Merge branch 'rs/sha1-name-readdir-optim'
[git] / vcs-svn / svndump.c
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
10 #include "cache.h"
11 #include "repo_tree.h"
12 #include "fast_export.h"
13 #include "line_buffer.h"
14 #include "strbuf.h"
15 #include "svndump.h"
16
17 /*
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
23 #define REPORT_FILENO 3
24
25 #define NODEACT_REPLACE 4
26 #define NODEACT_DELETE 3
27 #define NODEACT_ADD 2
28 #define NODEACT_CHANGE 1
29 #define NODEACT_UNKNOWN 0
30
31 /* States: */
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 */
36
37 #define DATE_RFC2822_LEN 31
38
39 static struct line_buffer input = LINE_BUFFER_INIT;
40
41 static struct {
42         uint32_t action, srcRev, type;
43         off_t prop_length, text_length;
44         struct strbuf src, dst;
45         uint32_t text_delta, prop_delta;
46 } node_ctx;
47
48 static struct {
49         uint32_t revision;
50         timestamp_t timestamp;
51         struct strbuf log, author, note;
52 } rev_ctx;
53
54 static struct {
55         uint32_t version;
56         struct strbuf uuid, url;
57 } dump_ctx;
58
59 static void reset_node_ctx(char *fname)
60 {
61         node_ctx.type = 0;
62         node_ctx.action = NODEACT_UNKNOWN;
63         node_ctx.prop_length = -1;
64         node_ctx.text_length = -1;
65         strbuf_reset(&node_ctx.src);
66         node_ctx.srcRev = 0;
67         strbuf_reset(&node_ctx.dst);
68         if (fname)
69                 strbuf_addstr(&node_ctx.dst, fname);
70         node_ctx.text_delta = 0;
71         node_ctx.prop_delta = 0;
72 }
73
74 static void reset_rev_ctx(uint32_t revision)
75 {
76         rev_ctx.revision = revision;
77         rev_ctx.timestamp = 0;
78         strbuf_reset(&rev_ctx.log);
79         strbuf_reset(&rev_ctx.author);
80         strbuf_reset(&rev_ctx.note);
81 }
82
83 static void reset_dump_ctx(const char *url)
84 {
85         strbuf_reset(&dump_ctx.url);
86         if (url)
87                 strbuf_addstr(&dump_ctx.url, url);
88         dump_ctx.version = 1;
89         strbuf_reset(&dump_ctx.uuid);
90 }
91
92 static void handle_property(const struct strbuf *key_buf,
93                                 struct strbuf *val,
94                                 uint32_t *type_set)
95 {
96         const char *key = key_buf->buf;
97         size_t keylen = key_buf->len;
98
99         switch (keylen + 1) {
100         case sizeof("svn:log"):
101                 if (constcmp(key, "svn:log"))
102                         break;
103                 if (!val)
104                         die("invalid dump: unsets svn:log");
105                 strbuf_swap(&rev_ctx.log, val);
106                 break;
107         case sizeof("svn:author"):
108                 if (constcmp(key, "svn:author"))
109                         break;
110                 if (!val)
111                         strbuf_reset(&rev_ctx.author);
112                 else
113                         strbuf_swap(&rev_ctx.author, val);
114                 break;
115         case sizeof("svn:date"):
116                 if (constcmp(key, "svn:date"))
117                         break;
118                 if (!val)
119                         die("invalid dump: unsets svn:date");
120                 if (parse_date_basic(val->buf, &rev_ctx.timestamp, NULL))
121                         warning("invalid timestamp: %s", val->buf);
122                 break;
123         case sizeof("svn:executable"):
124         case sizeof("svn:special"):
125                 if (keylen == strlen("svn:executable") &&
126                     constcmp(key, "svn:executable"))
127                         break;
128                 if (keylen == strlen("svn:special") &&
129                     constcmp(key, "svn:special"))
130                         break;
131                 if (*type_set) {
132                         if (!val)
133                                 return;
134                         die("invalid dump: sets type twice");
135                 }
136                 if (!val) {
137                         node_ctx.type = REPO_MODE_BLB;
138                         return;
139                 }
140                 *type_set = 1;
141                 node_ctx.type = keylen == strlen("svn:executable") ?
142                                 REPO_MODE_EXE :
143                                 REPO_MODE_LNK;
144         }
145 }
146
147 static void die_short_read(void)
148 {
149         if (buffer_ferror(&input))
150                 die_errno("error reading dump file");
151         die("invalid dump: unexpected end of file");
152 }
153
154 static void read_props(void)
155 {
156         static struct strbuf key = STRBUF_INIT;
157         static struct strbuf val = STRBUF_INIT;
158         const char *t;
159         /*
160          * NEEDSWORK: to support simple mode changes like
161          *      K 11
162          *      svn:special
163          *      V 1
164          *      *
165          *      D 14
166          *      svn:executable
167          * we keep track of whether a mode has been set and reset to
168          * plain file only if not.  We should be keeping track of the
169          * symlink and executable bits separately instead.
170          */
171         uint32_t type_set = 0;
172         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
173                 uint32_t len;
174                 const char type = t[0];
175                 int ch;
176
177                 if (!type || t[1] != ' ')
178                         die("invalid property line: %s", t);
179                 len = atoi(&t[2]);
180                 strbuf_reset(&val);
181                 buffer_read_binary(&input, &val, len);
182                 if (val.len < len)
183                         die_short_read();
184
185                 /* Discard trailing newline. */
186                 ch = buffer_read_char(&input);
187                 if (ch == EOF)
188                         die_short_read();
189                 if (ch != '\n')
190                         die("invalid dump: expected newline after %s", val.buf);
191
192                 switch (type) {
193                 case 'K':
194                         strbuf_swap(&key, &val);
195                         continue;
196                 case 'D':
197                         handle_property(&val, NULL, &type_set);
198                         continue;
199                 case 'V':
200                         handle_property(&key, &val, &type_set);
201                         strbuf_reset(&key);
202                         continue;
203                 default:
204                         die("invalid property line: %s", t);
205                 }
206         }
207 }
208
209 static void handle_node(void)
210 {
211         const uint32_t type = node_ctx.type;
212         const int have_props = node_ctx.prop_length != -1;
213         const int have_text = node_ctx.text_length != -1;
214         /*
215          * Old text for this node:
216          *  NULL        - directory or bug
217          *  empty_blob  - empty
218          *  "<dataref>" - data retrievable from fast-import
219          */
220         static const char *const empty_blob = "::empty::";
221         const char *old_data = NULL;
222         uint32_t old_mode = REPO_MODE_BLB;
223
224         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
243         /*
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                 old_mode = mode;
259         } else if (node_ctx.action == NODEACT_ADD) {
260                 if (type == REPO_MODE_DIR)
261                         old_data = NULL;
262                 else if (have_text)
263                         old_data = empty_blob;
264                 else
265                         die("invalid dump: adds node without text");
266         } else {
267                 die("invalid dump: Node-path block lacks Node-action");
268         }
269
270         /*
271          * Adjust mode to reflect properties.
272          */
273         if (have_props) {
274                 if (!node_ctx.prop_delta)
275                         node_ctx.type = type;
276                 if (node_ctx.prop_length)
277                         read_props();
278         }
279
280         /*
281          * Save the result.
282          */
283         if (type == REPO_MODE_DIR)      /* directories are not tracked. */
284                 return;
285         assert(old_data);
286         if (old_data == empty_blob)
287                 /* For the fast_export_* functions, NULL means empty. */
288                 old_data = NULL;
289         if (!have_text) {
290                 fast_export_modify(node_ctx.dst.buf, node_ctx.type, old_data);
291                 return;
292         }
293         if (!node_ctx.text_delta) {
294                 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
295                 fast_export_data(node_ctx.type, node_ctx.text_length, &input);
296                 return;
297         }
298         fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
299         fast_export_blob_delta(node_ctx.type, old_mode, old_data,
300                                 node_ctx.text_length, &input);
301 }
302
303 static void begin_revision(const char *remote_ref)
304 {
305         if (!rev_ctx.revision)  /* revision 0 gets no git commit. */
306                 return;
307         fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
308                 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
309                 rev_ctx.timestamp, remote_ref);
310 }
311
312 static void end_revision(const char *note_ref)
313 {
314         struct strbuf mark = STRBUF_INIT;
315         if (rev_ctx.revision) {
316                 fast_export_end_commit(rev_ctx.revision);
317                 fast_export_begin_note(rev_ctx.revision, "remote-svn",
318                                 "Note created by remote-svn.", rev_ctx.timestamp, note_ref);
319                 strbuf_addf(&mark, ":%"PRIu32, rev_ctx.revision);
320                 fast_export_note(mark.buf, "inline");
321                 fast_export_buf_to_data(&rev_ctx.note);
322         }
323 }
324
325 void svndump_read(const char *url, const char *local_ref, const char *notes_ref)
326 {
327         char *val;
328         char *t;
329         uint32_t active_ctx = DUMP_CTX;
330         uint32_t len;
331
332         reset_dump_ctx(url);
333         while ((t = buffer_read_line(&input))) {
334                 val = strchr(t, ':');
335                 if (!val)
336                         continue;
337                 val++;
338                 if (*val != ' ')
339                         continue;
340                 val++;
341
342                 /* strlen(key) + 1 */
343                 switch (val - t - 1) {
344                 case sizeof("SVN-fs-dump-format-version"):
345                         if (constcmp(t, "SVN-fs-dump-format-version"))
346                                 continue;
347                         dump_ctx.version = atoi(val);
348                         if (dump_ctx.version > 3)
349                                 die("expected svn dump format version <= 3, found %"PRIu32,
350                                     dump_ctx.version);
351                         break;
352                 case sizeof("UUID"):
353                         if (constcmp(t, "UUID"))
354                                 continue;
355                         strbuf_reset(&dump_ctx.uuid);
356                         strbuf_addstr(&dump_ctx.uuid, val);
357                         break;
358                 case sizeof("Revision-number"):
359                         if (constcmp(t, "Revision-number"))
360                                 continue;
361                         if (active_ctx == NODE_CTX)
362                                 handle_node();
363                         if (active_ctx == REV_CTX)
364                                 begin_revision(local_ref);
365                         if (active_ctx != DUMP_CTX)
366                                 end_revision(notes_ref);
367                         active_ctx = REV_CTX;
368                         reset_rev_ctx(atoi(val));
369                         strbuf_addf(&rev_ctx.note, "%s\n", t);
370                         break;
371                 case sizeof("Node-path"):
372                         if (constcmp(t, "Node-"))
373                                 continue;
374                         if (!constcmp(t + strlen("Node-"), "path")) {
375                                 if (active_ctx == NODE_CTX)
376                                         handle_node();
377                                 if (active_ctx == REV_CTX)
378                                         begin_revision(local_ref);
379                                 active_ctx = NODE_CTX;
380                                 reset_node_ctx(val);
381                                 strbuf_addf(&rev_ctx.note, "%s\n", t);
382                                 break;
383                         }
384                         if (constcmp(t + strlen("Node-"), "kind"))
385                                 continue;
386                         strbuf_addf(&rev_ctx.note, "%s\n", t);
387                         if (!strcmp(val, "dir"))
388                                 node_ctx.type = REPO_MODE_DIR;
389                         else if (!strcmp(val, "file"))
390                                 node_ctx.type = REPO_MODE_BLB;
391                         else
392                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
393                         break;
394                 case sizeof("Node-action"):
395                         if (constcmp(t, "Node-action"))
396                                 continue;
397                         strbuf_addf(&rev_ctx.note, "%s\n", t);
398                         if (!strcmp(val, "delete")) {
399                                 node_ctx.action = NODEACT_DELETE;
400                         } else if (!strcmp(val, "add")) {
401                                 node_ctx.action = NODEACT_ADD;
402                         } else if (!strcmp(val, "change")) {
403                                 node_ctx.action = NODEACT_CHANGE;
404                         } else if (!strcmp(val, "replace")) {
405                                 node_ctx.action = NODEACT_REPLACE;
406                         } else {
407                                 fprintf(stderr, "Unknown node-action: %s\n", val);
408                                 node_ctx.action = NODEACT_UNKNOWN;
409                         }
410                         break;
411                 case sizeof("Node-copyfrom-path"):
412                         if (constcmp(t, "Node-copyfrom-path"))
413                                 continue;
414                         strbuf_reset(&node_ctx.src);
415                         strbuf_addstr(&node_ctx.src, val);
416                         strbuf_addf(&rev_ctx.note, "%s\n", t);
417                         break;
418                 case sizeof("Node-copyfrom-rev"):
419                         if (constcmp(t, "Node-copyfrom-rev"))
420                                 continue;
421                         node_ctx.srcRev = atoi(val);
422                         strbuf_addf(&rev_ctx.note, "%s\n", t);
423                         break;
424                 case sizeof("Text-content-length"):
425                         if (constcmp(t, "Text") && constcmp(t, "Prop"))
426                                 continue;
427                         if (constcmp(t + 4, "-content-length"))
428                                 continue;
429                         {
430                                 char *end;
431                                 uintmax_t len;
432
433                                 len = strtoumax(val, &end, 10);
434                                 if (!isdigit(*val) || *end)
435                                         die("invalid dump: non-numeric length %s", val);
436                                 if (len > maximum_signed_value_of_type(off_t))
437                                         die("unrepresentable length in dump: %s", val);
438
439                                 if (*t == 'T')
440                                         node_ctx.text_length = (off_t) len;
441                                 else
442                                         node_ctx.prop_length = (off_t) len;
443                                 break;
444                         }
445                 case sizeof("Text-delta"):
446                         if (!constcmp(t, "Text-delta")) {
447                                 node_ctx.text_delta = !strcmp(val, "true");
448                                 break;
449                         }
450                         if (constcmp(t, "Prop-delta"))
451                                 continue;
452                         node_ctx.prop_delta = !strcmp(val, "true");
453                         break;
454                 case sizeof("Content-length"):
455                         if (constcmp(t, "Content-length"))
456                                 continue;
457                         len = atoi(val);
458                         t = buffer_read_line(&input);
459                         if (!t)
460                                 die_short_read();
461                         if (*t)
462                                 die("invalid dump: expected blank line after content length header");
463                         if (active_ctx == REV_CTX) {
464                                 read_props();
465                         } else if (active_ctx == NODE_CTX) {
466                                 handle_node();
467                                 active_ctx = INTERNODE_CTX;
468                         } else {
469                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
470                                 if (buffer_skip_bytes(&input, len) != len)
471                                         die_short_read();
472                         }
473                 }
474         }
475         if (buffer_ferror(&input))
476                 die_short_read();
477         if (active_ctx == NODE_CTX)
478                 handle_node();
479         if (active_ctx == REV_CTX)
480                 begin_revision(local_ref);
481         if (active_ctx != DUMP_CTX)
482                 end_revision(notes_ref);
483 }
484
485 static void init(int report_fd)
486 {
487         fast_export_init(report_fd);
488         strbuf_init(&dump_ctx.uuid, 4096);
489         strbuf_init(&dump_ctx.url, 4096);
490         strbuf_init(&rev_ctx.log, 4096);
491         strbuf_init(&rev_ctx.author, 4096);
492         strbuf_init(&rev_ctx.note, 4096);
493         strbuf_init(&node_ctx.src, 4096);
494         strbuf_init(&node_ctx.dst, 4096);
495         reset_dump_ctx(NULL);
496         reset_rev_ctx(0);
497         reset_node_ctx(NULL);
498         return;
499 }
500
501 int svndump_init(const char *filename)
502 {
503         if (buffer_init(&input, filename))
504                 return error_errno("cannot open %s", filename ? filename : "NULL");
505         init(REPORT_FILENO);
506         return 0;
507 }
508
509 int svndump_init_fd(int in_fd, int back_fd)
510 {
511         if(buffer_fdinit(&input, xdup(in_fd)))
512                 return error_errno("cannot open fd %d", in_fd);
513         init(xdup(back_fd));
514         return 0;
515 }
516
517 void svndump_deinit(void)
518 {
519         fast_export_deinit();
520         reset_dump_ctx(NULL);
521         reset_rev_ctx(0);
522         reset_node_ctx(NULL);
523         strbuf_release(&rev_ctx.log);
524         strbuf_release(&rev_ctx.author);
525         strbuf_release(&rev_ctx.note);
526         strbuf_release(&node_ctx.src);
527         strbuf_release(&node_ctx.dst);
528         if (buffer_deinit(&input))
529                 fprintf(stderr, "Input error\n");
530         if (ferror(stdout))
531                 fprintf(stderr, "Output error\n");
532 }
533
534 void svndump_reset(void)
535 {
536         strbuf_release(&dump_ctx.uuid);
537         strbuf_release(&dump_ctx.url);
538         strbuf_release(&rev_ctx.log);
539         strbuf_release(&rev_ctx.author);
540 }