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 "obj_pool.h"
  15 #include "string_pool.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 /* Create memory pool for log messages */
  41 obj_pool_gen(log, char, 4096)
  42
  43 static struct line_buffer input = LINE_BUFFER_INIT;
  44
  45 #define REPORT_FILENO 3
  46
  47 static char *log_copy(uint32_t length, const char *log)
  48 {
  49         char *buffer;
  50         log_free(log_pool.size);
  51         buffer = log_pointer(log_alloc(length));
  52         strncpy(buffer, log, length);
  53         return buffer;
  54 }
  55
  56 static struct {
  57         uint32_t action, propLength, textLength, srcRev, type;
  58         uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
  59         uint32_t text_delta, prop_delta;
  60 } node_ctx;
  61
  62 static struct {
  63         uint32_t revision, author;
  64         unsigned long timestamp;
  65         char *log;
  66 } rev_ctx;
  67
  68 static struct {
  69         uint32_t version, uuid, url;
  70 } dump_ctx;
  71
  72 static void reset_node_ctx(char *fname)
  73 {
  74         node_ctx.type = 0;
  75         node_ctx.action = NODEACT_UNKNOWN;
  76         node_ctx.propLength = LENGTH_UNKNOWN;
  77         node_ctx.textLength = LENGTH_UNKNOWN;
  78         node_ctx.src[0] = ~0;
  79         node_ctx.srcRev = 0;
  80         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  81         node_ctx.text_delta = 0;
  82         node_ctx.prop_delta = 0;
  83 }
  84
  85 static void reset_rev_ctx(uint32_t revision)
  86 {
  87         rev_ctx.revision = revision;
  88         rev_ctx.timestamp = 0;
  89         rev_ctx.log = NULL;
  90         rev_ctx.author = ~0;
  91 }
  92
  93 static void reset_dump_ctx(uint32_t url)
  94 {
  95         dump_ctx.url = url;
  96         dump_ctx.version = 1;
  97         dump_ctx.uuid = ~0;
  98 }
  99
 100 static void handle_property(const struct strbuf *key_buf,
 101                                 const char *val, uint32_t len,
 102                                 uint32_t *type_set)
 103 {
 104         const char *key = key_buf->buf;
 105         size_t keylen = key_buf->len;
 106
 107         switch (keylen + 1) {
 108         case sizeof("svn:log"):
 109                 if (constcmp(key, "svn:log"))
 110                         break;
 111                 if (!val)
 112                         die("invalid dump: unsets svn:log");
 113                 /* Value length excludes terminating nul. */
 114                 rev_ctx.log = log_copy(len + 1, val);
 115                 break;
 116         case sizeof("svn:author"):
 117                 if (constcmp(key, "svn:author"))
 118                         break;
 119                 rev_ctx.author = pool_intern(val);
 120                 break;
 121         case sizeof("svn:date"):
 122                 if (constcmp(key, "svn:date"))
 123                         break;
 124                 if (!val)
 125                         die("invalid dump: unsets svn:date");
 126                 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
 127                         warning("invalid timestamp: %s", val);
 128                 break;
 129         case sizeof("svn:executable"):
 130         case sizeof("svn:special"):
 131                 if (keylen == strlen("svn:executable") &&
 132                     constcmp(key, "svn:executable"))
 133                         break;
 134                 if (keylen == strlen("svn:special") &&
 135                     constcmp(key, "svn:special"))
 136                         break;
 137                 if (*type_set) {
 138                         if (!val)
 139                                 return;
 140                         die("invalid dump: sets type twice");
 141                 }
 142                 if (!val) {
 143                         node_ctx.type = REPO_MODE_BLB;
 144                         return;
 145                 }
 146                 *type_set = 1;
 147                 node_ctx.type = keylen == strlen("svn:executable") ?
 148                                 REPO_MODE_EXE :
 149                                 REPO_MODE_LNK;
 150         }
 151 }
 152
 153 static void die_short_read(void)
 154 {
 155         if (buffer_ferror(&input))
 156                 die_errno("error reading dump file");
 157         die("invalid dump: unexpected end of file");
 158 }
 159
 160 static void read_props(void)
 161 {
 162         static struct strbuf key = STRBUF_INIT;
 163         const char *t;
 164         /*
 165          * NEEDSWORK: to support simple mode changes like
 166          *      K 11
 167          *      svn:special
 168          *      V 1
 169          *      *
 170          *      D 14
 171          *      svn:executable
 172          * we keep track of whether a mode has been set and reset to
 173          * plain file only if not.  We should be keeping track of the
 174          * symlink and executable bits separately instead.
 175          */
 176         uint32_t type_set = 0;
 177         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 178                 uint32_t len;
 179                 const char *val;
 180                 const char type = t[0];
 181                 int ch;
 182
 183                 if (!type || t[1] != ' ')
 184                         die("invalid property line: %s\n", t);
 185                 len = atoi(&t[2]);
 186                 val = buffer_read_string(&input, len);
 187                 if (!val || strlen(val) != len)
 188                         die_short_read();
 189
 190                 /* Discard trailing newline. */
 191                 ch = buffer_read_char(&input);
 192                 if (ch == EOF)
 193                         die_short_read();
 194                 if (ch != '\n')
 195                         die("invalid dump: expected newline after %s", val);
 196
 197                 switch (type) {
 198                 case 'K':
 199                 case 'D':
 200                         strbuf_reset(&key);
 201                         if (val)
 202                                 strbuf_add(&key, val, len);
 203                         if (type == 'K')
 204                                 continue;
 205                         assert(type == 'D');
 206                         val = NULL;
 207                         len = 0;
 208                         /* fall through */
 209                 case 'V':
 210                         handle_property(&key, val, len, &type_set);
 211                         strbuf_reset(&key);
 212                         continue;
 213                 default:
 214                         die("invalid property line: %s\n", t);
 215                 }
 216         }
 217 }
 218
 219 static void handle_node(void)
 220 {
 221         const uint32_t type = node_ctx.type;
 222         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 223         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 224         /*
 225          * Old text for this node:
 226          *  NULL        - directory or bug
 227          *  empty_blob  - empty
 228          *  "<dataref>" - data retrievable from fast-import
 229          */
 230         static const char *const empty_blob = "::empty::";
 231         const char *old_data = NULL;
 232
 233         if (node_ctx.text_delta)
 234                 die("text deltas not supported");
 235
 236         if (node_ctx.action == NODEACT_DELETE) {
 237                 if (have_text || have_props || node_ctx.srcRev)
 238                         die("invalid dump: deletion node has "
 239                                 "copyfrom info, text, or properties");
 240                 return repo_delete(node_ctx.dst);
 241         }
 242         if (node_ctx.action == NODEACT_REPLACE) {
 243                 repo_delete(node_ctx.dst);
 244                 node_ctx.action = NODEACT_ADD;
 245         }
 246         if (node_ctx.srcRev) {
 247                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 248                 if (node_ctx.action == NODEACT_ADD)
 249                         node_ctx.action = NODEACT_CHANGE;
 250         }
 251         if (have_text && type == REPO_MODE_DIR)
 252                 die("invalid dump: directories cannot have text attached");
 253
 254         /*
 255          * Find old content (old_data) and decide on the new mode.
 256          */
 257         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 258                 if (type != REPO_MODE_DIR)
 259                         die("invalid dump: root of tree is not a regular file");
 260                 old_data = NULL;
 261         } else if (node_ctx.action == NODEACT_CHANGE) {
 262                 uint32_t mode;
 263                 old_data = repo_read_path(node_ctx.dst);
 264                 mode = repo_read_mode(node_ctx.dst);
 265                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 266                         die("invalid dump: cannot modify a directory into a file");
 267                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 268                         die("invalid dump: cannot modify a file into a directory");
 269                 node_ctx.type = mode;
 270         } else if (node_ctx.action == NODEACT_ADD) {
 271                 if (type == REPO_MODE_DIR)
 272                         old_data = NULL;
 273                 else if (have_text)
 274                         old_data = empty_blob;
 275                 else
 276                         die("invalid dump: adds node without text");
 277         } else {
 278                 die("invalid dump: Node-path block lacks Node-action");
 279         }
 280
 281         /*
 282          * Adjust mode to reflect properties.
 283          */
 284         if (have_props) {
 285                 if (!node_ctx.prop_delta)
 286                         node_ctx.type = type;
 287                 if (node_ctx.propLength)
 288                         read_props();
 289         }
 290
 291         /*
 292          * Save the result.
 293          */
 294         if (type == REPO_MODE_DIR)      /* directories are not tracked. */
 295                 return;
 296         assert(old_data);
 297         if (old_data == empty_blob)
 298                 /* For the fast_export_* functions, NULL means empty. */
 299                 old_data = NULL;
 300         if (!have_text) {
 301                 fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
 302                                         node_ctx.type, old_data);
 303                 return;
 304         }
 305         fast_export_modify(REPO_MAX_PATH_DEPTH, node_ctx.dst,
 306                                 node_ctx.type, "inline");
 307         fast_export_data(node_ctx.type, node_ctx.textLength, &input);
 308 }
 309
 310 static void begin_revision(void)
 311 {
 312         if (!rev_ctx.revision)  /* revision 0 gets no git commit. */
 313                 return;
 314         fast_export_begin_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 315                 dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 316 }
 317
 318 static void end_revision(void)
 319 {
 320         if (rev_ctx.revision)
 321                 fast_export_end_commit(rev_ctx.revision);
 322 }
 323
 324 void svndump_read(const char *url)
 325 {
 326         char *val;
 327         char *t;
 328         uint32_t active_ctx = DUMP_CTX;
 329         uint32_t len;
 330
 331         reset_dump_ctx(pool_intern(url));
 332         while ((t = buffer_read_line(&input))) {
 333                 val = strstr(t, ": ");
 334                 if (!val)
 335                         continue;
 336                 val += 2;
 337
 338                 /* strlen(key) + 1 */
 339                 switch (val - t - 1) {
 340                 case sizeof("SVN-fs-dump-format-version"):
 341                         if (constcmp(t, "SVN-fs-dump-format-version"))
 342                                 continue;
 343                         dump_ctx.version = atoi(val);
 344                         if (dump_ctx.version > 3)
 345                                 die("expected svn dump format version <= 3, found %"PRIu32,
 346                                     dump_ctx.version);
 347                         break;
 348                 case sizeof("UUID"):
 349                         if (constcmp(t, "UUID"))
 350                                 continue;
 351                         dump_ctx.uuid = pool_intern(val);
 352                         break;
 353                 case sizeof("Revision-number"):
 354                         if (constcmp(t, "Revision-number"))
 355                                 continue;
 356                         if (active_ctx == NODE_CTX)
 357                                 handle_node();
 358                         if (active_ctx == REV_CTX)
 359                                 begin_revision();
 360                         if (active_ctx != DUMP_CTX)
 361                                 end_revision();
 362                         active_ctx = REV_CTX;
 363                         reset_rev_ctx(atoi(val));
 364                         break;
 365                 case sizeof("Node-path"):
 366                         if (prefixcmp(t, "Node-"))
 367                                 continue;
 368                         if (!constcmp(t + strlen("Node-"), "path")) {
 369                                 if (active_ctx == NODE_CTX)
 370                                         handle_node();
 371                                 if (active_ctx == REV_CTX)
 372                                         begin_revision();
 373                                 active_ctx = NODE_CTX;
 374                                 reset_node_ctx(val);
 375                                 break;
 376                         }
 377                         if (constcmp(t + strlen("Node-"), "kind"))
 378                                 continue;
 379                         if (!strcmp(val, "dir"))
 380                                 node_ctx.type = REPO_MODE_DIR;
 381                         else if (!strcmp(val, "file"))
 382                                 node_ctx.type = REPO_MODE_BLB;
 383                         else
 384                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 385                         break;
 386                 case sizeof("Node-action"):
 387                         if (constcmp(t, "Node-action"))
 388                                 continue;
 389                         if (!strcmp(val, "delete")) {
 390                                 node_ctx.action = NODEACT_DELETE;
 391                         } else if (!strcmp(val, "add")) {
 392                                 node_ctx.action = NODEACT_ADD;
 393                         } else if (!strcmp(val, "change")) {
 394                                 node_ctx.action = NODEACT_CHANGE;
 395                         } else if (!strcmp(val, "replace")) {
 396                                 node_ctx.action = NODEACT_REPLACE;
 397                         } else {
 398                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 399                                 node_ctx.action = NODEACT_UNKNOWN;
 400                         }
 401                         break;
 402                 case sizeof("Node-copyfrom-path"):
 403                         if (constcmp(t, "Node-copyfrom-path"))
 404                                 continue;
 405                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 406                         break;
 407                 case sizeof("Node-copyfrom-rev"):
 408                         if (constcmp(t, "Node-copyfrom-rev"))
 409                                 continue;
 410                         node_ctx.srcRev = atoi(val);
 411                         break;
 412                 case sizeof("Text-content-length"):
 413                         if (!constcmp(t, "Text-content-length")) {
 414                                 node_ctx.textLength = atoi(val);
 415                                 break;
 416                         }
 417                         if (constcmp(t, "Prop-content-length"))
 418                                 continue;
 419                         node_ctx.propLength = atoi(val);
 420                         break;
 421                 case sizeof("Text-delta"):
 422                         if (!constcmp(t, "Text-delta")) {
 423                                 node_ctx.text_delta = !strcmp(val, "true");
 424                                 break;
 425                         }
 426                         if (constcmp(t, "Prop-delta"))
 427                                 continue;
 428                         node_ctx.prop_delta = !strcmp(val, "true");
 429                         break;
 430                 case sizeof("Content-length"):
 431                         if (constcmp(t, "Content-length"))
 432                                 continue;
 433                         len = atoi(val);
 434                         t = buffer_read_line(&input);
 435                         if (!t)
 436                                 die_short_read();
 437                         if (*t)
 438                                 die("invalid dump: expected blank line after content length header");
 439                         if (active_ctx == REV_CTX) {
 440                                 read_props();
 441                         } else if (active_ctx == NODE_CTX) {
 442                                 handle_node();
 443                                 active_ctx = INTERNODE_CTX;
 444                         } else {
 445                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 446                                 if (buffer_skip_bytes(&input, len) != len)
 447                                         die_short_read();
 448                         }
 449                 }
 450         }
 451         if (buffer_ferror(&input))
 452                 die_short_read();
 453         if (active_ctx == NODE_CTX)
 454                 handle_node();
 455         if (active_ctx == REV_CTX)
 456                 begin_revision();
 457         if (active_ctx != DUMP_CTX)
 458                 end_revision();
 459 }
 460
 461 int svndump_init(const char *filename)
 462 {
 463         if (buffer_init(&input, filename))
 464                 return error("cannot open %s: %s", filename, strerror(errno));
 465         fast_export_init(REPORT_FILENO);
 466         reset_dump_ctx(~0);
 467         reset_rev_ctx(0);
 468         reset_node_ctx(NULL);
 469         return 0;
 470 }
 471
 472 void svndump_deinit(void)
 473 {
 474         log_reset();
 475         fast_export_deinit();
 476         reset_dump_ctx(~0);
 477         reset_rev_ctx(0);
 478         reset_node_ctx(NULL);
 479         if (buffer_deinit(&input))
 480                 fprintf(stderr, "Input error\n");
 481         if (ferror(stdout))
 482                 fprintf(stderr, "Output error\n");
 483 }
 484
 485 void svndump_reset(void)
 486 {
 487         log_reset();
 488         fast_export_reset();
 489         buffer_reset(&input);
 490         reset_dump_ctx(~0);
 491         reset_rev_ctx(0);
 492         reset_node_ctx(NULL);
 493 }