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 struct {
  67         uint32_t uuid, revision_number, node_path, node_kind, node_action,
  68                 node_copyfrom_path, node_copyfrom_rev, text_content_length,
  69                 prop_content_length, content_length, svn_fs_dump_format_version,
  70                 /* version 3 format */
  71                 text_delta, prop_delta;
  72 } keys;
  73
  74 static void reset_node_ctx(char *fname)
  75 {
  76         node_ctx.type = 0;
  77         node_ctx.action = NODEACT_UNKNOWN;
  78         node_ctx.propLength = LENGTH_UNKNOWN;
  79         node_ctx.textLength = LENGTH_UNKNOWN;
  80         node_ctx.src[0] = ~0;
  81         node_ctx.srcRev = 0;
  82         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  83         node_ctx.text_delta = 0;
  84         node_ctx.prop_delta = 0;
  85 }
  86
  87 static void reset_rev_ctx(uint32_t revision)
  88 {
  89         rev_ctx.revision = revision;
  90         rev_ctx.timestamp = 0;
  91         rev_ctx.log = NULL;
  92         rev_ctx.author = ~0;
  93 }
  94
  95 static void reset_dump_ctx(uint32_t url)
  96 {
  97         dump_ctx.url = url;
  98         dump_ctx.version = 1;
  99         dump_ctx.uuid = ~0;
 100 }
 101
 102 static void init_keys(void)
 103 {
 104         keys.uuid = pool_intern("UUID");
 105         keys.revision_number = pool_intern("Revision-number");
 106         keys.node_path = pool_intern("Node-path");
 107         keys.node_kind = pool_intern("Node-kind");
 108         keys.node_action = pool_intern("Node-action");
 109         keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
 110         keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
 111         keys.text_content_length = pool_intern("Text-content-length");
 112         keys.prop_content_length = pool_intern("Prop-content-length");
 113         keys.content_length = pool_intern("Content-length");
 114         keys.svn_fs_dump_format_version = pool_intern("SVN-fs-dump-format-version");
 115         /* version 3 format (Subversion 1.1.0) */
 116         keys.text_delta = pool_intern("Text-delta");
 117         keys.prop_delta = pool_intern("Prop-delta");
 118 }
 119
 120 static void handle_property(const struct strbuf *key_buf,
 121                                 const char *val, uint32_t len,
 122                                 uint32_t *type_set)
 123 {
 124         const char *key = key_buf->buf;
 125         size_t keylen = key_buf->len;
 126
 127         switch (keylen + 1) {
 128         case sizeof("svn:log"):
 129                 if (constcmp(key, "svn:log"))
 130                         break;
 131                 if (!val)
 132                         die("invalid dump: unsets svn:log");
 133                 /* Value length excludes terminating nul. */
 134                 rev_ctx.log = log_copy(len + 1, val);
 135                 break;
 136         case sizeof("svn:author"):
 137                 if (constcmp(key, "svn:author"))
 138                         break;
 139                 rev_ctx.author = pool_intern(val);
 140                 break;
 141         case sizeof("svn:date"):
 142                 if (constcmp(key, "svn:date"))
 143                         break;
 144                 if (!val)
 145                         die("invalid dump: unsets svn:date");
 146                 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
 147                         warning("invalid timestamp: %s", val);
 148                 break;
 149         case sizeof("svn:executable"):
 150         case sizeof("svn:special"):
 151                 if (keylen == strlen("svn:executable") &&
 152                     constcmp(key, "svn:executable"))
 153                         break;
 154                 if (keylen == strlen("svn:special") &&
 155                     constcmp(key, "svn:special"))
 156                         break;
 157                 if (*type_set) {
 158                         if (!val)
 159                                 return;
 160                         die("invalid dump: sets type twice");
 161                 }
 162                 if (!val) {
 163                         node_ctx.type = REPO_MODE_BLB;
 164                         return;
 165                 }
 166                 *type_set = 1;
 167                 node_ctx.type = keylen == strlen("svn:executable") ?
 168                                 REPO_MODE_EXE :
 169                                 REPO_MODE_LNK;
 170         }
 171 }
 172
 173 static void die_short_read(void)
 174 {
 175         if (buffer_ferror(&input))
 176                 die_errno("error reading dump file");
 177         die("invalid dump: unexpected end of file");
 178 }
 179
 180 static void read_props(void)
 181 {
 182         static struct strbuf key = STRBUF_INIT;
 183         const char *t;
 184         /*
 185          * NEEDSWORK: to support simple mode changes like
 186          *      K 11
 187          *      svn:special
 188          *      V 1
 189          *      *
 190          *      D 14
 191          *      svn:executable
 192          * we keep track of whether a mode has been set and reset to
 193          * plain file only if not.  We should be keeping track of the
 194          * symlink and executable bits separately instead.
 195          */
 196         uint32_t type_set = 0;
 197         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 198                 uint32_t len;
 199                 const char *val;
 200                 const char type = t[0];
 201                 int ch;
 202
 203                 if (!type || t[1] != ' ')
 204                         die("invalid property line: %s\n", t);
 205                 len = atoi(&t[2]);
 206                 val = buffer_read_string(&input, len);
 207                 if (!val || strlen(val) != len)
 208                         die_short_read();
 209
 210                 /* Discard trailing newline. */
 211                 ch = buffer_read_char(&input);
 212                 if (ch == EOF)
 213                         die_short_read();
 214                 if (ch != '\n')
 215                         die("invalid dump: expected newline after %s", val);
 216
 217                 switch (type) {
 218                 case 'K':
 219                 case 'D':
 220                         strbuf_reset(&key);
 221                         if (val)
 222                                 strbuf_add(&key, val, len);
 223                         if (type == 'K')
 224                                 continue;
 225                         assert(type == 'D');
 226                         val = NULL;
 227                         len = 0;
 228                         /* fall through */
 229                 case 'V':
 230                         handle_property(&key, val, len, &type_set);
 231                         strbuf_reset(&key);
 232                         continue;
 233                 default:
 234                         die("invalid property line: %s\n", t);
 235                 }
 236         }
 237 }
 238
 239 static void handle_node(void)
 240 {
 241         uint32_t mark = 0;
 242         const uint32_t type = node_ctx.type;
 243         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 244         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 245
 246         if (node_ctx.text_delta)
 247                 die("text deltas not supported");
 248         if (have_text)
 249                 mark = next_blob_mark();
 250         if (node_ctx.action == NODEACT_DELETE) {
 251                 if (have_text || have_props || node_ctx.srcRev)
 252                         die("invalid dump: deletion node has "
 253                                 "copyfrom info, text, or properties");
 254                 return repo_delete(node_ctx.dst);
 255         }
 256         if (node_ctx.action == NODEACT_REPLACE) {
 257                 repo_delete(node_ctx.dst);
 258                 node_ctx.action = NODEACT_ADD;
 259         }
 260         if (node_ctx.srcRev) {
 261                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 262                 if (node_ctx.action == NODEACT_ADD)
 263                         node_ctx.action = NODEACT_CHANGE;
 264         }
 265         if (have_text && type == REPO_MODE_DIR)
 266                 die("invalid dump: directories cannot have text attached");
 267
 268         /*
 269          * Decide on the new content (mark) and mode (node_ctx.type).
 270          */
 271         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 272                 if (type != REPO_MODE_DIR)
 273                         die("invalid dump: root of tree is not a regular file");
 274         } else if (node_ctx.action == NODEACT_CHANGE) {
 275                 uint32_t mode;
 276                 if (!have_text)
 277                         mark = repo_read_path(node_ctx.dst);
 278                 mode = repo_read_mode(node_ctx.dst);
 279                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 280                         die("invalid dump: cannot modify a directory into a file");
 281                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 282                         die("invalid dump: cannot modify a file into a directory");
 283                 node_ctx.type = mode;
 284         } else if (node_ctx.action == NODEACT_ADD) {
 285                 if (!have_text && type != REPO_MODE_DIR)
 286                         die("invalid dump: adds node without text");
 287         } else {
 288                 die("invalid dump: Node-path block lacks Node-action");
 289         }
 290
 291         /*
 292          * Adjust mode to reflect properties.
 293          */
 294         if (have_props) {
 295                 if (!node_ctx.prop_delta)
 296                         node_ctx.type = type;
 297                 if (node_ctx.propLength)
 298                         read_props();
 299         }
 300
 301         /*
 302          * Save the result.
 303          */
 304         repo_add(node_ctx.dst, node_ctx.type, mark);
 305         if (have_text)
 306                 fast_export_blob(node_ctx.type, mark,
 307                                  node_ctx.textLength, &input);
 308 }
 309
 310 static void handle_revision(void)
 311 {
 312         if (rev_ctx.revision)
 313                 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 314                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 315 }
 316
 317 void svndump_read(const char *url)
 318 {
 319         char *val;
 320         char *t;
 321         uint32_t active_ctx = DUMP_CTX;
 322         uint32_t len;
 323         uint32_t key;
 324
 325         reset_dump_ctx(pool_intern(url));
 326         while ((t = buffer_read_line(&input))) {
 327                 val = strstr(t, ": ");
 328                 if (!val)
 329                         continue;
 330                 *val++ = '\0';
 331                 *val++ = '\0';
 332                 key = pool_intern(t);
 333
 334                 if (key == keys.svn_fs_dump_format_version) {
 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                 } else if (key == keys.uuid) {
 340                         dump_ctx.uuid = pool_intern(val);
 341                 } else if (key == keys.revision_number) {
 342                         if (active_ctx == NODE_CTX)
 343                                 handle_node();
 344                         if (active_ctx != DUMP_CTX)
 345                                 handle_revision();
 346                         active_ctx = REV_CTX;
 347                         reset_rev_ctx(atoi(val));
 348                 } else if (key == keys.node_path) {
 349                         if (active_ctx == NODE_CTX)
 350                                 handle_node();
 351                         active_ctx = NODE_CTX;
 352                         reset_node_ctx(val);
 353                 } else if (key == keys.node_kind) {
 354                         if (!strcmp(val, "dir"))
 355                                 node_ctx.type = REPO_MODE_DIR;
 356                         else if (!strcmp(val, "file"))
 357                                 node_ctx.type = REPO_MODE_BLB;
 358                         else
 359                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 360                 } else if (key == keys.node_action) {
 361                         if (!strcmp(val, "delete")) {
 362                                 node_ctx.action = NODEACT_DELETE;
 363                         } else if (!strcmp(val, "add")) {
 364                                 node_ctx.action = NODEACT_ADD;
 365                         } else if (!strcmp(val, "change")) {
 366                                 node_ctx.action = NODEACT_CHANGE;
 367                         } else if (!strcmp(val, "replace")) {
 368                                 node_ctx.action = NODEACT_REPLACE;
 369                         } else {
 370                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 371                                 node_ctx.action = NODEACT_UNKNOWN;
 372                         }
 373                 } else if (key == keys.node_copyfrom_path) {
 374                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 375                 } else if (key == keys.node_copyfrom_rev) {
 376                         node_ctx.srcRev = atoi(val);
 377                 } else if (key == keys.text_content_length) {
 378                         node_ctx.textLength = atoi(val);
 379                 } else if (key == keys.prop_content_length) {
 380                         node_ctx.propLength = atoi(val);
 381                 } else if (key == keys.text_delta) {
 382                         node_ctx.text_delta = !strcmp(val, "true");
 383                 } else if (key == keys.prop_delta) {
 384                         node_ctx.prop_delta = !strcmp(val, "true");
 385                 } else if (key == keys.content_length) {
 386                         len = atoi(val);
 387                         t = buffer_read_line(&input);
 388                         if (!t)
 389                                 die_short_read();
 390                         if (*t)
 391                                 die("invalid dump: expected blank line after content length header");
 392                         if (active_ctx == REV_CTX) {
 393                                 read_props();
 394                         } else if (active_ctx == NODE_CTX) {
 395                                 handle_node();
 396                                 active_ctx = REV_CTX;
 397                         } else {
 398                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 399                                 if (buffer_skip_bytes(&input, len) != len)
 400                                         die_short_read();
 401                         }
 402                 }
 403         }
 404         if (buffer_ferror(&input))
 405                 die_short_read();
 406         if (active_ctx == NODE_CTX)
 407                 handle_node();
 408         if (active_ctx != DUMP_CTX)
 409                 handle_revision();
 410 }
 411
 412 int svndump_init(const char *filename)
 413 {
 414         if (buffer_init(&input, filename))
 415                 return error("cannot open %s: %s", filename, strerror(errno));
 416         repo_init();
 417         reset_dump_ctx(~0);
 418         reset_rev_ctx(0);
 419         reset_node_ctx(NULL);
 420         init_keys();
 421         return 0;
 422 }
 423
 424 void svndump_deinit(void)
 425 {
 426         log_reset();
 427         repo_reset();
 428         reset_dump_ctx(~0);
 429         reset_rev_ctx(0);
 430         reset_node_ctx(NULL);
 431         if (buffer_deinit(&input))
 432                 fprintf(stderr, "Input error\n");
 433         if (ferror(stdout))
 434                 fprintf(stderr, "Output error\n");
 435 }
 436
 437 void svndump_reset(void)
 438 {
 439         log_reset();
 440         buffer_reset(&input);
 441         repo_reset();
 442         reset_dump_ctx(~0);
 443         reset_rev_ctx(0);
 444         reset_node_ctx(NULL);
 445 }