3cc4135892248469f4a9fb97b332950f89e587ba
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 "obj_pool.h"
15#include "string_pool.h"
16#define REPORT_FILENO 3
18#define NODEACT_REPLACE 4
20#define NODEACT_DELETE 3
21#define NODEACT_ADD 2
22#define NODEACT_CHANGE 1
23#define NODEACT_UNKNOWN 0
24/* States: */
26#define DUMP_CTX 0 /* dump metadata */
27#define REV_CTX 1 /* revision metadata */
28#define NODE_CTX 2 /* node metadata */
29#define INTERNODE_CTX 3 /* between nodes */
30#define LENGTH_UNKNOWN (~0)
32#define DATE_RFC2822_LEN 31
33/* Create memory pool for log messages */
35obj_pool_gen(log, char, 4096)
36static struct line_buffer input = LINE_BUFFER_INIT;
38static char *log_copy(uint32_t length, const char *log)
40{
41 char *buffer;
42 log_free(log_pool.size);
43 buffer = log_pointer(log_alloc(length));
44 strncpy(buffer, log, length);
45 return buffer;
46}
47static struct {
49 uint32_t action, propLength, textLength, srcRev, type;
50 uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
51 uint32_t text_delta, prop_delta;
52} node_ctx;
53static struct {
55 uint32_t revision, author;
56 unsigned long timestamp;
57 char *log;
58} rev_ctx;
59static struct {
61 uint32_t version, uuid, url;
62} dump_ctx;
63static struct {
65 uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
66 revision_number, node_path, node_kind, node_action,
67 node_copyfrom_path, node_copyfrom_rev, text_content_length,
68 prop_content_length, content_length, svn_fs_dump_format_version,
69 /* version 3 format */
70 text_delta, prop_delta;
71} keys;
72static void reset_node_ctx(char *fname)
74{
75 node_ctx.type = 0;
76 node_ctx.action = NODEACT_UNKNOWN;
77 node_ctx.propLength = LENGTH_UNKNOWN;
78 node_ctx.textLength = LENGTH_UNKNOWN;
79 node_ctx.src[0] = ~0;
80 node_ctx.srcRev = 0;
81 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
82 node_ctx.text_delta = 0;
83 node_ctx.prop_delta = 0;
84}
85static void reset_rev_ctx(uint32_t revision)
87{
88 rev_ctx.revision = revision;
89 rev_ctx.timestamp = 0;
90 rev_ctx.log = NULL;
91 rev_ctx.author = ~0;
92}
93static void reset_dump_ctx(uint32_t url)
95{
96 dump_ctx.url = url;
97 dump_ctx.version = 1;
98 dump_ctx.uuid = ~0;
99}
100static void init_keys(void)
102{
103 keys.svn_log = pool_intern("svn:log");
104 keys.svn_author = pool_intern("svn:author");
105 keys.svn_date = pool_intern("svn:date");
106 keys.svn_executable = pool_intern("svn:executable");
107 keys.svn_special = pool_intern("svn:special");
108 keys.uuid = pool_intern("UUID");
109 keys.revision_number = pool_intern("Revision-number");
110 keys.node_path = pool_intern("Node-path");
111 keys.node_kind = pool_intern("Node-kind");
112 keys.node_action = pool_intern("Node-action");
113 keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
114 keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
115 keys.text_content_length = pool_intern("Text-content-length");
116 keys.prop_content_length = pool_intern("Prop-content-length");
117 keys.content_length = pool_intern("Content-length");
118 keys.svn_fs_dump_format_version = pool_intern("SVN-fs-dump-format-version");
119 /* version 3 format (Subversion 1.1.0) */
120 keys.text_delta = pool_intern("Text-delta");
121 keys.prop_delta = pool_intern("Prop-delta");
122}
123static void handle_property(uint32_t key, const char *val, uint32_t len,
125 uint32_t *type_set)
126{
127 if (key == keys.svn_log) {
128 if (!val)
129 die("invalid dump: unsets svn:log");
130 /* Value length excludes terminating nul. */
131 rev_ctx.log = log_copy(len + 1, val);
132 } else if (key == keys.svn_author) {
133 rev_ctx.author = pool_intern(val);
134 } else if (key == keys.svn_date) {
135 if (!val)
136 die("invalid dump: unsets svn:date");
137 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
138 warning("invalid timestamp: %s", val);
139 } else if (key == keys.svn_executable || key == keys.svn_special) {
140 if (*type_set) {
141 if (!val)
142 return;
143 die("invalid dump: sets type twice");
144 }
145 if (!val) {
146 node_ctx.type = REPO_MODE_BLB;
147 return;
148 }
149 *type_set = 1;
150 node_ctx.type = key == keys.svn_executable ?
151 REPO_MODE_EXE :
152 REPO_MODE_LNK;
153 }
154}
155static void read_props(void)
157{
158 uint32_t key = ~0;
159 const char *t;
160 /*
161 * NEEDSWORK: to support simple mode changes like
162 * K 11
163 * svn:special
164 * V 1
165 * *
166 * D 14
167 * svn:executable
168 * we keep track of whether a mode has been set and reset to
169 * plain file only if not. We should be keeping track of the
170 * symlink and executable bits separately instead.
171 */
172 uint32_t type_set = 0;
173 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
174 uint32_t len;
175 const char *val;
176 const char type = t[0];
177 if (!type || t[1] != ' ')
179 die("invalid property line: %s\n", t);
180 len = atoi(&t[2]);
181 val = buffer_read_string(&input, len);
182 buffer_skip_bytes(&input, 1); /* Discard trailing newline. */
183 switch (type) {
185 case 'K':
186 key = pool_intern(val);
187 continue;
188 case 'D':
189 key = pool_intern(val);
190 val = NULL;
191 len = 0;
192 /* fall through */
193 case 'V':
194 handle_property(key, val, len, &type_set);
195 key = ~0;
196 continue;
197 default:
198 die("invalid property line: %s\n", t);
199 }
200 }
201}
202static void handle_node(void)
204{
205 uint32_t mark = 0;
206 const uint32_t type = node_ctx.type;
207 const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
208 const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
209 if (node_ctx.text_delta)
211 die("text deltas not supported");
212 if (have_text)
213 mark = next_blob_mark();
214 if (node_ctx.action == NODEACT_DELETE) {
215 if (have_text || have_props || node_ctx.srcRev)
216 die("invalid dump: deletion node has "
217 "copyfrom info, text, or properties");
218 return repo_delete(node_ctx.dst);
219 }
220 if (node_ctx.action == NODEACT_REPLACE) {
221 repo_delete(node_ctx.dst);
222 node_ctx.action = NODEACT_ADD;
223 }
224 if (node_ctx.srcRev) {
225 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
226 if (node_ctx.action == NODEACT_ADD)
227 node_ctx.action = NODEACT_CHANGE;
228 }
229 if (have_text && type == REPO_MODE_DIR)
230 die("invalid dump: directories cannot have text attached");
231 /*
233 * Decide on the new content (mark) and mode (node_ctx.type).
234 */
235 if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
236 if (type != REPO_MODE_DIR)
237 die("invalid dump: root of tree is not a regular file");
238 } else if (node_ctx.action == NODEACT_CHANGE) {
239 uint32_t mode;
240 if (!have_text)
241 mark = repo_read_path(node_ctx.dst);
242 mode = repo_read_mode(node_ctx.dst);
243 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
244 die("invalid dump: cannot modify a directory into a file");
245 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
246 die("invalid dump: cannot modify a file into a directory");
247 node_ctx.type = mode;
248 } else if (node_ctx.action == NODEACT_ADD) {
249 if (!have_text && type != REPO_MODE_DIR)
250 die("invalid dump: adds node without text");
251 } else {
252 die("invalid dump: Node-path block lacks Node-action");
253 }
254 /*
256 * Adjust mode to reflect properties.
257 */
258 if (have_props) {
259 if (!node_ctx.prop_delta)
260 node_ctx.type = type;
261 if (node_ctx.propLength)
262 read_props();
263 }
264 /*
266 * Save the result.
267 */
268 repo_add(node_ctx.dst, node_ctx.type, mark);
269 if (have_text)
270 fast_export_blob(node_ctx.type, mark,
271 node_ctx.textLength, &input);
272}
273static void begin_revision(void)
275{
276 if (!rev_ctx.revision) /* revision 0 gets no git commit. */
277 return;
278 fast_export_begin_commit(rev_ctx.revision);
279}
280static void end_revision(void)
282{
283 if (rev_ctx.revision)
284 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
285 dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
286}
287void svndump_read(const char *url)
289{
290 char *val;
291 char *t;
292 uint32_t active_ctx = DUMP_CTX;
293 uint32_t len;
294 uint32_t key;
295 reset_dump_ctx(pool_intern(url));
297 while ((t = buffer_read_line(&input))) {
298 val = strstr(t, ": ");
299 if (!val)
300 continue;
301 *val++ = '\0';
302 *val++ = '\0';
303 key = pool_intern(t);
304 if (key == keys.svn_fs_dump_format_version) {
306 dump_ctx.version = atoi(val);
307 if (dump_ctx.version > 3)
308 die("expected svn dump format version <= 3, found %"PRIu32,
309 dump_ctx.version);
310 } else if (key == keys.uuid) {
311 dump_ctx.uuid = pool_intern(val);
312 } else if (key == keys.revision_number) {
313 if (active_ctx == NODE_CTX)
314 handle_node();
315 if (active_ctx == REV_CTX)
316 begin_revision();
317 if (active_ctx != DUMP_CTX)
318 end_revision();
319 active_ctx = REV_CTX;
320 reset_rev_ctx(atoi(val));
321 } else if (key == keys.node_path) {
322 if (active_ctx == NODE_CTX)
323 handle_node();
324 if (active_ctx == REV_CTX)
325 begin_revision();
326 active_ctx = NODE_CTX;
327 reset_node_ctx(val);
328 } else if (key == keys.node_kind) {
329 if (!strcmp(val, "dir"))
330 node_ctx.type = REPO_MODE_DIR;
331 else if (!strcmp(val, "file"))
332 node_ctx.type = REPO_MODE_BLB;
333 else
334 fprintf(stderr, "Unknown node-kind: %s\n", val);
335 } else if (key == keys.node_action) {
336 if (!strcmp(val, "delete")) {
337 node_ctx.action = NODEACT_DELETE;
338 } else if (!strcmp(val, "add")) {
339 node_ctx.action = NODEACT_ADD;
340 } else if (!strcmp(val, "change")) {
341 node_ctx.action = NODEACT_CHANGE;
342 } else if (!strcmp(val, "replace")) {
343 node_ctx.action = NODEACT_REPLACE;
344 } else {
345 fprintf(stderr, "Unknown node-action: %s\n", val);
346 node_ctx.action = NODEACT_UNKNOWN;
347 }
348 } else if (key == keys.node_copyfrom_path) {
349 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
350 } else if (key == keys.node_copyfrom_rev) {
351 node_ctx.srcRev = atoi(val);
352 } else if (key == keys.text_content_length) {
353 node_ctx.textLength = atoi(val);
354 } else if (key == keys.prop_content_length) {
355 node_ctx.propLength = atoi(val);
356 } else if (key == keys.text_delta) {
357 node_ctx.text_delta = !strcmp(val, "true");
358 } else if (key == keys.prop_delta) {
359 node_ctx.prop_delta = !strcmp(val, "true");
360 } else if (key == keys.content_length) {
361 len = atoi(val);
362 buffer_read_line(&input);
363 if (active_ctx == REV_CTX) {
364 read_props();
365 } else if (active_ctx == NODE_CTX) {
366 handle_node();
367 active_ctx = INTERNODE_CTX;
368 } else {
369 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
370 buffer_skip_bytes(&input, len);
371 }
372 }
373 }
374 if (active_ctx == NODE_CTX)
375 handle_node();
376 if (active_ctx == REV_CTX)
377 begin_revision();
378 if (active_ctx != DUMP_CTX)
379 end_revision();
380}
381int svndump_init(const char *filename)
383{
384 if (buffer_init(&input, filename))
385 return error("cannot open %s: %s", filename, strerror(errno));
386 repo_init();
387 fast_export_init(REPORT_FILENO);
388 reset_dump_ctx(~0);
389 reset_rev_ctx(0);
390 reset_node_ctx(NULL);
391 init_keys();
392 return 0;
393}
394void svndump_deinit(void)
396{
397 log_reset();
398 fast_export_deinit();
399 repo_reset();
400 reset_dump_ctx(~0);
401 reset_rev_ctx(0);
402 reset_node_ctx(NULL);
403 if (buffer_deinit(&input))
404 fprintf(stderr, "Input error\n");
405 if (ferror(stdout))
406 fprintf(stderr, "Output error\n");
407}
408void svndump_reset(void)
410{
411 log_reset();
412 fast_export_reset();
413 buffer_reset(&input);
414 repo_reset();
415 reset_dump_ctx(~0);
416 reset_rev_ctx(0);
417 reset_node_ctx(NULL);
418}