source-filter_dest-filter.diff

   1 CAUTION:  This patch compiles, but is otherwise totally untested!
   2
   3 This patch also implements --times-only.
   4
   5 Implementation details for the --source-filter and -dest-filter options:
   6
   7  - These options open a *HUGE* security hole in daemon mode unless they
   8    are refused in your rsyncd.conf!
   9
  10  - Filtering disables rsync alogrithm. (This should be fixed.)
  11
  12  - Source filter makes temporary files in /tmp. (Should be overridable.)
  13
  14  - If source filter fails, data is send unfiltered. (Should be changed
  15    to abort.)
  16
  17  - Failure of destination filter, causes data loss!!! (Should be changed
  18    to abort.)
  19
  20  - If filter changes size of file, you should use --times-only option to
  21    prevent repeated transfers of unchanged files.
  22
  23  - If the COMMAND contains single quotes, option-passing breaks.  (Needs
  24    to be fixed.)
  25
  26 You should run "make proto" before running "make".
  27
  28 --- orig/generator.c    2005-08-17 06:45:07
  29 +++ generator.c 2005-08-17 07:28:01
  30 @@ -58,6 +58,7 @@ extern int append_mode;
  31  extern int make_backups;
  32  extern int csum_length;
  33  extern int ignore_times;
  34 +extern int times_only;
  35  extern int size_only;
  36  extern OFF_T max_size;
  37  extern int io_error;
  38 @@ -360,7 +361,7 @@ void itemize(struct file_struct *file, i
  39  /* Perform our quick-check heuristic for determining if a file is unchanged. */
  40  static int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
  41  {
  42 -       if (st->st_size != file->length)
  43 +       if (!times_only && st->st_size != file->length)
  44                 return 0;
  45
  46         /* if always checksum is set then we use the checksum instead
  47 --- orig/options.c      2005-08-27 21:11:26
  48 +++ options.c   2005-08-27 21:27:17
  49 @@ -90,6 +90,7 @@ int keep_partial = 0;
  50  int safe_symlinks = 0;
  51  int copy_unsafe_links = 0;
  52  int size_only = 0;
  53 +int times_only = 0;
  54  int daemon_bwlimit = 0;
  55  int bwlimit = 0;
  56  int fuzzy_basis = 0;
  57 @@ -138,6 +139,8 @@ char *basis_dir[MAX_BASIS_DIRS+1];
  58  char *config_file = NULL;
  59  char *shell_cmd = NULL;
  60  char *log_format = NULL;
  61 +char *source_filter = NULL;
  62 +char *dest_filter = NULL;
  63  char *password_file = NULL;
  64  char *rsync_path = RSYNC_PATH;
  65  char *backup_dir = NULL;
  66 @@ -316,6 +319,7 @@ void usage(enum logcode F)
  67    rprintf(F,"     --timeout=TIME          set I/O timeout in seconds\n");
  68    rprintf(F," -I, --ignore-times          don't skip files that match in size and mod-time\n");
  69    rprintf(F,"     --size-only             skip files that match in size\n");
  70 +  rprintf(F,"     --times-only            skip files that match in mod-time\n");
  71    rprintf(F,"     --modify-window=NUM     compare mod-times with reduced accuracy\n");
  72    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
  73    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
  74 @@ -347,6 +351,8 @@ void usage(enum logcode F)
  75    rprintf(F,"     --write-batch=FILE      write a batched update to FILE\n");
  76    rprintf(F,"     --only-write-batch=FILE like --write-batch but w/o updating destination\n");
  77    rprintf(F,"     --read-batch=FILE       read a batched update from FILE\n");
  78 +  rprintf(F,"     --source-filter=COMMAND filter file through COMMAND at source\n");
  79 +  rprintf(F,"     --dest-filter=COMMAND   filter file through COMMAND at destination\n");
  80    rprintf(F,"     --protocol=NUM          force an older protocol version to be used\n");
  81  #ifdef INET6
  82    rprintf(F," -4, --ipv4                  prefer IPv4\n");
  83 @@ -413,6 +419,7 @@ static struct poptOption long_options[]
  84    {"no-implied-dirs",  0,  POPT_ARG_VAL,    &implied_dirs, 0, 0, 0 },
  85    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
  86    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
  87 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
  88    {"one-file-system", 'x', POPT_ARG_NONE,   &one_file_system, 0, 0, 0 },
  89    {"update",          'u', POPT_ARG_NONE,   &update_only, 0, 0, 0 },
  90    {"existing",         0,  POPT_ARG_NONE,   &only_existing, 0, 0, 0 },
  91 @@ -481,6 +488,8 @@ static struct poptOption long_options[]
  92    {"password-file",    0,  POPT_ARG_STRING, &password_file, 0, 0, 0 },
  93    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
  94    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
  95 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
  96 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
  97    {"protocol",         0,  POPT_ARG_INT,    &protocol_version, 0, 0, 0 },
  98    {"checksum-seed",    0,  POPT_ARG_INT,    &checksum_seed, 0, 0, 0 },
  99    {"server",           0,  POPT_ARG_NONE,   &am_server, 0, 0, 0 },
 100 @@ -1219,6 +1228,16 @@ int parse_arguments(int *argc, const cha
 101                 }
 102         }
 103
 104 +       if (source_filter || dest_filter) {
 105 +               if (whole_file == 0) {
 106 +                       snprintf(err_buf, sizeof err_buf,
 107 +                                "--no-whole-file cannot be used with --%s-filter\n",
 108 +                                source_filter ? "source" : "dest");
 109 +                       return 0;
 110 +               }
 111 +               whole_file = 1;
 112 +       }
 113 +
 114         if (files_from) {
 115                 char *h, *p;
 116                 int q;
 117 @@ -1442,6 +1461,25 @@ void server_options(char **args,int *arg
 118                         args[ac++] = "--only-write-batch=X";
 119         }
 120
 121 +       if (times_only && am_sender)
 122 +               args[ac++] = "--times-only";
 123 +
 124 +       if (source_filter && !am_sender) {
 125 +               /* Need to single quote the arg to keep the remote shell
 126 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 127 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 128 +                       goto oom;
 129 +               args[ac++] = arg;
 130 +       }
 131 +
 132 +       if (dest_filter && am_sender) {
 133 +               /* Need to single quote the arg to keep the remote shell
 134 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 135 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 136 +                       goto oom;
 137 +               args[ac++] = arg;
 138 +       }
 139 +
 140         if (size_only)
 141                 args[ac++] = "--size-only";
 142
 143 --- orig/pipe.c 2005-04-09 18:00:29
 144 +++ pipe.c      2005-08-17 07:24:44
 145 @@ -158,3 +158,79 @@ pid_t local_child(int argc, char **argv,
 146
 147         return pid;
 148  }
 149 +
 150 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 151 +{
 152 +       pid_t pid;
 153 +       int pipefds[2];
 154 +
 155 +       if (verbose >= 2)
 156 +               print_child_argv(command);
 157 +
 158 +       if (pipe(pipefds) < 0) {
 159 +               rsyserr(FERROR, errno, "pipe");
 160 +               exit_cleanup(RERR_IPC);
 161 +       }
 162 +
 163 +       pid = do_fork();
 164 +       if (pid == -1) {
 165 +               rsyserr(FERROR, errno, "fork");
 166 +               exit_cleanup(RERR_IPC);
 167 +       }
 168 +
 169 +       if (pid == 0) {
 170 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 171 +                || close(pipefds[1]) < 0
 172 +                || dup2(out, STDOUT_FILENO) < 0) {
 173 +                       rsyserr(FERROR, errno, "Failed dup/close");
 174 +                       exit_cleanup(RERR_IPC);
 175 +               }
 176 +               umask(orig_umask);
 177 +               set_blocking(STDIN_FILENO);
 178 +               if (blocking_io)
 179 +                       set_blocking(STDOUT_FILENO);
 180 +               execvp(command[0], command);
 181 +               rsyserr(FERROR, errno, "Failed to exec %s",
 182 +                       safe_fname(command[0]));
 183 +               exit_cleanup(RERR_IPC);
 184 +       }
 185 +
 186 +       if (close(pipefds[0]) < 0) {
 187 +               rsyserr(FERROR, errno, "Failed to close");
 188 +               exit_cleanup(RERR_IPC);
 189 +       }
 190 +
 191 +       *pipe_to_filter = pipefds[1];
 192 +
 193 +       return pid;
 194 +}
 195 +
 196 +pid_t run_filter_on_file(char *command[], int out, int in)
 197 +{
 198 +       pid_t pid;
 199 +
 200 +       if (verbose >= 2)
 201 +               print_child_argv(command);
 202 +
 203 +       pid = do_fork();
 204 +       if (pid == -1) {
 205 +               rsyserr(FERROR, errno, "fork");
 206 +               exit_cleanup(RERR_IPC);
 207 +       }
 208 +
 209 +       if (pid == 0) {
 210 +               if (dup2(in, STDIN_FILENO) < 0
 211 +                || dup2(out, STDOUT_FILENO) < 0) {
 212 +                       rsyserr(FERROR, errno, "Failed to dup2");
 213 +                       exit_cleanup(RERR_IPC);
 214 +               }
 215 +               if (blocking_io)
 216 +                       set_blocking(STDOUT_FILENO);
 217 +               execvp(command[0], command);
 218 +               rsyserr(FERROR, errno, "Failed to exec %s",
 219 +                       safe_fname(command[0]));
 220 +               exit_cleanup(RERR_IPC);
 221 +       }
 222 +
 223 +       return pid;
 224 +}
 225 --- orig/receiver.c     2005-08-17 06:45:08
 226 +++ receiver.c  2005-08-17 07:57:33
 227 @@ -52,6 +52,7 @@ extern int inplace;
 228  extern int delay_updates;
 229  extern struct stats stats;
 230  extern char *log_format;
 231 +extern char *dest_filter;
 232  extern char *tmpdir;
 233  extern char *partial_dir;
 234  extern char *basis_dir[];
 235 @@ -412,6 +413,8 @@ int recv_files(int f_in, struct file_lis
 236                       : !am_server && log_format_has_i;
 237         int max_phase = protocol_version >= 29 ? 2 : 1;
 238         int i, recv_ok;
 239 +       pid_t pid = 0;
 240 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 241
 242         if (verbose > 2)
 243                 rprintf(FINFO,"recv_files(%d) starting\n",flist->count);
 244 @@ -424,6 +427,23 @@ int recv_files(int f_in, struct file_lis
 245         if (delay_updates)
 246                 init_delayed_bits(flist->count);
 247
 248 +       if (dest_filter) {
 249 +               char *p;
 250 +               char *sep = " \t";
 251 +               int i;
 252 +               for (p = strtok(dest_filter, sep), i = 0;
 253 +                    p && i < MAX_FILTER_ARGS;
 254 +                    p = strtok(0, sep))
 255 +                       filter_argv[i++] = p;
 256 +               filter_argv[i] = NULL;
 257 +               if (p) {
 258 +                       rprintf(FERROR,
 259 +                               "Too many arguments to dest-filter (> %d)\n",
 260 +                               MAX_FILTER_ARGS);
 261 +                       exit_cleanup(RERR_SYNTAX);
 262 +               }
 263 +       }
 264 +
 265         while (1) {
 266                 cleanup_disable();
 267
 268 @@ -662,6 +682,9 @@ int recv_files(int f_in, struct file_lis
 269                 else if (!am_server && verbose && do_progress)
 270                         rprintf(FINFO, "%s\n", safe_fname(fname));
 271
 272 +               if (dest_filter)
 273 +                       pid = run_filter(filter_argv, fd2, &fd2);
 274 +
 275                 /* recv file data */
 276                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 277                                        fname, fd2, file->length);
 278 @@ -677,6 +700,16 @@ int recv_files(int f_in, struct file_lis
 279                         exit_cleanup(RERR_FILEIO);
 280                 }
 281
 282 +               if (dest_filter) {
 283 +                       int status;
 284 +                       wait_process(pid, &status);
 285 +                       if (status != 0) {
 286 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 287 +                                       dest_filter, status);
 288 +                               continue;
 289 +                       }
 290 +               }
 291 +
 292                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 293                         finish_transfer(fname, fnametmp, file, recv_ok, 1);
 294                         if (partialptr != fname && fnamecmp == partialptr) {
 295 --- orig/rsync.h        2005-08-17 06:45:08
 296 +++ rsync.h     2005-08-17 07:10:11
 297 @@ -103,6 +103,7 @@
 298  #define IOERR_DEL_LIMIT (1<<2)
 299
 300  #define MAX_ARGS 1000
 301 +#define MAX_FILTER_ARGS 100
 302  #define MAX_BASIS_DIRS 20
 303  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 304
 305 --- orig/rsync.yo       2005-08-27 21:05:12
 306 +++ rsync.yo    2005-08-17 07:08:21
 307 @@ -350,6 +350,7 @@ to the detailed description below for a
 308       --timeout=TIME          set I/O timeout in seconds
 309   -I, --ignore-times          don't skip files that match size and time
 310       --size-only             skip files that match in size
 311 +     --times-only            skip files that match in mod-time
 312       --modify-window=NUM     compare mod-times with reduced accuracy
 313   -T, --temp-dir=DIR          create temporary files in directory DIR
 314   -y, --fuzzy                 find similar file for basis if no dest file
 315 @@ -381,6 +382,8 @@ to the detailed description below for a
 316       --write-batch=FILE      write a batched update to FILE
 317       --only-write-batch=FILE like --write-batch but w/o updating dest
 318       --read-batch=FILE       read a batched update from FILE
 319 +     --source-filter=COMMAND filter file through COMMAND at source
 320 +     --dest-filter=COMMAND   filter file through COMMAND at destination
 321       --protocol=NUM          force an older protocol version to be used
 322       --checksum-seed=NUM     set block/file checksum seed (advanced)
 323   -4, --ipv4                  prefer IPv4
 324 @@ -1346,6 +1349,33 @@ file previously generated by bf(--write-
 325  If em(FILE) is "-" the batch data will be read from standard input.
 326  See the "BATCH MODE" section for details.
 327
 328 +dit(bf(--source-filter=COMMAND)) This option allows the user to specify a
 329 +filter program that will be applied to the contents of all transferred
 330 +regular files before the data is sent to destination.  COMMAND will receive
 331 +the data on its standard input and it should write the filtered data to
 332 +standard output.  COMMAND should exit non-zero if it cannot process the
 333 +data or if it encounters an error when writing the data to stdout.
 334 +
 335 +Example: --source-filter="gzip -9" will cause remote files to be
 336 +compressed.
 337 +Use of --source-filter automatically enables --whole-file.
 338 +If your filter does not output the same number of bytes that it received on
 339 +input, you should use --times-only to disable size and content checks on
 340 +subsequent rsync runs.
 341 +
 342 +dit(bf(--dest-filter=COMMAND)) This option allows you to specify a filter
 343 +program that will be applied to the contents of all transferred regular
 344 +files before the data is written to disk.  COMMAND will receive the data on
 345 +its standard input and it should write the filtered data to standard
 346 +output.  COMMAND should exit non-zero if it cannot process the data or if
 347 +it encounters an error when writing the data to stdout.
 348 +
 349 +Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 350 +Use of --dest-filter automatically enables --whole-file.
 351 +If your filter does not output the same number of bytes that it
 352 +received on input, you should use --times-only to disable size and
 353 +content checks on subsequent rsync runs.
 354 +
 355  dit(bf(--protocol=NUM)) Force an older protocol version to be used.  This
 356  is useful for creating a batch file that is compatible with an older
 357  version of rsync.  For instance, if rsync 2.6.4 is being used with the
 358 --- orig/sender.c       2005-07-28 01:46:25
 359 +++ sender.c    2005-08-17 07:40:49
 360 @@ -41,6 +41,7 @@ extern int write_batch;
 361  extern struct stats stats;
 362  extern struct file_list *the_file_list;
 363  extern char *log_format;
 364 +extern char *source_filter;
 365
 366
 367  /**
 368 @@ -221,6 +222,26 @@ void send_files(struct file_list *flist,
 369                       : !am_server && log_format_has_i;
 370         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 371         int i, j;
 372 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 373 +       char *tmp = 0;
 374 +       int unlink_tmp = 0;
 375 +
 376 +       if (source_filter) {
 377 +               char *p;
 378 +               char *sep = " \t";
 379 +               int i;
 380 +               for (p = strtok(source_filter, sep), i = 0;
 381 +                    p && i < MAX_FILTER_ARGS;
 382 +                    p = strtok(0, sep))
 383 +                       filter_argv[i++] = p;
 384 +               filter_argv[i] = NULL;
 385 +               if (p) {
 386 +                       rprintf(FERROR,
 387 +                               "Too many arguments to source-filter (> %d)\n",
 388 +                               MAX_FILTER_ARGS);
 389 +                       exit_cleanup(RERR_SYNTAX);
 390 +               }
 391 +       }
 392
 393         if (verbose > 2)
 394                 rprintf(FINFO, "send_files starting\n");
 395 @@ -295,6 +316,7 @@ void send_files(struct file_list *flist,
 396                         return;
 397                 }
 398
 399 +               unlink_tmp = 0;
 400                 fd = do_open(fname, O_RDONLY, 0);
 401                 if (fd == -1) {
 402                         if (errno == ENOENT) {
 403 @@ -323,6 +345,33 @@ void send_files(struct file_list *flist,
 404                         return;
 405                 }
 406
 407 +               if (source_filter) {
 408 +                       int fd2;
 409 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 410 +
 411 +                       tmp = strdup(tmpl);
 412 +                       fd2 = mkstemp(tmp);
 413 +                       if (fd2 == -1) {
 414 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 415 +                                       tmp, strerror(errno));
 416 +                       } else {
 417 +                               int status;
 418 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 419 +                               close(fd);
 420 +                               close(fd2);
 421 +                               wait_process(pid, &status);
 422 +                               if (status != 0) {
 423 +                                       rprintf(FERROR,
 424 +                                           "bypassing source filter %s; exited with code: %d\n",
 425 +                                           source_filter, status);
 426 +                                       fd = do_open(fname, O_RDONLY, 0);
 427 +                               } else {
 428 +                                       fd = do_open(tmp, O_RDONLY, 0);
 429 +                                       unlink_tmp = 1;
 430 +                               }
 431 +                       }
 432 +               }
 433 +
 434                 if (st.st_size) {
 435                         int32 read_size = MAX(s->blength * 3, MAX_MAP_SIZE);
 436                         mbuf = map_file(fd, st.st_size, read_size, s->blength);
 437 @@ -367,6 +416,8 @@ void send_files(struct file_list *flist,
 438                         }
 439                 }
 440                 close(fd);
 441 +               if (unlink_tmp)
 442 +                       unlink(tmp);
 443
 444                 free_sums(s);
 445