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 "string_pool.h"
15#include "strbuf.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 NODEACT_REPLACE 4
24#define NODEACT_DELETE 3
25#define NODEACT_ADD 2
26#define NODEACT_CHANGE 1
27#define NODEACT_UNKNOWN 0
28#define DUMP_CTX 0
30#define REV_CTX 1
31#define NODE_CTX 2
32#define LENGTH_UNKNOWN (~0)
34#define DATE_RFC2822_LEN 31
35static struct line_buffer input = LINE_BUFFER_INIT;
37static struct {
39 uint32_t action, propLength, textLength, srcRev, type;
40 uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
41 uint32_t text_delta, prop_delta;
42} node_ctx;
43static struct {
45 uint32_t revision;
46 unsigned long timestamp;
47 struct strbuf log, author;
48} rev_ctx;
49static struct {
51 uint32_t version;
52 struct strbuf uuid, url;
53} dump_ctx;
54static void reset_node_ctx(char *fname)
56{
57 node_ctx.type = 0;
58 node_ctx.action = NODEACT_UNKNOWN;
59 node_ctx.propLength = LENGTH_UNKNOWN;
60 node_ctx.textLength = LENGTH_UNKNOWN;
61 node_ctx.src[0] = ~0;
62 node_ctx.srcRev = 0;
63 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
64 node_ctx.text_delta = 0;
65 node_ctx.prop_delta = 0;
66}
67static void reset_rev_ctx(uint32_t revision)
69{
70 rev_ctx.revision = revision;
71 rev_ctx.timestamp = 0;
72 strbuf_reset(&rev_ctx.log);
73 strbuf_reset(&rev_ctx.author);
74}
75static void reset_dump_ctx(const char *url)
77{
78 strbuf_reset(&dump_ctx.url);
79 if (url)
80 strbuf_addstr(&dump_ctx.url, url);
81 dump_ctx.version = 1;
82 strbuf_reset(&dump_ctx.uuid);
83}
84static void handle_property(const struct strbuf *key_buf,
86 const char *val, uint32_t len,
87 uint32_t *type_set)
88{
89 const char *key = key_buf->buf;
90 size_t keylen = key_buf->len;
91 switch (keylen + 1) {
93 case sizeof("svn:log"):
94 if (constcmp(key, "svn:log"))
95 break;
96 if (!val)
97 die("invalid dump: unsets svn:log");
98 strbuf_reset(&rev_ctx.log);
99 strbuf_add(&rev_ctx.log, val, len);
100 break;
101 case sizeof("svn:author"):
102 if (constcmp(key, "svn:author"))
103 break;
104 strbuf_reset(&rev_ctx.author);
105 if (val)
106 strbuf_add(&rev_ctx.author, val, len);
107 break;
108 case sizeof("svn:date"):
109 if (constcmp(key, "svn:date"))
110 break;
111 if (!val)
112 die("invalid dump: unsets svn:date");
113 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
114 warning("invalid timestamp: %s", val);
115 break;
116 case sizeof("svn:executable"):
117 case sizeof("svn:special"):
118 if (keylen == strlen("svn:executable") &&
119 constcmp(key, "svn:executable"))
120 break;
121 if (keylen == strlen("svn:special") &&
122 constcmp(key, "svn:special"))
123 break;
124 if (*type_set) {
125 if (!val)
126 return;
127 die("invalid dump: sets type twice");
128 }
129 if (!val) {
130 node_ctx.type = REPO_MODE_BLB;
131 return;
132 }
133 *type_set = 1;
134 node_ctx.type = keylen == strlen("svn:executable") ?
135 REPO_MODE_EXE :
136 REPO_MODE_LNK;
137 }
138}
139static void die_short_read(void)
141{
142 if (buffer_ferror(&input))
143 die_errno("error reading dump file");
144 die("invalid dump: unexpected end of file");
145}
146static void read_props(void)
148{
149 static struct strbuf key = STRBUF_INIT;
150 const char *t;
151 /*
152 * NEEDSWORK: to support simple mode changes like
153 * K 11
154 * svn:special
155 * V 1
156 * *
157 * D 14
158 * svn:executable
159 * we keep track of whether a mode has been set and reset to
160 * plain file only if not. We should be keeping track of the
161 * symlink and executable bits separately instead.
162 */
163 uint32_t type_set = 0;
164 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
165 uint32_t len;
166 const char *val;
167 const char type = t[0];
168 int ch;
169 if (!type || t[1] != ' ')
171 die("invalid property line: %s\n", t);
172 len = atoi(&t[2]);
173 val = buffer_read_string(&input, len);
174 if (!val || strlen(val) != len)
175 die_short_read();
176 /* Discard trailing newline. */
178 ch = buffer_read_char(&input);
179 if (ch == EOF)
180 die_short_read();
181 if (ch != '\n')
182 die("invalid dump: expected newline after %s", val);
183 switch (type) {
185 case 'K':
186 case 'D':
187 strbuf_reset(&key);
188 if (val)
189 strbuf_add(&key, val, len);
190 if (type == 'K')
191 continue;
192 assert(type == 'D');
193 val = NULL;
194 len = 0;
195 /* fall through */
196 case 'V':
197 handle_property(&key, val, len, &type_set);
198 strbuf_reset(&key);
199 continue;
200 default:
201 die("invalid property line: %s\n", t);
202 }
203 }
204}
205static void handle_node(void)
207{
208 uint32_t mark = 0;
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 if (node_ctx.text_delta)
214 die("text deltas not supported");
215 if (have_text)
216 mark = next_blob_mark();
217 if (node_ctx.action == NODEACT_DELETE) {
218 if (have_text || have_props || node_ctx.srcRev)
219 die("invalid dump: deletion node has "
220 "copyfrom info, text, or properties");
221 return repo_delete(node_ctx.dst);
222 }
223 if (node_ctx.action == NODEACT_REPLACE) {
224 repo_delete(node_ctx.dst);
225 node_ctx.action = NODEACT_ADD;
226 }
227 if (node_ctx.srcRev) {
228 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
229 if (node_ctx.action == NODEACT_ADD)
230 node_ctx.action = NODEACT_CHANGE;
231 }
232 if (have_text && type == REPO_MODE_DIR)
233 die("invalid dump: directories cannot have text attached");
234 /*
236 * Decide on the new content (mark) and mode (node_ctx.type).
237 */
238 if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
239 if (type != REPO_MODE_DIR)
240 die("invalid dump: root of tree is not a regular file");
241 } else if (node_ctx.action == NODEACT_CHANGE) {
242 uint32_t mode;
243 if (!have_text)
244 mark = repo_read_path(node_ctx.dst);
245 mode = repo_read_mode(node_ctx.dst);
246 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
247 die("invalid dump: cannot modify a directory into a file");
248 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
249 die("invalid dump: cannot modify a file into a directory");
250 node_ctx.type = mode;
251 } else if (node_ctx.action == NODEACT_ADD) {
252 if (!have_text && type != REPO_MODE_DIR)
253 die("invalid dump: adds node without text");
254 } else {
255 die("invalid dump: Node-path block lacks Node-action");
256 }
257 /*
259 * Adjust mode to reflect properties.
260 */
261 if (have_props) {
262 if (!node_ctx.prop_delta)
263 node_ctx.type = type;
264 if (node_ctx.propLength)
265 read_props();
266 }
267 /*
269 * Save the result.
270 */
271 repo_add(node_ctx.dst, node_ctx.type, mark);
272 if (have_text)
273 fast_export_blob(node_ctx.type, mark,
274 node_ctx.textLength, &input);
275}
276static void handle_revision(void)
278{
279 if (rev_ctx.revision)
280 repo_commit(rev_ctx.revision, rev_ctx.author.buf,
281 rev_ctx.log.buf, dump_ctx.uuid.buf, dump_ctx.url.buf,
282 rev_ctx.timestamp);
283}
284void svndump_read(const char *url)
286{
287 char *val;
288 char *t;
289 uint32_t active_ctx = DUMP_CTX;
290 uint32_t len;
291 reset_dump_ctx(url);
293 while ((t = buffer_read_line(&input))) {
294 val = strchr(t, ':');
295 if (!val)
296 continue;
297 val++;
298 if (*val != ' ')
299 continue;
300 val++;
301 /* strlen(key) + 1 */
303 switch (val - t - 1) {
304 case sizeof("SVN-fs-dump-format-version"):
305 if (constcmp(t, "SVN-fs-dump-format-version"))
306 continue;
307 dump_ctx.version = atoi(val);
308 if (dump_ctx.version > 3)
309 die("expected svn dump format version <= 3, found %"PRIu32,
310 dump_ctx.version);
311 break;
312 case sizeof("UUID"):
313 if (constcmp(t, "UUID"))
314 continue;
315 strbuf_reset(&dump_ctx.uuid);
316 strbuf_addstr(&dump_ctx.uuid, val);
317 break;
318 case sizeof("Revision-number"):
319 if (constcmp(t, "Revision-number"))
320 continue;
321 if (active_ctx == NODE_CTX)
322 handle_node();
323 if (active_ctx != DUMP_CTX)
324 handle_revision();
325 active_ctx = REV_CTX;
326 reset_rev_ctx(atoi(val));
327 break;
328 case sizeof("Node-path"):
329 if (prefixcmp(t, "Node-"))
330 continue;
331 if (!constcmp(t + strlen("Node-"), "path")) {
332 if (active_ctx == NODE_CTX)
333 handle_node();
334 active_ctx = NODE_CTX;
335 reset_node_ctx(val);
336 break;
337 }
338 if (constcmp(t + strlen("Node-"), "kind"))
339 continue;
340 if (!strcmp(val, "dir"))
341 node_ctx.type = REPO_MODE_DIR;
342 else if (!strcmp(val, "file"))
343 node_ctx.type = REPO_MODE_BLB;
344 else
345 fprintf(stderr, "Unknown node-kind: %s\n", val);
346 break;
347 case sizeof("Node-action"):
348 if (constcmp(t, "Node-action"))
349 continue;
350 if (!strcmp(val, "delete")) {
351 node_ctx.action = NODEACT_DELETE;
352 } else if (!strcmp(val, "add")) {
353 node_ctx.action = NODEACT_ADD;
354 } else if (!strcmp(val, "change")) {
355 node_ctx.action = NODEACT_CHANGE;
356 } else if (!strcmp(val, "replace")) {
357 node_ctx.action = NODEACT_REPLACE;
358 } else {
359 fprintf(stderr, "Unknown node-action: %s\n", val);
360 node_ctx.action = NODEACT_UNKNOWN;
361 }
362 break;
363 case sizeof("Node-copyfrom-path"):
364 if (constcmp(t, "Node-copyfrom-path"))
365 continue;
366 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
367 break;
368 case sizeof("Node-copyfrom-rev"):
369 if (constcmp(t, "Node-copyfrom-rev"))
370 continue;
371 node_ctx.srcRev = atoi(val);
372 break;
373 case sizeof("Text-content-length"):
374 if (!constcmp(t, "Text-content-length")) {
375 node_ctx.textLength = atoi(val);
376 break;
377 }
378 if (constcmp(t, "Prop-content-length"))
379 continue;
380 node_ctx.propLength = atoi(val);
381 break;
382 case sizeof("Text-delta"):
383 if (!constcmp(t, "Text-delta")) {
384 node_ctx.text_delta = !strcmp(val, "true");
385 break;
386 }
387 if (constcmp(t, "Prop-delta"))
388 continue;
389 node_ctx.prop_delta = !strcmp(val, "true");
390 break;
391 case sizeof("Content-length"):
392 if (constcmp(t, "Content-length"))
393 continue;
394 len = atoi(val);
395 t = buffer_read_line(&input);
396 if (!t)
397 die_short_read();
398 if (*t)
399 die("invalid dump: expected blank line after content length header");
400 if (active_ctx == REV_CTX) {
401 read_props();
402 } else if (active_ctx == NODE_CTX) {
403 handle_node();
404 active_ctx = REV_CTX;
405 } else {
406 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
407 if (buffer_skip_bytes(&input, len) != len)
408 die_short_read();
409 }
410 }
411 }
412 if (buffer_ferror(&input))
413 die_short_read();
414 if (active_ctx == NODE_CTX)
415 handle_node();
416 if (active_ctx != DUMP_CTX)
417 handle_revision();
418}
419int svndump_init(const char *filename)
421{
422 if (buffer_init(&input, filename))
423 return error("cannot open %s: %s", filename, strerror(errno));
424 repo_init();
425 strbuf_init(&dump_ctx.uuid, 4096);
426 strbuf_init(&dump_ctx.url, 4096);
427 strbuf_init(&rev_ctx.log, 4096);
428 strbuf_init(&rev_ctx.author, 4096);
429 reset_dump_ctx(NULL);
430 reset_rev_ctx(0);
431 reset_node_ctx(NULL);
432 return 0;
433}
434void svndump_deinit(void)
436{
437 repo_reset();
438 reset_dump_ctx(NULL);
439 reset_rev_ctx(0);
440 reset_node_ctx(NULL);
441 strbuf_release(&rev_ctx.log);
442 if (buffer_deinit(&input))
443 fprintf(stderr, "Input error\n");
444 if (ferror(stdout))
445 fprintf(stderr, "Output error\n");
446}
447void svndump_reset(void)
449{
450 buffer_reset(&input);
451 repo_reset();
452 strbuf_release(&dump_ctx.uuid);
453 strbuf_release(&dump_ctx.url);
454 strbuf_release(&rev_ctx.log);
455 strbuf_release(&rev_ctx.author);
456}