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 To use this patch, run these commands for a successful build:
  27
  28     patch -p1 <patches/source-filter_dest-filter.diff
  29     ./prepare-source
  30     ./configure                                (optional if already run)
  31     make
  32
  33 --- old/generator.c
  34 +++ new/generator.c
  35 @@ -61,6 +61,7 @@ extern int append_mode;
  36  extern int make_backups;
  37  extern int csum_length;
  38  extern int ignore_times;
  39 +extern int times_only;
  40  extern int size_only;
  41  extern OFF_T max_size;
  42  extern OFF_T min_size;
  43 @@ -619,7 +620,7 @@ void itemize(const char *fnamecmp, struc
  44  /* Perform our quick-check heuristic for determining if a file is unchanged. */
  45  int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
  46  {
  47 -       if (st->st_size != F_LENGTH(file))
  48 +       if (!times_only && st->st_size != F_LENGTH(file))
  49                 return 0;
  50
  51         /* if always checksum is set then we use the checksum instead
  52 --- old/main.c
  53 +++ new/main.c
  54 @@ -133,7 +133,7 @@ pid_t wait_process(pid_t pid, int *statu
  55  }
  56
  57  /* Wait for a process to exit, calling io_flush while waiting. */
  58 -static void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  59 +void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  60  {
  61         pid_t waited_pid;
  62         int status;
  63 --- old/options.c
  64 +++ new/options.c
  65 @@ -103,6 +103,7 @@ int keep_partial = 0;
  66  int safe_symlinks = 0;
  67  int copy_unsafe_links = 0;
  68  int size_only = 0;
  69 +int times_only = 0;
  70  int daemon_bwlimit = 0;
  71  int bwlimit = 0;
  72  int fuzzy_basis = 0;
  73 @@ -160,6 +161,8 @@ char *logfile_name = NULL;
  74  char *logfile_format = NULL;
  75  char *stdout_format = NULL;
  76  char *password_file = NULL;
  77 +char *source_filter = NULL;
  78 +char *dest_filter = NULL;
  79  char *rsync_path = RSYNC_PATH;
  80  char *backup_dir = NULL;
  81  char backup_dir_buf[MAXPATHLEN];
  82 @@ -380,6 +383,7 @@ void usage(enum logcode F)
  83    rprintf(F,"     --timeout=TIME          set I/O timeout in seconds\n");
  84    rprintf(F," -I, --ignore-times          don't skip files that match in size and mod-time\n");
  85    rprintf(F,"     --size-only             skip files that match in size\n");
  86 +  rprintf(F,"     --times-only            skip files that match in mod-time\n");
  87    rprintf(F,"     --modify-window=NUM     compare mod-times with reduced accuracy\n");
  88    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
  89    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
  90 @@ -419,6 +423,8 @@ void usage(enum logcode F)
  91    rprintf(F,"     --write-batch=FILE      write a batched update to FILE\n");
  92    rprintf(F,"     --only-write-batch=FILE like --write-batch but w/o updating destination\n");
  93    rprintf(F,"     --read-batch=FILE       read a batched update from FILE\n");
  94 +  rprintf(F,"     --source-filter=COMMAND filter file through COMMAND at source\n");
  95 +  rprintf(F,"     --dest-filter=COMMAND   filter file through COMMAND at destination\n");
  96    rprintf(F,"     --protocol=NUM          force an older protocol version to be used\n");
  97  #ifdef ICONV_OPTION
  98    rprintf(F,"     --iconv=CONVERT_SPEC    request charset conversion of filesnames\n");
  99 @@ -520,6 +526,7 @@ static struct poptOption long_options[]
 100    {"chmod",            0,  POPT_ARG_STRING, 0, OPT_CHMOD, 0, 0 },
 101    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
 102    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
 103 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
 104    {"one-file-system", 'x', POPT_ARG_NONE,   0, 'x', 0, 0 },
 105    {"update",          'u', POPT_ARG_NONE,   &update_only, 0, 0, 0 },
 106    {"existing",         0,  POPT_ARG_NONE,   &ignore_non_existing, 0, 0, 0 },
 107 @@ -619,6 +626,8 @@ static struct poptOption long_options[]
 108    {"password-file",    0,  POPT_ARG_STRING, &password_file, 0, 0, 0 },
 109    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
 110    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
 111 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
 112 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
 113    {"protocol",         0,  POPT_ARG_INT,    &protocol_version, 0, 0, 0 },
 114    {"checksum-seed",    0,  POPT_ARG_INT,    &checksum_seed, 0, 0, 0 },
 115    {"server",           0,  POPT_ARG_NONE,   0, OPT_SERVER, 0, 0 },
 116 @@ -1577,6 +1586,16 @@ int parse_arguments(int *argc_p, const c
 117                 }
 118         }
 119
 120 +       if (source_filter || dest_filter) {
 121 +               if (whole_file == 0) {
 122 +                       snprintf(err_buf, sizeof err_buf,
 123 +                                "--no-whole-file cannot be used with --%s-filter\n",
 124 +                                source_filter ? "source" : "dest");
 125 +                       return 0;
 126 +               }
 127 +               whole_file = 1;
 128 +       }
 129 +
 130         if (files_from) {
 131                 char *h, *p;
 132                 int q;
 133 @@ -1897,6 +1916,25 @@ void server_options(char **args, int *ar
 134                 }
 135         }
 136
 137 +       if (times_only && am_sender)
 138 +               args[ac++] = "--times-only";
 139 +
 140 +       if (source_filter && !am_sender) {
 141 +               /* Need to single quote the arg to keep the remote shell
 142 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 143 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 144 +                       goto oom;
 145 +               args[ac++] = arg;
 146 +       }
 147 +
 148 +       if (dest_filter && am_sender) {
 149 +               /* Need to single quote the arg to keep the remote shell
 150 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 151 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 152 +                       goto oom;
 153 +               args[ac++] = arg;
 154 +       }
 155 +
 156         if (modify_window_set) {
 157                 if (asprintf(&arg, "--modify-window=%d", modify_window) < 0)
 158                         goto oom;
 159 --- old/pipe.c
 160 +++ new/pipe.c
 161 @@ -164,3 +164,77 @@ pid_t local_child(int argc, char **argv,
 162
 163         return pid;
 164  }
 165 +
 166 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 167 +{
 168 +       pid_t pid;
 169 +       int pipefds[2];
 170 +
 171 +       if (verbose >= 2)
 172 +               print_child_argv("opening connection using:", command);
 173 +
 174 +       if (pipe(pipefds) < 0) {
 175 +               rsyserr(FERROR, errno, "pipe");
 176 +               exit_cleanup(RERR_IPC);
 177 +       }
 178 +
 179 +       pid = do_fork();
 180 +       if (pid == -1) {
 181 +               rsyserr(FERROR, errno, "fork");
 182 +               exit_cleanup(RERR_IPC);
 183 +       }
 184 +
 185 +       if (pid == 0) {
 186 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 187 +                || close(pipefds[1]) < 0
 188 +                || dup2(out, STDOUT_FILENO) < 0) {
 189 +                       rsyserr(FERROR, errno, "Failed dup/close");
 190 +                       exit_cleanup(RERR_IPC);
 191 +               }
 192 +               umask(orig_umask);
 193 +               set_blocking(STDIN_FILENO);
 194 +               if (blocking_io)
 195 +                       set_blocking(STDOUT_FILENO);
 196 +               execvp(command[0], command);
 197 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 198 +               exit_cleanup(RERR_IPC);
 199 +       }
 200 +
 201 +       if (close(pipefds[0]) < 0) {
 202 +               rsyserr(FERROR, errno, "Failed to close");
 203 +               exit_cleanup(RERR_IPC);
 204 +       }
 205 +
 206 +       *pipe_to_filter = pipefds[1];
 207 +
 208 +       return pid;
 209 +}
 210 +
 211 +pid_t run_filter_on_file(char *command[], int out, int in)
 212 +{
 213 +       pid_t pid;
 214 +
 215 +       if (verbose >= 2)
 216 +               print_child_argv("opening connection using:", command);
 217 +
 218 +       pid = do_fork();
 219 +       if (pid == -1) {
 220 +               rsyserr(FERROR, errno, "fork");
 221 +               exit_cleanup(RERR_IPC);
 222 +       }
 223 +
 224 +       if (pid == 0) {
 225 +               if (dup2(in, STDIN_FILENO) < 0
 226 +                || dup2(out, STDOUT_FILENO) < 0) {
 227 +                       rsyserr(FERROR, errno, "Failed to dup2");
 228 +                       exit_cleanup(RERR_IPC);
 229 +               }
 230 +               if (blocking_io)
 231 +                       set_blocking(STDOUT_FILENO);
 232 +               execvp(command[0], command);
 233 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 234 +               exit_cleanup(RERR_IPC);
 235 +       }
 236 +
 237 +       return pid;
 238 +}
 239 --- old/receiver.c
 240 +++ new/receiver.c
 241 @@ -52,6 +52,7 @@ extern int delay_updates;
 242  extern mode_t orig_umask;
 243  extern struct stats stats;
 244  extern char *tmpdir;
 245 +extern char *dest_filter;
 246  extern char *partial_dir;
 247  extern char *basis_dir[];
 248  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 249 @@ -357,6 +358,8 @@ int recv_files(int f_in, char *local_nam
 250         const char *parent_dirname = "";
 251  #endif
 252         int ndx, recv_ok;
 253 +       pid_t pid = 0;
 254 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 255
 256         if (verbose > 2)
 257                 rprintf(FINFO, "recv_files(%d) starting\n", cur_flist->used);
 258 @@ -366,6 +369,23 @@ int recv_files(int f_in, char *local_nam
 259
 260         updating_basis = inplace;
 261
 262 +       if (dest_filter) {
 263 +               char *p;
 264 +               char *sep = " \t";
 265 +               int i;
 266 +               for (p = strtok(dest_filter, sep), i = 0;
 267 +                    p && i < MAX_FILTER_ARGS;
 268 +                    p = strtok(0, sep))
 269 +                       filter_argv[i++] = p;
 270 +               filter_argv[i] = NULL;
 271 +               if (p) {
 272 +                       rprintf(FERROR,
 273 +                               "Too many arguments to dest-filter (> %d)\n",
 274 +                               MAX_FILTER_ARGS);
 275 +                       exit_cleanup(RERR_SYNTAX);
 276 +               }
 277 +       }
 278 +
 279         while (1) {
 280                 cleanup_disable();
 281
 282 @@ -663,6 +683,9 @@ int recv_files(int f_in, char *local_nam
 283                 else if (!am_server && verbose && do_progress)
 284                         rprintf(FINFO, "%s\n", fname);
 285
 286 +               if (dest_filter)
 287 +                       pid = run_filter(filter_argv, fd2, &fd2);
 288 +
 289                 /* recv file data */
 290                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 291                                        fname, fd2, F_LENGTH(file));
 292 @@ -677,6 +700,16 @@ int recv_files(int f_in, char *local_nam
 293                         exit_cleanup(RERR_FILEIO);
 294                 }
 295
 296 +               if (dest_filter) {
 297 +                       int status;
 298 +                       wait_process_with_flush(pid, &status);
 299 +                       if (status != 0) {
 300 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 301 +                                       dest_filter, status);
 302 +                               continue;
 303 +                       }
 304 +               }
 305 +
 306                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 307                         char *temp_copy_name;
 308                         if (partialptr == fname)
 309 --- old/rsync.h
 310 +++ new/rsync.h
 311 @@ -128,6 +128,7 @@
 312  #define IOERR_DEL_LIMIT (1<<2)
 313
 314  #define MAX_ARGS 1000
 315 +#define MAX_FILTER_ARGS 100
 316  #define MAX_BASIS_DIRS 20
 317  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 318
 319 --- old/rsync.yo
 320 +++ new/rsync.yo
 321 @@ -386,6 +386,7 @@ to the detailed description below for a
 322       --timeout=TIME          set I/O timeout in seconds
 323   -I, --ignore-times          don't skip files that match size and time
 324       --size-only             skip files that match in size
 325 +     --times-only            skip files that match in mod-time
 326       --modify-window=NUM     compare mod-times with reduced accuracy
 327   -T, --temp-dir=DIR          create temporary files in directory DIR
 328   -y, --fuzzy                 find similar file for basis if no dest file
 329 @@ -425,6 +426,8 @@ to the detailed description below for a
 330       --write-batch=FILE      write a batched update to FILE
 331       --only-write-batch=FILE like --write-batch but w/o updating dest
 332       --read-batch=FILE       read a batched update from FILE
 333 +     --source-filter=COMMAND filter file through COMMAND at source
 334 +     --dest-filter=COMMAND   filter file through COMMAND at destination
 335       --protocol=NUM          force an older protocol version to be used
 336       --iconv=CONVERT_SPEC    request charset conversion of filesnames
 337       --checksum-seed=NUM     set block/file checksum seed (advanced)
 338 @@ -1959,6 +1962,33 @@ file previously generated by bf(--write-
 339  If em(FILE) is bf(-), the batch data will be read from standard input.
 340  See the "BATCH MODE" section for details.
 341
 342 +dit(bf(--source-filter=COMMAND)) This option allows the user to specify a
 343 +filter program that will be applied to the contents of all transferred
 344 +regular files before the data is sent to destination.  COMMAND will receive
 345 +the data on its standard input and it should write the filtered data to
 346 +standard output.  COMMAND should exit non-zero if it cannot process the
 347 +data or if it encounters an error when writing the data to stdout.
 348 +
 349 +Example: --source-filter="gzip -9" will cause remote files to be
 350 +compressed.
 351 +Use of --source-filter automatically enables --whole-file.
 352 +If your filter does not output the same number of bytes that it received on
 353 +input, you should use --times-only to disable size and content checks on
 354 +subsequent rsync runs.
 355 +
 356 +dit(bf(--dest-filter=COMMAND)) This option allows you to specify a filter
 357 +program that will be applied to the contents of all transferred regular
 358 +files before the data is written to disk.  COMMAND will receive the data on
 359 +its standard input and it should write the filtered data to standard
 360 +output.  COMMAND should exit non-zero if it cannot process the data or if
 361 +it encounters an error when writing the data to stdout.
 362 +
 363 +Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 364 +Use of --dest-filter automatically enables --whole-file.
 365 +If your filter does not output the same number of bytes that it
 366 +received on input, you should use --times-only to disable size and
 367 +content checks on subsequent rsync runs.
 368 +
 369  dit(bf(--protocol=NUM)) Force an older protocol version to be used.  This
 370  is useful for creating a batch file that is compatible with an older
 371  version of rsync.  For instance, if rsync 2.6.4 is being used with the
 372 --- old/sender.c
 373 +++ new/sender.c
 374 @@ -43,6 +43,7 @@ extern int do_progress;
 375  extern int inplace;
 376  extern int batch_fd;
 377  extern int write_batch;
 378 +extern char *source_filter;
 379  extern struct stats stats;
 380  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 381
 382 @@ -181,6 +182,26 @@ void send_files(int f_in, int f_out)
 383         enum logcode log_code = log_before_transfer ? FLOG : FINFO;
 384         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 385         int ndx, j;
 386 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 387 +       char *tmp = 0;
 388 +       int unlink_tmp = 0;
 389 +
 390 +       if (source_filter) {
 391 +               char *p;
 392 +               char *sep = " \t";
 393 +               int i;
 394 +               for (p = strtok(source_filter, sep), i = 0;
 395 +                    p && i < MAX_FILTER_ARGS;
 396 +                    p = strtok(0, sep))
 397 +                       filter_argv[i++] = p;
 398 +               filter_argv[i] = NULL;
 399 +               if (p) {
 400 +                       rprintf(FERROR,
 401 +                               "Too many arguments to source-filter (> %d)\n",
 402 +                               MAX_FILTER_ARGS);
 403 +                       exit_cleanup(RERR_SYNTAX);
 404 +               }
 405 +       }
 406
 407         if (verbose > 2)
 408                 rprintf(FINFO, "send_files starting\n");
 409 @@ -284,6 +305,7 @@ void send_files(int f_in, int f_out)
 410                         exit_cleanup(RERR_PROTOCOL);
 411                 }
 412
 413 +               unlink_tmp = 0;
 414                 fd = do_open(fname, O_RDONLY, 0);
 415                 if (fd == -1) {
 416                         if (errno == ENOENT) {
 417 @@ -305,6 +327,33 @@ void send_files(int f_in, int f_out)
 418                         continue;
 419                 }
 420
 421 +               if (source_filter) {
 422 +                       int fd2;
 423 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 424 +
 425 +                       tmp = strdup(tmpl);
 426 +                       fd2 = mkstemp(tmp);
 427 +                       if (fd2 == -1) {
 428 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 429 +                                       tmp, strerror(errno));
 430 +                       } else {
 431 +                               int status;
 432 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 433 +                               close(fd);
 434 +                               close(fd2);
 435 +                               wait_process_with_flush(pid, &status);
 436 +                               if (status != 0) {
 437 +                                       rprintf(FERROR,
 438 +                                           "bypassing source filter %s; exited with code: %d\n",
 439 +                                           source_filter, status);
 440 +                                       fd = do_open(fname, O_RDONLY, 0);
 441 +                               } else {
 442 +                                       fd = do_open(tmp, O_RDONLY, 0);
 443 +                                       unlink_tmp = 1;
 444 +                               }
 445 +                       }
 446 +               }
 447 +
 448                 /* map the local file */
 449                 if (do_fstat(fd, &st) != 0) {
 450                         io_error |= IOERR_GENERAL;
 451 @@ -355,6 +404,8 @@ void send_files(int f_in, int f_out)
 452                         }
 453                 }
 454                 close(fd);
 455 +               if (unlink_tmp)
 456 +                       unlink(tmp);
 457
 458                 free_sums(s);
 459