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 After applying this patch, run these commands for a successful build:
  27
  28     ./prepare-source
  29     ./configure                      (optional if already run)
  30     make
  31
  32 --- old/generator.c
  33 +++ new/generator.c
  34 @@ -60,6 +60,7 @@ extern int append_mode;
  35  extern int make_backups;
  36  extern int csum_length;
  37  extern int ignore_times;
  38 +extern int times_only;
  39  extern int size_only;
  40  extern OFF_T max_size;
  41  extern OFF_T min_size;
  42 @@ -378,7 +379,7 @@ void itemize(struct file_struct *file, i
  43  /* Perform our quick-check heuristic for determining if a file is unchanged. */
  44  int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
  45  {
  46 -       if (st->st_size != file->length)
  47 +       if (!times_only && st->st_size != file->length)
  48                 return 0;
  49
  50         /* if always checksum is set then we use the checksum instead
  51 --- old/main.c
  52 +++ new/main.c
  53 @@ -122,7 +122,7 @@ pid_t wait_process(pid_t pid, int *statu
  54  }
  55
  56  /* Wait for a process to exit, calling io_flush while waiting. */
  57 -static void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  58 +void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  59  {
  60         pid_t waited_pid;
  61         int status;
  62 --- old/options.c
  63 +++ new/options.c
  64 @@ -99,6 +99,7 @@ int keep_partial = 0;
  65  int safe_symlinks = 0;
  66  int copy_unsafe_links = 0;
  67  int size_only = 0;
  68 +int times_only = 0;
  69  int daemon_bwlimit = 0;
  70  int bwlimit = 0;
  71  int fuzzy_basis = 0;
  72 @@ -151,6 +152,8 @@ char *logfile_name = NULL;
  73  char *logfile_format = NULL;
  74  char *stdout_format = NULL;
  75  char *password_file = NULL;
  76 +char *source_filter = NULL;
  77 +char *dest_filter = NULL;
  78  char *rsync_path = RSYNC_PATH;
  79  char *backup_dir = NULL;
  80  char backup_dir_buf[MAXPATHLEN];
  81 @@ -341,6 +344,7 @@ void usage(enum logcode F)
  82    rprintf(F,"     --timeout=TIME          set I/O timeout in seconds\n");
  83    rprintf(F," -I, --ignore-times          don't skip files that match in size and mod-time\n");
  84    rprintf(F,"     --size-only             skip files that match in size\n");
  85 +  rprintf(F,"     --times-only            skip files that match in mod-time\n");
  86    rprintf(F,"     --modify-window=NUM     compare mod-times with reduced accuracy\n");
  87    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
  88    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
  89 @@ -378,6 +382,8 @@ void usage(enum logcode F)
  90    rprintf(F,"     --write-batch=FILE      write a batched update to FILE\n");
  91    rprintf(F,"     --only-write-batch=FILE like --write-batch but w/o updating destination\n");
  92    rprintf(F,"     --read-batch=FILE       read a batched update from FILE\n");
  93 +  rprintf(F,"     --source-filter=COMMAND filter file through COMMAND at source\n");
  94 +  rprintf(F,"     --dest-filter=COMMAND   filter file through COMMAND at destination\n");
  95    rprintf(F,"     --protocol=NUM          force an older protocol version to be used\n");
  96  #ifdef INET6
  97    rprintf(F," -4, --ipv4                  prefer IPv4\n");
  98 @@ -459,6 +465,7 @@ static struct poptOption long_options[]
  99    {"chmod",            0,  POPT_ARG_STRING, 0, OPT_CHMOD, 0, 0 },
 100    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
 101    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
 102 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
 103    {"one-file-system", 'x', POPT_ARG_NONE,   0, 'x', 0, 0 },
 104    {"update",          'u', POPT_ARG_NONE,   &update_only, 0, 0, 0 },
 105    {"existing",         0,  POPT_ARG_NONE,   &ignore_non_existing, 0, 0, 0 },
 106 @@ -537,6 +544,8 @@ static struct poptOption long_options[]
 107    {"password-file",    0,  POPT_ARG_STRING, &password_file, 0, 0, 0 },
 108    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
 109    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
 110 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
 111 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
 112    {"protocol",         0,  POPT_ARG_INT,    &protocol_version, 0, 0, 0 },
 113    {"checksum-seed",    0,  POPT_ARG_INT,    &checksum_seed, 0, 0, 0 },
 114    {"server",           0,  POPT_ARG_NONE,   0, OPT_SERVER, 0, 0 },
 115 @@ -1403,6 +1412,16 @@ int parse_arguments(int *argc, const cha
 116                 }
 117         }
 118
 119 +       if (source_filter || dest_filter) {
 120 +               if (whole_file == 0) {
 121 +                       snprintf(err_buf, sizeof err_buf,
 122 +                                "--no-whole-file cannot be used with --%s-filter\n",
 123 +                                source_filter ? "source" : "dest");
 124 +                       return 0;
 125 +               }
 126 +               whole_file = 1;
 127 +       }
 128 +
 129         if (files_from) {
 130                 char *h, *p;
 131                 int q;
 132 @@ -1669,6 +1688,25 @@ void server_options(char **args,int *arg
 133                         args[ac++] = "--size-only";
 134         }
 135
 136 +       if (times_only && am_sender)
 137 +               args[ac++] = "--times-only";
 138 +
 139 +       if (source_filter && !am_sender) {
 140 +               /* Need to single quote the arg to keep the remote shell
 141 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 142 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 143 +                       goto oom;
 144 +               args[ac++] = arg;
 145 +       }
 146 +
 147 +       if (dest_filter && am_sender) {
 148 +               /* Need to single quote the arg to keep the remote shell
 149 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 150 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 151 +                       goto oom;
 152 +               args[ac++] = arg;
 153 +       }
 154 +
 155         if (modify_window_set) {
 156                 if (asprintf(&arg, "--modify-window=%d", modify_window) < 0)
 157                         goto oom;
 158 --- old/pipe.c
 159 +++ new/pipe.c
 160 @@ -166,3 +166,77 @@ pid_t local_child(int argc, char **argv,
 161
 162         return pid;
 163  }
 164 +
 165 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 166 +{
 167 +       pid_t pid;
 168 +       int pipefds[2];
 169 +
 170 +       if (verbose >= 2)
 171 +               print_child_argv(command);
 172 +
 173 +       if (pipe(pipefds) < 0) {
 174 +               rsyserr(FERROR, errno, "pipe");
 175 +               exit_cleanup(RERR_IPC);
 176 +       }
 177 +
 178 +       pid = do_fork();
 179 +       if (pid == -1) {
 180 +               rsyserr(FERROR, errno, "fork");
 181 +               exit_cleanup(RERR_IPC);
 182 +       }
 183 +
 184 +       if (pid == 0) {
 185 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 186 +                || close(pipefds[1]) < 0
 187 +                || dup2(out, STDOUT_FILENO) < 0) {
 188 +                       rsyserr(FERROR, errno, "Failed dup/close");
 189 +                       exit_cleanup(RERR_IPC);
 190 +               }
 191 +               umask(orig_umask);
 192 +               set_blocking(STDIN_FILENO);
 193 +               if (blocking_io)
 194 +                       set_blocking(STDOUT_FILENO);
 195 +               execvp(command[0], command);
 196 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 197 +               exit_cleanup(RERR_IPC);
 198 +       }
 199 +
 200 +       if (close(pipefds[0]) < 0) {
 201 +               rsyserr(FERROR, errno, "Failed to close");
 202 +               exit_cleanup(RERR_IPC);
 203 +       }
 204 +
 205 +       *pipe_to_filter = pipefds[1];
 206 +
 207 +       return pid;
 208 +}
 209 +
 210 +pid_t run_filter_on_file(char *command[], int out, int in)
 211 +{
 212 +       pid_t pid;
 213 +
 214 +       if (verbose >= 2)
 215 +               print_child_argv(command);
 216 +
 217 +       pid = do_fork();
 218 +       if (pid == -1) {
 219 +               rsyserr(FERROR, errno, "fork");
 220 +               exit_cleanup(RERR_IPC);
 221 +       }
 222 +
 223 +       if (pid == 0) {
 224 +               if (dup2(in, STDIN_FILENO) < 0
 225 +                || dup2(out, STDOUT_FILENO) < 0) {
 226 +                       rsyserr(FERROR, errno, "Failed to dup2");
 227 +                       exit_cleanup(RERR_IPC);
 228 +               }
 229 +               if (blocking_io)
 230 +                       set_blocking(STDOUT_FILENO);
 231 +               execvp(command[0], command);
 232 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 233 +               exit_cleanup(RERR_IPC);
 234 +       }
 235 +
 236 +       return pid;
 237 +}
 238 --- old/receiver.c
 239 +++ new/receiver.c
 240 @@ -48,6 +48,7 @@ extern int checksum_seed;
 241  extern int inplace;
 242  extern int delay_updates;
 243  extern struct stats stats;
 244 +extern char *dest_filter;
 245  extern char *stdout_format;
 246  extern char *tmpdir;
 247  extern char *partial_dir;
 248 @@ -350,6 +351,8 @@ int recv_files(int f_in, struct file_lis
 249         enum logcode log_code = log_before_transfer ? FLOG : FINFO;
 250         int max_phase = protocol_version >= 29 ? 2 : 1;
 251         int i, recv_ok;
 252 +       pid_t pid = 0;
 253 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 254
 255         if (verbose > 2)
 256                 rprintf(FINFO,"recv_files(%d) starting\n",flist->count);
 257 @@ -364,6 +367,23 @@ int recv_files(int f_in, struct file_lis
 258
 259         updating_basis = inplace;
 260
 261 +       if (dest_filter) {
 262 +               char *p;
 263 +               char *sep = " \t";
 264 +               int i;
 265 +               for (p = strtok(dest_filter, sep), i = 0;
 266 +                    p && i < MAX_FILTER_ARGS;
 267 +                    p = strtok(0, sep))
 268 +                       filter_argv[i++] = p;
 269 +               filter_argv[i] = NULL;
 270 +               if (p) {
 271 +                       rprintf(FERROR,
 272 +                               "Too many arguments to dest-filter (> %d)\n",
 273 +                               MAX_FILTER_ARGS);
 274 +                       exit_cleanup(RERR_SYNTAX);
 275 +               }
 276 +       }
 277 +
 278         while (1) {
 279                 cleanup_disable();
 280
 281 @@ -606,6 +626,9 @@ int recv_files(int f_in, struct file_lis
 282                 else if (!am_server && verbose && do_progress)
 283                         rprintf(FINFO, "%s\n", fname);
 284
 285 +               if (dest_filter)
 286 +                       pid = run_filter(filter_argv, fd2, &fd2);
 287 +
 288                 /* recv file data */
 289                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 290                                        fname, fd2, file->length);
 291 @@ -620,6 +643,16 @@ int recv_files(int f_in, struct file_lis
 292                         exit_cleanup(RERR_FILEIO);
 293                 }
 294
 295 +               if (dest_filter) {
 296 +                       int status;
 297 +                       wait_process_with_flush(pid, &status);
 298 +                       if (status != 0) {
 299 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 300 +                                       dest_filter, status);
 301 +                               continue;
 302 +                       }
 303 +               }
 304 +
 305                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 306                         char *temp_copy_name;
 307                         if (partialptr == fname)
 308 --- old/rsync.h
 309 +++ new/rsync.h
 310 @@ -104,6 +104,7 @@
 311  #define IOERR_DEL_LIMIT (1<<2)
 312
 313  #define MAX_ARGS 1000
 314 +#define MAX_FILTER_ARGS 100
 315  #define MAX_BASIS_DIRS 20
 316  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 317
 318 --- old/rsync.yo
 319 +++ new/rsync.yo
 320 @@ -359,6 +359,7 @@ to the detailed description below for a
 321       --timeout=TIME          set I/O timeout in seconds
 322   -I, --ignore-times          don't skip files that match size and time
 323       --size-only             skip files that match in size
 324 +     --times-only            skip files that match in mod-time
 325       --modify-window=NUM     compare mod-times with reduced accuracy
 326   -T, --temp-dir=DIR          create temporary files in directory DIR
 327   -y, --fuzzy                 find similar file for basis if no dest file
 328 @@ -396,6 +397,8 @@ to the detailed description below for a
 329       --write-batch=FILE      write a batched update to FILE
 330       --only-write-batch=FILE like --write-batch but w/o updating dest
 331       --read-batch=FILE       read a batched update from FILE
 332 +     --source-filter=COMMAND filter file through COMMAND at source
 333 +     --dest-filter=COMMAND   filter file through COMMAND at destination
 334       --protocol=NUM          force an older protocol version to be used
 335       --checksum-seed=NUM     set block/file checksum seed (advanced)
 336   -4, --ipv4                  prefer IPv4
 337 @@ -1753,6 +1756,33 @@ file previously generated by bf(--write-
 338  If em(FILE) is bf(-), the batch data will be read from standard input.
 339  See the "BATCH MODE" section for details.
 340
 341 +dit(bf(--source-filter=COMMAND)) This option allows the user to specify a
 342 +filter program that will be applied to the contents of all transferred
 343 +regular files before the data is sent to destination.  COMMAND will receive
 344 +the data on its standard input and it should write the filtered data to
 345 +standard output.  COMMAND should exit non-zero if it cannot process the
 346 +data or if it encounters an error when writing the data to stdout.
 347 +
 348 +Example: --source-filter="gzip -9" will cause remote files to be
 349 +compressed.
 350 +Use of --source-filter automatically enables --whole-file.
 351 +If your filter does not output the same number of bytes that it received on
 352 +input, you should use --times-only to disable size and content checks on
 353 +subsequent rsync runs.
 354 +
 355 +dit(bf(--dest-filter=COMMAND)) This option allows you to specify a filter
 356 +program that will be applied to the contents of all transferred regular
 357 +files before the data is written to disk.  COMMAND will receive the data on
 358 +its standard input and it should write the filtered data to standard
 359 +output.  COMMAND should exit non-zero if it cannot process the data or if
 360 +it encounters an error when writing the data to stdout.
 361 +
 362 +Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 363 +Use of --dest-filter automatically enables --whole-file.
 364 +If your filter does not output the same number of bytes that it
 365 +received on input, you should use --times-only to disable size and
 366 +content checks on subsequent rsync runs.
 367 +
 368  dit(bf(--protocol=NUM)) Force an older protocol version to be used.  This
 369  is useful for creating a batch file that is compatible with an older
 370  version of rsync.  For instance, if rsync 2.6.4 is being used with the
 371 --- old/sender.c
 372 +++ new/sender.c
 373 @@ -43,6 +43,7 @@ extern int batch_fd;
 374  extern int write_batch;
 375  extern struct stats stats;
 376  extern struct file_list *the_file_list;
 377 +extern char *source_filter;
 378  extern char *stdout_format;
 379
 380
 381 @@ -224,6 +225,26 @@ void send_files(struct file_list *flist,
 382         enum logcode log_code = log_before_transfer ? FLOG : FINFO;
 383         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 384         int i, j;
 385 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 386 +       char *tmp = 0;
 387 +       int unlink_tmp = 0;
 388 +
 389 +       if (source_filter) {
 390 +               char *p;
 391 +               char *sep = " \t";
 392 +               int i;
 393 +               for (p = strtok(source_filter, sep), i = 0;
 394 +                    p && i < MAX_FILTER_ARGS;
 395 +                    p = strtok(0, sep))
 396 +                       filter_argv[i++] = p;
 397 +               filter_argv[i] = NULL;
 398 +               if (p) {
 399 +                       rprintf(FERROR,
 400 +                               "Too many arguments to source-filter (> %d)\n",
 401 +                               MAX_FILTER_ARGS);
 402 +                       exit_cleanup(RERR_SYNTAX);
 403 +               }
 404 +       }
 405
 406         if (verbose > 2)
 407                 rprintf(FINFO, "send_files starting\n");
 408 @@ -297,6 +318,7 @@ void send_files(struct file_list *flist,
 409                         return;
 410                 }
 411
 412 +               unlink_tmp = 0;
 413                 fd = do_open(fname, O_RDONLY, 0);
 414                 if (fd == -1) {
 415                         if (errno == ENOENT) {
 416 @@ -325,6 +347,33 @@ void send_files(struct file_list *flist,
 417                         return;
 418                 }
 419
 420 +               if (source_filter) {
 421 +                       int fd2;
 422 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 423 +
 424 +                       tmp = strdup(tmpl);
 425 +                       fd2 = mkstemp(tmp);
 426 +                       if (fd2 == -1) {
 427 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 428 +                                       tmp, strerror(errno));
 429 +                       } else {
 430 +                               int status;
 431 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 432 +                               close(fd);
 433 +                               close(fd2);
 434 +                               wait_process_with_flush(pid, &status);
 435 +                               if (status != 0) {
 436 +                                       rprintf(FERROR,
 437 +                                           "bypassing source filter %s; exited with code: %d\n",
 438 +                                           source_filter, status);
 439 +                                       fd = do_open(fname, O_RDONLY, 0);
 440 +                               } else {
 441 +                                       fd = do_open(tmp, O_RDONLY, 0);
 442 +                                       unlink_tmp = 1;
 443 +                               }
 444 +                       }
 445 +               }
 446 +
 447                 if (st.st_size) {
 448                         int32 read_size = MAX(s->blength * 3, MAX_MAP_SIZE);
 449                         mbuf = map_file(fd, st.st_size, read_size, s->blength);
 450 @@ -366,6 +415,8 @@ void send_files(struct file_list *flist,
 451                         }
 452                 }
 453                 close(fd);
 454 +               if (unlink_tmp)
 455 +                       unlink(tmp);
 456
 457                 free_sums(s);
 458