Content-length"))
429 continue;
430 len = atoi(val);
431 t = buffer_read_line(&input);
432 if (!t)
433 die_short_read();
434 if (*t)
435 die("invalid dump: expected blank line after content length header");
436 if (active_ctx == REV_CTX) {
437 read_props();
438 } else if (active_ctx == NODE_CTX) {
439 handle_node();
440 active_ctx = INTERNODE_CTX;
441 } else {
442 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
443 if (buffer_skip_bytes(&input, len) != len)
444 die_short_read();
445 }
446 }
447 }
448 if (buffer_ferror(&input))
449 die_short_read();
450 if (active_ctx == NODE_CTX)
451 handle_node();
452 if (active_ctx == REV_CTX)
453 begin_revision();
454 if (active_ctx != DUMP_CTX)
455 end_revision();
456}
457int svndump_init(const char *filename)
459{
460 if (buffer_init(&input, filename))
461 return error("cannot open %s: %s", filename, strerror(errno));
462 fast_export_init(REPORT_FILENO);
463 strbuf_init(&dump_ctx.uuid, 4096);
464 strbuf_init(&dump_ctx.url, 4096);
465 strbuf_init(&rev_ctx.log, 4096);
466 strbuf_init(&rev_ctx.author, 4096);
467 strbuf_init(&node_ctx.src, 4096);
468 strbuf_init(&node_ctx.dst, 4096);
469 reset_dump_ctx(NULL);
470 reset_rev_ctx(0);
471 reset_node_ctx(NULL);
472 return 0;
473}
474void svndump_deinit(void)
476{
477 fast_export_deinit();
478 reset_dump_ctx(NULL);
479 reset_rev_ctx(0);
480 reset_node_ctx(NULL);
481 strbuf_release(&rev_ctx.log);
482 strbuf_release(&node_ctx.src);
483 strbuf_release(&node_ctx.dst);
484 if (buffer_deinit(&input))
485 fprintf(stderr, "Input error\n");
486 if (ferror(stdout))
487 fprintf(stderr, "Output error\n");
488}
489void svndump_reset(void)
491{
492 fast_export_reset();
493 buffer_reset(&input);
494 strbuf_release(&dump_ctx.uuid);
495 strbuf_release(&dump_ctx.url);
496 strbuf_release(&rev_ctx.log);
497 strbuf_release(&rev_ctx.author);
498}