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