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, 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 read_props(void)
 119 {
 120         uint32_t len;
 121         uint32_t key = ~0;
 122         char *val = NULL;
 123         char *t;
 124         while ((t = buffer_read_line()) && strcmp(t, "PROPS-END")) {
 125                 if (!strncmp(t, "K ", 2)) {
 126                         len = atoi(&t[2]);
 127                         key = pool_intern(buffer_read_string(len));
 128                         buffer_read_line();
 129                 } else if (!strncmp(t, "V ", 2)) {
 130                         len = atoi(&t[2]);
 131                         val = buffer_read_string(len);
 132                         if (key == keys.svn_log) {
 133                                 /* Value length excludes terminating nul. */
 134                                 rev_ctx.log = log_copy(len + 1, val);
 135                         } else if (key == keys.svn_author) {
 136                                 rev_ctx.author = pool_intern(val);
 137                         } else if (key == keys.svn_date) {
 138                                 if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
 139                                         fprintf(stderr, "Invalid timestamp: %s\n", val);
 140                         } else if (key == keys.svn_executable) {
 141                                 node_ctx.type = REPO_MODE_EXE;
 142                         } else if (key == keys.svn_special) {
 143                                 node_ctx.type = REPO_MODE_LNK;
 144                         }
 145                         key = ~0;
 146                         buffer_read_line();
 147                 }
 148         }
 149 }
 150
 151 static void handle_node(void)
 152 {
 153         uint32_t mark = 0;
 154         const uint32_t type = node_ctx.type;
 155         const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
 156
 157         if (node_ctx.text_delta || node_ctx.prop_delta)
 158                 die("text and property deltas not supported");
 159
 160         if (node_ctx.textLength != LENGTH_UNKNOWN)
 161                 mark = next_blob_mark();
 162
 163         if (node_ctx.action == NODEACT_DELETE) {
 164                 if (mark || have_props || node_ctx.srcRev)
 165                         die("invalid dump: deletion node has "
 166                                 "copyfrom info, text, or properties");
 167                 return repo_delete(node_ctx.dst);
 168         }
 169
 170         if (node_ctx.action == NODEACT_REPLACE) {
 171                 repo_delete(node_ctx.dst);
 172                 node_ctx.action = NODEACT_ADD;
 173         }
 174
 175         if (node_ctx.srcRev) {
 176                 repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 177                 if (node_ctx.action == NODEACT_ADD)
 178                         node_ctx.action = NODEACT_CHANGE;
 179         }
 180
 181         if (mark && type == REPO_MODE_DIR)
 182                 die("invalid dump: directories cannot have text attached");
 183
 184         if (node_ctx.action == NODEACT_CHANGE) {
 185                 uint32_t mode = repo_modify_path(node_ctx.dst, 0, mark);
 186                 if (!mode)
 187                         die("invalid dump: path to be modified is missing");
 188                 if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
 189                         die("invalid dump: cannot modify a directory into a file");
 190                 if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
 191                         die("invalid dump: cannot modify a file into a directory");
 192                 node_ctx.type = mode;
 193         } else if (node_ctx.action == NODEACT_ADD) {
 194                 if (!mark && type != REPO_MODE_DIR)
 195                         die("invalid dump: adds node without text");
 196                 repo_add(node_ctx.dst, type, mark);
 197         } else {
 198                 die("invalid dump: Node-path block lacks Node-action");
 199         }
 200
 201         if (have_props) {
 202                 const uint32_t old_mode = node_ctx.type;
 203                 node_ctx.type = type;
 204                 if (node_ctx.propLength)
 205                         read_props();
 206                 if (node_ctx.type != old_mode)
 207                         repo_modify_path(node_ctx.dst, node_ctx.type, mark);
 208         }
 209
 210         if (mark)
 211                 fast_export_blob(node_ctx.type, mark, node_ctx.textLength);
 212 }
 213
 214 static void handle_revision(void)
 215 {
 216         if (rev_ctx.revision)
 217                 repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log,
 218                         dump_ctx.uuid, dump_ctx.url, rev_ctx.timestamp);
 219 }
 220
 221 void svndump_read(const char *url)
 222 {
 223         char *val;
 224         char *t;
 225         uint32_t active_ctx = DUMP_CTX;
 226         uint32_t len;
 227         uint32_t key;
 228
 229         reset_dump_ctx(pool_intern(url));
 230         while ((t = buffer_read_line())) {
 231                 val = strstr(t, ": ");
 232                 if (!val)
 233                         continue;
 234                 *val++ = '\0';
 235                 *val++ = '\0';
 236                 key = pool_intern(t);
 237
 238                 if (key == keys.svn_fs_dump_format_version) {
 239                         dump_ctx.version = atoi(val);
 240                         if (dump_ctx.version > 3)
 241                                 die("expected svn dump format version <= 3, found %d",
 242                                     dump_ctx.version);
 243                 } else if (key == keys.uuid) {
 244                         dump_ctx.uuid = pool_intern(val);
 245                 } else if (key == keys.revision_number) {
 246                         if (active_ctx == NODE_CTX)
 247                                 handle_node();
 248                         if (active_ctx != DUMP_CTX)
 249                                 handle_revision();
 250                         active_ctx = REV_CTX;
 251                         reset_rev_ctx(atoi(val));
 252                 } else if (key == keys.node_path) {
 253                         if (active_ctx == NODE_CTX)
 254                                 handle_node();
 255                         active_ctx = NODE_CTX;
 256                         reset_node_ctx(val);
 257                 } else if (key == keys.node_kind) {
 258                         if (!strcmp(val, "dir"))
 259                                 node_ctx.type = REPO_MODE_DIR;
 260                         else if (!strcmp(val, "file"))
 261                                 node_ctx.type = REPO_MODE_BLB;
 262                         else
 263                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 264                 } else if (key == keys.node_action) {
 265                         if (!strcmp(val, "delete")) {
 266                                 node_ctx.action = NODEACT_DELETE;
 267                         } else if (!strcmp(val, "add")) {
 268                                 node_ctx.action = NODEACT_ADD;
 269                         } else if (!strcmp(val, "change")) {
 270                                 node_ctx.action = NODEACT_CHANGE;
 271                         } else if (!strcmp(val, "replace")) {
 272                                 node_ctx.action = NODEACT_REPLACE;
 273                         } else {
 274                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 275                                 node_ctx.action = NODEACT_UNKNOWN;
 276                         }
 277                 } else if (key == keys.node_copyfrom_path) {
 278                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 279                 } else if (key == keys.node_copyfrom_rev) {
 280                         node_ctx.srcRev = atoi(val);
 281                 } else if (key == keys.text_content_length) {
 282                         node_ctx.textLength = atoi(val);
 283                 } else if (key == keys.prop_content_length) {
 284                         node_ctx.propLength = atoi(val);
 285                 } else if (key == keys.text_delta) {
 286                         node_ctx.text_delta = !strcmp(val, "true");
 287                 } else if (key == keys.prop_delta) {
 288                         node_ctx.prop_delta = !strcmp(val, "true");
 289                 } else if (key == keys.content_length) {
 290                         len = atoi(val);
 291                         buffer_read_line();
 292                         if (active_ctx == REV_CTX) {
 293                                 read_props();
 294                         } else if (active_ctx == NODE_CTX) {
 295                                 handle_node();
 296                                 active_ctx = REV_CTX;
 297                         } else {
 298                                 fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
 299                                 buffer_skip_bytes(len);
 300                         }
 301                 }
 302         }
 303         if (active_ctx == NODE_CTX)
 304                 handle_node();
 305         if (active_ctx != DUMP_CTX)
 306                 handle_revision();
 307 }
 308
 309 int svndump_init(const char *filename)
 310 {
 311         if (buffer_init(filename))
 312                 return error("cannot open %s: %s", filename, strerror(errno));
 313         repo_init();
 314         reset_dump_ctx(~0);
 315         reset_rev_ctx(0);
 316         reset_node_ctx(NULL);
 317         init_keys();
 318         return 0;
 319 }
 320
 321 void svndump_deinit(void)
 322 {
 323         log_reset();
 324         repo_reset();
 325         reset_dump_ctx(~0);
 326         reset_rev_ctx(0);
 327         reset_node_ctx(NULL);
 328         if (buffer_deinit())
 329                 fprintf(stderr, "Input error\n");
 330         if (ferror(stdout))
 331                 fprintf(stderr, "Output error\n");
 332 }
 333
 334 void svndump_reset(void)
 335 {
 336         log_reset();
 337         buffer_reset();
 338         repo_reset();
 339         reset_dump_ctx(~0);
 340         reset_rev_ctx(0);
 341         reset_node_ctx(NULL);
 342 }