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 char *log_copy(uint32_t length, const char *log)
  34 {
  35         char *buffer;
  36         log_free(log_pool.size);
  37         buffer = log_pointer(log_alloc(length));
  38         strncpy(buffer, log, length);
  39         return buffer;
  40 }
  41
  42 static struct {
  43         uint32_t action, propLength, textLength, srcRev, type;
  44         uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
  45         uint32_t text_delta, prop_delta;
  46 } node_ctx;
  47
  48 static struct {
  49         uint32_t revision, author;
  50         unsigned long timestamp;
  51         char *log;
  52 } rev_ctx;
  53
  54 static struct {
  55         uint32_t version, uuid, url;
  56 } dump_ctx;
  57
  58 static struct {
  59         uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
  60                 revision_number, node_path, node_kind, node_action,
  61                 node_copyfrom_path, node_copyfrom_rev, text_content_length,
  62                 prop_content_length, content_length, svn_fs_dump_format_version,
  63                 /* version 3 format */
  64                 text_delta, prop_delta;
  65 } keys;
  66
  67 static void reset_node_ctx(char *fname)
  68 {
  69         node_ctx.type = 0;
  70         node_ctx.action = NODEACT_UNKNOWN;
  71         node_ctx.propLength = LENGTH_UNKNOWN;
  72         node_ctx.textLength = LENGTH_UNKNOWN;
  73         node_ctx.src[0] = ~0;
  74         node_ctx.srcRev = 0;
  75         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  76         node_ctx.text_delta = 0;
  77         node_ctx.prop_delta = 0;
  78 }
  79
  80 static void reset_rev_ctx(uint32_t revision)
  81 {
  82         rev_ctx.revision = revision;
  83         rev_ctx.timestamp = 0;
  84         rev_ctx.log = NULL;
  85         rev_ctx.author = ~0;
  86 }
  87
  88 static void reset_dump_ctx(uint32_t url)
  89 {
  90         dump_ctx.url = url;
  91         dump_ctx.version = 1;
  92         dump_ctx.uuid = ~0;
  93 }
  94
  95 static void init_keys(void)
  96 {
  97         keys.svn_log = pool_intern("svn:log");
  98         keys.svn_author = pool_intern("svn:author");
  99         keys.svn_date = pool_intern("svn:date");
 100         keys.svn_executable = pool_intern("svn:executable");
 101         keys.svn_special = pool_intern("svn:special");
 102         keys.uuid = pool_intern("UUID");
 103         keys.revision_number = pool_intern("Revision-number");
 104         keys.node_path = pool_intern("Node-path");
 105         keys.node_kind = pool_intern("Node-kind");
 106         keys.node_action = pool_intern("Node-action");
 107         keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
 108         keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
 109         keys.text_content_length = pool_intern("Text-content-length");
 110         keys.prop_content_length = pool_intern("Prop-content-length");
 111         keys.content_length = pool_intern("Content-length");
 112         keys.svn_fs_dump_format_version = pool_intern("SVN-fs-dump-format-version");
 113         /* version 3 format (Subversion 1.1.0) */
 114         keys.text_delta = pool_intern("Text-delta");
 115         keys.prop_delta = pool_intern("Prop-delta");
 116 }
 117
 118 static void handle_property(uint32_t key, const char *val, uint32_t len,
 119                                 uint32_t *type_set)
 120 {
 121         if (key == keys.svn_log) {
 122                 if (!val)
 123                         die("invalid dump: unsets svn:log");
 124                 /* Value length excludes terminating nul. */
 125                 rev_ctx.log = log_copy(len + 1, val);
 126         } else if (key == keys.svn_author) {
 127                 rev_ctx.author = pool_intern(val);
 128         } else if (key == keys.svn_date) {
 129                 if (!val)
 130                         die("invalid dump: unsets svn:date");
 131                 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
 132                         warning("invalid timestamp: %s", val);
 133         } else if (key == keys.svn_executable || key == keys.svn_special) {
 134                 if (*type_set) {
 135                         if (!val)
 136                                 return;
 137                         die("invalid dump: sets type twice");
 138                 }
 139                 if (!val) {
 140                         node_ctx.type = REPO_MODE_BLB;
 141                         return;
 142                 }
 143                 *type_set = 1;
 144                 node_ctx.type = key == keys.svn_executable ?
 145                                 REPO_MODE_EXE :
 146                                 REPO_MODE_LNK;
 147         }
 148 }
 149
 150 static void read_props(void)
 151 {
 152         uint32_t key = ~0;
 153         const char *t;
 154         /*
 155          * NEEDSWORK: to support simple mode changes like
 156          *      K 11
 157          *      svn:special
 158          *      V 1
 159          *      *
 160          *      D 14
 161          *      svn:executable
 162          * we keep track of whether a mode has been set and reset to
 163          * plain file only if not.  We should be keeping track of the
 164          * symlink and executable bits separately instead.
 165          */
 166         uint32_t type_set = 0;
 167         while ((t = buffer_read_line()) && strcmp(t, "PROPS-END")) {
 168                 uint32_t len;
 169                 const char *val;
 170                 const char type = t[0];
 171
 172                 if (!type || t[1] != ' ')
 173                         die("invalid property line: %s\n", t);
 174                 len = atoi(&t[2]);
 175                 val = buffer_read_string(len);
 176                 buffer_skip_bytes(1);   /* Discard trailing newline. */
 177
 178                 switch (type) {
 179                 case 'K':
 180                         key = pool_intern(val);
 181                         continue;
 182                 case 'D':
 183                         key = pool_intern(val);
 184                         val = NULL;
 185                         len = 0;
 186                         /* fall through */
 187                 case 'V':
 188                         handle_property(key, val, len, &type_set);
 189                         key = ~0;
 190                         continue;
 191                 default:
 192                         die("invalid property line: %s\n", t);
 193                 }
 194         }
 195 }
 196
 197 static void handle_node(void)
 198 {
 199         uint32_t mark = 0;
 200         const uint32_t type = node_ctx.type;
 201         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 202
 203         if (node_ctx.text_delta)
 204                 die("text deltas not supported");
 205         if (node_ctx.textLength != LENGTH_UNKNOWN)
 206                 mark = next_blob_mark();
 207         if (node_ctx.action == NODEACT_DELETE) {
 208                 if (mark || have_props || node_ctx.srcRev)
 209                         die("invalid dump: deletion node has "
 210                                 "copyfrom info, text, or properties");
 211                 return repo_delete(node_ctx.dst);
 212         }
 213         if (node_ctx.action == NODEACT_REPLACE) {
 214                 repo_delete(node_ctx.dst);
 215                 node_ctx.action = NODEACT_ADD;
 216         }
 217         if (node_ctx.srcRev) {
 218                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 219                 if (node_ctx.action == NODEACT_ADD)
 220                         node_ctx.action = NODEACT_CHANGE;
 221         }
 222         if (mark && type == REPO_MODE_DIR)
 223                 die("invalid dump: directories cannot have text attached");
 224         if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
 225                 if (type != REPO_MODE_DIR)
 226                         die("invalid dump: root of tree is not a regular file");
 227         } else if (node_ctx.action == NODEACT_CHANGE) {
 228                 uint32_t mode = repo_modify_path(node_ctx.dst, 0, mark);
 229                 if (!mode)
 230                         die("invalid dump: path to be modified is missing");
 231                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 232                         die("invalid dump: cannot modify a directory into a file");
 233                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 234                         die("invalid dump: cannot modify a file into a directory");
 235                 node_ctx.type = mode;
 236         } else if (node_ctx.action == NODEACT_ADD) {
 237                 if (!mark && type != REPO_MODE_DIR)
 238                         die("invalid dump: adds node without text");
 239                 repo_add(node_ctx.dst, type, mark);
 240         } else {
 241                 die("invalid dump: Node-path block lacks Node-action");
 242         }
 243         if (have_props) {
 244                 const uint32_t old_mode = node_ctx.type;
 245                 if (!node_ctx.prop_delta)
 246                         node_ctx.type = type;
 247                 if (node_ctx.propLength)
 248                         read_props();
 249                 if (node_ctx.type != old_mode)
 250                         repo_modify_path(node_ctx.dst, node_ctx.type, mark);
 251         }
 252         if (mark)
 253                 fast_export_blob(node_ctx.type, mark, node_ctx.textLength);
 254 }
 255
 256 static void handle_revision(void)
 257 {
 258         if (rev_ctx.revision)
 259                 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 260                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 261 }
 262
 263 void svndump_read(const char *url)
 264 {
 265         char *val;
 266         char *t;
 267         uint32_t active_ctx = DUMP_CTX;
 268         uint32_t len;
 269         uint32_t key;
 270
 271         reset_dump_ctx(pool_intern(url));
 272         while ((t = buffer_read_line())) {
 273                 val = strstr(t, ": ");
 274                 if (!val)
 275                         continue;
 276                 *val++ = '\0';
 277                 *val++ = '\0';
 278                 key = pool_intern(t);
 279
 280                 if (key == keys.svn_fs_dump_format_version) {
 281                         dump_ctx.version = atoi(val);
 282                         if (dump_ctx.version > 3)
 283                                 die("expected svn dump format version <= 3, found %d",
 284                                     dump_ctx.version);
 285                 } else if (key == keys.uuid) {
 286                         dump_ctx.uuid = pool_intern(val);
 287                 } else if (key == keys.revision_number) {
 288                         if (active_ctx == NODE_CTX)
 289                                 handle_node();
 290                         if (active_ctx != DUMP_CTX)
 291                                 handle_revision();
 292                         active_ctx = REV_CTX;
 293                         reset_rev_ctx(atoi(val));
 294                 } else if (key == keys.node_path) {
 295                         if (active_ctx == NODE_CTX)
 296                                 handle_node();
 297                         active_ctx = NODE_CTX;
 298                         reset_node_ctx(val);
 299                 } else if (key == keys.node_kind) {
 300                         if (!strcmp(val, "dir"))
 301                                 node_ctx.type = REPO_MODE_DIR;
 302                         else if (!strcmp(val, "file"))
 303                                 node_ctx.type = REPO_MODE_BLB;
 304                         else
 305                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 306                 } else if (key == keys.node_action) {
 307                         if (!strcmp(val, "delete")) {
 308                                 node_ctx.action = NODEACT_DELETE;
 309                         } else if (!strcmp(val, "add")) {
 310                                 node_ctx.action = NODEACT_ADD;
 311                         } else if (!strcmp(val, "change")) {
 312                                 node_ctx.action = NODEACT_CHANGE;
 313                         } else if (!strcmp(val, "replace")) {
 314                                 node_ctx.action = NODEACT_REPLACE;
 315                         } else {
 316                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 317                                 node_ctx.action = NODEACT_UNKNOWN;
 318                         }
 319                 } else if (key == keys.node_copyfrom_path) {
 320                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 321                 } else if (key == keys.node_copyfrom_rev) {
 322                         node_ctx.srcRev = atoi(val);
 323                 } else if (key == keys.text_content_length) {
 324                         node_ctx.textLength = atoi(val);
 325                 } else if (key == keys.prop_content_length) {
 326                         node_ctx.propLength = atoi(val);
 327                 } else if (key == keys.text_delta) {
 328                         node_ctx.text_delta = !strcmp(val, "true");
 329                 } else if (key == keys.prop_delta) {
 330                         node_ctx.prop_delta = !strcmp(val, "true");
 331                 } else if (key == keys.content_length) {
 332                         len = atoi(val);
 333                         buffer_read_line();
 334                         if (active_ctx == REV_CTX) {
 335                                 read_props();
 336                         } else if (active_ctx == NODE_CTX) {
 337                                 handle_node();
 338                                 active_ctx = REV_CTX;
 339                         } else {
 340                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 341                                 buffer_skip_bytes(len);
 342                         }
 343                 }
 344         }
 345         if (active_ctx == NODE_CTX)
 346                 handle_node();
 347         if (active_ctx != DUMP_CTX)
 348                 handle_revision();
 349 }
 350
 351 int svndump_init(const char *filename)
 352 {
 353         if (buffer_init(filename))
 354                 return error("cannot open %s: %s", filename, strerror(errno));
 355         repo_init();
 356         reset_dump_ctx(~0);
 357         reset_rev_ctx(0);
 358         reset_node_ctx(NULL);
 359         init_keys();
 360         return 0;
 361 }
 362
 363 void svndump_deinit(void)
 364 {
 365         log_reset();
 366         repo_reset();
 367         reset_dump_ctx(~0);
 368         reset_rev_ctx(0);
 369         reset_node_ctx(NULL);
 370         if (buffer_deinit())
 371                 fprintf(stderr, "Input error\n");
 372         if (ferror(stdout))
 373                 fprintf(stderr, "Output error\n");
 374 }
 375
 376 void svndump_reset(void)
 377 {
 378         log_reset();
 379         buffer_reset();
 380         repo_reset();
 381         reset_dump_ctx(~0);
 382         reset_rev_ctx(0);
 383         reset_node_ctx(NULL);
 384 }