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 "string_pool.h"
  15 #include "strbuf.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 NODEACT_REPLACE 4
  24 #define NODEACT_DELETE 3
  25 #define NODEACT_ADD 2
  26 #define NODEACT_CHANGE 1
  27 #define NODEACT_UNKNOWN 0
  28
  29 #define DUMP_CTX 0
  30 #define REV_CTX  1
  31 #define NODE_CTX 2
  32
  33 #define LENGTH_UNKNOWN (~0)
  34 #define DATE_RFC2822_LEN 31
  35
  36 static struct line_buffer input = LINE_BUFFER_INIT;
  37
  38 static 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;
  43
  44 static struct {
  45         uint32_t revision;
  46         unsigned long timestamp;
  47         struct strbuf log, author;
  48 } rev_ctx;
  49
  50 static struct {
  51         uint32_t version;
  52         struct strbuf uuid, url;
  53 } dump_ctx;
  54
  55 static 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 }
  67
  68 static 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 }
  75
  76 static 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 }
  84
  85 static 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
  92         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 }
 139
 140 static 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 }
 146
 147 static void read_props(void)
 148 {
 149         static struct strbuf key = STRBUF_INIT;
 150         static struct strbuf val = STRBUF_INIT;
 151         const char *t;
 152         /*
 153          * NEEDSWORK: to support simple mode changes like
 154          *      K 11
 155          *      svn:special
 156          *      V 1
 157          *      *
 158          *      D 14
 159          *      svn:executable
 160          * we keep track of whether a mode has been set and reset to
 161          * plain file only if not.  We should be keeping track of the
 162          * symlink and executable bits separately instead.
 163          */
 164         uint32_t type_set = 0;
 165         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 166                 uint32_t len;
 167                 const char type = t[0];
 168                 int ch;
 169
 170                 if (!type || t[1] != ' ')
 171                         die("invalid property line: %s\n", t);
 172                 len = atoi(&t[2]);
 173                 strbuf_reset(&val);
 174                 buffer_read_binary(&input, &val, len);
 175                 if (val.len < len)
 176                         die_short_read();
 177
 178                 /* Discard trailing newline. */
 179                 ch = buffer_read_char(&input);
 180                 if (ch == EOF)
 181                         die_short_read();
 182                 if (ch != '\n')
 183                         die("invalid dump: expected newline after %s", val.buf);
 184
 185                 switch (type) {
 186                 case 'K':
 187                         strbuf_swap(&key, &val);
 188                         continue;
 189                 case 'D':
 190                         handle_property(&val, NULL, 0, &type_set);
 191                         continue;
 192                 case 'V':
 193                         handle_property(&key, val.buf, len, &type_set);
 194                         strbuf_reset(&key);
 195                         continue;
 196                 default:
 197                         die("invalid property line: %s\n", t);
 198                 }
 199         }
 200 }
 201
 202 static void handle_node(void)
 203 {
 204         uint32_t mark = 0;
 205         const uint32_t type = node_ctx.type;
 206         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 207         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 208
 209         if (node_ctx.text_delta)
 210                 die("text deltas not supported");
 211         if (have_text)
 212                 mark = next_blob_mark();
 213         if (node_ctx.action == NODEACT_DELETE) {
 214                 if (have_text || have_props || node_ctx.srcRev)
 215                         die("invalid dump: deletion node has "
 216                                 "copyfrom info, text, or properties");
 217                 return repo_delete(node_ctx.dst);
 218         }
 219         if (node_ctx.action == NODEACT_REPLACE) {
 220                 repo_delete(node_ctx.dst);
 221                 node_ctx.action = NODEACT_ADD;
 222         }
 223         if (node_ctx.srcRev) {
 224                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 225                 if (node_ctx.action == NODEACT_ADD)
 226                         node_ctx.action = NODEACT_CHANGE;
 227         }
 228         if (have_text && type == REPO_MODE_DIR)
 229                 die("invalid dump: directories cannot have text attached");
 230
 231         /*
 232          * Decide on the new content (mark) and mode (node_ctx.type).
 233          */
 234         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 235                 if (type != REPO_MODE_DIR)
 236                         die("invalid dump: root of tree is not a regular file");
 237         } else if (node_ctx.action == NODEACT_CHANGE) {
 238                 uint32_t mode;
 239                 if (!have_text)
 240                         mark = repo_read_path(node_ctx.dst);
 241                 mode = repo_read_mode(node_ctx.dst);
 242                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 243                         die("invalid dump: cannot modify a directory into a file");
 244                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 245                         die("invalid dump: cannot modify a file into a directory");
 246                 node_ctx.type = mode;
 247         } else if (node_ctx.action == NODEACT_ADD) {
 248                 if (!have_text && type != REPO_MODE_DIR)
 249                         die("invalid dump: adds node without text");
 250         } else {
 251                 die("invalid dump: Node-path block lacks Node-action");
 252         }
 253
 254         /*
 255          * Adjust mode to reflect properties.
 256          */
 257         if (have_props) {
 258                 if (!node_ctx.prop_delta)
 259                         node_ctx.type = type;
 260                 if (node_ctx.propLength)
 261                         read_props();
 262         }
 263
 264         /*
 265          * Save the result.
 266          */
 267         repo_add(node_ctx.dst, node_ctx.type, mark);
 268         if (have_text)
 269                 fast_export_blob(node_ctx.type, mark,
 270                                  node_ctx.textLength, &input);
 271 }
 272
 273 static void handle_revision(void)
 274 {
 275         if (rev_ctx.revision)
 276                 repo_commit(rev_ctx.revision, rev_ctx.author.buf,
 277                         rev_ctx.log.buf, dump_ctx.uuid.buf, dump_ctx.url.buf,
 278                         rev_ctx.timestamp);
 279 }
 280
 281 void svndump_read(const char *url)
 282 {
 283         char *val;
 284         char *t;
 285         uint32_t active_ctx = DUMP_CTX;
 286         uint32_t len;
 287
 288         reset_dump_ctx(url);
 289         while ((t = buffer_read_line(&input))) {
 290                 val = strchr(t, ':');
 291                 if (!val)
 292                         continue;
 293                 val++;
 294                 if (*val != ' ')
 295                         continue;
 296                 val++;
 297
 298                 /* strlen(key) + 1 */
 299                 switch (val - t - 1) {
 300                 case sizeof("SVN-fs-dump-format-version"):
 301                         if (constcmp(t, "SVN-fs-dump-format-version"))
 302                                 continue;
 303                         dump_ctx.version = atoi(val);
 304                         if (dump_ctx.version > 3)
 305                                 die("expected svn dump format version <= 3, found %"PRIu32,
 306                                     dump_ctx.version);
 307                         break;
 308                 case sizeof("UUID"):
 309                         if (constcmp(t, "UUID"))
 310                                 continue;
 311                         strbuf_reset(&dump_ctx.uuid);
 312                         strbuf_addstr(&dump_ctx.uuid, val);
 313                         break;
 314                 case sizeof("Revision-number"):
 315                         if (constcmp(t, "Revision-number"))
 316                                 continue;
 317                         if (active_ctx == NODE_CTX)
 318                                 handle_node();
 319                         if (active_ctx != DUMP_CTX)
 320                                 handle_revision();
 321                         active_ctx = REV_CTX;
 322                         reset_rev_ctx(atoi(val));
 323                         break;
 324                 case sizeof("Node-path"):
 325                         if (prefixcmp(t, "Node-"))
 326                                 continue;
 327                         if (!constcmp(t + strlen("Node-"), "path")) {
 328                                 if (active_ctx == NODE_CTX)
 329                                         handle_node();
 330                                 active_ctx = NODE_CTX;
 331                                 reset_node_ctx(val);
 332                                 break;
 333                         }
 334                         if (constcmp(t + strlen("Node-"), "kind"))
 335                                 continue;
 336                         if (!strcmp(val, "dir"))
 337                                 node_ctx.type = REPO_MODE_DIR;
 338                         else if (!strcmp(val, "file"))
 339                                 node_ctx.type = REPO_MODE_BLB;
 340                         else
 341                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 342                         break;
 343                 case sizeof("Node-action"):
 344                         if (constcmp(t, "Node-action"))
 345                                 continue;
 346                         if (!strcmp(val, "delete")) {
 347                                 node_ctx.action = NODEACT_DELETE;
 348                         } else if (!strcmp(val, "add")) {
 349                                 node_ctx.action = NODEACT_ADD;
 350                         } else if (!strcmp(val, "change")) {
 351                                 node_ctx.action = NODEACT_CHANGE;
 352                         } else if (!strcmp(val, "replace")) {
 353                                 node_ctx.action = NODEACT_REPLACE;
 354                         } else {
 355                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 356                                 node_ctx.action = NODEACT_UNKNOWN;
 357                         }
 358                         break;
 359                 case sizeof("Node-copyfrom-path"):
 360                         if (constcmp(t, "Node-copyfrom-path"))
 361                                 continue;
 362                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 363                         break;
 364                 case sizeof("Node-copyfrom-rev"):
 365                         if (constcmp(t, "Node-copyfrom-rev"))
 366                                 continue;
 367                         node_ctx.srcRev = atoi(val);
 368                         break;
 369                 case sizeof("Text-content-length"):
 370                         if (!constcmp(t, "Text-content-length")) {
 371                                 node_ctx.textLength = atoi(val);
 372                                 break;
 373                         }
 374                         if (constcmp(t, "Prop-content-length"))
 375                                 continue;
 376                         node_ctx.propLength = atoi(val);
 377                         break;
 378                 case sizeof("Text-delta"):
 379                         if (!constcmp(t, "Text-delta")) {
 380                                 node_ctx.text_delta = !strcmp(val, "true");
 381                                 break;
 382                         }
 383                         if (constcmp(t, "Prop-delta"))
 384                                 continue;
 385                         node_ctx.prop_delta = !strcmp(val, "true");
 386                         break;
 387                 case sizeof("Content-length"):
 388                         if (constcmp(t, "Content-length"))
 389                                 continue;
 390                         len = atoi(val);
 391                         t = buffer_read_line(&input);
 392                         if (!t)
 393                                 die_short_read();
 394                         if (*t)
 395                                 die("invalid dump: expected blank line after content length header");
 396                         if (active_ctx == REV_CTX) {
 397                                 read_props();
 398                         } else if (active_ctx == NODE_CTX) {
 399                                 handle_node();
 400                                 active_ctx = REV_CTX;
 401                         } else {
 402                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 403                                 if (buffer_skip_bytes(&input, len) != len)
 404                                         die_short_read();
 405                         }
 406                 }
 407         }
 408         if (buffer_ferror(&input))
 409                 die_short_read();
 410         if (active_ctx == NODE_CTX)
 411                 handle_node();
 412         if (active_ctx != DUMP_CTX)
 413                 handle_revision();
 414 }
 415
 416 int svndump_init(const char *filename)
 417 {
 418         if (buffer_init(&input, filename))
 419                 return error("cannot open %s: %s", filename, strerror(errno));
 420         repo_init();
 421         strbuf_init(&dump_ctx.uuid, 4096);
 422         strbuf_init(&dump_ctx.url, 4096);
 423         strbuf_init(&rev_ctx.log, 4096);
 424         strbuf_init(&rev_ctx.author, 4096);
 425         reset_dump_ctx(NULL);
 426         reset_rev_ctx(0);
 427         reset_node_ctx(NULL);
 428         return 0;
 429 }
 430
 431 void svndump_deinit(void)
 432 {
 433         repo_reset();
 434         reset_dump_ctx(NULL);
 435         reset_rev_ctx(0);
 436         reset_node_ctx(NULL);
 437         strbuf_release(&rev_ctx.log);
 438         if (buffer_deinit(&input))
 439                 fprintf(stderr, "Input error\n");
 440         if (ferror(stdout))
 441                 fprintf(stderr, "Output error\n");
 442 }
 443
 444 void svndump_reset(void)
 445 {
 446         buffer_reset(&input);
 447         repo_reset();
 448         strbuf_release(&dump_ctx.uuid);
 449         strbuf_release(&dump_ctx.url);
 450         strbuf_release(&rev_ctx.log);
 451         strbuf_release(&rev_ctx.author);
 452 }