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 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
 174                 if (!type || t[1] != ' ')
 175                         die("invalid property line: %s\n", t);
 176                 len = atoi(&t[2]);
 177                 val = buffer_read_string(&input, len);
 178                 buffer_skip_bytes(&input, 1);   /* Discard trailing newline. */
 179
 180                 switch (type) {
 181                 case 'K':
 182                         key = pool_intern(val);
 183                         continue;
 184                 case 'D':
 185                         key = pool_intern(val);
 186                         val = NULL;
 187                         len = 0;
 188                         /* fall through */
 189                 case 'V':
 190                         handle_property(key, val, len, &type_set);
 191                         key = ~0;
 192                         continue;
 193                 default:
 194                         die("invalid property line: %s\n", t);
 195                 }
 196         }
 197 }
 198
 199 static void handle_node(void)
 200 {
 201         uint32_t mark = 0;
 202         const uint32_t type = node_ctx.type;
 203         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 204         const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 205
 206         if (node_ctx.text_delta)
 207                 die("text deltas not supported");
 208         if (have_text)
 209                 mark = next_blob_mark();
 210         if (node_ctx.action == NODEACT_DELETE) {
 211                 if (have_text || have_props || node_ctx.srcRev)
 212                         die("invalid dump: deletion node has "
 213                                 "copyfrom info, text, or properties");
 214                 return repo_delete(node_ctx.dst);
 215         }
 216         if (node_ctx.action == NODEACT_REPLACE) {
 217                 repo_delete(node_ctx.dst);
 218                 node_ctx.action = NODEACT_ADD;
 219         }
 220         if (node_ctx.srcRev) {
 221                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 222                 if (node_ctx.action == NODEACT_ADD)
 223                         node_ctx.action = NODEACT_CHANGE;
 224         }
 225         if (have_text && type == REPO_MODE_DIR)
 226                 die("invalid dump: directories cannot have text attached");
 227
 228         /*
 229          * Decide on the new content (mark) and mode (node_ctx.type).
 230          */
 231         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 232                 if (type != REPO_MODE_DIR)
 233                         die("invalid dump: root of tree is not a regular file");
 234         } else if (node_ctx.action == NODEACT_CHANGE) {
 235                 uint32_t mode;
 236                 if (!have_text)
 237                         mark = repo_read_path(node_ctx.dst);
 238                 mode = repo_read_mode(node_ctx.dst);
 239                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 240                         die("invalid dump: cannot modify a directory into a file");
 241                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 242                         die("invalid dump: cannot modify a file into a directory");
 243                 node_ctx.type = mode;
 244         } else if (node_ctx.action == NODEACT_ADD) {
 245                 if (!have_text && type != REPO_MODE_DIR)
 246                         die("invalid dump: adds node without text");
 247         } else {
 248                 die("invalid dump: Node-path block lacks Node-action");
 249         }
 250
 251         /*
 252          * Adjust mode to reflect properties.
 253          */
 254         if (have_props) {
 255                 if (!node_ctx.prop_delta)
 256                         node_ctx.type = type;
 257                 if (node_ctx.propLength)
 258                         read_props();
 259         }
 260
 261         /*
 262          * Save the result.
 263          */
 264         repo_add(node_ctx.dst, node_ctx.type, mark);
 265         if (have_text)
 266                 fast_export_blob(node_ctx.type, mark,
 267                                  node_ctx.textLength, &input);
 268 }
 269
 270 static void handle_revision(void)
 271 {
 272         if (rev_ctx.revision)
 273                 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 274                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 275 }
 276
 277 void svndump_read(const char *url)
 278 {
 279         char *val;
 280         char *t;
 281         uint32_t active_ctx = DUMP_CTX;
 282         uint32_t len;
 283         uint32_t key;
 284
 285         reset_dump_ctx(pool_intern(url));
 286         while ((t = buffer_read_line(&input))) {
 287                 val = strstr(t, ": ");
 288                 if (!val)
 289                         continue;
 290                 *val++ = '\0';
 291                 *val++ = '\0';
 292                 key = pool_intern(t);
 293
 294                 if (key == keys.svn_fs_dump_format_version) {
 295                         dump_ctx.version = atoi(val);
 296                         if (dump_ctx.version > 3)
 297                                 die("expected svn dump format version <= 3, found %"PRIu32,
 298                                     dump_ctx.version);
 299                 } else if (key == keys.uuid) {
 300                         dump_ctx.uuid = pool_intern(val);
 301                 } else if (key == keys.revision_number) {
 302                         if (active_ctx == NODE_CTX)
 303                                 handle_node();
 304                         if (active_ctx != DUMP_CTX)
 305                                 handle_revision();
 306                         active_ctx = REV_CTX;
 307                         reset_rev_ctx(atoi(val));
 308                 } else if (key == keys.node_path) {
 309                         if (active_ctx == NODE_CTX)
 310                                 handle_node();
 311                         active_ctx = NODE_CTX;
 312                         reset_node_ctx(val);
 313                 } else if (key == keys.node_kind) {
 314                         if (!strcmp(val, "dir"))
 315                                 node_ctx.type = REPO_MODE_DIR;
 316                         else if (!strcmp(val, "file"))
 317                                 node_ctx.type = REPO_MODE_BLB;
 318                         else
 319                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 320                 } else if (key == keys.node_action) {
 321                         if (!strcmp(val, "delete")) {
 322                                 node_ctx.action = NODEACT_DELETE;
 323                         } else if (!strcmp(val, "add")) {
 324                                 node_ctx.action = NODEACT_ADD;
 325                         } else if (!strcmp(val, "change")) {
 326                                 node_ctx.action = NODEACT_CHANGE;
 327                         } else if (!strcmp(val, "replace")) {
 328                                 node_ctx.action = NODEACT_REPLACE;
 329                         } else {
 330                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 331                                 node_ctx.action = NODEACT_UNKNOWN;
 332                         }
 333                 } else if (key == keys.node_copyfrom_path) {
 334                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 335                 } else if (key == keys.node_copyfrom_rev) {
 336                         node_ctx.srcRev = atoi(val);
 337                 } else if (key == keys.text_content_length) {
 338                         node_ctx.textLength = atoi(val);
 339                 } else if (key == keys.prop_content_length) {
 340                         node_ctx.propLength = atoi(val);
 341                 } else if (key == keys.text_delta) {
 342                         node_ctx.text_delta = !strcmp(val, "true");
 343                 } else if (key == keys.prop_delta) {
 344                         node_ctx.prop_delta = !strcmp(val, "true");
 345                 } else if (key == keys.content_length) {
 346                         len = atoi(val);
 347                         buffer_read_line(&input);
 348                         if (active_ctx == REV_CTX) {
 349                                 read_props();
 350                         } else if (active_ctx == NODE_CTX) {
 351                                 handle_node();
 352                                 active_ctx = REV_CTX;
 353                         } else {
 354                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 355                                 buffer_skip_bytes(&input, len);
 356                         }
 357                 }
 358         }
 359         if (active_ctx == NODE_CTX)
 360                 handle_node();
 361         if (active_ctx != DUMP_CTX)
 362                 handle_revision();
 363 }
 364
 365 int svndump_init(const char *filename)
 366 {
 367         if (buffer_init(&input, filename))
 368                 return error("cannot open %s: %s", filename, strerror(errno));
 369         repo_init();
 370         reset_dump_ctx(~0);
 371         reset_rev_ctx(0);
 372         reset_node_ctx(NULL);
 373         init_keys();
 374         return 0;
 375 }
 376
 377 void svndump_deinit(void)
 378 {
 379         log_reset();
 380         repo_reset();
 381         reset_dump_ctx(~0);
 382         reset_rev_ctx(0);
 383         reset_node_ctx(NULL);
 384         if (buffer_deinit(&input))
 385                 fprintf(stderr, "Input error\n");
 386         if (ferror(stdout))
 387                 fprintf(stderr, "Output error\n");
 388 }
 389
 390 void svndump_reset(void)
 391 {
 392         log_reset();
 393         buffer_reset(&input);
 394         repo_reset();
 395         reset_dump_ctx(~0);
 396         reset_rev_ctx(0);
 397         reset_node_ctx(NULL);
 398 }