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 #define REPORT_FILENO 3
  18
  19 /*
  20  * Compare start of string to literal of equal length;
  21  * must be guarded by length test.
  22  */
  23 #define constcmp(s, ref) memcmp(s, ref, sizeof(ref) - 1)
  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                                 const char *val, uint32_t len,
  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_reset(&rev_ctx.log);
 103                 strbuf_add(&rev_ctx.log, val, len);
 104                 break;
 105         case sizeof("svn:author"):
 106                 if (constcmp(key, "svn:author"))
 107                         break;
 108                 strbuf_reset(&rev_ctx.author);
 109                 if (val)
 110                         strbuf_add(&rev_ctx.author, val, len);
 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, &rev_ctx.timestamp, NULL))
 118                         warning("invalid timestamp: %s", val);
 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         const char *t;
 155         /*
 156          * NEEDSWORK: to support simple mode changes like
 157          *      K 11
 158          *      svn:special
 159          *      V 1
 160          *      *
 161          *      D 14
 162          *      svn:executable
 163          * we keep track of whether a mode has been set and reset to
 164          * plain file only if not.  We should be keeping track of the
 165          * symlink and executable bits separately instead.
 166          */
 167         uint32_t type_set = 0;
 168         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 169                 uint32_t len;
 170                 const char *val;
 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                 val = buffer_read_string(&input, len);
 178                 if (!val || strlen(val) != len)
 179                         die_short_read();
 180
 181                 /* Discard trailing newline. */
 182                 ch = buffer_read_char(&input);
 183                 if (ch == EOF)
 184                         die_short_read();
 185                 if (ch != '\n')
 186                         die("invalid dump: expected newline after %s", val);
 187
 188                 switch (type) {
 189                 case 'K':
 190                 case 'D':
 191                         strbuf_reset(&key);
 192                         if (val)
 193                                 strbuf_add(&key, val, len);
 194                         if (type == 'K')
 195                                 continue;
 196                         assert(type == 'D');
 197                         val = NULL;
 198                         len = 0;
 199                         /* fall through */
 200                 case 'V':
 201                         handle_property(&key, val, len, &type_set);
 202                         strbuf_reset(&key);
 203                         continue;
 204                 default:
 205                         die("invalid property line: %s\n", t);
 206                 }
 207         }
 208 }
 209
 210 static void handle_node(void)
 211 {
 212         const uint32_t type = node_ctx.type;
 213         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 214         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 215         /*
 216          * Old text for this node:
 217          *  NULL        - directory or bug
 218          *  empty_blob  - empty
 219          *  "<dataref>" - data retrievable from fast-import
 220          */
 221         static const char *const empty_blob = "::empty::";
 222         const char *old_data = NULL;
 223
 224         if (node_ctx.text_delta)
 225                 die("text deltas not supported");
 226
 227         if (node_ctx.action == NODEACT_DELETE) {
 228                 if (have_text || have_props || node_ctx.srcRev)
 229                         die("invalid dump: deletion node has "
 230                                 "copyfrom info, text, or properties");
 231                 return repo_delete(node_ctx.dst);
 232         }
 233         if (node_ctx.action == NODEACT_REPLACE) {
 234                 repo_delete(node_ctx.dst);
 235                 node_ctx.action = NODEACT_ADD;
 236         }
 237         if (node_ctx.srcRev) {
 238                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 239                 if (node_ctx.action == NODEACT_ADD)
 240                         node_ctx.action = NODEACT_CHANGE;
 241         }
 242         if (have_text && type == REPO_MODE_DIR)
 243                 die("invalid dump: directories cannot have text attached");
 244
 245         /*
 246          * Find old content (old_data) and decide on the new mode.
 247          */
 248         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 249                 if (type != REPO_MODE_DIR)
 250                         die("invalid dump: root of tree is not a regular file");
 251                 old_data = NULL;
 252         } else if (node_ctx.action == NODEACT_CHANGE) {
 253                 uint32_t mode;
 254                 old_data = repo_read_path(node_ctx.dst);
 255                 mode = repo_read_mode(node_ctx.dst);
 256                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 257                         die("invalid dump: cannot modify a directory into a file");
 258                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 259                         die("invalid dump: cannot modify a file into a directory");
 260                 node_ctx.type = mode;
 261         } else if (node_ctx.action == NODEACT_ADD) {
 262                 if (type == REPO_MODE_DIR)
 263                         old_data = NULL;
 264                 else if (have_text)
 265                         old_data = empty_blob;
 266                 else
 267                         die("invalid dump: adds node without text");
 268         } else {
 269                 die("invalid dump: Node-path block lacks Node-action");
 270         }
 271
 272         /*
 273          * Adjust mode to reflect properties.
 274          */
 275         if (have_props) {
 276                 if (!node_ctx.prop_delta)
 277                         node_ctx.type = type;
 278                 if (node_ctx.propLength)
 279                         read_props();
 280         }
 281
 282         /*
 283          * Save the result.
 284          */
 285         if (type == REPO_MODE_DIR)      /* directories are not tracked. */
 286                 return;
 287         assert(old_data);
 288         if (old_data == empty_blob)
 289                 /* For the fast_export_* functions, NULL means empty. */
 290                 old_data = NULL;
 291         if (!have_text) {
 292                 fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
 293                                         node_ctx.type, old_data);
 294                 return;
 295         }
 296         fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
 297                                 node_ctx.type, "inline");
 298         fast_export_data(node_ctx.type, node_ctx.textLength, &input);
 299 }
 300
 301 static void begin_revision(void)
 302 {
 303         if (!rev_ctx.revision)  /* revision 0 gets no git commit. */
 304                 return;
 305         fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
 306                 rev_ctx.log.buf, dump_ctx.uuid.buf, dump_ctx.url.buf,
 307                 rev_ctx.timestamp);
 308 }
 309
 310 static void end_revision(void)
 311 {
 312         if (rev_ctx.revision)
 313                 fast_export_end_commit(rev_ctx.revision);
 314 }
 315
 316 void svndump_read(const char *url)
 317 {
 318         char *val;
 319         char *t;
 320         uint32_t active_ctx = DUMP_CTX;
 321         uint32_t len;
 322
 323         reset_dump_ctx(url);
 324         while ((t = buffer_read_line(&input))) {
 325                 val = strstr(t, ": ");
 326                 if (!val)
 327                         continue;
 328                 val += 2;
 329
 330                 /* strlen(key) + 1 */
 331                 switch (val - t - 1) {
 332                 case sizeof("SVN-fs-dump-format-version"):
 333                         if (constcmp(t, "SVN-fs-dump-format-version"))
 334                                 continue;
 335                         dump_ctx.version = atoi(val);
 336                         if (dump_ctx.version > 3)
 337                                 die("expected svn dump format version <= 3, found %"PRIu32,
 338                                     dump_ctx.version);
 339                         break;
 340                 case sizeof("UUID"):
 341                         if (constcmp(t, "UUID"))
 342                                 continue;
 343                         strbuf_reset(&dump_ctx.uuid);
 344                         strbuf_addstr(&dump_ctx.uuid, val);
 345                         break;
 346                 case sizeof("Revision-number"):
 347                         if (constcmp(t, "Revision-number"))
 348                                 continue;
 349                         if (active_ctx == NODE_CTX)
 350                                 handle_node();
 351                         if (active_ctx == REV_CTX)
 352                                 begin_revision();
 353                         if (active_ctx != DUMP_CTX)
 354                                 end_revision();
 355                         active_ctx = REV_CTX;
 356                         reset_rev_ctx(atoi(val));
 357                         break;
 358                 case sizeof("Node-path"):
 359                         if (prefixcmp(t, "Node-"))
 360                                 continue;
 361                         if (!constcmp(t + strlen("Node-"), "path")) {
 362                                 if (active_ctx == NODE_CTX)
 363                                         handle_node();
 364                                 if (active_ctx == REV_CTX)
 365                                         begin_revision();
 366                                 active_ctx = NODE_CTX;
 367                                 reset_node_ctx(val);
 368                                 break;
 369                         }
 370                         if (constcmp(t + strlen("Node-"), "kind"))
 371                                 continue;
 372                         if (!strcmp(val, "dir"))
 373                                 node_ctx.type = REPO_MODE_DIR;
 374                         else if (!strcmp(val, "file"))
 375                                 node_ctx.type = REPO_MODE_BLB;
 376                         else
 377                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 378                         break;
 379                 case sizeof("Node-action"):
 380                         if (constcmp(t, "Node-action"))
 381                                 continue;
 382                         if (!strcmp(val, "delete")) {
 383                                 node_ctx.action = NODEACT_DELETE;
 384                         } else if (!strcmp(val, "add")) {
 385                                 node_ctx.action = NODEACT_ADD;
 386                         } else if (!strcmp(val, "change")) {
 387                                 node_ctx.action = NODEACT_CHANGE;
 388                         } else if (!strcmp(val, "replace")) {
 389                                 node_ctx.action = NODEACT_REPLACE;
 390                         } else {
 391                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 392                                 node_ctx.action = NODEACT_UNKNOWN;
 393                         }
 394                         break;
 395                 case sizeof("Node-copyfrom-path"):
 396                         if (constcmp(t, "Node-copyfrom-path"))
 397                                 continue;
 398                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 399                         break;
 400                 case sizeof("Node-copyfrom-rev"):
 401                         if (constcmp(t, "Node-copyfrom-rev"))
 402                                 continue;
 403                         node_ctx.srcRev = atoi(val);
 404                         break;
 405                 case sizeof("Text-content-length"):
 406                         if (!constcmp(t, "Text-content-length")) {
 407                                 node_ctx.textLength = atoi(val);
 408                                 break;
 409                         }
 410                         if (constcmp(t, "Prop-content-length"))
 411                                 continue;
 412                         node_ctx.propLength = atoi(val);
 413                         break;
 414                 case sizeof("Text-delta"):
 415                         if (!constcmp(t, "Text-delta")) {
 416                                 node_ctx.text_delta = !strcmp(val, "true");
 417                                 break;
 418                         }
 419                         if (constcmp(t, "Prop-delta"))
 420                                 continue;
 421                         node_ctx.prop_delta = !strcmp(val, "true");
 422                         break;
 423                 case sizeof("Content-length"):
 424                         if (constcmp(t, "Content-length"))
 425                                 continue;
 426                         len = atoi(val);
 427                         t = buffer_read_line(&input);
 428                         if (!t)
 429                                 die_short_read();
 430                         if (*t)
 431                                 die("invalid dump: expected blank line after content length header");
 432                         if (active_ctx == REV_CTX) {
 433                                 read_props();
 434                         } else if (active_ctx == NODE_CTX) {
 435                                 handle_node();
 436                                 active_ctx = INTERNODE_CTX;
 437                         } else {
 438                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 439                                 if (buffer_skip_bytes(&input, len) != len)
 440                                         die_short_read();
 441                         }
 442                 }
 443         }
 444         if (buffer_ferror(&input))
 445                 die_short_read();
 446         if (active_ctx == NODE_CTX)
 447                 handle_node();
 448         if (active_ctx == REV_CTX)
 449                 begin_revision();
 450         if (active_ctx != DUMP_CTX)
 451                 end_revision();
 452 }
 453
 454 int svndump_init(const char *filename)
 455 {
 456         if (buffer_init(&input, filename))
 457                 return error("cannot open %s: %s", filename, strerror(errno));
 458         fast_export_init(REPORT_FILENO);
 459         strbuf_init(&dump_ctx.uuid, 4096);
 460         strbuf_init(&dump_ctx.url, 4096);
 461         strbuf_init(&rev_ctx.log, 4096);
 462         strbuf_init(&rev_ctx.author, 4096);
 463         reset_dump_ctx(NULL);
 464         reset_rev_ctx(0);
 465         reset_node_ctx(NULL);
 466         return 0;
 467 }
 468
 469 void svndump_deinit(void)
 470 {
 471         fast_export_deinit();
 472         reset_dump_ctx(NULL);
 473         reset_rev_ctx(0);
 474         reset_node_ctx(NULL);
 475         strbuf_release(&rev_ctx.log);
 476         if (buffer_deinit(&input))
 477                 fprintf(stderr, "Input error\n");
 478         if (ferror(stdout))
 479                 fprintf(stderr, "Output error\n");
 480 }
 481
 482 void svndump_reset(void)
 483 {
 484         fast_export_reset();
 485         buffer_reset(&input);
 486         strbuf_release(&dump_ctx.uuid);
 487         strbuf_release(&dump_ctx.url);
 488         strbuf_release(&rev_ctx.log);
 489         strbuf_release(&rev_ctx.author);
 490 }