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 "obj_pool.h"
  15 #include "string_pool.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 /* Create memory pool for log messages */
  31 obj_pool_gen(log, char, 4096)
  32
  33 static struct line_buffer input = LINE_BUFFER_INIT;
  34
  35 static char *log_copy(uint32_t length, const char *log)
  36 {
  37         char *buffer;
  38         log_free(log_pool.size);
  39         buffer = log_pointer(log_alloc(length));
  40         strncpy(buffer, log, length);
  41         return buffer;
  42 }
  43
  44 static struct {
  45         uint32_t action, propLength, textLength, srcRev, type;
  46         uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
  47         uint32_t text_delta, prop_delta;
  48 } node_ctx;
  49
  50 static struct {
  51         uint32_t revision, author;
  52         unsigned long timestamp;
  53         char *log;
  54 } rev_ctx;
  55
  56 static struct {
  57         uint32_t version, uuid, url;
  58 } dump_ctx;
  59
  60 static struct {
  61         uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
  62                 revision_number, node_path, node_kind, node_action,
  63                 node_copyfrom_path, node_copyfrom_rev, text_content_length,
  64                 prop_content_length, content_length, svn_fs_dump_format_version,
  65                 /* version 3 format */
  66                 text_delta, prop_delta;
  67 } keys;
  68
  69 static void reset_node_ctx(char *fname)
  70 {
  71         node_ctx.type = 0;
  72         node_ctx.action = NODEACT_UNKNOWN;
  73         node_ctx.propLength = LENGTH_UNKNOWN;
  74         node_ctx.textLength = LENGTH_UNKNOWN;
  75         node_ctx.src[0] = ~0;
  76         node_ctx.srcRev = 0;
  77         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  78         node_ctx.text_delta = 0;
  79         node_ctx.prop_delta = 0;
  80 }
  81
  82 static void reset_rev_ctx(uint32_t revision)
  83 {
  84         rev_ctx.revision = revision;
  85         rev_ctx.timestamp = 0;
  86         rev_ctx.log = NULL;
  87         rev_ctx.author = ~0;
  88 }
  89
  90 static void reset_dump_ctx(uint32_t url)
  91 {
  92         dump_ctx.url = url;
  93         dump_ctx.version = 1;
  94         dump_ctx.uuid = ~0;
  95 }
  96
  97 static void init_keys(void)
  98 {
  99         keys.svn_log = pool_intern("svn:log");
 100         keys.svn_author = pool_intern("svn:author");
 101         keys.svn_date = pool_intern("svn:date");
 102         keys.svn_executable = pool_intern("svn:executable");
 103         keys.svn_special = pool_intern("svn:special");
 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(uint32_t key, const char *val, uint32_t len,
 121                                 uint32_t *type_set)
 122 {
 123         if (key == keys.svn_log) {
 124                 if (!val)
 125                         die("invalid dump: unsets svn:log");
 126                 /* Value length excludes terminating nul. */
 127                 rev_ctx.log = log_copy(len + 1, val);
 128         } else if (key == keys.svn_author) {
 129                 rev_ctx.author = pool_intern(val);
 130         } else if (key == keys.svn_date) {
 131                 if (!val)
 132                         die("invalid dump: unsets svn:date");
 133                 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
 134                         warning("invalid timestamp: %s", val);
 135         } else if (key == keys.svn_executable || key == keys.svn_special) {
 136                 if (*type_set) {
 137                         if (!val)
 138                                 return;
 139                         die("invalid dump: sets type twice");
 140                 }
 141                 if (!val) {
 142                         node_ctx.type = REPO_MODE_BLB;
 143                         return;
 144                 }
 145                 *type_set = 1;
 146                 node_ctx.type = key == keys.svn_executable ?
 147                                 REPO_MODE_EXE :
 148                                 REPO_MODE_LNK;
 149         }
 150 }
 151
 152 static void die_short_read(void)
 153 {
 154         if (buffer_ferror(&input))
 155                 die_errno("error reading dump file");
 156         die("invalid dump: unexpected end of file");
 157 }
 158
 159 static void read_props(void)
 160 {
 161         uint32_t key = ~0;
 162         const char *t;
 163         /*
 164          * NEEDSWORK: to support simple mode changes like
 165          *      K 11
 166          *      svn:special
 167          *      V 1
 168          *      *
 169          *      D 14
 170          *      svn:executable
 171          * we keep track of whether a mode has been set and reset to
 172          * plain file only if not.  We should be keeping track of the
 173          * symlink and executable bits separately instead.
 174          */
 175         uint32_t type_set = 0;
 176         while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
 177                 uint32_t len;
 178                 const char *val;
 179                 const char type = t[0];
 180                 int ch;
 181
 182                 if (!type || t[1] != ' ')
 183                         die("invalid property line: %s\n", t);
 184                 len = atoi(&t[2]);
 185                 val = buffer_read_string(&input, len);
 186                 if (!val || strlen(val) != len)
 187                         die_short_read();
 188
 189                 /* Discard trailing newline. */
 190                 ch = buffer_read_char(&input);
 191                 if (ch == EOF)
 192                         die_short_read();
 193                 if (ch != '\n')
 194                         die("invalid dump: expected newline after %s", val);
 195
 196                 switch (type) {
 197                 case 'K':
 198                         key = pool_intern(val);
 199                         continue;
 200                 case 'D':
 201                         key = pool_intern(val);
 202                         val = NULL;
 203                         len = 0;
 204                         /* fall through */
 205                 case 'V':
 206                         handle_property(key, val, len, &type_set);
 207                         key = ~0;
 208                         continue;
 209                 default:
 210                         die("invalid property line: %s\n", t);
 211                 }
 212         }
 213 }
 214
 215 static void handle_node(void)
 216 {
 217         uint32_t mark = 0;
 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.textLength != LENGTH_UNKNOWN;
 221
 222         if (node_ctx.text_delta)
 223                 die("text deltas not supported");
 224         if (have_text)
 225                 mark = next_blob_mark();
 226         if (node_ctx.action == NODEACT_DELETE) {
 227                 if (have_text || have_props || node_ctx.srcRev)
 228                         die("invalid dump: deletion node has "
 229                                 "copyfrom info, text, or properties");
 230                 return repo_delete(node_ctx.dst);
 231         }
 232         if (node_ctx.action == NODEACT_REPLACE) {
 233                 repo_delete(node_ctx.dst);
 234                 node_ctx.action = NODEACT_ADD;
 235         }
 236         if (node_ctx.srcRev) {
 237                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 238                 if (node_ctx.action == NODEACT_ADD)
 239                         node_ctx.action = NODEACT_CHANGE;
 240         }
 241         if (have_text && type == REPO_MODE_DIR)
 242                 die("invalid dump: directories cannot have text attached");
 243
 244         /*
 245          * Decide on the new content (mark) and mode (node_ctx.type).
 246          */
 247         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 248                 if (type != REPO_MODE_DIR)
 249                         die("invalid dump: root of tree is not a regular file");
 250         } else if (node_ctx.action == NODEACT_CHANGE) {
 251                 uint32_t mode;
 252                 if (!have_text)
 253                         mark = repo_read_path(node_ctx.dst);
 254                 mode = repo_read_mode(node_ctx.dst);
 255                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 256                         die("invalid dump: cannot modify a directory into a file");
 257                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 258                         die("invalid dump: cannot modify a file into a directory");
 259                 node_ctx.type = mode;
 260         } else if (node_ctx.action == NODEACT_ADD) {
 261                 if (!have_text && type != REPO_MODE_DIR)
 262                         die("invalid dump: adds node without text");
 263         } else {
 264                 die("invalid dump: Node-path block lacks Node-action");
 265         }
 266
 267         /*
 268          * Adjust mode to reflect properties.
 269          */
 270         if (have_props) {
 271                 if (!node_ctx.prop_delta)
 272                         node_ctx.type = type;
 273                 if (node_ctx.propLength)
 274                         read_props();
 275         }
 276
 277         /*
 278          * Save the result.
 279          */
 280         repo_add(node_ctx.dst, node_ctx.type, mark);
 281         if (have_text)
 282                 fast_export_blob(node_ctx.type, mark,
 283                                  node_ctx.textLength, &input);
 284 }
 285
 286 static void handle_revision(void)
 287 {
 288         if (rev_ctx.revision)
 289                 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 290                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 291 }
 292
 293 void svndump_read(const char *url)
 294 {
 295         char *val;
 296         char *t;
 297         uint32_t active_ctx = DUMP_CTX;
 298         uint32_t len;
 299         uint32_t key;
 300
 301         reset_dump_ctx(pool_intern(url));
 302         while ((t = buffer_read_line(&input))) {
 303                 val = strstr(t, ": ");
 304                 if (!val)
 305                         continue;
 306                 *val++ = '\0';
 307                 *val++ = '\0';
 308                 key = pool_intern(t);
 309
 310                 if (key == keys.svn_fs_dump_format_version) {
 311                         dump_ctx.version = atoi(val);
 312                         if (dump_ctx.version > 3)
 313                                 die("expected svn dump format version <= 3, found %"PRIu32,
 314                                     dump_ctx.version);
 315                 } else if (key == keys.uuid) {
 316                         dump_ctx.uuid = pool_intern(val);
 317                 } else if (key == keys.revision_number) {
 318                         if (active_ctx == NODE_CTX)
 319                                 handle_node();
 320                         if (active_ctx != DUMP_CTX)
 321                                 handle_revision();
 322                         active_ctx = REV_CTX;
 323                         reset_rev_ctx(atoi(val));
 324                 } else if (key == keys.node_path) {
 325                         if (active_ctx == NODE_CTX)
 326                                 handle_node();
 327                         active_ctx = NODE_CTX;
 328                         reset_node_ctx(val);
 329                 } else if (key == keys.node_kind) {
 330                         if (!strcmp(val, "dir"))
 331                                 node_ctx.type = REPO_MODE_DIR;
 332                         else if (!strcmp(val, "file"))
 333                                 node_ctx.type = REPO_MODE_BLB;
 334                         else
 335                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 336                 } else if (key == keys.node_action) {
 337                         if (!strcmp(val, "delete")) {
 338                                 node_ctx.action = NODEACT_DELETE;
 339                         } else if (!strcmp(val, "add")) {
 340                                 node_ctx.action = NODEACT_ADD;
 341                         } else if (!strcmp(val, "change")) {
 342                                 node_ctx.action = NODEACT_CHANGE;
 343                         } else if (!strcmp(val, "replace")) {
 344                                 node_ctx.action = NODEACT_REPLACE;
 345                         } else {
 346                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 347                                 node_ctx.action = NODEACT_UNKNOWN;
 348                         }
 349                 } else if (key == keys.node_copyfrom_path) {
 350                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 351                 } else if (key == keys.node_copyfrom_rev) {
 352                         node_ctx.srcRev = atoi(val);
 353                 } else if (key == keys.text_content_length) {
 354                         node_ctx.textLength = atoi(val);
 355                 } else if (key == keys.prop_content_length) {
 356                         node_ctx.propLength = atoi(val);
 357                 } else if (key == keys.text_delta) {
 358                         node_ctx.text_delta = !strcmp(val, "true");
 359                 } else if (key == keys.prop_delta) {
 360                         node_ctx.prop_delta = !strcmp(val, "true");
 361                 } else if (key == keys.content_length) {
 362                         len = atoi(val);
 363                         t = buffer_read_line(&input);
 364                         if (!t)
 365                                 die_short_read();
 366                         if (*t)
 367                                 die("invalid dump: expected blank line after content length header");
 368                         if (active_ctx == REV_CTX) {
 369                                 read_props();
 370                         } else if (active_ctx == NODE_CTX) {
 371                                 handle_node();
 372                                 active_ctx = REV_CTX;
 373                         } else {
 374                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 375                                 if (buffer_skip_bytes(&input, len) != len)
 376                                         die_short_read();
 377                         }
 378                 }
 379         }
 380         if (buffer_ferror(&input))
 381                 die_short_read();
 382         if (active_ctx == NODE_CTX)
 383                 handle_node();
 384         if (active_ctx != DUMP_CTX)
 385                 handle_revision();
 386 }
 387
 388 int svndump_init(const char *filename)
 389 {
 390         if (buffer_init(&input, filename))
 391                 return error("cannot open %s: %s", filename, strerror(errno));
 392         repo_init();
 393         reset_dump_ctx(~0);
 394         reset_rev_ctx(0);
 395         reset_node_ctx(NULL);
 396         init_keys();
 397         return 0;
 398 }
 399
 400 void svndump_deinit(void)
 401 {
 402         log_reset();
 403         repo_reset();
 404         reset_dump_ctx(~0);
 405         reset_rev_ctx(0);
 406         reset_node_ctx(NULL);
 407         if (buffer_deinit(&input))
 408                 fprintf(stderr, "Input error\n");
 409         if (ferror(stdout))
 410                 fprintf(stderr, "Output error\n");
 411 }
 412
 413 void svndump_reset(void)
 414 {
 415         log_reset();
 416         buffer_reset(&input);
 417         repo_reset();
 418         reset_dump_ctx(~0);
 419         reset_rev_ctx(0);
 420         reset_node_ctx(NULL);
 421 }