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 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 LENGTH_UNKNOWN (~0)
  38 #define DATE_RFC2822_LEN 31
  39
  40 static struct line_buffer input = LINE_BUFFER_INIT;
  41
  42 static struct {
  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;
  46 } node_ctx;
  47
  48 static struct {
  49         uint32_t revision;
  50         unsigned long timestamp;
  51         struct strbuf log, author;
  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.propLength = LENGTH_UNKNOWN;
  64         node_ctx.textLength = LENGTH_UNKNOWN;
  65         node_ctx.src[0] = ~0;
  66         node_ctx.srcRev = 0;
  67         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  68         node_ctx.text_delta = 0;
  69         node_ctx.prop_delta = 0;
  70 }
  71
  72 static void reset_rev_ctx(uint32_t revision)
  73 {
  74         rev_ctx.revision = revision;
  75         rev_ctx.timestamp = 0;
  76         strbuf_reset(&rev_ctx.log);
  77         strbuf_reset(&rev_ctx.author);
  78 }
  79
  80 static void reset_dump_ctx(const char *url)
  81 {
  82         strbuf_reset(&dump_ctx.url);
  83         if (url)
  84                 strbuf_addstr(&dump_ctx.url, url);
  85         dump_ctx.version = 1;
  86         strbuf_reset(&dump_ctx.uuid);
  87 }
  88
  89 static void handle_property(const struct strbuf *key_buf,
  90                                 struct strbuf *val,
  91                                 uint32_t *type_set)
  92 {
  93         const char *key = key_buf->buf;
  94         size_t keylen = key_buf->len;
  95
  96         switch (keylen + 1) {
  97         case sizeof("svn:log"):
  98                 if (constcmp(key, "svn:log"))
  99                         break;
 100                 if (!val)
 101                         die("invalid dump: unsets svn:log");
 102                 strbuf_swap(&rev_ctx.log, val);
 103                 break;
 104         case sizeof("svn:author"):
 105                 if (constcmp(key, "svn:author"))
 106                         break;
 107                 if (!val)
 108                         strbuf_reset(&rev_ctx.author);
 109                 else
 110                         strbuf_swap(&rev_ctx.author, val);
 111                 break;
 112         case sizeof("svn:date"):
 113                 if (constcmp(key, "svn:date"))
 114                         break;
 115                 if (!val)
 116                         die("invalid dump: unsets svn:date");
 117                 if (parse_date_basic(val->buf, &rev_ctx.timestamp, NULL))
 118                         warning("invalid timestamp: %s", val->buf);
 119                 break;
 120         case sizeof("svn:executable"):
 121         case sizeof("svn:special"):
 122                 if (keylen == strlen("svn:executable") &&
 123                     constcmp(key, "svn:executable"))
 124                         break;
 125                 if (keylen == strlen("svn:special") &&
 126                     constcmp(key, "svn:special"))
 127                         break;
 128                 if (*type_set) {
 129                         if (!val)
 130                                 return;
 131                         die("invalid dump: sets type twice");
 132                 }
 133                 if (!val) {
 134                         node_ctx.type = REPO_MODE_BLB;
 135                         return;
 136                 }
 137                 *type_set = 1;
 138                 node_ctx.type = keylen == strlen("svn:executable") ?
 139                                 REPO_MODE_EXE :
 140                                 REPO_MODE_LNK;
 141         }
 142 }
 143
 144 static void die_short_read(void)
 145 {
 146         if (buffer_ferror(&input))
 147                 die_errno("error reading dump file");
 148         die("invalid dump: unexpected end of file");
 149 }
 150
 151 static void read_props(void)
 152 {
 153         static struct strbuf key = STRBUF_INIT;
 154         static struct strbuf val = STRBUF_INIT;
 155         const char *t;
 156         /*
 157          * NEEDSWORK: to support simple mode changes like
 158          *      K 11
 159          *      svn:special
 160          *      V 1
 161          *      *
 162          *      D 14
 163          *      svn:executable
 164          * we keep track of whether a mode has been set and reset to
 165          * plain file only if not.  We should be keeping track of the
 166          * symlink and executable bits separately instead.
 167          */
 168         uint32_t type_set = 0;
 169         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 170                 uint32_t len;
 171                 const char type = t[0];
 172                 int ch;
 173
 174                 if (!type || t[1] != ' ')
 175                         die("invalid property line: %s\n", t);
 176                 len = atoi(&t[2]);
 177                 strbuf_reset(&val);
 178                 buffer_read_binary(&input, &val, len);
 179                 if (val.len < len)
 180                         die_short_read();
 181
 182                 /* Discard trailing newline. */
 183                 ch = buffer_read_char(&input);
 184                 if (ch == EOF)
 185                         die_short_read();
 186                 if (ch != '\n')
 187                         die("invalid dump: expected newline after %s", val.buf);
 188
 189                 switch (type) {
 190                 case 'K':
 191                         strbuf_swap(&key, &val);
 192                         continue;
 193                 case 'D':
 194                         handle_property(&val, NULL, &type_set);
 195                         continue;
 196                 case 'V':
 197                         handle_property(&key, &val, &type_set);
 198                         strbuf_reset(&key);
 199                         continue;
 200                 default:
 201                         die("invalid property line: %s\n", t);
 202                 }
 203         }
 204 }
 205
 206 static void handle_node(void)
 207 {
 208         const uint32_t type = node_ctx.type;
 209         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 210         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 211         /*
 212          * Old text for this node:
 213          *  NULL        - directory or bug
 214          *  empty_blob  - empty
 215          *  "<dataref>" - data retrievable from fast-import
 216          */
 217         static const char *const empty_blob = "::empty::";
 218         const char *old_data = NULL;
 219
 220         if (node_ctx.text_delta)
 221                 die("text deltas not supported");
 222
 223         if (node_ctx.action == NODEACT_DELETE) {
 224                 if (have_text || have_props || node_ctx.srcRev)
 225                         die("invalid dump: deletion node has "
 226                                 "copyfrom info, text, or properties");
 227                 repo_delete(node_ctx.dst);
 228                 return;
 229         }
 230         if (node_ctx.action == NODEACT_REPLACE) {
 231                 repo_delete(node_ctx.dst);
 232                 node_ctx.action = NODEACT_ADD;
 233         }
 234         if (node_ctx.srcRev) {
 235                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 236                 if (node_ctx.action == NODEACT_ADD)
 237                         node_ctx.action = NODEACT_CHANGE;
 238         }
 239         if (have_text && type == REPO_MODE_DIR)
 240                 die("invalid dump: directories cannot have text attached");
 241
 242         /*
 243          * Find old content (old_data) and decide on the new mode.
 244          */
 245         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 246                 if (type != REPO_MODE_DIR)
 247                         die("invalid dump: root of tree is not a regular file");
 248                 old_data = NULL;
 249         } else if (node_ctx.action == NODEACT_CHANGE) {
 250                 uint32_t mode;
 251                 old_data = repo_read_path(node_ctx.dst, &mode);
 252                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 253                         die("invalid dump: cannot modify a directory into a file");
 254                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 255                         die("invalid dump: cannot modify a file into a directory");
 256                 node_ctx.type = mode;
 257         } else if (node_ctx.action == NODEACT_ADD) {
 258                 if (type == REPO_MODE_DIR)
 259                         old_data = NULL;
 260                 else if (have_text)
 261                         old_data = empty_blob;
 262                 else
 263                         die("invalid dump: adds node without text");
 264         } else {
 265                 die("invalid dump: Node-path block lacks Node-action");
 266         }
 267
 268         /*
 269          * Adjust mode to reflect properties.
 270          */
 271         if (have_props) {
 272                 if (!node_ctx.prop_delta)
 273                         node_ctx.type = type;
 274                 if (node_ctx.propLength)
 275                         read_props();
 276         }
 277
 278         /*
 279          * Save the result.
 280          */
 281         if (type == REPO_MODE_DIR)      /* directories are not tracked. */
 282                 return;
 283         assert(old_data);
 284         if (old_data == empty_blob)
 285                 /* For the fast_export_* functions, NULL means empty. */
 286                 old_data = NULL;
 287         if (!have_text) {
 288                 fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
 289                                         node_ctx.type, old_data);
 290                 return;
 291         }
 292         fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
 293                                 node_ctx.type, "inline");
 294         fast_export_data(node_ctx.type, node_ctx.textLength, &input);
 295 }
 296
 297 static void begin_revision(void)
 298 {
 299         if (!rev_ctx.revision)  /* revision 0 gets no git commit. */
 300                 return;
 301         fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
 302                 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
 303                 rev_ctx.timestamp);
 304 }
 305
 306 static void end_revision(void)
 307 {
 308         if (rev_ctx.revision)
 309                 fast_export_end_commit(rev_ctx.revision);
 310 }
 311
 312 void svndump_read(const char *url)
 313 {
 314         char *val;
 315         char *t;
 316         uint32_t active_ctx = DUMP_CTX;
 317         uint32_t len;
 318
 319         reset_dump_ctx(url);
 320         while ((t = buffer_read_line(&input))) {
 321                 val = strchr(t, ':');
 322                 if (!val)
 323                         continue;
 324                 val++;
 325                 if (*val != ' ')
 326                         continue;
 327                 val++;
 328
 329                 /* strlen(key) + 1 */
 330                 switch (val - t - 1) {
 331                 case sizeof("SVN-fs-dump-format-version"):
 332                         if (constcmp(t, "SVN-fs-dump-format-version"))
 333                                 continue;
 334                         dump_ctx.version = atoi(val);
 335                         if (dump_ctx.version > 3)
 336                                 die("expected svn dump format version <= 3, found %"PRIu32,
 337                                     dump_ctx.version);
 338                         break;
 339                 case sizeof("UUID"):
 340                         if (constcmp(t, "UUID"))
 341                                 continue;
 342                         strbuf_reset(&dump_ctx.uuid);
 343                         strbuf_addstr(&dump_ctx.uuid, val);
 344                         break;
 345                 case sizeof("Revision-number"):
 346                         if (constcmp(t, "Revision-number"))
 347                                 continue;
 348                         if (active_ctx == NODE_CTX)
 349                                 handle_node();
 350                         if (active_ctx == REV_CTX)
 351                                 begin_revision();
 352                         if (active_ctx != DUMP_CTX)
 353                                 end_revision();
 354                         active_ctx = REV_CTX;
 355                         reset_rev_ctx(atoi(val));
 356                         break;
 357                 case sizeof("Node-path"):
 358                         if (prefixcmp(t, "Node-"))
 359                                 continue;
 360                         if (!constcmp(t + strlen("Node-"), "path")) {
 361                                 if (active_ctx == NODE_CTX)
 362                                         handle_node();
 363                                 if (active_ctx == REV_CTX)
 364                                         begin_revision();
 365                                 active_ctx = NODE_CTX;
 366                                 reset_node_ctx(val);
 367                                 break;
 368                         }
 369                         if (constcmp(t + strlen("Node-"), "kind"))
 370                                 continue;
 371                         if (!strcmp(val, "dir"))
 372                                 node_ctx.type = REPO_MODE_DIR;
 373                         else if (!strcmp(val, "file"))
 374                                 node_ctx.type = REPO_MODE_BLB;
 375                         else
 376                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 377                         break;
 378                 case sizeof("Node-action"):
 379                         if (constcmp(t, "Node-action"))
 380                                 continue;
 381                         if (!strcmp(val, "delete")) {
 382                                 node_ctx.action = NODEACT_DELETE;
 383                         } else if (!strcmp(val, "add")) {
 384                                 node_ctx.action = NODEACT_ADD;
 385                         } else if (!strcmp(val, "change")) {
 386                                 node_ctx.action = NODEACT_CHANGE;
 387                         } else if (!strcmp(val, "replace")) {
 388                                 node_ctx.action = NODEACT_REPLACE;
 389                         } else {
 390                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 391                                 node_ctx.action = NODEACT_UNKNOWN;
 392                         }
 393                         break;
 394                 case sizeof("Node-copyfrom-path"):
 395                         if (constcmp(t, "Node-copyfrom-path"))
 396                                 continue;
 397                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 398                         break;
 399                 case sizeof("Node-copyfrom-rev"):
 400                         if (constcmp(t, "Node-copyfrom-rev"))
 401                                 continue;
 402                         node_ctx.srcRev = atoi(val);
 403                         break;
 404                 case sizeof("Text-content-length"):
 405                         if (!constcmp(t, "Text-content-length")) {
 406                                 node_ctx.textLength = atoi(val);
 407                                 break;
 408                         }
 409                         if (constcmp(t, "Prop-content-length"))
 410                                 continue;
 411                         node_ctx.propLength = atoi(val);
 412                         break;
 413                 case sizeof("Text-delta"):
 414                         if (!constcmp(t, "Text-delta")) {
 415                                 node_ctx.text_delta = !strcmp(val, "true");
 416                                 break;
 417                         }
 418                         if (constcmp(t, "Prop-delta"))
 419                                 continue;
 420                         node_ctx.prop_delta = !strcmp(val, "true");
 421                         break;
 422                 case sizeof("Content-length"):
 423                         if (constcmp(t, "Content-length"))
 424                                 continue;
 425                         len = atoi(val);
 426                         t = buffer_read_line(&input);
 427                         if (!t)
 428                                 die_short_read();
 429                         if (*t)
 430                                 die("invalid dump: expected blank line after content length header");
 431                         if (active_ctx == REV_CTX) {
 432                                 read_props();
 433                         } else if (active_ctx == NODE_CTX) {
 434                                 handle_node();
 435                                 active_ctx = INTERNODE_CTX;
 436                         } else {
 437                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 438                                 if (buffer_skip_bytes(&input, len) != len)
 439                                         die_short_read();
 440                         }
 441                 }
 442         }
 443         if (buffer_ferror(&input))
 444                 die_short_read();
 445         if (active_ctx == NODE_CTX)
 446                 handle_node();
 447         if (active_ctx == REV_CTX)
 448                 begin_revision();
 449         if (active_ctx != DUMP_CTX)
 450                 end_revision();
 451 }
 452
 453 int svndump_init(const char *filename)
 454 {
 455         if (buffer_init(&input, filename))
 456                 return error("cannot open %s: %s", filename, strerror(errno));
 457         fast_export_init(REPORT_FILENO);
 458         strbuf_init(&dump_ctx.uuid, 4096);
 459         strbuf_init(&dump_ctx.url, 4096);
 460         strbuf_init(&rev_ctx.log, 4096);
 461         strbuf_init(&rev_ctx.author, 4096);
 462         reset_dump_ctx(NULL);
 463         reset_rev_ctx(0);
 464         reset_node_ctx(NULL);
 465         return 0;
 466 }
 467
 468 void svndump_deinit(void)
 469 {
 470         fast_export_deinit();
 471         reset_dump_ctx(NULL);
 472         reset_rev_ctx(0);
 473         reset_node_ctx(NULL);
 474         strbuf_release(&rev_ctx.log);
 475         if (buffer_deinit(&input))
 476                 fprintf(stderr, "Input error\n");
 477         if (ferror(stdout))
 478                 fprintf(stderr, "Output error\n");
 479 }
 480
 481 void svndump_reset(void)
 482 {
 483         fast_export_reset();
 484         buffer_reset(&input);
 485         strbuf_release(&dump_ctx.uuid);
 486         strbuf_release(&dump_ctx.url);
 487         strbuf_release(&rev_ctx.log);
 488         strbuf_release(&rev_ctx.author);
 489 }