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, author;
  40         unsigned long timestamp;
  41         struct strbuf log;
  42 } rev_ctx;
  43
  44 static struct {
  45         uint32_t version, uuid, url;
  46 } dump_ctx;
  47
  48 static struct {
  49         uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
  50                 revision_number, node_path, node_kind, node_action,
  51                 node_copyfrom_path, node_copyfrom_rev, text_content_length,
  52                 prop_content_length, content_length, svn_fs_dump_format_version,
  53                 /* version 3 format */
  54                 text_delta, prop_delta;
  55 } keys;
  56
  57 static void reset_node_ctx(char *fname)
  58 {
  59         node_ctx.type = 0;
  60         node_ctx.action = NODEACT_UNKNOWN;
  61         node_ctx.propLength = LENGTH_UNKNOWN;
  62         node_ctx.textLength = LENGTH_UNKNOWN;
  63         node_ctx.src[0] = ~0;
  64         node_ctx.srcRev = 0;
  65         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  66         node_ctx.text_delta = 0;
  67         node_ctx.prop_delta = 0;
  68 }
  69
  70 static void reset_rev_ctx(uint32_t revision)
  71 {
  72         rev_ctx.revision = revision;
  73         rev_ctx.timestamp = 0;
  74         strbuf_reset(&rev_ctx.log);
  75         rev_ctx.author = ~0;
  76 }
  77
  78 static void reset_dump_ctx(uint32_t url)
  79 {
  80         dump_ctx.url = url;
  81         dump_ctx.version = 1;
  82         dump_ctx.uuid = ~0;
  83 }
  84
  85 static void init_keys(void)
  86 {
  87         keys.svn_log = pool_intern("svn:log");
  88         keys.svn_author = pool_intern("svn:author");
  89         keys.svn_date = pool_intern("svn:date");
  90         keys.svn_executable = pool_intern("svn:executable");
  91         keys.svn_special = pool_intern("svn:special");
  92         keys.uuid = pool_intern("UUID");
  93         keys.revision_number = pool_intern("Revision-number");
  94         keys.node_path = pool_intern("Node-path");
  95         keys.node_kind = pool_intern("Node-kind");
  96         keys.node_action = pool_intern("Node-action");
  97         keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
  98         keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
  99         keys.text_content_length = pool_intern("Text-content-length");
 100         keys.prop_content_length = pool_intern("Prop-content-length");
 101         keys.content_length = pool_intern("Content-length");
 102         keys.svn_fs_dump_format_version = pool_intern("SVN-fs-dump-format-version");
 103         /* version 3 format (Subversion 1.1.0) */
 104         keys.text_delta = pool_intern("Text-delta");
 105         keys.prop_delta = pool_intern("Prop-delta");
 106 }
 107
 108 static void handle_property(uint32_t key, const char *val, uint32_t len,
 109                                 uint32_t *type_set)
 110 {
 111         if (key == keys.svn_log) {
 112                 if (!val)
 113                         die("invalid dump: unsets svn:log");
 114                 strbuf_reset(&rev_ctx.log);
 115                 strbuf_add(&rev_ctx.log, val, len);
 116         } else if (key == keys.svn_author) {
 117                 rev_ctx.author = pool_intern(val);
 118         } else if (key == keys.svn_date) {
 119                 if (!val)
 120                         die("invalid dump: unsets svn:date");
 121                 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
 122                         warning("invalid timestamp: %s", val);
 123         } else if (key == keys.svn_executable || key == keys.svn_special) {
 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 = key == keys.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         uint32_t key = ~0;
 150         const char *t;
 151         /*
 152          * NEEDSWORK: to support simple mode changes like
 153          *      K 11
 154          *      svn:special
 155          *      V 1
 156          *      *
 157          *      D 14
 158          *      svn:executable
 159          * we keep track of whether a mode has been set and reset to
 160          * plain file only if not.  We should be keeping track of the
 161          * symlink and executable bits separately instead.
 162          */
 163         uint32_t type_set = 0;
 164         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 165                 uint32_t len;
 166                 const char *val;
 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                 val = buffer_read_string(&input, len);
 174                 if (!val || strlen(val) != len)
 175                         die_short_read();
 176
 177                 /* Discard trailing newline. */
 178                 ch = buffer_read_char(&input);
 179                 if (ch == EOF)
 180                         die_short_read();
 181                 if (ch != '\n')
 182                         die("invalid dump: expected newline after %s", val);
 183
 184                 switch (type) {
 185                 case 'K':
 186                         key = pool_intern(val);
 187                         continue;
 188                 case 'D':
 189                         key = pool_intern(val);
 190                         val = NULL;
 191                         len = 0;
 192                         /* fall through */
 193                 case 'V':
 194                         handle_property(key, val, len, &type_set);
 195                         key = ~0;
 196                         continue;
 197                 default:
 198                         die("invalid property line: %s\n", t);
 199                 }
 200         }
 201 }
 202
 203 static void handle_node(void)
 204 {
 205         uint32_t mark = 0;
 206         const uint32_t type = node_ctx.type;
 207         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 208         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 209
 210         if (node_ctx.text_delta)
 211                 die("text deltas not supported");
 212         if (have_text)
 213                 mark = next_blob_mark();
 214         if (node_ctx.action == NODEACT_DELETE) {
 215                 if (have_text || have_props || node_ctx.srcRev)
 216                         die("invalid dump: deletion node has "
 217                                 "copyfrom info, text, or properties");
 218                 return repo_delete(node_ctx.dst);
 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, rev_ctx.log.buf,
 278                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 279 }
 280
 281 void svndump_read(const char *url)
 282 {
 283         char *val;
 284         char *t;
 285         uint32_t active_ctx = DUMP_CTX;
 286         uint32_t len;
 287         uint32_t key;
 288
 289         reset_dump_ctx(pool_intern(url));
 290         while ((t = buffer_read_line(&input))) {
 291                 val = strstr(t, ": ");
 292                 if (!val)
 293                         continue;
 294                 *val++ = '\0';
 295                 *val++ = '\0';
 296                 key = pool_intern(t);
 297
 298                 if (key == keys.svn_fs_dump_format_version) {
 299                         dump_ctx.version = atoi(val);
 300                         if (dump_ctx.version > 3)
 301                                 die("expected svn dump format version <= 3, found %"PRIu32,
 302                                     dump_ctx.version);
 303                 } else if (key == keys.uuid) {
 304                         dump_ctx.uuid = pool_intern(val);
 305                 } else if (key == keys.revision_number) {
 306                         if (active_ctx == NODE_CTX)
 307                                 handle_node();
 308                         if (active_ctx != DUMP_CTX)
 309                                 handle_revision();
 310                         active_ctx = REV_CTX;
 311                         reset_rev_ctx(atoi(val));
 312                 } else if (key == keys.node_path) {
 313                         if (active_ctx == NODE_CTX)
 314                                 handle_node();
 315                         active_ctx = NODE_CTX;
 316                         reset_node_ctx(val);
 317                 } else if (key == keys.node_kind) {
 318                         if (!strcmp(val, "dir"))
 319                                 node_ctx.type = REPO_MODE_DIR;
 320                         else if (!strcmp(val, "file"))
 321                                 node_ctx.type = REPO_MODE_BLB;
 322                         else
 323                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 324                 } else if (key == keys.node_action) {
 325                         if (!strcmp(val, "delete")) {
 326                                 node_ctx.action = NODEACT_DELETE;
 327                         } else if (!strcmp(val, "add")) {
 328                                 node_ctx.action = NODEACT_ADD;
 329                         } else if (!strcmp(val, "change")) {
 330                                 node_ctx.action = NODEACT_CHANGE;
 331                         } else if (!strcmp(val, "replace")) {
 332                                 node_ctx.action = NODEACT_REPLACE;
 333                         } else {
 334                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 335                                 node_ctx.action = NODEACT_UNKNOWN;
 336                         }
 337                 } else if (key == keys.node_copyfrom_path) {
 338                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 339                 } else if (key == keys.node_copyfrom_rev) {
 340                         node_ctx.srcRev = atoi(val);
 341                 } else if (key == keys.text_content_length) {
 342                         node_ctx.textLength = atoi(val);
 343                 } else if (key == keys.prop_content_length) {
 344                         node_ctx.propLength = atoi(val);
 345                 } else if (key == keys.text_delta) {
 346                         node_ctx.text_delta = !strcmp(val, "true");
 347                 } else if (key == keys.prop_delta) {
 348                         node_ctx.prop_delta = !strcmp(val, "true");
 349                 } else if (key == keys.content_length) {
 350                         len = atoi(val);
 351                         t = buffer_read_line(&input);
 352                         if (!t)
 353                                 die_short_read();
 354                         if (*t)
 355                                 die("invalid dump: expected blank line after content length header");
 356                         if (active_ctx == REV_CTX) {
 357                                 read_props();
 358                         } else if (active_ctx == NODE_CTX) {
 359                                 handle_node();
 360                                 active_ctx = REV_CTX;
 361                         } else {
 362                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 363                                 if (buffer_skip_bytes(&input, len) != len)
 364                                         die_short_read();
 365                         }
 366                 }
 367         }
 368         if (buffer_ferror(&input))
 369                 die_short_read();
 370         if (active_ctx == NODE_CTX)
 371                 handle_node();
 372         if (active_ctx != DUMP_CTX)
 373                 handle_revision();
 374 }
 375
 376 int svndump_init(const char *filename)
 377 {
 378         if (buffer_init(&input, filename))
 379                 return error("cannot open %s: %s", filename, strerror(errno));
 380         repo_init();
 381         strbuf_init(&rev_ctx.log, 4096);
 382         reset_dump_ctx(~0);
 383         reset_rev_ctx(0);
 384         reset_node_ctx(NULL);
 385         init_keys();
 386         return 0;
 387 }
 388
 389 void svndump_deinit(void)
 390 {
 391         repo_reset();
 392         reset_dump_ctx(~0);
 393         reset_rev_ctx(0);
 394         reset_node_ctx(NULL);
 395         strbuf_release(&rev_ctx.log);
 396         if (buffer_deinit(&input))
 397                 fprintf(stderr, "Input error\n");
 398         if (ferror(stdout))
 399                 fprintf(stderr, "Output error\n");
 400 }
 401
 402 void svndump_reset(void)
 403 {
 404         buffer_reset(&input);
 405         repo_reset();
 406         reset_dump_ctx(~0);
 407         reset_rev_ctx(0);
 408         reset_node_ctx(NULL);
 409 }