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
   7 #define _GNU_SOURCE
   8 #include <string.h>
   9 #include <stdio.h>
  10 #include <stdlib.h>
  11 #include <time.h>
  12
  13 #include "mkgmtime.h"
  14
  15 #include "repo_tree.h"
  16 #include "fast_export.h"
  17 #include "line_buffer.h"
  18 #include "obj_pool.h"
  19 #include "string_pool.h"
  20
  21 #define NODEACT_REPLACE 4
  22 #define NODEACT_DELETE 3
  23 #define NODEACT_ADD 2
  24 #define NODEACT_CHANGE 1
  25 #define NODEACT_UNKNOWN 0
  26
  27 #define DUMP_CTX 0
  28 #define REV_CTX  1
  29 #define NODE_CTX 2
  30
  31 #define LENGTH_UNKNOWN (~0)
  32
  33 #define BLOB_MARK_OFFSET 1000000000
  34
  35 /* Create memory pool for log messages */
  36 obj_pool_gen(log, char, 4096);
  37
  38 static char* log_copy(uint32_t length, char *log)
  39 {
  40         char *buffer;
  41         log_free(log_pool.size);
  42         buffer = log_pointer(log_alloc(length));
  43         strncpy(buffer, log, length);
  44         return buffer;
  45 }
  46
  47 static struct {
  48         uint32_t action, propLength, textLength, srcRev, srcMode, mark, type;
  49         uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
  50 } node_ctx;
  51
  52 static struct {
  53         uint32_t revision, author;
  54         time_t timestamp;
  55         char *log;
  56 } rev_ctx;
  57
  58 static struct {
  59         uint32_t uuid, url;
  60 } dump_ctx;
  61
  62 static struct {
  63         uint32_t svn_log, svn_author, svn_date, svn_executable, svn_special, uuid,
  64                 revision_number, node_path, node_kind, node_action,
  65                 node_copyfrom_path, node_copyfrom_rev, text_content_length,
  66                 prop_content_length, content_length;
  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         node_ctx.srcMode = 0;
  78         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
  79         node_ctx.mark = 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.uuid = ~0;
  94 }
  95
  96 static void init_keys(void)
  97 {
  98         keys.svn_log = pool_intern("svn:log");
  99         keys.svn_author = pool_intern("svn:author");
 100         keys.svn_date = pool_intern("svn:date");
 101         keys.svn_executable = pool_intern("svn:executable");
 102         keys.svn_special = pool_intern("svn:special");
 103         keys.uuid = pool_intern("UUID");
 104         keys.revision_number = pool_intern("Revision-number");
 105         keys.node_path = pool_intern("Node-path");
 106         keys.node_kind = pool_intern("Node-kind");
 107         keys.node_action = pool_intern("Node-action");
 108         keys.node_copyfrom_path = pool_intern("Node-copyfrom-path");
 109         keys.node_copyfrom_rev = pool_intern("Node-copyfrom-rev");
 110         keys.text_content_length = pool_intern("Text-content-length");
 111         keys.prop_content_length = pool_intern("Prop-content-length");
 112         keys.content_length = pool_intern("Content-length");
 113 }
 114
 115 static uint32_t next_blob_mark(void)
 116 {
 117         static uint32_t mark = BLOB_MARK_OFFSET;
 118         return mark++;
 119 }
 120
 121 static void read_props(void)
 122 {
 123         struct tm tm;
 124         uint32_t len;
 125         uint32_t key = ~0;
 126         char *val = NULL;
 127         char *t;
 128         while ((t = buffer_read_line()) && strcmp(t, "PROPS-END")) {
 129                 if (!strncmp(t, "K ", 2)) {
 130                         len = atoi(&t[2]);
 131                         key = pool_intern(buffer_read_string(len));
 132                         buffer_read_line();
 133                 } else if (!strncmp(t, "V ", 2)) {
 134                         len = atoi(&t[2]);
 135                         val = buffer_read_string(len);
 136                         if (key == keys.svn_log) {
 137                                 /* Value length excludes terminating nul. */
 138                                 rev_ctx.log = log_copy(len + 1, val);
 139                         } else if (key == keys.svn_author) {
 140                                 rev_ctx.author = pool_intern(val);
 141                         } else if (key == keys.svn_date) {
 142                                 strptime(val, "%FT%T", &tm);
 143                                 rev_ctx.timestamp = mkgmtime(&tm);
 144                         } else if (key == keys.svn_executable) {
 145                                 node_ctx.type = REPO_MODE_EXE;
 146                         } else if (key == keys.svn_special) {
 147                                 node_ctx.type = REPO_MODE_LNK;
 148                         }
 149                         key = ~0;
 150                         buffer_read_line();
 151                 }
 152         }
 153 }
 154
 155 static void handle_node(void)
 156 {
 157         if (node_ctx.propLength != LENGTH_UNKNOWN && node_ctx.propLength) {
 158                 read_props();
 159         }
 160
 161         if (node_ctx.srcRev) {
 162                 node_ctx.srcMode = repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
 163         }
 164
 165         if (node_ctx.textLength != LENGTH_UNKNOWN &&
 166                 node_ctx.type != REPO_MODE_DIR) {
 167                 node_ctx.mark = next_blob_mark();
 168         }
 169
 170         if (node_ctx.action == NODEACT_DELETE) {
 171                 repo_delete(node_ctx.dst);
 172         } else if (node_ctx.action == NODEACT_CHANGE ||
 173                            node_ctx.action == NODEACT_REPLACE) {
 174                 if (node_ctx.action == NODEACT_REPLACE &&
 175                         node_ctx.type == REPO_MODE_DIR) {
 176                         repo_replace(node_ctx.dst, node_ctx.mark);
 177                 } else if (node_ctx.propLength != LENGTH_UNKNOWN ) {
 178                         repo_modify(node_ctx.dst, node_ctx.type, node_ctx.mark);
 179                 } else if (node_ctx.textLength != LENGTH_UNKNOWN) {
 180                         node_ctx.srcMode = repo_replace(node_ctx.dst, node_ctx.mark);
 181                 }
 182         } else if (node_ctx.action == NODEACT_ADD) {
 183                 if (node_ctx.srcRev &&
 184                         node_ctx.propLength == LENGTH_UNKNOWN &&
 185                         node_ctx.textLength != LENGTH_UNKNOWN) {
 186                         node_ctx.srcMode = repo_replace(node_ctx.dst, node_ctx.mark);
 187                 } else if ((node_ctx.type == REPO_MODE_DIR && !node_ctx.srcRev) ||
 188                                    node_ctx.textLength != LENGTH_UNKNOWN){
 189                         repo_add(node_ctx.dst, node_ctx.type, node_ctx.mark);
 190                 }
 191         }
 192
 193         if (node_ctx.propLength == LENGTH_UNKNOWN && node_ctx.srcMode) {
 194                 node_ctx.type = node_ctx.srcMode;
 195         }
 196
 197         if (node_ctx.mark) {
 198                 fast_export_blob(node_ctx.type, node_ctx.mark, node_ctx.textLength);
 199         } else if (node_ctx.textLength != LENGTH_UNKNOWN) {
 200                 buffer_skip_bytes(node_ctx.textLength);
 201         }
 202 }
 203
 204 static void handle_revision(void)
 205 {
 206         repo_commit(rev_ctx.revision, rev_ctx.author, rev_ctx.log, dump_ctx.uuid,
 207                     dump_ctx.url, rev_ctx.timestamp);
 208 }
 209
 210 static void svndump_read(uint32_t url)
 211 {
 212         char *val;
 213         char *t;
 214         uint32_t active_ctx = DUMP_CTX;
 215         uint32_t len;
 216         uint32_t key;
 217
 218         reset_dump_ctx(url);
 219         while ((t = buffer_read_line())) {
 220                 val = strstr(t, ": ");
 221                 if (!val) continue;
 222                 *val++ = '\0';
 223                 *val++ = '\0';
 224                 key = pool_intern(t);
 225
 226                 if(key == keys.uuid) {
 227                         dump_ctx.uuid = pool_intern(val);
 228                 } else if (key == keys.revision_number) {
 229                         if (active_ctx == NODE_CTX) handle_node();
 230                         if (active_ctx != DUMP_CTX) handle_revision();
 231                         active_ctx = REV_CTX;
 232                         reset_rev_ctx(atoi(val));
 233                 } else if (key == keys.node_path) {
 234                         if (active_ctx == NODE_CTX)
 235                                 handle_node();
 236                         active_ctx = NODE_CTX;
 237                         reset_node_ctx(val);
 238                 } else if (key == keys.node_kind) {
 239                         if (!strcmp(val, "dir")) {
 240                                 node_ctx.type = REPO_MODE_DIR;
 241                         } else if (!strcmp(val, "file")) {
 242                                 node_ctx.type = REPO_MODE_BLB;
 243                         } else {
 244                                 fprintf(stderr, "Unknown node-kind: %s\n", val);
 245                         }
 246                 } else if (key == keys.node_action) {
 247                         if (!strcmp(val, "delete")) {
 248                                 node_ctx.action = NODEACT_DELETE;
 249                         } else if (!strcmp(val, "add")) {
 250                                 node_ctx.action = NODEACT_ADD;
 251                         } else if (!strcmp(val, "change")) {
 252                                 node_ctx.action = NODEACT_CHANGE;
 253                         } else if (!strcmp(val, "replace")) {
 254                                 node_ctx.action = NODEACT_REPLACE;
 255                         } else {
 256                                 fprintf(stderr, "Unknown node-action: %s\n", val);
 257                                 node_ctx.action = NODEACT_UNKNOWN;
 258                         }
 259                 } else if (key == keys.node_copyfrom_path) {
 260                         pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.src, "/", val);
 261                 } else if (key == keys.node_copyfrom_rev) {
 262                         node_ctx.srcRev = atoi(val);
 263                 } else if (key == keys.text_content_length) {
 264                         node_ctx.textLength = atoi(val);
 265                 } else if (key == keys.prop_content_length) {
 266                         node_ctx.propLength = atoi(val);
 267                 } else if (key == keys.content_length) {
 268                         len = atoi(val);
 269                         buffer_read_line();
 270                         if (active_ctx == REV_CTX) {
 271                                 read_props();
 272                         } else if (active_ctx == NODE_CTX) {
 273                                 handle_node();
 274                                 active_ctx = REV_CTX;
 275                         } else {
 276                                 fprintf(stderr, "Unexpected content length header: %d\n", len);
 277                                 buffer_skip_bytes(len);
 278                         }
 279                 }
 280         }
 281         if (active_ctx == NODE_CTX) handle_node();
 282         if (active_ctx != DUMP_CTX) handle_revision();
 283 }
 284
 285 static void svndump_reset(void)
 286 {
 287         log_reset();
 288         buffer_reset();
 289         repo_reset();
 290         reset_dump_ctx(~0);
 291         reset_rev_ctx(0);
 292         reset_node_ctx(NULL);
 293 }
 294
 295 int main(int argc, char **argv)
 296 {
 297     init_keys();
 298     svndump_read((argc > 1) ? pool_intern(argv[1]) : ~0);
 299     svndump_reset();
 300     return 0;
 301 }