2 * Parse and rearrange a svnadmin dump.
3 * Create the dump with:
4 * svnadmin dump --incremental -r<startrev>:<endrev> <repository> >outfile
6 * Licensed under a two-clause BSD-style license.
7 * See LICENSE for details.
11 #include "repo_tree.h"
12 #include "fast_export.h"
13 #include "line_buffer.h"
14 #include "string_pool.h"
17 #define REPORT_FILENO 3
20 * Compare start of string to literal of equal length;
21 * must be guarded by length test.
23 #define constcmp(s, ref) memcmp(s, ref, sizeof(ref) - 1)
25 #define NODEACT_REPLACE 4
26 #define NODEACT_DELETE 3
28 #define NODEACT_CHANGE 1
29 #define NODEACT_UNKNOWN 0
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 */
37 #define LENGTH_UNKNOWN (~0)
38 #define DATE_RFC2822_LEN 31
40 static struct line_buffer input = LINE_BUFFER_INIT;
43 uint32_t action, propLength, textLength, srcRev, type;
44 uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
45 uint32_t text_delta, prop_delta;
50 unsigned long timestamp;
51 struct strbuf log, author;
56 struct strbuf uuid, url;
59 static void reset_node_ctx(char *fname)
62 node_ctx.action = NODEACT_UNKNOWN;
63 node_ctx.propLength = LENGTH_UNKNOWN;
64 node_ctx.textLength = LENGTH_UNKNOWN;
67 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
68 node_ctx.text_delta = 0;
69 node_ctx.prop_delta = 0;
72 static void reset_rev_ctx(uint32_t revision)
74 rev_ctx.revision = revision;
75 rev_ctx.timestamp = 0;
76 strbuf_reset(&rev_ctx.log);
77 strbuf_reset(&rev_ctx.author);
80 static void reset_dump_ctx(const char *url)
82 strbuf_reset(&dump_ctx.url);
84 strbuf_addstr(&dump_ctx.url, url);
86 strbuf_reset(&dump_ctx.uuid);
89 static void handle_property(const struct strbuf *key_buf,
90 const char *val, uint32_t len,
93 const char *key = key_buf->buf;
94 size_t keylen = key_buf->len;
97 case sizeof("svn:log"):
98 if (constcmp(key, "svn:log"))
101 die("invalid dump: unsets svn:log");
102 strbuf_reset(&rev_ctx.log);
103 strbuf_add(&rev_ctx.log, val, len);
105 case sizeof("svn:author"):
106 if (constcmp(key, "svn:author"))
108 strbuf_reset(&rev_ctx.author);
110 strbuf_add(&rev_ctx.author, val, len);
112 case sizeof("svn:date"):
113 if (constcmp(key, "svn:date"))
116 die("invalid dump: unsets svn:date");
117 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
118 warning("invalid timestamp: %s", val);
120 case sizeof("svn:executable"):
121 case sizeof("svn:special"):
122 if (keylen == strlen("svn:executable") &&
123 constcmp(key, "svn:executable"))
125 if (keylen == strlen("svn:special") &&
126 constcmp(key, "svn:special"))
131 die("invalid dump: sets type twice");
134 node_ctx.type = REPO_MODE_BLB;
138 node_ctx.type = keylen == strlen("svn:executable") ?
144 static void die_short_read(void)
146 if (buffer_ferror(&input))
147 die_errno("error reading dump file");
148 die("invalid dump: unexpected end of file");
151 static void read_props(void)
153 static struct strbuf key = STRBUF_INIT;
156 * NEEDSWORK: to support simple mode changes like
163 * we keep track of whether a mode has been set and reset to
164 * plain file only if not. We should be keeping track of the
165 * symlink and executable bits separately instead.
167 uint32_t type_set = 0;
168 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
171 const char type = t[0];
174 if (!type || t[1] != ' ')
175 die("invalid property line: %s\n", t);
177 val = buffer_read_string(&input, len);
178 if (!val || strlen(val) != len)
181 /* Discard trailing newline. */
182 ch = buffer_read_char(&input);
186 die("invalid dump: expected newline after %s", val);
193 strbuf_add(&key, val, len);
201 handle_property(&key, val, len, &type_set);
205 die("invalid property line: %s\n", t);
210 static void handle_node(void)
212 const uint32_t type = node_ctx.type;
213 const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
214 const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
216 * Old text for this node:
217 * NULL - directory or bug
219 * "<dataref>" - data retrievable from fast-import
221 static const char *const empty_blob = "::empty::";
222 const char *old_data = NULL;
224 if (node_ctx.text_delta)
225 die("text deltas not supported");
227 if (node_ctx.action == NODEACT_DELETE) {
228 if (have_text || have_props || node_ctx.srcRev)
229 die("invalid dump: deletion node has "
230 "copyfrom info, text, or properties");
231 return repo_delete(node_ctx.dst);
233 if (node_ctx.action == NODEACT_REPLACE) {
234 repo_delete(node_ctx.dst);
235 node_ctx.action = NODEACT_ADD;
237 if (node_ctx.srcRev) {
238 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
239 if (node_ctx.action == NODEACT_ADD)
240 node_ctx.action = NODEACT_CHANGE;
242 if (have_text && type == REPO_MODE_DIR)
243 die("invalid dump: directories cannot have text attached");
246 * Find old content (old_data) and decide on the new mode.
248 if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
249 if (type != REPO_MODE_DIR)
250 die("invalid dump: root of tree is not a regular file");
252 } else if (node_ctx.action == NODEACT_CHANGE) {
254 old_data = repo_read_path(node_ctx.dst);
255 mode = repo_read_mode(node_ctx.dst);
256 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
257 die("invalid dump: cannot modify a directory into a file");
258 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
259 die("invalid dump: cannot modify a file into a directory");
260 node_ctx.type = mode;
261 } else if (node_ctx.action == NODEACT_ADD) {
262 if (type == REPO_MODE_DIR)
265 old_data = empty_blob;
267 die("invalid dump: adds node without text");
269 die("invalid dump: Node-path block lacks Node-action");
273 * Adjust mode to reflect properties.
276 if (!node_ctx.prop_delta)
277 node_ctx.type = type;
278 if (node_ctx.propLength)
285 if (type == REPO_MODE_DIR) /* directories are not tracked. */
288 if (old_data == empty_blob)
289 /* For the fast_export_* functions, NULL means empty. */
292 fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
293 node_ctx.type, old_data);
296 fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
297 node_ctx.type, "inline");
298 fast_export_data(node_ctx.type, node_ctx.textLength, &input);
301 static void begin_revision(void)
303 if (!rev_ctx.revision) /* revision 0 gets no git commit. */
305 fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
306 rev_ctx.log.buf, dump_ctx.uuid.buf, dump_ctx.url.buf,
310 static void end_revision(void)
312 if (rev_ctx.revision)
313 fast_export_end_commit(rev_ctx.revision);
316 void svndump_read(const char *url)
320 uint32_t active_ctx = DUMP_CTX;
324 while ((t = buffer_read_line(&input))) {
325 val = strstr(t, ": ");
330 /* strlen(key) + 1 */
331 switch (val - t - 1) {
332 case sizeof("SVN-fs-dump-format-version"):
333 if (constcmp(t, "SVN-fs-dump-format-version"))
335 dump_ctx.version = atoi(val);
336 if (dump_ctx.version > 3)
337 die("expected svn dump format version <= 3, found %"PRIu32,
341 if (constcmp(t, "UUID"))
343 strbuf_reset(&dump_ctx.uuid);
344 strbuf_addstr(&dump_ctx.uuid, val);
346 case sizeof("Revision-number"):
347 if (constcmp(t, "Revision-number"))
349 if (active_ctx == NODE_CTX)
351 if (active_ctx == REV_CTX)
353 if (active_ctx != DUMP_CTX)
355 active_ctx = REV_CTX;
356 reset_rev_ctx(atoi(val));
358 case sizeof("Node-path"):
359 if (prefixcmp(t, "Node-"))
361 if (!constcmp(t + strlen("Node-"), "path")) {
362 if (active_ctx == NODE_CTX)
364 if (active_ctx == REV_CTX)
366 active_ctx = NODE_CTX;
370 if (constcmp(t + strlen("Node-"), "kind"))
372 if (!strcmp(val, "dir"))
373 node_ctx.type = REPO_MODE_DIR;
374 else if (!strcmp(val, "file"))
375 node_ctx.type = REPO_MODE_BLB;
377 fprintf(stderr, "Unknown node-kind: %s\n", val);
379 case sizeof("Node-action"):
380 if (constcmp(t, "Node-action"))
382 if (!strcmp(val, "delete")) {
383 node_ctx.action = NODEACT_DELETE;
384 } else if (!strcmp(val, "add")) {
385 node_ctx.action = NODEACT_ADD;
386 } else if (!strcmp(val, "change")) {
387 node_ctx.action = NODEACT_CHANGE;
388 } else if (!strcmp(val, "replace")) {
389 node_ctx.action = NODEACT_REPLACE;
391 fprintf(stderr, "Unknown node-action: %s\n", val);
392 node_ctx.action = NODEACT_UNKNOWN;
395 case sizeof("Node-copyfrom-path"):
396 if (constcmp(t, "Node-copyfrom-path"))
398 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
400 case sizeof("Node-copyfrom-rev"):
401 if (constcmp(t, "Node-copyfrom-rev"))
403 node_ctx.srcRev = atoi(val);
405 case sizeof("Text-content-length"):
406 if (!constcmp(t, "Text-content-length")) {
407 node_ctx.textLength = atoi(val);
410 if (constcmp(t, "Prop-content-length"))
412 node_ctx.propLength = atoi(val);
414 case sizeof("Text-delta"):
415 if (!constcmp(t, "Text-delta")) {
416 node_ctx.text_delta = !strcmp(val, "true");
419 if (constcmp(t, "Prop-delta"))
421 node_ctx.prop_delta = !strcmp(val, "true");
423 case sizeof("Content-length"):
424 if (constcmp(t, "Content-length"))
427 t = buffer_read_line(&input);
431 die("invalid dump: expected blank line after content length header");
432 if (active_ctx == REV_CTX) {
434 } else if (active_ctx == NODE_CTX) {
436 active_ctx = INTERNODE_CTX;
438 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
439 if (buffer_skip_bytes(&input, len) != len)
444 if (buffer_ferror(&input))
446 if (active_ctx == NODE_CTX)
448 if (active_ctx == REV_CTX)
450 if (active_ctx != DUMP_CTX)
454 int svndump_init(const char *filename)
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 reset_dump_ctx(NULL);
465 reset_node_ctx(NULL);
469 void svndump_deinit(void)
471 fast_export_deinit();
472 reset_dump_ctx(NULL);
474 reset_node_ctx(NULL);
475 strbuf_release(&rev_ctx.log);
476 if (buffer_deinit(&input))
477 fprintf(stderr, "Input error\n");
479 fprintf(stderr, "Output error\n");
482 void svndump_reset(void)
485 buffer_reset(&input);
486 strbuf_release(&dump_ctx.uuid);
487 strbuf_release(&dump_ctx.url);
488 strbuf_release(&rev_ctx.log);
489 strbuf_release(&rev_ctx.author);