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                                 struct strbuf *val,
  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_swap(&rev_ctx.log, val);
  99                 break;
 100         case sizeof("svn:author"):
 101                 if (constcmp(key, "svn:author"))
 102                         break;
 103                 if (!val)
 104                         strbuf_reset(&rev_ctx.author);
 105                 else
 106                         strbuf_swap(&rev_ctx.author, val);
 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->buf, &rev_ctx.timestamp, NULL))
 114                         warning("invalid timestamp: %s", val->buf);
 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, &type_set);
 191                         continue;
 192                 case 'V':
 193                         handle_property(&key, &val, &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                 repo_delete(node_ctx.dst);
 218                 return;
 219         }
 220         if (node_ctx.action == NODEACT_REPLACE) {
 221                 repo_delete(node_ctx.dst);
 222                 node_ctx.action = NODEACT_ADD;
 223         }
 224         if (node_ctx.srcRev) {
 225                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 226                 if (node_ctx.action == NODEACT_ADD)
 227                         node_ctx.action = NODEACT_CHANGE;
 228         }
 229         if (have_text && type == REPO_MODE_DIR)
 230                 die("invalid dump: directories cannot have text attached");
 231
 232         /*
 233          * Decide on the new content (mark) and mode (node_ctx.type).
 234          */
 235         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 236                 if (type != REPO_MODE_DIR)
 237                         die("invalid dump: root of tree is not a regular file");
 238         } else if (node_ctx.action == NODEACT_CHANGE) {
 239                 uint32_t mode;
 240                 if (!have_text)
 241                         mark = repo_read_path(node_ctx.dst);
 242                 mode = repo_read_mode(node_ctx.dst);
 243                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 244                         die("invalid dump: cannot modify a directory into a file");
 245                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 246                         die("invalid dump: cannot modify a file into a directory");
 247                 node_ctx.type = mode;
 248         } else if (node_ctx.action == NODEACT_ADD) {
 249                 if (!have_text && type != REPO_MODE_DIR)
 250                         die("invalid dump: adds node without text");
 251         } else {
 252                 die("invalid dump: Node-path block lacks Node-action");
 253         }
 254
 255         /*
 256          * Adjust mode to reflect properties.
 257          */
 258         if (have_props) {
 259                 if (!node_ctx.prop_delta)
 260                         node_ctx.type = type;
 261                 if (node_ctx.propLength)
 262                         read_props();
 263         }
 264
 265         /*
 266          * Save the result.
 267          */
 268         repo_add(node_ctx.dst, node_ctx.type, mark);
 269         if (have_text)
 270                 fast_export_blob(node_ctx.type, mark,
 271                                  node_ctx.textLength, &input);
 272 }
 273
 274 static void handle_revision(void)
 275 {
 276         if (rev_ctx.revision)
 277                 repo_commit(rev_ctx.revision, rev_ctx.author.buf,
 278                         &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
 279                         rev_ctx.timestamp);
 280 }
 281
 282 void svndump_read(const char *url)
 283 {
 284         char *val;
 285         char *t;
 286         uint32_t active_ctx = DUMP_CTX;
 287         uint32_t len;
 288
 289         reset_dump_ctx(url);
 290         while ((t = buffer_read_line(&input))) {
 291                 val = strchr(t, ':');
 292                 if (!val)
 293                         continue;
 294                 val++;
 295                 if (*val != ' ')
 296                         continue;
 297                 val++;
 298
 299                 /* strlen(key) + 1 */
 300                 switch (val - t - 1) {
 301                 case sizeof("SVN-fs-dump-format-version"):
 302                         if (constcmp(t, "SVN-fs-dump-format-version"))
 303                                 continue;
 304                         dump_ctx.version = atoi(val);
 305                         if (dump_ctx.version > 3)
 306                                 die("expected svn dump format version <= 3, found %"PRIu32,
 307                                     dump_ctx.version);
 308                         break;
 309                 case sizeof("UUID"):
 310                         if (constcmp(t, "UUID"))
 311                                 continue;
 312                         strbuf_reset(&dump_ctx.uuid);
 313                         strbuf_addstr(&dump_ctx.uuid, val);
 314                         break;
 315                 case sizeof("Revision-number"):
 316                         if (constcmp(t, "Revision-number"))
 317                                 continue;
 318                         if (active_ctx == NODE_CTX)
 319                                 handle_node();
 320                         if (active_ctx != DUMP_CTX)
 321                                 handle_revision();
 322                         active_ctx = REV_CTX;
 323                         reset_rev_ctx(atoi(val));
 324                         break;
 325                 case sizeof("Node-path"):
 326                         if (prefixcmp(t, "Node-"))
 327                                 continue;
 328                         if (!constcmp(t + strlen("Node-"), "path")) {
 329                                 if (active_ctx == NODE_CTX)
 330                                         handle_node();
 331                                 active_ctx = NODE_CTX;
 332                                 reset_node_ctx(val);
 333                                 break;
 334                         }
 335                         if (constcmp(t + strlen("Node-"), "kind"))
 336                                 continue;
 337                         if (!strcmp(val, "dir"))
 338                                 node_ctx.type = REPO_MODE_DIR;
 339                         else if (!strcmp(val, "file"))
 340                                 node_ctx.type = REPO_MODE_BLB;
 341                         else
 342                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 343                         break;
 344                 case sizeof("Node-action"):
 345                         if (constcmp(t, "Node-action"))
 346                                 continue;
 347                         if (!strcmp(val, "delete")) {
 348                                 node_ctx.action = NODEACT_DELETE;
 349                         } else if (!strcmp(val, "add")) {
 350                                 node_ctx.action = NODEACT_ADD;
 351                         } else if (!strcmp(val, "change")) {
 352                                 node_ctx.action = NODEACT_CHANGE;
 353                         } else if (!strcmp(val, "replace")) {
 354                                 node_ctx.action = NODEACT_REPLACE;
 355                         } else {
 356                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 357                                 node_ctx.action = NODEACT_UNKNOWN;
 358                         }
 359                         break;
 360                 case sizeof("Node-copyfrom-path"):
 361                         if (constcmp(t, "Node-copyfrom-path"))
 362                                 continue;
 363                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 364                         break;
 365                 case sizeof("Node-copyfrom-rev"):
 366                         if (constcmp(t, "Node-copyfrom-rev"))
 367                                 continue;
 368                         node_ctx.srcRev = atoi(val);
 369                         break;
 370                 case sizeof("Text-content-length"):
 371                         if (!constcmp(t, "Text-content-length")) {
 372                                 node_ctx.textLength = atoi(val);
 373                                 break;
 374                         }
 375                         if (constcmp(t, "Prop-content-length"))
 376                                 continue;
 377                         node_ctx.propLength = atoi(val);
 378                         break;
 379                 case sizeof("Text-delta"):
 380                         if (!constcmp(t, "Text-delta")) {
 381                                 node_ctx.text_delta = !strcmp(val, "true");
 382                                 break;
 383                         }
 384                         if (constcmp(t, "Prop-delta"))
 385                                 continue;
 386                         node_ctx.prop_delta = !strcmp(val, "true");
 387                         break;
 388                 case sizeof("Content-length"):
 389                         if (constcmp(t, "Content-length"))
 390                                 continue;
 391                         len = atoi(val);
 392                         t = buffer_read_line(&input);
 393                         if (!t)
 394                                 die_short_read();
 395                         if (*t)
 396                                 die("invalid dump: expected blank line after content length header");
 397                         if (active_ctx == REV_CTX) {
 398                                 read_props();
 399                         } else if (active_ctx == NODE_CTX) {
 400                                 handle_node();
 401                                 active_ctx = REV_CTX;
 402                         } else {
 403                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 404                                 if (buffer_skip_bytes(&input, len) != len)
 405                                         die_short_read();
 406                         }
 407                 }
 408         }
 409         if (buffer_ferror(&input))
 410                 die_short_read();
 411         if (active_ctx == NODE_CTX)
 412                 handle_node();
 413         if (active_ctx != DUMP_CTX)
 414                 handle_revision();
 415 }
 416
 417 int svndump_init(const char *filename)
 418 {
 419         if (buffer_init(&input, filename))
 420                 return error("cannot open %s: %s", filename, strerror(errno));
 421         repo_init();
 422         strbuf_init(&dump_ctx.uuid, 4096);
 423         strbuf_init(&dump_ctx.url, 4096);
 424         strbuf_init(&rev_ctx.log, 4096);
 425         strbuf_init(&rev_ctx.author, 4096);
 426         reset_dump_ctx(NULL);
 427         reset_rev_ctx(0);
 428         reset_node_ctx(NULL);
 429         return 0;
 430 }
 431
 432 void svndump_deinit(void)
 433 {
 434         repo_reset();
 435         reset_dump_ctx(NULL);
 436         reset_rev_ctx(0);
 437         reset_node_ctx(NULL);
 438         strbuf_release(&rev_ctx.log);
 439         if (buffer_deinit(&input))
 440                 fprintf(stderr, "Input error\n");
 441         if (ferror(stdout))
 442                 fprintf(stderr, "Output error\n");
 443 }
 444
 445 void svndump_reset(void)
 446 {
 447         buffer_reset(&input);
 448         repo_reset();
 449         strbuf_release(&dump_ctx.uuid);
 450         strbuf_release(&dump_ctx.url);
 451         strbuf_release(&rev_ctx.log);
 452         strbuf_release(&rev_ctx.author);
 453 }