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