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 if (node_ctx.text_delta)
222 die("text deltas not supported");
223 if (node_ctx.action == NODEACT_DELETE) {
225 if (have_text || have_props || node_ctx.srcRev)
226 die("invalid dump: deletion node has "
227 "copyfrom info, text, or properties");
228 repo_delete(node_ctx.dst.buf);
229 return;
230 }
231 if (node_ctx.action == NODEACT_REPLACE) {
232 repo_delete(node_ctx.dst.buf);
233 node_ctx.action = NODEACT_ADD;
234 }
235 if (node_ctx.srcRev) {
236 repo_copy(node_ctx.srcRev, node_ctx.src.buf, node_ctx.dst.buf);
237 if (node_ctx.action == NODEACT_ADD)
238 node_ctx.action = NODEACT_CHANGE;
239 }
240 if (have_text && type == REPO_MODE_DIR)
241 die("invalid dump: directories cannot have text attached");
242 /*
244 * Find old content (old_data) and decide on the new mode.
245 */
246 if (node_ctx.action == NODEACT_CHANGE && !*node_ctx.dst.buf) {
247 if (type != REPO_MODE_DIR)
248 die("invalid dump: root of tree is not a regular file");
249 old_data = NULL;
250 } else if (node_ctx.action == NODEACT_CHANGE) {
251 uint32_t mode;
252 old_data = repo_read_path(node_ctx.dst.buf, &mode);
253 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
254 die("invalid dump: cannot modify a directory into a file");
255 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
256 die("invalid dump: cannot modify a file into a directory");
257 node_ctx.type = 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.propLength)
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 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
293 fast_export_data(node_ctx.type, node_ctx.textLength, &input);
294}
295static void begin_revision(void)
297{
298 if (!rev_ctx.revision) /* revision 0 gets no git commit. */
299 return;
300 fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
301 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
302 rev_ctx.timestamp);
303}
304static void end_revision(void)
306{
307 if (rev_ctx.revision)
308 fast_export_end_commit(rev_ctx.revision);
309}
310void svndump_read(const char *url)
312{
313 char *val;
314 char *t;
315 uint32_t active_ctx = DUMP_CTX;
316 uint32_t len;
317 reset_dump_ctx(url);
319 while ((t = buffer_read_line(&input))) {
320 val = strchr(t, ':');
321 if (!val)
322 continue;
323 val++;
324 if (*val != ' ')
325 continue;
326 val++;
327 /* strlen(key) + 1 */
329 switch (val - t - 1) {
330 case sizeof("SVN-fs-dump-format-version"):
331 if (constcmp(t, "SVN-fs-dump-format-version"))
332 continue;
333 dump_ctx.version = atoi(val);
334 if (dump_ctx.version > 3)
335 die("expected svn dump format version <= 3, found %"PRIu32,
336 dump_ctx.version);
337 break;
338 case sizeof("UUID"):
339 if (constcmp(t, "UUID"))
340 continue;
341 strbuf_reset(&dump_ctx.uuid);
342 strbuf_addstr(&dump_ctx.uuid, val);
343 break;
344 case sizeof("Revision-number"):
345 if (constcmp(t, "Revision-number"))
346 continue;
347 if (active_ctx == NODE_CTX)
348 handle_node();
349 if (active_ctx == REV_CTX)
350 begin_revision();
351 if (active_ctx != DUMP_CTX)
352 end_revision();
353 active_ctx = REV_CTX;
354 reset_rev_ctx(atoi(val));
355 break;
356 case sizeof("Node-path"):
357 if (prefixcmp(t, "Node-"))
358 continue;
359 if (!constcmp(t + strlen("Node-"), "path")) {
360 if (active_ctx == NODE_CTX)
361 handle_node();
362 if (active_ctx == REV_CTX)
363 begin_revision();
364 active_ctx = NODE_CTX;
365 reset_node_ctx(val);
366 break;
367 }
368 if (constcmp(t + strlen("Node-"), "kind"))
369 continue;
370 if (!strcmp(val, "dir"))
371 node_ctx.type = REPO_MODE_DIR;
372 else if (!strcmp(val, "file"))
373 node_ctx.type = REPO_MODE_BLB;
374 else
375 fprintf(stderr, "Unknown node-kind: %s\n", val);
376 break;
377 case sizeof("Node-action"):
378 if (constcmp(t, "Node-action"))
379 continue;
380 if (!strcmp(val, "delete")) {
381 node_ctx.action = NODEACT_DELETE;
382 } else if (!strcmp(val, "add")) {
383 node_ctx.action = NODEACT_ADD;
384 } else if (!strcmp(val, "change")) {
385 node_ctx.action = NODEACT_CHANGE;
386 } else if (!strcmp(val, "replace")) {
387 node_ctx.action = NODEACT_REPLACE;
388 } else {
389 fprintf(stderr, "Unknown node-action: %s\n", val);
390 node_ctx.action = NODEACT_UNKNOWN;
391 }
392 break;
393 case sizeof("Node-copyfrom-path"):
394 if (constcmp(t, "Node-copyfrom-path"))
395 continue;
396 strbuf_reset(&node_ctx.src);
397 strbuf_addstr(&node_ctx.src, val);
398 break;
399 case sizeof("Node-copyfrom-rev"):
400 if (constcmp(t, "Node-copyfrom-rev"))
401 continue;
402 node_ctx.srcRev = atoi(val);
403 break;
404 case sizeof("Text-content-length"):
405 if (!constcmp(t, "Text-content-length")) {
406 node_ctx.textLength = atoi(val);
407 break;
408 }
409 if (constcmp(t, "Prop-content-length"))
410 continue;
411 node_ctx.propLength = atoi(val);
412 break;
413 case sizeof("Text-delta"):
414 if (!constcmp(t, "Text-delta")) {
415 node_ctx.text_delta = !strcmp(val, "true");
416 break;
417 }
418 if (constcmp(t, "Prop-delta"))
419 continue;
420 node_ctx.prop_delta = !strcmp(val, "true");
421 break;
422 case sizeof("Content-length"):
423 if (constcmp(t, "Content-length"))
424 continue;
425 len = atoi(val);
426 t = buffer_read_line(&input);
427 if (!t)
428 die_short_read();
429 if (*t)
430 die("invalid dump: expected blank line after content length header");
431 if (active_ctx == REV_CTX) {
432 read_props();
433 } else if (active_ctx == NODE_CTX) {
434 handle_node();
435 active_ctx = INTERNODE_CTX;
436 } else {
437 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
438 if (buffer_skip_bytes(&input, len) != len)
439 die_short_read();
440 }
441 }
442 }
443 if (buffer_ferror(&input))
444 die_short_read();
445 if (active_ctx == NODE_CTX)
446 handle_node();
447 if (active_ctx == REV_CTX)
448 begin_revision();
449 if (active_ctx != DUMP_CTX)
450 end_revision();
451}
452int svndump_init(const char *filename)
454{
455 if (buffer_init(&input, filename))
456 return error("cannot open %s: %s", filename, strerror(errno));
457 fast_export_init(REPORT_FILENO);
458 strbuf_init(&dump_ctx.uuid, 4096);
459 strbuf_init(&dump_ctx.url, 4096);
460 strbuf_init(&rev_ctx.log, 4096);
461 strbuf_init(&rev_ctx.author, 4096);
462 strbuf_init(&node_ctx.src, 4096);
463 strbuf_init(&node_ctx.dst, 4096);
464 reset_dump_ctx(NULL);
465 reset_rev_ctx(0);
466 reset_node_ctx(NULL);
467 return 0;
468}
469void svndump_deinit(void)
471{
472 fast_export_deinit();
473 reset_dump_ctx(NULL);
474 reset_rev_ctx(0);
475 reset_node_ctx(NULL);
476 strbuf_release(&rev_ctx.log);
477 strbuf_release(&node_ctx.src);
478 strbuf_release(&node_ctx.dst);
479 if (buffer_deinit(&input))
480 fprintf(stderr, "Input error\n");
481 if (ferror(stdout))
482 fprintf(stderr, "Output error\n");
483}
484void svndump_reset(void)
486{
487 fast_export_reset();
488 buffer_reset(&input);
489 strbuf_release(&dump_ctx.uuid);
490 strbuf_release(&dump_ctx.url);
491 strbuf_release(&rev_ctx.log);
492 strbuf_release(&rev_ctx.author);
493}