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 /*
  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 /* Create memory pool for log messages */
  37 obj_pool_gen(log, char, 4096)
  38
  39 static struct line_buffer input = LINE_BUFFER_INIT;
  40
  41 static char *log_copy(uint32_t length, const char *log)
  42 {
  43         char *buffer;
  44         log_free(log_pool.size);
  45         buffer = log_pointer(log_alloc(length));
  46         strncpy(buffer, log, length);
  47         return buffer;
  48 }
  49
  50 static struct {
  51         uint32_t action, propLength, textLength, srcRev, type;
  52         uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
  53         uint32_t text_delta, prop_delta;
  54 } node_ctx;
  55
  56 static struct {
  57         uint32_t revision, author;
  58         unsigned long timestamp;
  59         char *log;
  60 } rev_ctx;
  61
  62 static struct {
  63         uint32_t version, uuid, url;
  64 } dump_ctx;
  65
  66 static void reset_node_ctx(char *fname)
  67 {
  68         node_ctx.type = 0;
  69         node_ctx.action = NODEACT_UNKNOWN;
  70         node_ctx.propLength = LENGTH_UNKNOWN;
  71         node_ctx.textLength = LENGTH_UNKNOWN;
  72         node_ctx.src[0] = ~0;
  73         node_ctx.srcRev = 0;
  74         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  75         node_ctx.text_delta = 0;
  76         node_ctx.prop_delta = 0;
  77 }
  78
  79 static void reset_rev_ctx(uint32_t revision)
  80 {
  81         rev_ctx.revision = revision;
  82         rev_ctx.timestamp = 0;
  83         rev_ctx.log = NULL;
  84         rev_ctx.author = ~0;
  85 }
  86
  87 static void reset_dump_ctx(uint32_t url)
  88 {
  89         dump_ctx.url = url;
  90         dump_ctx.version = 1;
  91         dump_ctx.uuid = ~0;
  92 }
  93
  94 static void handle_property(const struct strbuf *key_buf,
  95                                 const char *val, uint32_t len,
  96                                 uint32_t *type_set)
  97 {
  98         const char *key = key_buf->buf;
  99         size_t keylen = key_buf->len;
 100
 101         switch (keylen + 1) {
 102         case sizeof("svn:log"):
 103                 if (constcmp(key, "svn:log"))
 104                         break;
 105                 if (!val)
 106                         die("invalid dump: unsets svn:log");
 107                 /* Value length excludes terminating nul. */
 108                 rev_ctx.log = log_copy(len + 1, val);
 109                 break;
 110         case sizeof("svn:author"):
 111                 if (constcmp(key, "svn:author"))
 112                         break;
 113                 rev_ctx.author = pool_intern(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, &rev_ctx.timestamp, NULL))
 121                         warning("invalid timestamp: %s", val);
 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         const char *t;
 158         /*
 159          * NEEDSWORK: to support simple mode changes like
 160          *      K 11
 161          *      svn:special
 162          *      V 1
 163          *      *
 164          *      D 14
 165          *      svn:executable
 166          * we keep track of whether a mode has been set and reset to
 167          * plain file only if not.  We should be keeping track of the
 168          * symlink and executable bits separately instead.
 169          */
 170         uint32_t type_set = 0;
 171         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 172                 uint32_t len;
 173                 const char *val;
 174                 const char type = t[0];
 175                 int ch;
 176
 177                 if (!type || t[1] != ' ')
 178                         die("invalid property line: %s\n", t);
 179                 len = atoi(&t[2]);
 180                 val = buffer_read_string(&input, len);
 181                 if (!val || strlen(val) != len)
 182                         die_short_read();
 183
 184                 /* Discard trailing newline. */
 185                 ch = buffer_read_char(&input);
 186                 if (ch == EOF)
 187                         die_short_read();
 188                 if (ch != '\n')
 189                         die("invalid dump: expected newline after %s", val);
 190
 191                 switch (type) {
 192                 case 'K':
 193                 case 'D':
 194                         strbuf_reset(&key);
 195                         if (val)
 196                                 strbuf_add(&key, val, len);
 197                         if (type == 'K')
 198                                 continue;
 199                         assert(type == 'D');
 200                         val = NULL;
 201                         len = 0;
 202                         /* fall through */
 203                 case 'V':
 204                         handle_property(&key, val, len, &type_set);
 205                         strbuf_reset(&key);
 206                         continue;
 207                 default:
 208                         die("invalid property line: %s\n", t);
 209                 }
 210         }
 211 }
 212
 213 static void handle_node(void)
 214 {
 215         uint32_t mark = 0;
 216         const uint32_t type = node_ctx.type;
 217         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 218         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 219
 220         if (node_ctx.text_delta)
 221                 die("text deltas not supported");
 222         if (have_text)
 223                 mark = next_blob_mark();
 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                 return repo_delete(node_ctx.dst);
 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          * Decide on the new content (mark) and mode (node_ctx.type).
 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         } else if (node_ctx.action == NODEACT_CHANGE) {
 249                 uint32_t mode;
 250                 if (!have_text)
 251                         mark = repo_read_path(node_ctx.dst);
 252                 mode = repo_read_mode(node_ctx.dst);
 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         } else if (node_ctx.action == NODEACT_ADD) {
 259                 if (!have_text && type != REPO_MODE_DIR)
 260                         die("invalid dump: adds node without text");
 261         } else {
 262                 die("invalid dump: Node-path block lacks Node-action");
 263         }
 264
 265         /*
 266          * Adjust mode to reflect properties.
 267          */
 268         if (have_props) {
 269                 if (!node_ctx.prop_delta)
 270                         node_ctx.type = type;
 271                 if (node_ctx.propLength)
 272                         read_props();
 273         }
 274
 275         /*
 276          * Save the result.
 277          */
 278         repo_add(node_ctx.dst, node_ctx.type, mark);
 279         if (have_text)
 280                 fast_export_blob(node_ctx.type, mark,
 281                                  node_ctx.textLength, &input);
 282 }
 283
 284 static void handle_revision(void)
 285 {
 286         if (rev_ctx.revision)
 287                 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 288                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 289 }
 290
 291 void svndump_read(const char *url)
 292 {
 293         char *val;
 294         char *t;
 295         uint32_t active_ctx = DUMP_CTX;
 296         uint32_t len;
 297
 298         reset_dump_ctx(pool_intern(url));
 299         while ((t = buffer_read_line(&input))) {
 300                 val = strchr(t, ':');
 301                 if (!val)
 302                         continue;
 303                 val++;
 304                 if (*val != ' ')
 305                         continue;
 306                 val++;
 307
 308                 /* strlen(key) + 1 */
 309                 switch (val - t - 1) {
 310                 case sizeof("SVN-fs-dump-format-version"):
 311                         if (constcmp(t, "SVN-fs-dump-format-version"))
 312                                 continue;
 313                         dump_ctx.version = atoi(val);
 314                         if (dump_ctx.version > 3)
 315                                 die("expected svn dump format version <= 3, found %"PRIu32,
 316                                     dump_ctx.version);
 317                         break;
 318                 case sizeof("UUID"):
 319                         if (constcmp(t, "UUID"))
 320                                 continue;
 321                         dump_ctx.uuid = pool_intern(val);
 322                         break;
 323                 case sizeof("Revision-number"):
 324                         if (constcmp(t, "Revision-number"))
 325                                 continue;
 326                         if (active_ctx == NODE_CTX)
 327                                 handle_node();
 328                         if (active_ctx != DUMP_CTX)
 329                                 handle_revision();
 330                         active_ctx = REV_CTX;
 331                         reset_rev_ctx(atoi(val));
 332                         break;
 333                 case sizeof("Node-path"):
 334                         if (prefixcmp(t, "Node-"))
 335                                 continue;
 336                         if (!constcmp(t + strlen("Node-"), "path")) {
 337                                 if (active_ctx == NODE_CTX)
 338                                         handle_node();
 339                                 active_ctx = NODE_CTX;
 340                                 reset_node_ctx(val);
 341                                 break;
 342                         }
 343                         if (constcmp(t + strlen("Node-"), "kind"))
 344                                 continue;
 345                         if (!strcmp(val, "dir"))
 346                                 node_ctx.type = REPO_MODE_DIR;
 347                         else if (!strcmp(val, "file"))
 348                                 node_ctx.type = REPO_MODE_BLB;
 349                         else
 350                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 351                         break;
 352                 case sizeof("Node-action"):
 353                         if (constcmp(t, "Node-action"))
 354                                 continue;
 355                         if (!strcmp(val, "delete")) {
 356                                 node_ctx.action = NODEACT_DELETE;
 357                         } else if (!strcmp(val, "add")) {
 358                                 node_ctx.action = NODEACT_ADD;
 359                         } else if (!strcmp(val, "change")) {
 360                                 node_ctx.action = NODEACT_CHANGE;
 361                         } else if (!strcmp(val, "replace")) {
 362                                 node_ctx.action = NODEACT_REPLACE;
 363                         } else {
 364                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 365                                 node_ctx.action = NODEACT_UNKNOWN;
 366                         }
 367                         break;
 368                 case sizeof("Node-copyfrom-path"):
 369                         if (constcmp(t, "Node-copyfrom-path"))
 370                                 continue;
 371                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 372                         break;
 373                 case sizeof("Node-copyfrom-rev"):
 374                         if (constcmp(t, "Node-copyfrom-rev"))
 375                                 continue;
 376                         node_ctx.srcRev = atoi(val);
 377                         break;
 378                 case sizeof("Text-content-length"):
 379                         if (!constcmp(t, "Text-content-length")) {
 380                                 node_ctx.textLength = atoi(val);
 381                                 break;
 382                         }
 383                         if (constcmp(t, "Prop-content-length"))
 384                                 continue;
 385                         node_ctx.propLength = atoi(val);
 386                         break;
 387                 case sizeof("Text-delta"):
 388                         if (!constcmp(t, "Text-delta")) {
 389                                 node_ctx.text_delta = !strcmp(val, "true");
 390                                 break;
 391                         }
 392                         if (constcmp(t, "Prop-delta"))
 393                                 continue;
 394                         node_ctx.prop_delta = !strcmp(val, "true");
 395                         break;
 396                 case sizeof("Content-length"):
 397                         if (constcmp(t, "Content-length"))
 398                                 continue;
 399                         len = atoi(val);
 400                         t = buffer_read_line(&input);
 401                         if (!t)
 402                                 die_short_read();
 403                         if (*t)
 404                                 die("invalid dump: expected blank line after content length header");
 405                         if (active_ctx == REV_CTX) {
 406                                 read_props();
 407                         } else if (active_ctx == NODE_CTX) {
 408                                 handle_node();
 409                                 active_ctx = REV_CTX;
 410                         } else {
 411                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 412                                 if (buffer_skip_bytes(&input, len) != len)
 413                                         die_short_read();
 414                         }
 415                 }
 416         }
 417         if (buffer_ferror(&input))
 418                 die_short_read();
 419         if (active_ctx == NODE_CTX)
 420                 handle_node();
 421         if (active_ctx != DUMP_CTX)
 422                 handle_revision();
 423 }
 424
 425 int svndump_init(const char *filename)
 426 {
 427         if (buffer_init(&input, filename))
 428                 return error("cannot open %s: %s", filename, strerror(errno));
 429         repo_init();
 430         reset_dump_ctx(~0);
 431         reset_rev_ctx(0);
 432         reset_node_ctx(NULL);
 433         return 0;
 434 }
 435
 436 void svndump_deinit(void)
 437 {
 438         log_reset();
 439         repo_reset();
 440         reset_dump_ctx(~0);
 441         reset_rev_ctx(0);
 442         reset_node_ctx(NULL);
 443         if (buffer_deinit(&input))
 444                 fprintf(stderr, "Input error\n");
 445         if (ferror(stdout))
 446                 fprintf(stderr, "Output error\n");
 447 }
 448
 449 void svndump_reset(void)
 450 {
 451         log_reset();
 452         buffer_reset(&input);
 453         repo_reset();
 454         reset_dump_ctx(~0);
 455         reset_rev_ctx(0);
 456         reset_node_ctx(NULL);
 457 }