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 DATE_RFC2822_LEN 31
  42
  43 static struct line_buffer input = LINE_BUFFER_INIT;
  44
  45 static struct {
  46         uint32_t action, srcRev, type;
  47         off_t prop_length, text_length;
  48         struct strbuf src, dst;
  49         uint32_t text_delta, prop_delta;
  50 } node_ctx;
  51
  52 static struct {
  53         uint32_t revision;
  54         unsigned long timestamp;
  55         struct strbuf log, author;
  56 } rev_ctx;
  57
  58 static struct {
  59         uint32_t version;
  60         struct strbuf uuid, url;
  61 } dump_ctx;
  62
  63 static void reset_node_ctx(char *fname)
  64 {
  65         node_ctx.type = 0;
  66         node_ctx.action = NODEACT_UNKNOWN;
  67         node_ctx.prop_length = -1;
  68         node_ctx.text_length = -1;
  69         strbuf_reset(&node_ctx.src);
  70         node_ctx.srcRev = 0;
  71         strbuf_reset(&node_ctx.dst);
  72         if (fname)
  73                 strbuf_addstr(&node_ctx.dst, fname);
  74         node_ctx.text_delta = 0;
  75         node_ctx.prop_delta = 0;
  76 }
  77
  78 static void reset_rev_ctx(uint32_t revision)
  79 {
  80         rev_ctx.revision = revision;
  81         rev_ctx.timestamp = 0;
  82         strbuf_reset(&rev_ctx.log);
  83         strbuf_reset(&rev_ctx.author);
  84 }
  85
  86 static void reset_dump_ctx(const char *url)
  87 {
  88         strbuf_reset(&dump_ctx.url);
  89         if (url)
  90                 strbuf_addstr(&dump_ctx.url, url);
  91         dump_ctx.version = 1;
  92         strbuf_reset(&dump_ctx.uuid);
  93 }
  94
  95 static void handle_property(const struct strbuf *key_buf,
  96                                 struct strbuf *val,
  97                                 uint32_t *type_set)
  98 {
  99         struct tm tm;
 100         const char *key = key_buf->buf;
 101         size_t keylen = key_buf->len;
 102
 103         switch (keylen + 1) {
 104         case sizeof("svn:log"):
 105                 if (constcmp(key, "svn:log"))
 106                         break;
 107                 if (!val)
 108                         die("invalid dump: unsets svn:log");
 109                 strbuf_swap(&rev_ctx.log, val);
 110                 break;
 111         case sizeof("svn:author"):
 112                 if (constcmp(key, "svn:author"))
 113                         break;
 114                 if (!val)
 115                         strbuf_reset(&rev_ctx.author);
 116                 else
 117                         strbuf_swap(&rev_ctx.author, val);
 118                 break;
 119         case sizeof("svn:date"):
 120                 if (constcmp(key, "svn:date"))
 121                         break;
 122                 if (!val)
 123                         die("invalid dump: unsets svn:date");
 124                 if (!strptime(val->buf, "%FT%T", &tm))
 125                         fprintf(stderr, "warning: " "invalid timestamp: %s", val->buf);
 126                 else
 127                         rev_ctx.timestamp = mkgmtime(&tm);
 128                 break;
 129         case sizeof("svn:executable"):
 130         case sizeof("svn:special"):
 131                 if (keylen == strlen("svn:executable") &&
 132                     constcmp(key, "svn:executable"))
 133                         break;
 134                 if (keylen == strlen("svn:special") &&
 135                     constcmp(key, "svn:special"))
 136                         break;
 137                 if (*type_set) {
 138                         if (!val)
 139                                 return;
 140                         die("invalid dump: sets type twice");
 141                 }
 142                 if (!val) {
 143                         node_ctx.type = REPO_MODE_BLB;
 144                         return;
 145                 }
 146                 *type_set = 1;
 147                 node_ctx.type = keylen == strlen("svn:executable") ?
 148                                 REPO_MODE_EXE :
 149                                 REPO_MODE_LNK;
 150         }
 151 }
 152
 153 static void die_short_read(void)
 154 {
 155         if (buffer_ferror(&input))
 156                 die_errno("error reading dump file");
 157         die("invalid dump: unexpected end of file");
 158 }
 159
 160 static void read_props(void)
 161 {
 162         static struct strbuf key = STRBUF_INIT;
 163         static struct strbuf val = STRBUF_INIT;
 164         const char *t;
 165         /*
 166          * NEEDSWORK: to support simple mode changes like
 167          *      K 11
 168          *      svn:special
 169          *      V 1
 170          *      *
 171          *      D 14
 172          *      svn:executable
 173          * we keep track of whether a mode has been set and reset to
 174          * plain file only if not.  We should be keeping track of the
 175          * symlink and executable bits separately instead.
 176          */
 177         uint32_t type_set = 0;
 178         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 179                 uint32_t len;
 180                 const char type = t[0];
 181                 int ch;
 182
 183                 if (!type || t[1] != ' ')
 184                         die("invalid property line: %s\n", t);
 185                 len = atoi(&t[2]);
 186                 strbuf_reset(&val);
 187                 buffer_read_binary(&input, &val, len);
 188                 if (val.len < len)
 189                         die_short_read();
 190
 191                 /* Discard trailing newline. */
 192                 ch = buffer_read_char(&input);
 193                 if (ch == EOF)
 194                         die_short_read();
 195                 if (ch != '\n')
 196                         die("invalid dump: expected newline after %s", val.buf);
 197
 198                 switch (type) {
 199                 case 'K':
 200                         strbuf_swap(&key, &val);
 201                         continue;
 202                 case 'D':
 203                         handle_property(&val, NULL, &type_set);
 204                         continue;
 205                 case 'V':
 206                         handle_property(&key, &val, &type_set);
 207                         strbuf_reset(&key);
 208                         continue;
 209                 default:
 210                         die("invalid property line: %s\n", t);
 211                 }
 212         }
 213 }
 214
 215 static void handle_node(void)
 216 {
 217         const uint32_t type = node_ctx.type;
 218         const int have_props = node_ctx.prop_length != -1;
 219         const int have_text = node_ctx.text_length != -1;
 220         /*
 221          * Old text for this node:
 222          *  NULL        - directory or bug
 223          *  empty_blob  - empty
 224          *  "<dataref>" - data retrievable from fast-import
 225          */
 226         static const char *const empty_blob = "::empty::";
 227         const char *old_data = NULL;
 228         uint32_t old_mode = REPO_MODE_BLB;
 229
 230         if (node_ctx.action == NODEACT_DELETE) {
 231                 if (have_text || have_props || node_ctx.srcRev)
 232                         die("invalid dump: deletion node has "
 233                                 "copyfrom info, text, or properties");
 234                 repo_delete(node_ctx.dst.buf);
 235                 return;
 236         }
 237         if (node_ctx.action == NODEACT_REPLACE) {
 238                 repo_delete(node_ctx.dst.buf);
 239                 node_ctx.action = NODEACT_ADD;
 240         }
 241         if (node_ctx.srcRev) {
 242                 repo_copy(node_ctx.srcRev, node_ctx.src.buf, node_ctx.dst.buf);
 243                 if (node_ctx.action == NODEACT_ADD)
 244                         node_ctx.action = NODEACT_CHANGE;
 245         }
 246         if (have_text && type == REPO_MODE_DIR)
 247                 die("invalid dump: directories cannot have text attached");
 248
 249         /*
 250          * Find old content (old_data) and decide on the new mode.
 251          */
 252         if (node_ctx.action == NODEACT_CHANGE && !*node_ctx.dst.buf) {
 253                 if (type != REPO_MODE_DIR)
 254                         die("invalid dump: root of tree is not a regular file");
 255                 old_data = NULL;
 256         } else if (node_ctx.action == NODEACT_CHANGE) {
 257                 uint32_t mode;
 258                 old_data = repo_read_path(node_ctx.dst.buf, &mode);
 259                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 260                         die("invalid dump: cannot modify a directory into a file");
 261                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 262                         die("invalid dump: cannot modify a file into a directory");
 263                 node_ctx.type = mode;
 264                 old_mode = mode;
 265         } else if (node_ctx.action == NODEACT_ADD) {
 266                 if (type == REPO_MODE_DIR)
 267                         old_data = NULL;
 268                 else if (have_text)
 269                         old_data = empty_blob;
 270                 else
 271                         die("invalid dump: adds node without text");
 272         } else {
 273                 die("invalid dump: Node-path block lacks Node-action");
 274         }
 275
 276         /*
 277          * Adjust mode to reflect properties.
 278          */
 279         if (have_props) {
 280                 if (!node_ctx.prop_delta)
 281                         node_ctx.type = type;
 282                 if (node_ctx.prop_length)
 283                         read_props();
 284         }
 285
 286         /*
 287          * Save the result.
 288          */
 289         if (type == REPO_MODE_DIR)      /* directories are not tracked. */
 290                 return;
 291         assert(old_data);
 292         if (old_data == empty_blob)
 293                 /* For the fast_export_* functions, NULL means empty. */
 294                 old_data = NULL;
 295         if (!have_text) {
 296                 fast_export_modify(node_ctx.dst.buf, node_ctx.type, old_data);
 297                 return;
 298         }
 299         if (!node_ctx.text_delta) {
 300                 fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
 301                 fast_export_data(node_ctx.type, node_ctx.text_length, &input);
 302                 return;
 303         }
 304         fast_export_modify(node_ctx.dst.buf, node_ctx.type, "inline");
 305         fast_export_blob_delta(node_ctx.type, old_mode, old_data,
 306                                 node_ctx.text_length, &input);
 307 }
 308
 309 static void begin_revision(void)
 310 {
 311         if (!rev_ctx.revision)  /* revision 0 gets no git commit. */
 312                 return;
 313         fast_export_begin_commit(rev_ctx.revision, rev_ctx.author.buf,
 314                 &rev_ctx.log, dump_ctx.uuid.buf, dump_ctx.url.buf,
 315                 rev_ctx.timestamp);
 316 }
 317
 318 static void end_revision(void)
 319 {
 320         if (rev_ctx.revision)
 321                 fast_export_end_commit(rev_ctx.revision);
 322 }
 323
 324 void svndump_read(const char *url)
 325 {
 326         char *val;
 327         char *t;
 328         uint32_t active_ctx = DUMP_CTX;
 329         uint32_t len;
 330
 331         reset_dump_ctx(url);
 332         while ((t = buffer_read_line(&input))) {
 333                 val = strchr(t, ':');
 334                 if (!val)
 335                         continue;
 336                 val++;
 337                 if (*val != ' ')
 338                         continue;
 339                 val++;
 340
 341                 /* strlen(key) + 1 */
 342                 switch (val - t - 1) {
 343                 case sizeof("SVN-fs-dump-format-version"):
 344                         if (constcmp(t, "SVN-fs-dump-format-version"))
 345                                 continue;
 346                         dump_ctx.version = atoi(val);
 347                         if (dump_ctx.version > 3)
 348                                 die("expected svn dump format version <= 3, found %"PRIu32,
 349                                     dump_ctx.version);
 350                         break;
 351                 case sizeof("UUID"):
 352                         if (constcmp(t, "UUID"))
 353                                 continue;
 354                         strbuf_reset(&dump_ctx.uuid);
 355                         strbuf_addstr(&dump_ctx.uuid, val);
 356                         break;
 357                 case sizeof("Revision-number"):
 358                         if (constcmp(t, "Revision-number"))
 359                                 continue;
 360                         if (active_ctx == NODE_CTX)
 361                                 handle_node();
 362                         if (active_ctx == REV_CTX)
 363                                 begin_revision();
 364                         if (active_ctx != DUMP_CTX)
 365                                 end_revision();
 366                         active_ctx = REV_CTX;
 367                         reset_rev_ctx(atoi(val));
 368                         break;
 369                 case sizeof("Node-path"):
 370                         if (constcmp(t, "Node-"))
 371                                 continue;
 372                         if (!constcmp(t + strlen("Node-"), "path")) {
 373                                 if (active_ctx == NODE_CTX)
 374                                         handle_node();
 375                                 if (active_ctx == REV_CTX)
 376                                         begin_revision();
 377                                 active_ctx = NODE_CTX;
 378                                 reset_node_ctx(val);
 379                                 break;
 380                         }
 381                         if (constcmp(t + strlen("Node-"), "kind"))
 382                                 continue;
 383                         if (!strcmp(val, "dir"))
 384                                 node_ctx.type = REPO_MODE_DIR;
 385                         else if (!strcmp(val, "file"))
 386                                 node_ctx.type = REPO_MODE_BLB;
 387                         else
 388                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 389                         break;
 390                 case sizeof("Node-action"):
 391                         if (constcmp(t, "Node-action"))
 392                                 continue;
 393                         if (!strcmp(val, "delete")) {
 394                                 node_ctx.action = NODEACT_DELETE;
 395                         } else if (!strcmp(val, "add")) {
 396                                 node_ctx.action = NODEACT_ADD;
 397                         } else if (!strcmp(val, "change")) {
 398                                 node_ctx.action = NODEACT_CHANGE;
 399                         } else if (!strcmp(val, "replace")) {
 400                                 node_ctx.action = NODEACT_REPLACE;
 401                         } else {
 402                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 403                                 node_ctx.action = NODEACT_UNKNOWN;
 404                         }
 405                         break;
 406                 case sizeof("Node-copyfrom-path"):
 407                         if (constcmp(t, "Node-copyfrom-path"))
 408                                 continue;
 409                         strbuf_reset(&node_ctx.src);
 410                         strbuf_addstr(&node_ctx.src, val);
 411                         break;
 412                 case sizeof("Node-copyfrom-rev"):
 413                         if (constcmp(t, "Node-copyfrom-rev"))
 414                                 continue;
 415                         node_ctx.srcRev = atoi(val);
 416                         break;
 417                 case sizeof("Text-content-length"):
 418                         if (constcmp(t, "Text") && constcmp(t, "Prop"))
 419                                 continue;
 420                         if (constcmp(t + 4, "-content-length"))
 421                                 continue;
 422                         {
 423                                 char *end;
 424                                 uintmax_t len;
 425
 426                                 len = strtoumax(val, &end, 10);
 427                                 if (!isdigit(*val) || *end)
 428                                         die("invalid dump: non-numeric length %s", val);
 429                                 if (len > maximum_signed_value_of_type(off_t))
 430                                         die("unrepresentable length in dump: %s", val);
 431
 432                                 if (*t == 'T')
 433                                         node_ctx.text_length = (off_t) len;
 434                                 else
 435                                         node_ctx.prop_length = (off_t) len;
 436                                 break;
 437                         }
 438                 case sizeof("Text-delta"):
 439                         if (!constcmp(t, "Text-delta")) {
 440                                 node_ctx.text_delta = !strcmp(val, "true");
 441                                 break;
 442                         }
 443                         if (constcmp(t, "Prop-delta"))
 444                                 continue;
 445                         node_ctx.prop_delta = !strcmp(val, "true");
 446                         break;
 447                 case sizeof("Content-length"):
 448                         if (constcmp(t, "Content-length"))
 449                                 continue;
 450                         len = atoi(val);
 451                         t = buffer_read_line(&input);
 452                         if (!t)
 453                                 die_short_read();
 454                         if (*t)
 455                                 die("invalid dump: expected blank line after content length header");
 456                         if (active_ctx == REV_CTX) {
 457                                 read_props();
 458                         } else if (active_ctx == NODE_CTX) {
 459                                 handle_node();
 460                                 active_ctx = INTERNODE_CTX;
 461                         } else {
 462                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 463                                 if (buffer_skip_bytes(&input, len) != len)
 464                                         die_short_read();
 465                         }
 466                 }
 467         }
 468         if (buffer_ferror(&input))
 469                 die_short_read();
 470         if (active_ctx == NODE_CTX)
 471                 handle_node();
 472         if (active_ctx == REV_CTX)
 473                 begin_revision();
 474         if (active_ctx != DUMP_CTX)
 475                 end_revision();
 476 }
 477
 478 int svndump_init(const char *filename)
 479 {
 480         if (buffer_init(&input, filename))
 481                 return error("cannot open %s: %s", filename, strerror(errno));
 482         fast_export_init(REPORT_FILENO);
 483         strbuf_init(&dump_ctx.uuid, 4096);
 484         strbuf_init(&dump_ctx.url, 4096);
 485         strbuf_init(&rev_ctx.log, 4096);
 486         strbuf_init(&rev_ctx.author, 4096);
 487         strbuf_init(&node_ctx.src, 4096);
 488         strbuf_init(&node_ctx.dst, 4096);
 489         reset_dump_ctx(NULL);
 490         reset_rev_ctx(0);
 491         reset_node_ctx(NULL);
 492         return 0;
 493 }
 494
 495 void svndump_deinit(void)
 496 {
 497         fast_export_deinit();
 498         reset_dump_ctx(NULL);
 499         reset_rev_ctx(0);
 500         reset_node_ctx(NULL);
 501         strbuf_release(&rev_ctx.log);
 502         strbuf_release(&node_ctx.src);
 503         strbuf_release(&node_ctx.dst);
 504         if (buffer_deinit(&input))
 505                 fprintf(stderr, "Input error\n");
 506         if (ferror(stdout))
 507                 fprintf(stderr, "Output error\n");
 508 }
 509
 510 void svndump_reset(void)
 511 {
 512         strbuf_release(&dump_ctx.uuid);
 513         strbuf_release(&dump_ctx.url);
 514         strbuf_release(&rev_ctx.log);
 515         strbuf_release(&rev_ctx.author);
 516 }