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 <time.h>
  11 #include <ctype.h>
  12
  13 #include "compat-util.h"
  14 #include "repo_tree.h"
  15 #include "fast_export.h"
  16 #include "line_buffer.h"
  17 #include "strbuf.h"
  18 #include "mkgmtime.h"
  19 #include "svndump.h"
  20
  21 /*
  22  * Compare start of string to literal of equal length;
  23  * must be guarded by length test.
  24  */
  25 #define constcmp(s, ref) memcmp(s, ref, sizeof(ref) - 1)
  26
  27 #define REPORT_FILENO 3
  28
  29 #define NODEACT_REPLACE 4
  30 #define NODEACT_DELETE 3
  31 #define NODEACT_ADD 2
  32 #define NODEACT_CHANGE 1
  33 #define NODEACT_UNKNOWN 0
  34
  35 /* States: */
  36 #define DUMP_CTX 0      /* dump metadata */
  37 #define REV_CTX  1      /* revision metadata */
  38 #define NODE_CTX 2      /* node metadata */
  39 #define INTERNODE_CTX 3 /* between nodes */
  40
  41 #define LENGTH_UNKNOWN (~0)
  42 #define DATE_RFC2822_LEN 31
  43
  44 static struct line_buffer input = LINE_BUFFER_INIT;
  45
  46 static struct {
  47         uint32_t action, propLength, srcRev, type;
  48         off_t text_length;
  49         struct strbuf src, dst;
  50         uint32_t text_delta, prop_delta;
  51 } node_ctx;
  52
  53 static struct {
  54         uint32_t revision;
  55         unsigned long timestamp;
  56         struct strbuf log, author;
  57 } rev_ctx;
  58
  59 static struct {
  60         uint32_t version;
  61         struct strbuf uuid, url;
  62 } dump_ctx;
  63
  64 static void reset_node_ctx(char *fname)
  65 {
  66         node_ctx.type = 0;
  67         node_ctx.action = NODEACT_UNKNOWN;
  68         node_ctx.propLength = LENGTH_UNKNOWN;
  69         node_ctx.text_length = -1;
  70         strbuf_reset(&node_ctx.src);
  71         node_ctx.srcRev = 0;
  72         strbuf_reset(&node_ctx.dst);
  73         if (fname)
  74                 strbuf_addstr(&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         strbuf_reset(&rev_ctx.log);
  84         strbuf_reset(&rev_ctx.author);
  85 }
  86
  87 static void reset_dump_ctx(const char *url)
  88 {
  89         strbuf_reset(&dump_ctx.url);
  90         if (url)
  91                 strbuf_addstr(&dump_ctx.url, url);
  92         dump_ctx.version = 1;
  93         strbuf_reset(&dump_ctx.uuid);
  94 }
  95
  96 static void handle_property(const struct strbuf *key_buf,
  97                                 struct strbuf *val,
  98                                 uint32_t *type_set)
  99 {
 100         struct tm tm;
 101         const char *key = key_buf->buf;
 102         size_t keylen = key_buf->len;
 103
 104         switch (keylen + 1) {
 105         case sizeof("svn:log"):
 106                 if (constcmp(key, "svn:log"))
 107                         break;
 108                 if (!val)
 109                         die("invalid dump: unsets svn:log");
 110                 strbuf_swap(&rev_ctx.log, val);
 111                 break;
 112         case sizeof("svn:author"):
 113                 if (constcmp(key, "svn:author"))
 114                         break;
 115                 if (!val)
 116                         strbuf_reset(&rev_ctx.author);
 117                 else
 118                         strbuf_swap(&rev_ctx.author, val);
 119                 break;
 120         case sizeof("svn:date"):
 121                 if (constcmp(key, "svn:date"))
 122                         break;
 123                 if (!val)
 124                         die("invalid dump: unsets svn:date");
 125                 if (!strptime(val->buf, "%FT%T", &tm))
 126                         fprintf(stderr, "warning: " "invalid timestamp: %s", val->buf);
 127                 else
 128                         rev_ctx.timestamp = mkgmtime(&tm);
 129                 break;
 130         case sizeof("svn:executable"):
 131         case sizeof("svn:special"):
 132                 if (keylen == strlen("svn:executable") &&
 133                     constcmp(key, "svn:executable"))
 134                         break;
 135                 if (keylen == strlen("svn:special") &&
 136                     constcmp(key, "svn:special"))
 137                         break;
 138                 if (*type_set) {
 139                         if (!val)
 140                                 return;
 141                         die("invalid dump: sets type twice");
 142                 }
 143                 if (!val) {
 144                         node_ctx.type = REPO_MODE_BLB;
 145                         return;
 146                 }
 147                 *type_set = 1;
 148                 node_ctx.type = keylen == strlen("svn:executable") ?
 149                                 REPO_MODE_EXE :
 150                                 REPO_MODE_LNK;
 151         }
 152 }
 153
 154 static void die_short_read(void)
 155 {
 156         if (buffer_ferror(&input))
 157                 die_errno("error reading dump file");
 158         die("invalid dump: unexpected end of file");
 159 }
 160
 161 static void read_props(void)
 162 {
 163         static struct strbuf key = STRBUF_INIT;
 164         static struct strbuf val = STRBUF_INIT;
 165         const char *t;
 166         /*
 167          * NEEDSWORK: to support simple mode changes like
 168          *      K 11
 169          *      svn:special
 170          *      V 1
 171          *      *
 172          *      D 14
 173          *      svn:executable
 174          * we keep track of whether a mode has been set and reset to
 175          * plain file only if not.  We should be keeping track of the
 176          * symlink and executable bits separately instead.
 177          */
 178         uint32_t type_set = 0;
 179         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 180                 uint32_t len;
 181                 const char type = t[0];
 182                 int ch;
 183
 184                 if (!type || t[1] != ' ')
 185                         die("invalid property line: %s\n", t);
 186                 len = atoi(&t[2]);
 187                 strbuf_reset(&val);
 188                 buffer_read_binary(&input, &val, len);
 189                 if (val.len < len)
 190                         die_short_read();
 191
 192                 /* Discard trailing newline. */
 193                 ch = buffer_read_char(&input);
 194                 if (ch == EOF)
 195                         die_short_read();
 196                 if (ch != '\n')
 197                         die("invalid dump: expected newline after %s", val.buf);
 198
 199                 switch (type) {
 200                 case 'K':
 201                         strbuf_swap(&key, &val);
 202                         continue;
 203                 case 'D':
 204                         handle_property(&val, NULL, &type_set);
 205                         continue;
 206                 case 'V':
 207                         handle_property(&key, &val, &type_set);
 208                         strbuf_reset(&key);
 209                         continue;
 210                 default:
 211                         die("invalid property line: %s\n", t);
 212                 }
 213         }
 214 }
 215
 216 static void handle_node(void)
 217 {
 218         const uint32_t type = node_ctx.type;
 219         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 220         const int have_text = node_ctx.text_length != -1;
 221         /*
 222          * Old text for this node:
 223          *  NULL        - directory or bug
 224          *  empty_blob  - empty
 225          *  "<dataref>" - data retrievable from fast-import
 226          */
 227         static const char *const empty_blob = "::empty::";
 228         const char *old_data = NULL;
 229         uint32_t old_mode = REPO_MODE_BLB;
 230
 231         if (node_ctx.action == NODEACT_DELETE) {
 232                 if (have_text || have_props || node_ctx.srcRev)
 233                         die("invalid dump: deletion node has "
 234                                 "copyfrom info, text, or properties");
 235                 repo_delete(node_ctx.dst.buf);
 236                 return;
 237         }
 238         if (node_ctx.action == NODEACT_REPLACE) {
 239                 repo_delete(node_ctx.dst.buf);
 240                 node_ctx.action = NODEACT_ADD;
 241         }
 242         if (node_ctx.srcRev) {
 243                 repo_copy(node_ctx.srcRev, node_ctx.src.buf, node_ctx.dst.buf);
 244                 if (node_ctx.action == NODEACT_ADD)
 245                         node_ctx.action = NODEACT_CHANGE;
 246         }
 247         if (have_text && type == REPO_MODE_DIR)
 248                 die("invalid dump: directories cannot have text attached");
 249
 250         /*
 251          * Find old content (old_data) and decide on the new mode.
 252          */
 253         if (node_ctx.action == NODEACT_CHANGE && !*node_ctx.dst.buf) {
 254                 if (type != REPO_MODE_DIR)
 255                         die("invalid dump: root of tree is not a regular file");
 256                 old_data = NULL;
 257         } else if (node_ctx.action == NODEACT_CHANGE) {
 258                 uint32_t mode;
 259                 old_data = repo_read_path(node_ctx.dst.buf, &mode);
 260                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 261                         die("invalid dump: cannot modify a directory into a file");
 262                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 263                         die("invalid dump: cannot modify a file into a directory");
 264                 node_ctx.type = mode;
 265                 old_mode = mode;
 266         } else if (node_ctx.action == NODEACT_ADD) {
 267                 if (type == REPO_MODE_DIR)
 268                         old_data = NULL;
 269                 else if (have_text)
 270                         old_data = empty_blob;
 271                 else
 272                         die("invalid dump: adds node without text");
 273         } else {
 274                 die("invalid dump: Node-path block lacks Node-action");
 275         }
 276
 277         /*
 278          * Adjust mode to reflect properties.
 279          */
 280         if (have_props) {
 281                 if (!node_ctx.prop_delta)
 282                         node_ctx.type = type;
 283                 if (node_ctx.propLength)
 284                         read_props();
 285         }
 286
 287         /*
 288          * Save the result.
 289          */
 290         if (type == REPO_MODE_DIR)      /* directories are not tracked. */
 291                 return;
 292         assert(old_data);
 293         if (old_data == empty_blob)
 294                 /* For the fast_export_* functions, NULL means empty. */
 295                 old_data = NULL;
 296         if (!have_text) {
 297                 fast_export_modify(node_ctx.dst.buf, node_ctx.type, old_data);
 298                 return;
 299         }
 300         if (!node_ctx.text_delta) {
 301                 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
 302                 fast_export_data(node_ctx.type, node_ctx.text_length, &input);
 303                 return;
 304         }
 305         fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
 306         fast_export_blob_delta(node_ctx.type, old_mode, old_data,
 307                                 node_ctx.text_length, &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.buf,
 315                 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
 316                 rev_ctx.timestamp);
 317 }
 318
 319 static void end_revision(void)
 320 {
 321         if (rev_ctx.revision)
 322                 fast_export_end_commit(rev_ctx.revision);
 323 }
 324
 325 void svndump_read(const char *url)
 326 {
 327         char *val;
 328         char *t;
 329         uint32_t active_ctx = DUMP_CTX;
 330         uint32_t len;
 331
 332         reset_dump_ctx(url);
 333         while ((t = buffer_read_line(&input))) {
 334                 val = strchr(t, ':');
 335                 if (!val)
 336                         continue;
 337                 val++;
 338                 if (*val != ' ')
 339                         continue;
 340                 val++;
 341
 342                 /* strlen(key) + 1 */
 343                 switch (val - t - 1) {
 344                 case sizeof("SVN-fs-dump-format-version"):
 345                         if (constcmp(t, "SVN-fs-dump-format-version"))
 346                                 continue;
 347                         dump_ctx.version = atoi(val);
 348                         if (dump_ctx.version > 3)
 349                                 die("expected svn dump format version <= 3, found %"PRIu32,
 350                                     dump_ctx.version);
 351                         break;
 352                 case sizeof("UUID"):
 353                         if (constcmp(t, "UUID"))
 354                                 continue;
 355                         strbuf_reset(&dump_ctx.uuid);
 356                         strbuf_addstr(&dump_ctx.uuid, val);
 357                         break;
 358                 case sizeof("Revision-number"):
 359                         if (constcmp(t, "Revision-number"))
 360                                 continue;
 361                         if (active_ctx == NODE_CTX)
 362                                 handle_node();
 363                         if (active_ctx == REV_CTX)
 364                                 begin_revision();
 365                         if (active_ctx != DUMP_CTX)
 366                                 end_revision();
 367                         active_ctx = REV_CTX;
 368                         reset_rev_ctx(atoi(val));
 369                         break;
 370                 case sizeof("Node-path"):
 371                         if (prefixcmp(t, "Node-"))
 372                                 continue;
 373                         if (!constcmp(t + strlen("Node-"), "path")) {
 374                                 if (active_ctx == NODE_CTX)
 375                                         handle_node();
 376                                 if (active_ctx == REV_CTX)
 377                                         begin_revision();
 378                                 active_ctx = NODE_CTX;
 379                                 reset_node_ctx(val);
 380                                 break;
 381                         }
 382                         if (constcmp(t + strlen("Node-"), "kind"))
 383                                 continue;
 384                         if (!strcmp(val, "dir"))
 385                                 node_ctx.type = REPO_MODE_DIR;
 386                         else if (!strcmp(val, "file"))
 387                                 node_ctx.type = REPO_MODE_BLB;
 388                         else
 389                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 390                         break;
 391                 case sizeof("Node-action"):
 392                         if (constcmp(t, "Node-action"))
 393                                 continue;
 394                         if (!strcmp(val, "delete")) {
 395                                 node_ctx.action = NODEACT_DELETE;
 396                         } else if (!strcmp(val, "add")) {
 397                                 node_ctx.action = NODEACT_ADD;
 398                         } else if (!strcmp(val, "change")) {
 399                                 node_ctx.action = NODEACT_CHANGE;
 400                         } else if (!strcmp(val, "replace")) {
 401                                 node_ctx.action = NODEACT_REPLACE;
 402                         } else {
 403                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 404                                 node_ctx.action = NODEACT_UNKNOWN;
 405                         }
 406                         break;
 407                 case sizeof("Node-copyfrom-path"):
 408                         if (constcmp(t, "Node-copyfrom-path"))
 409                                 continue;
 410                         strbuf_reset(&node_ctx.src);
 411                         strbuf_addstr(&node_ctx.src, val);
 412                         break;
 413                 case sizeof("Node-copyfrom-rev"):
 414                         if (constcmp(t, "Node-copyfrom-rev"))
 415                                 continue;
 416                         node_ctx.srcRev = atoi(val);
 417                         break;
 418                 case sizeof("Text-content-length"):
 419                         if (!constcmp(t, "Text-content-length")) {
 420                                 char *end;
 421                                 uintmax_t textlen;
 422
 423                                 textlen = strtoumax(val, &end, 10);
 424                                 if (!isdigit(*val) || *end)
 425                                         die("invalid dump: non-numeric length %s", val);
 426                                 if (textlen > maximum_signed_value_of_type(off_t))
 427                                         die("unrepresentable length in dump: %s", val);
 428                                 node_ctx.text_length = (off_t) textlen;
 429                                 break;
 430                         }
 431                         if (constcmp(t, "Prop-content-length"))
 432                                 continue;
 433                         node_ctx.propLength = atoi(val);
 434                         break;
 435                 case sizeof("Text-delta"):
 436                         if (!constcmp(t, "Text-delta")) {
 437                                 node_ctx.text_delta = !strcmp(val, "true");
 438                                 break;
 439                         }
 440                         if (constcmp(t, "Prop-delta"))
 441                                 continue;
 442                         node_ctx.prop_delta = !strcmp(val, "true");
 443                         break;
 444                 case sizeof("Content-length"):
 445                         if (constcmp(t, "Content-length"))
 446                                 continue;
 447                         len = atoi(val);
 448                         t = buffer_read_line(&input);
 449                         if (!t)
 450                                 die_short_read();
 451                         if (*t)
 452                                 die("invalid dump: expected blank line after content length header");
 453                         if (active_ctx == REV_CTX) {
 454                                 read_props();
 455                         } else if (active_ctx == NODE_CTX) {
 456                                 handle_node();
 457                                 active_ctx = INTERNODE_CTX;
 458                         } else {
 459                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 460                                 if (buffer_skip_bytes(&input, len) != len)
 461                                         die_short_read();
 462                         }
 463                 }
 464         }
 465         if (buffer_ferror(&input))
 466                 die_short_read();
 467         if (active_ctx == NODE_CTX)
 468                 handle_node();
 469         if (active_ctx == REV_CTX)
 470                 begin_revision();
 471         if (active_ctx != DUMP_CTX)
 472                 end_revision();
 473 }
 474
 475 int svndump_init(const char *filename)
 476 {
 477         if (buffer_init(&input, filename))
 478                 return error("cannot open %s: %s", filename, strerror(errno));
 479         fast_export_init(REPORT_FILENO);
 480         strbuf_init(&dump_ctx.uuid, 4096);
 481         strbuf_init(&dump_ctx.url, 4096);
 482         strbuf_init(&rev_ctx.log, 4096);
 483         strbuf_init(&rev_ctx.author, 4096);
 484         strbuf_init(&node_ctx.src, 4096);
 485         strbuf_init(&node_ctx.dst, 4096);
 486         reset_dump_ctx(NULL);
 487         reset_rev_ctx(0);
 488         reset_node_ctx(NULL);
 489         return 0;
 490 }
 491
 492 void svndump_deinit(void)
 493 {
 494         fast_export_deinit();
 495         reset_dump_ctx(NULL);
 496         reset_rev_ctx(0);
 497         reset_node_ctx(NULL);
 498         strbuf_release(&rev_ctx.log);
 499         strbuf_release(&node_ctx.src);
 500         strbuf_release(&node_ctx.dst);
 501         if (buffer_deinit(&input))
 502                 fprintf(stderr, "Input error\n");
 503         if (ferror(stdout))
 504                 fprintf(stderr, "Output error\n");
 505 }
 506
 507 void svndump_reset(void)
 508 {
 509         fast_export_reset();
 510         buffer_reset(&input);
 511         strbuf_release(&dump_ctx.uuid);
 512         strbuf_release(&dump_ctx.url);
 513         strbuf_release(&rev_ctx.log);
 514         strbuf_release(&rev_ctx.author);
 515 }