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 NODEACT_REPLACE 4
  18 #define NODEACT_DELETE 3
  19 #define NODEACT_ADD 2
  20 #define NODEACT_CHANGE 1
  21 #define NODEACT_UNKNOWN 0
  22
  23 #define DUMP_CTX 0
  24 #define REV_CTX  1
  25 #define NODE_CTX 2
  26
  27 #define LENGTH_UNKNOWN (~0)
  28 #define DATE_RFC2822_LEN 31
  29
  30 static struct line_buffer input = LINE_BUFFER_INIT;
  31
  32 static struct {
  33         uint32_t action, propLength, textLength, srcRev, type;
  34         uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
  35         uint32_t text_delta, prop_delta;
  36 } node_ctx;
  37
  38 static struct {
  39         uint32_t revision;
  40         unsigned long timestamp;
  41         struct strbuf log, author;
  42 } rev_ctx;
  43
  44 static struct {
  45         uint32_t version;
  46         struct strbuf uuid, url;
  47 } dump_ctx;
  48
  49 static struct {
  50         uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
  51                 revision_number, node_path, node_kind, node_action,
  52                 node_copyfrom_path, node_copyfrom_rev, text_content_length,
  53                 prop_content_length, content_length, svn_fs_dump_format_version,
  54                 /* version 3 format */
  55                 text_delta, prop_delta;
  56 } keys;
  57
  58 static void reset_node_ctx(char *fname)
  59 {
  60         node_ctx.type = 0;
  61         node_ctx.action = NODEACT_UNKNOWN;
  62         node_ctx.propLength = LENGTH_UNKNOWN;
  63         node_ctx.textLength = LENGTH_UNKNOWN;
  64         node_ctx.src[0] = ~0;
  65         node_ctx.srcRev = 0;
  66         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  67         node_ctx.text_delta = 0;
  68         node_ctx.prop_delta = 0;
  69 }
  70
  71 static void reset_rev_ctx(uint32_t revision)
  72 {
  73         rev_ctx.revision = revision;
  74         rev_ctx.timestamp = 0;
  75         strbuf_reset(&rev_ctx.log);
  76         strbuf_reset(&rev_ctx.author);
  77 }
  78
  79 static void reset_dump_ctx(const char *url)
  80 {
  81         strbuf_reset(&dump_ctx.url);
  82         if (url)
  83                 strbuf_addstr(&dump_ctx.url, url);
  84         dump_ctx.version = 1;
  85         strbuf_reset(&dump_ctx.uuid);
  86 }
  87
  88 static void init_keys(void)
  89 {
  90         keys.svn_log = pool_intern("svn:log");
  91         keys.svn_author = pool_intern("svn:author");
  92         keys.svn_date = pool_intern("svn:date");
  93         keys.svn_executable = pool_intern("svn:executable");
  94         keys.svn_special = pool_intern("svn:special");
  95         keys.uuid = pool_intern("UUID");
  96         keys.revision_number = pool_intern("Revision-number");
  97         keys.node_path = pool_intern("Node-path");
  98         keys.node_kind = pool_intern("Node-kind");
  99         keys.node_action = pool_intern("Node-action");
 100         keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
 101         keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
 102         keys.text_content_length = pool_intern("Text-content-length");
 103         keys.prop_content_length = pool_intern("Prop-content-length");
 104         keys.content_length = pool_intern("Content-length");
 105         keys.svn_fs_dump_format_version = pool_intern("SVN-fs-dump-format-version");
 106         /* version 3 format (Subversion 1.1.0) */
 107         keys.text_delta = pool_intern("Text-delta");
 108         keys.prop_delta = pool_intern("Prop-delta");
 109 }
 110
 111 static void handle_property(uint32_t key, const char *val, uint32_t len,
 112                                 uint32_t *type_set)
 113 {
 114         if (key == keys.svn_log) {
 115                 if (!val)
 116                         die("invalid dump: unsets svn:log");
 117                 strbuf_reset(&rev_ctx.log);
 118                 strbuf_add(&rev_ctx.log, val, len);
 119         } else if (key == keys.svn_author) {
 120                 strbuf_reset(&rev_ctx.author);
 121                 if (val)
 122                         strbuf_add(&rev_ctx.author, val, len);
 123         } else if (key == keys.svn_date) {
 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         } else if (key == keys.svn_executable || key == keys.svn_special) {
 129                 if (*type_set) {
 130                         if (!val)
 131                                 return;
 132                         die("invalid dump: sets type twice");
 133                 }
 134                 if (!val) {
 135                         node_ctx.type = REPO_MODE_BLB;
 136                         return;
 137                 }
 138                 *type_set = 1;
 139                 node_ctx.type = key == keys.svn_executable ?
 140                                 REPO_MODE_EXE :
 141                                 REPO_MODE_LNK;
 142         }
 143 }
 144
 145 static void die_short_read(void)
 146 {
 147         if (buffer_ferror(&input))
 148                 die_errno("error reading dump file");
 149         die("invalid dump: unexpected end of file");
 150 }
 151
 152 static void read_props(void)
 153 {
 154         uint32_t key = ~0;
 155         const char *t;
 156         /*
 157          * NEEDSWORK: to support simple mode changes like
 158          *      K 11
 159          *      svn:special
 160          *      V 1
 161          *      *
 162          *      D 14
 163          *      svn:executable
 164          * we keep track of whether a mode has been set and reset to
 165          * plain file only if not.  We should be keeping track of the
 166          * symlink and executable bits separately instead.
 167          */
 168         uint32_t type_set = 0;
 169         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 170                 uint32_t len;
 171                 const char *val;
 172                 const char type = t[0];
 173                 int ch;
 174
 175                 if (!type || t[1] != ' ')
 176                         die("invalid property line: %s\n", t);
 177                 len = atoi(&t[2]);
 178                 val = buffer_read_string(&input, len);
 179                 if (!val || strlen(val) != len)
 180                         die_short_read();
 181
 182                 /* Discard trailing newline. */
 183                 ch = buffer_read_char(&input);
 184                 if (ch == EOF)
 185                         die_short_read();
 186                 if (ch != '\n')
 187                         die("invalid dump: expected newline after %s", val);
 188
 189                 switch (type) {
 190                 case 'K':
 191                         key = pool_intern(val);
 192                         continue;
 193                 case 'D':
 194                         key = pool_intern(val);
 195                         val = NULL;
 196                         len = 0;
 197                         /* fall through */
 198                 case 'V':
 199                         handle_property(key, val, len, &type_set);
 200                         key = ~0;
 201                         continue;
 202                 default:
 203                         die("invalid property line: %s\n", t);
 204                 }
 205         }
 206 }
 207
 208 static void handle_node(void)
 209 {
 210         uint32_t mark = 0;
 211         const uint32_t type = node_ctx.type;
 212         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 213         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 214
 215         if (node_ctx.text_delta)
 216                 die("text deltas not supported");
 217         if (have_text)
 218                 mark = next_blob_mark();
 219         if (node_ctx.action == NODEACT_DELETE) {
 220                 if (have_text || have_props || node_ctx.srcRev)
 221                         die("invalid dump: deletion node has "
 222                                 "copyfrom info, text, or properties");
 223                 return repo_delete(node_ctx.dst);
 224         }
 225         if (node_ctx.action == NODEACT_REPLACE) {
 226                 repo_delete(node_ctx.dst);
 227                 node_ctx.action = NODEACT_ADD;
 228         }
 229         if (node_ctx.srcRev) {
 230                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 231                 if (node_ctx.action == NODEACT_ADD)
 232                         node_ctx.action = NODEACT_CHANGE;
 233         }
 234         if (have_text && type == REPO_MODE_DIR)
 235                 die("invalid dump: directories cannot have text attached");
 236
 237         /*
 238          * Decide on the new content (mark) and mode (node_ctx.type).
 239          */
 240         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 241                 if (type != REPO_MODE_DIR)
 242                         die("invalid dump: root of tree is not a regular file");
 243         } else if (node_ctx.action == NODEACT_CHANGE) {
 244                 uint32_t mode;
 245                 if (!have_text)
 246                         mark = repo_read_path(node_ctx.dst);
 247                 mode = repo_read_mode(node_ctx.dst);
 248                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 249                         die("invalid dump: cannot modify a directory into a file");
 250                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 251                         die("invalid dump: cannot modify a file into a directory");
 252                 node_ctx.type = mode;
 253         } else if (node_ctx.action == NODEACT_ADD) {
 254                 if (!have_text && type != REPO_MODE_DIR)
 255                         die("invalid dump: adds node without text");
 256         } else {
 257                 die("invalid dump: Node-path block lacks Node-action");
 258         }
 259
 260         /*
 261          * Adjust mode to reflect properties.
 262          */
 263         if (have_props) {
 264                 if (!node_ctx.prop_delta)
 265                         node_ctx.type = type;
 266                 if (node_ctx.propLength)
 267                         read_props();
 268         }
 269
 270         /*
 271          * Save the result.
 272          */
 273         repo_add(node_ctx.dst, node_ctx.type, mark);
 274         if (have_text)
 275                 fast_export_blob(node_ctx.type, mark,
 276                                  node_ctx.textLength, &input);
 277 }
 278
 279 static void handle_revision(void)
 280 {
 281         if (rev_ctx.revision)
 282                 repo_commit(rev_ctx.revision, rev_ctx.author.buf,
 283                         rev_ctx.log.buf, dump_ctx.uuid.buf, dump_ctx.url.buf,
 284                         rev_ctx.timestamp);
 285 }
 286
 287 void svndump_read(const char *url)
 288 {
 289         char *val;
 290         char *t;
 291         uint32_t active_ctx = DUMP_CTX;
 292         uint32_t len;
 293         uint32_t key;
 294
 295         reset_dump_ctx(url);
 296         while ((t = buffer_read_line(&input))) {
 297                 val = strstr(t, ": ");
 298                 if (!val)
 299                         continue;
 300                 *val++ = '\0';
 301                 *val++ = '\0';
 302                 key = pool_intern(t);
 303
 304                 if (key == keys.svn_fs_dump_format_version) {
 305                         dump_ctx.version = atoi(val);
 306                         if (dump_ctx.version > 3)
 307                                 die("expected svn dump format version <= 3, found %"PRIu32,
 308                                     dump_ctx.version);
 309                 } else if (key == keys.uuid) {
 310                         strbuf_reset(&dump_ctx.uuid);
 311                         strbuf_addstr(&dump_ctx.uuid, val);
 312                 } else if (key == keys.revision_number) {
 313                         if (active_ctx == NODE_CTX)
 314                                 handle_node();
 315                         if (active_ctx != DUMP_CTX)
 316                                 handle_revision();
 317                         active_ctx = REV_CTX;
 318                         reset_rev_ctx(atoi(val));
 319                 } else if (key == keys.node_path) {
 320                         if (active_ctx == NODE_CTX)
 321                                 handle_node();
 322                         active_ctx = NODE_CTX;
 323                         reset_node_ctx(val);
 324                 } else if (key == keys.node_kind) {
 325                         if (!strcmp(val, "dir"))
 326                                 node_ctx.type = REPO_MODE_DIR;
 327                         else if (!strcmp(val, "file"))
 328                                 node_ctx.type = REPO_MODE_BLB;
 329                         else
 330                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 331                 } else if (key == keys.node_action) {
 332                         if (!strcmp(val, "delete")) {
 333                                 node_ctx.action = NODEACT_DELETE;
 334                         } else if (!strcmp(val, "add")) {
 335                                 node_ctx.action = NODEACT_ADD;
 336                         } else if (!strcmp(val, "change")) {
 337                                 node_ctx.action = NODEACT_CHANGE;
 338                         } else if (!strcmp(val, "replace")) {
 339                                 node_ctx.action = NODEACT_REPLACE;
 340                         } else {
 341                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 342                                 node_ctx.action = NODEACT_UNKNOWN;
 343                         }
 344                 } else if (key == keys.node_copyfrom_path) {
 345                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 346                 } else if (key == keys.node_copyfrom_rev) {
 347                         node_ctx.srcRev = atoi(val);
 348                 } else if (key == keys.text_content_length) {
 349                         node_ctx.textLength = atoi(val);
 350                 } else if (key == keys.prop_content_length) {
 351                         node_ctx.propLength = atoi(val);
 352                 } else if (key == keys.text_delta) {
 353                         node_ctx.text_delta = !strcmp(val, "true");
 354                 } else if (key == keys.prop_delta) {
 355                         node_ctx.prop_delta = !strcmp(val, "true");
 356                 } else if (key == keys.content_length) {
 357                         len = atoi(val);
 358                         t = buffer_read_line(&input);
 359                         if (!t)
 360                                 die_short_read();
 361                         if (*t)
 362                                 die("invalid dump: expected blank line after content length header");
 363                         if (active_ctx == REV_CTX) {
 364                                 read_props();
 365                         } else if (active_ctx == NODE_CTX) {
 366                                 handle_node();
 367                                 active_ctx = REV_CTX;
 368                         } else {
 369                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 370                                 if (buffer_skip_bytes(&input, len) != len)
 371                                         die_short_read();
 372                         }
 373                 }
 374         }
 375         if (buffer_ferror(&input))
 376                 die_short_read();
 377         if (active_ctx == NODE_CTX)
 378                 handle_node();
 379         if (active_ctx != DUMP_CTX)
 380                 handle_revision();
 381 }
 382
 383 int svndump_init(const char *filename)
 384 {
 385         if (buffer_init(&input, filename))
 386                 return error("cannot open %s: %s", filename, strerror(errno));
 387         repo_init();
 388         strbuf_init(&dump_ctx.uuid, 4096);
 389         strbuf_init(&dump_ctx.url, 4096);
 390         strbuf_init(&rev_ctx.log, 4096);
 391         strbuf_init(&rev_ctx.author, 4096);
 392         reset_dump_ctx(NULL);
 393         reset_rev_ctx(0);
 394         reset_node_ctx(NULL);
 395         init_keys();
 396         return 0;
 397 }
 398
 399 void svndump_deinit(void)
 400 {
 401         repo_reset();
 402         reset_dump_ctx(NULL);
 403         reset_rev_ctx(0);
 404         reset_node_ctx(NULL);
 405         strbuf_release(&rev_ctx.log);
 406         if (buffer_deinit(&input))
 407                 fprintf(stderr, "Input error\n");
 408         if (ferror(stdout))
 409                 fprintf(stderr, "Output error\n");
 410 }
 411
 412 void svndump_reset(void)
 413 {
 414         buffer_reset(&input);
 415         repo_reset();
 416         strbuf_release(&dump_ctx.uuid);
 417         strbuf_release(&dump_ctx.url);
 418         strbuf_release(&rev_ctx.log);
 419         strbuf_release(&rev_ctx.author);
 420 }