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#define DUMP_CTX 0
26#define REV_CTX 1
27#define NODE_CTX 2
28#define LENGTH_UNKNOWN (~0)
30#define DATE_RFC2822_LEN 31
31/* Create memory pool for log messages */
33obj_pool_gen(log, char, 4096)
34static struct line_buffer input = LINE_BUFFER_INIT;
36static char *log_copy(uint32_t length, const char *log)
38{
39 char *buffer;
40 log_free(log_pool.size);
41 buffer = log_pointer(log_alloc(length));
42 strncpy(buffer, log, length);
43 return buffer;
44}
45static struct {
47 uint32_t action, propLength, textLength, srcRev, type;
48 uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
49 uint32_t text_delta, prop_delta;
50} node_ctx;
51static struct {
53 uint32_t revision, author;
54 unsigned long timestamp;
55 char *log;
56} rev_ctx;
57static struct {
59 uint32_t version, uuid, url;
60} dump_ctx;
61static struct {
63 uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
64 revision_number, node_path, node_kind, node_action,
65 node_copyfrom_path, node_copyfrom_rev, text_content_length,
66 prop_content_length, content_length, svn_fs_dump_format_version,
67 /* version 3 format */
68 text_delta, prop_delta;
69} keys;
70static void reset_node_ctx(char *fname)
72{
73 node_ctx.type = 0;
74 node_ctx.action = NODEACT_UNKNOWN;
75 node_ctx.propLength = LENGTH_UNKNOWN;
76 node_ctx.textLength = LENGTH_UNKNOWN;
77 node_ctx.src[0] = ~0;
78 node_ctx.srcRev = 0;
79 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
80 node_ctx.text_delta = 0;
81 node_ctx.prop_delta = 0;
82}
83static void reset_rev_ctx(uint32_t revision)
85{
86 rev_ctx.revision = revision;
87 rev_ctx.timestamp = 0;
88 rev_ctx.log = NULL;
89 rev_ctx.author = ~0;
90}
91static void reset_dump_ctx(uint32_t url)
93{
94 dump_ctx.url = url;
95 dump_ctx.version = 1;
96 dump_ctx.uuid = ~0;
97}
98static void init_keys(void)
100{
101 keys.svn_log = pool_intern("svn:log");
102 keys.svn_author = pool_intern("svn:author");
103 keys.svn_date = pool_intern("svn:date");
104 keys.svn_executable = pool_intern("svn:executable");
105 keys.svn_special = pool_intern("svn:special");
106 keys.uuid = pool_intern("UUID");
107 keys.revision_number = pool_intern("Revision-number");
108 keys.node_path = pool_intern("Node-path");
109 keys.node_kind = pool_intern("Node-kind");
110 keys.node_action = pool_intern("Node-action");
111 keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
112 keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
113 keys.text_content_length = pool_intern("Text-content-length");
114 keys.prop_content_length = pool_intern("Prop-content-length");
115 keys.content_length = pool_intern("Content-length");
116 keys.svn_fs_dump_format_version = pool_intern("SVN-fs-dump-format-version");
117 /* version 3 format (Subversion 1.1.0) */
118 keys.text_delta = pool_intern("Text-delta");
119 keys.prop_delta = pool_intern("Prop-delta");
120}
121static void handle_property(uint32_t key, const char *val, uint32_t len,
123 uint32_t *type_set)
124{
125 if (key == keys.svn_log) {
126 if (!val)
127 die("invalid dump: unsets svn:log");
128 /* Value length excludes terminating nul. */
129 rev_ctx.log = log_copy(len + 1, val);
130 } else if (key == keys.svn_author) {
131 rev_ctx.author = pool_intern(val);
132 } else if (key == keys.svn_date) {
133 if (!val)
134 die("invalid dump: unsets svn:date");
135 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
136 warning("invalid timestamp: %s", val);
137 } else if (key == keys.svn_executable || key == keys.svn_special) {
138 if (*type_set) {
139 if (!val)
140 return;
141 die("invalid dump: sets type twice");
142 }
143 if (!val) {
144 node_ctx.type = REPO_MODE_BLB;
145 return;
146 }
147 *type_set = 1;
148 node_ctx.type = key == keys.svn_executable ?
149 REPO_MODE_EXE :
150 REPO_MODE_LNK;
151 }
152}
153static void read_props(void)
155{
156 uint32_t key = ~0;
157 const char *t;
158 /*
159 * NEEDSWORK: to support simple mode changes like
160 * K 11
161 * svn:special
162 * V 1
163 * *
164 * D 14
165 * svn:executable
166 * we keep track of whether a mode has been set and reset to
167 * plain file only if not. We should be keeping track of the
168 * symlink and executable bits separately instead.
169 */
170 uint32_t type_set = 0;
171 while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
172 uint32_t len;
173 const char *val;
174 const char type = t[0];
175 if (!type || t[1] != ' ')
177 die("invalid property line: %s\n", t);
178 len = atoi(&t[2]);
179 val = buffer_read_string(&input, len);
180 buffer_skip_bytes(&input, 1); /* Discard trailing newline. */
181 switch (type) {
183 case 'K':
184 key = pool_intern(val);
185 continue;
186 case 'D':
187 key = pool_intern(val);
188 val = NULL;
189 len = 0;
190 /* fall through */
191 case 'V':
192 handle_property(key, val, len, &type_set);
193 key = ~0;
194 continue;
195 default:
196 die("invalid property line: %s\n", t);
197 }
198 }
199}
200static void handle_node(void)
202{
203 uint32_t mark = 0;
204 const uint32_t type = node_ctx.type;
205 const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
206 const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
207 if (node_ctx.text_delta)
209 die("text deltas not supported");
210 if (have_text)
211 mark = next_blob_mark();
212 if (node_ctx.action == NODEACT_DELETE) {
213 if (have_text || have_props || node_ctx.srcRev)
214 die("invalid dump: deletion node has "
215 "copyfrom info, text, or properties");
216 return repo_delete(node_ctx.dst);
217 }
218 if (node_ctx.action == NODEACT_REPLACE) {
219 repo_delete(node_ctx.dst);
220 node_ctx.action = NODEACT_ADD;
221 }
222 if (node_ctx.srcRev) {
223 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
224 if (node_ctx.action == NODEACT_ADD)
225 node_ctx.action = NODEACT_CHANGE;
226 }
227 if (have_text && type == REPO_MODE_DIR)
228 die("invalid dump: directories cannot have text attached");
229 /*
231 * Decide on the new content (mark) and mode (node_ctx.type).
232 */
233 if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
234 if (type != REPO_MODE_DIR)
235 die("invalid dump: root of tree is not a regular file");
236 } else if (node_ctx.action == NODEACT_CHANGE) {
237 uint32_t mode;
238 if (!have_text)
239 mark = repo_read_path(node_ctx.dst);
240 mode = repo_read_mode(node_ctx.dst);
241 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
242 die("invalid dump: cannot modify a directory into a file");
243 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
244 die("invalid dump: cannot modify a file into a directory");
245 node_ctx.type = mode;
246 } else if (node_ctx.action == NODEACT_ADD) {
247 if (!have_text && type != REPO_MODE_DIR)
248 die("invalid dump: adds node without text");
249 } else {
250 die("invalid dump: Node-path block lacks Node-action");
251 }
252 /*
254 * Adjust mode to reflect properties.
255 */
256 if (have_props) {
257 if (!node_ctx.prop_delta)
258 node_ctx.type = type;
259 if (node_ctx.propLength)
260 read_props();
261 }
262 /*
264 * Save the result.
265 */
266 repo_add(node_ctx.dst, node_ctx.type, mark);
267 if (have_text)
268 fast_export_blob(node_ctx.type, mark,
269 node_ctx.textLength, &input);
270}
271static void handle_revision(void)
273{
274 if (rev_ctx.revision)
275 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
276 dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
277}
278void svndump_read(const char *url)
280{
281 char *val;
282 char *t;
283 uint32_t active_ctx = DUMP_CTX;
284 uint32_t len;
285 uint32_t key;
286 reset_dump_ctx(pool_intern(url));
288 while ((t = buffer_read_line(&input))) {
289 val = strstr(t, ": ");
290 if (!val)
291 continue;
292 *val++ = '\0';
293 *val++ = '\0';
294 key = pool_intern(t);
295 if (key == keys.svn_fs_dump_format_version) {
297 dump_ctx.version = atoi(val);
298 if (dump_ctx.version > 3)
299 die("expected svn dump format version <= 3, found %"PRIu32,
300 dump_ctx.version);
301 } else if (key == keys.uuid) {
302 dump_ctx.uuid = pool_intern(val);
303 } else if (key == keys.revision_number) {
304 if (active_ctx == NODE_CTX)
305 handle_node();
306 if (active_ctx != DUMP_CTX)
307 handle_revision();
308 active_ctx = REV_CTX;
309 reset_rev_ctx(atoi(val));
310 } else if (key == keys.node_path) {
311 if (active_ctx == NODE_CTX)
312 handle_node();
313 active_ctx = NODE_CTX;
314 reset_node_ctx(val);
315 } else if (key == keys.node_kind) {
316 if (!strcmp(val, "dir"))
317 node_ctx.type = REPO_MODE_DIR;
318 else if (!strcmp(val, "file"))
319 node_ctx.type = REPO_MODE_BLB;
320 else
321 fprintf(stderr, "Unknown node-kind: %s\n", val);
322 } else if (key == keys.node_action) {
323 if (!strcmp(val, "delete")) {
324 node_ctx.action = NODEACT_DELETE;
325 } else if (!strcmp(val, "add")) {
326 node_ctx.action = NODEACT_ADD;
327 } else if (!strcmp(val, "change")) {
328 node_ctx.action = NODEACT_CHANGE;
329 } else if (!strcmp(val, "replace")) {
330 node_ctx.action = NODEACT_REPLACE;
331 } else {
332 fprintf(stderr, "Unknown node-action: %s\n", val);
333 node_ctx.action = NODEACT_UNKNOWN;
334 }
335 } else if (key == keys.node_copyfrom_path) {
336 pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
337 } else if (key == keys.node_copyfrom_rev) {
338 node_ctx.srcRev = atoi(val);
339 } else if (key == keys.text_content_length) {
340 node_ctx.textLength = atoi(val);
341 } else if (key == keys.prop_content_length) {
342 node_ctx.propLength = atoi(val);
343 } else if (key == keys.text_delta) {
344 node_ctx.text_delta = !strcmp(val, "true");
345 } else if (key == keys.prop_delta) {
346 node_ctx.prop_delta = !strcmp(val, "true");
347 } else if (key == keys.content_length) {
348 len = atoi(val);
349 buffer_read_line(&input);
350 if (active_ctx == REV_CTX) {
351 read_props();
352 } else if (active_ctx == NODE_CTX) {
353 handle_node();
354 active_ctx = REV_CTX;
355 } else {
356 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
357 buffer_skip_bytes(&input, len);
358 }
359 }
360 }
361 if (active_ctx == NODE_CTX)
362 handle_node();
363 if (active_ctx != DUMP_CTX)
364 handle_revision();
365}
366int svndump_init(const char *filename)
368{
369 if (buffer_init(&input, filename))
370 return error("cannot open %s: %s", filename, strerror(errno));
371 repo_init();
372 fast_export_init(REPORT_FILENO);
373 reset_dump_ctx(~0);
374 reset_rev_ctx(0);
375 reset_node_ctx(NULL);
376 init_keys();
377 return 0;
378}
379void svndump_deinit(void)
381{
382 log_reset();
383 fast_export_deinit();
384 repo_reset();
385 reset_dump_ctx(~0);
386 reset_rev_ctx(0);
387 reset_node_ctx(NULL);
388 if (buffer_deinit(&input))
389 fprintf(stderr, "Input error\n");
390 if (ferror(stdout))
391 fprintf(stderr, "Output error\n");
392}
393void svndump_reset(void)
395{
396 log_reset();
397 fast_export_reset();
398 buffer_reset(&input);
399 repo_reset();
400 reset_dump_ctx(~0);
401 reset_rev_ctx(0);
402 reset_node_ctx(NULL);
403}