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 based-on: 5b19cf787515fc2388dd070a85e86ced5d80510b
  34 diff --git a/generator.c b/generator.c
  35 --- a/generator.c
  36 +++ b/generator.c
  37 @@ -64,6 +64,7 @@ extern int append_mode;
  38  extern int make_backups;
  39  extern int csum_length;
  40  extern int ignore_times;
  41 +extern int times_only;
  42  extern int size_only;
  43  extern OFF_T max_size;
  44  extern OFF_T min_size;
  45 @@ -580,7 +581,7 @@ void itemize(const char *fnamecmp, struct file_struct *file, int ndx, int statre
  46  /* Perform our quick-check heuristic for determining if a file is unchanged. */
  47  int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
  48  {
  49 -       if (st->st_size != F_LENGTH(file))
  50 +       if (!times_only && st->st_size != F_LENGTH(file))
  51                 return 0;
  52
  53         /* if always checksum is set then we use the checksum instead
  54 diff --git a/main.c b/main.c
  55 --- a/main.c
  56 +++ b/main.c
  57 @@ -180,7 +180,7 @@ int shell_exec(const char *cmd)
  58  }
  59
  60  /* Wait for a process to exit, calling io_flush while waiting. */
  61 -static void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  62 +void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  63  {
  64         pid_t waited_pid;
  65         int status;
  66 diff --git a/options.c b/options.c
  67 --- a/options.c
  68 +++ b/options.c
  69 @@ -111,6 +111,7 @@ int safe_symlinks = 0;
  70  int copy_unsafe_links = 0;
  71  int munge_symlinks = 0;
  72  int size_only = 0;
  73 +int times_only = 0;
  74  int daemon_bwlimit = 0;
  75  int bwlimit = 0;
  76  int fuzzy_basis = 0;
  77 @@ -170,6 +171,8 @@ char *logfile_name = NULL;
  78  char *logfile_format = NULL;
  79  char *stdout_format = NULL;
  80  char *password_file = NULL;
  81 +char *source_filter = NULL;
  82 +char *dest_filter = NULL;
  83  char *rsync_path = RSYNC_PATH;
  84  char *backup_dir = NULL;
  85  char backup_dir_buf[MAXPATHLEN];
  86 @@ -823,6 +826,7 @@ void usage(enum logcode F)
  87    rprintf(F," -I, --ignore-times          don't skip files that match in size and mod-time\n");
  88    rprintf(F," -M, --remote-option=OPTION  send OPTION to the remote side only\n");
  89    rprintf(F,"     --size-only             skip files that match in size\n");
  90 +  rprintf(F,"     --times-only            skip files that match in mod-time\n");
  91    rprintf(F," -@, --modify-window=NUM     set the accuracy for mod-time comparisons\n");
  92    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
  93    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
  94 @@ -866,6 +870,8 @@ void usage(enum logcode F)
  95    rprintf(F,"     --write-batch=FILE      write a batched update to FILE\n");
  96    rprintf(F,"     --only-write-batch=FILE like --write-batch but w/o updating destination\n");
  97    rprintf(F,"     --read-batch=FILE       read a batched update from FILE\n");
  98 +  rprintf(F,"     --source-filter=COMMAND filter file through COMMAND at source\n");
  99 +  rprintf(F,"     --dest-filter=COMMAND   filter file through COMMAND at destination\n");
 100    rprintf(F,"     --protocol=NUM          force an older protocol version to be used\n");
 101  #ifdef ICONV_OPTION
 102    rprintf(F,"     --iconv=CONVERT_SPEC    request charset conversion of filenames\n");
 103 @@ -988,6 +994,7 @@ static struct poptOption long_options[] = {
 104    {"chmod",            0,  POPT_ARG_STRING, 0, OPT_CHMOD, 0, 0 },
 105    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
 106    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
 107 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
 108    {"one-file-system", 'x', POPT_ARG_NONE,   0, 'x', 0, 0 },
 109    {"no-one-file-system",0, POPT_ARG_VAL,    &one_file_system, 0, 0, 0 },
 110    {"no-x",             0,  POPT_ARG_VAL,    &one_file_system, 0, 0, 0 },
 111 @@ -1116,6 +1123,8 @@ static struct poptOption long_options[] = {
 112    {"password-file",    0,  POPT_ARG_STRING, &password_file, 0, 0, 0 },
 113    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
 114    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
 115 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
 116 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
 117  #ifdef HAVE_SETVBUF
 118    {"outbuf",           0,  POPT_ARG_STRING, &outbuf_mode, 0, 0, 0 },
 119  #endif
 120 @@ -2484,6 +2493,16 @@ int parse_arguments(int *argc_p, const char ***argv_p)
 121                 }
 122         }
 123
 124 +       if (source_filter || dest_filter) {
 125 +               if (whole_file == 0) {
 126 +                       snprintf(err_buf, sizeof err_buf,
 127 +                                "--no-whole-file cannot be used with --%s-filter\n",
 128 +                                source_filter ? "source" : "dest");
 129 +                       return 0;
 130 +               }
 131 +               whole_file = 1;
 132 +       }
 133 +
 134         if (files_from) {
 135                 char *h, *p;
 136                 int q;
 137 @@ -2868,6 +2887,25 @@ void server_options(char **args, int *argc_p)
 138         else if (missing_args == 1 && !am_sender)
 139                 args[ac++] = "--ignore-missing-args";
 140
 141 +       if (times_only && am_sender)
 142 +               args[ac++] = "--times-only";
 143 +
 144 +       if (source_filter && !am_sender) {
 145 +               /* Need to single quote the arg to keep the remote shell
 146 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 147 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 148 +                       goto oom;
 149 +               args[ac++] = arg;
 150 +       }
 151 +
 152 +       if (dest_filter && am_sender) {
 153 +               /* Need to single quote the arg to keep the remote shell
 154 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 155 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 156 +                       goto oom;
 157 +               args[ac++] = arg;
 158 +       }
 159 +
 160         if (modify_window_set && am_sender) {
 161                 char *fmt = modify_window < 0 ? "-@%d" : "--modify-window=%d";
 162                 if (asprintf(&arg, fmt, modify_window) < 0)
 163 diff --git a/pipe.c b/pipe.c
 164 --- a/pipe.c
 165 +++ b/pipe.c
 166 @@ -27,6 +27,7 @@ extern int am_server;
 167  extern int blocking_io;
 168  extern int filesfrom_fd;
 169  extern int munge_symlinks;
 170 +extern mode_t orig_umask;
 171  extern char *logfile_name;
 172  extern int remote_option_cnt;
 173  extern const char **remote_options;
 174 @@ -178,3 +179,77 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
 175
 176         return pid;
 177  }
 178 +
 179 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 180 +{
 181 +       pid_t pid;
 182 +       int pipefds[2];
 183 +
 184 +       if (DEBUG_GTE(CMD, 1))
 185 +               print_child_argv("opening connection using:", command);
 186 +
 187 +       if (pipe(pipefds) < 0) {
 188 +               rsyserr(FERROR, errno, "pipe");
 189 +               exit_cleanup(RERR_IPC);
 190 +       }
 191 +
 192 +       pid = do_fork();
 193 +       if (pid == -1) {
 194 +               rsyserr(FERROR, errno, "fork");
 195 +               exit_cleanup(RERR_IPC);
 196 +       }
 197 +
 198 +       if (pid == 0) {
 199 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 200 +                || close(pipefds[1]) < 0
 201 +                || dup2(out, STDOUT_FILENO) < 0) {
 202 +                       rsyserr(FERROR, errno, "Failed dup/close");
 203 +                       exit_cleanup(RERR_IPC);
 204 +               }
 205 +               umask(orig_umask);
 206 +               set_blocking(STDIN_FILENO);
 207 +               if (blocking_io)
 208 +                       set_blocking(STDOUT_FILENO);
 209 +               execvp(command[0], command);
 210 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 211 +               exit_cleanup(RERR_IPC);
 212 +       }
 213 +
 214 +       if (close(pipefds[0]) < 0) {
 215 +               rsyserr(FERROR, errno, "Failed to close");
 216 +               exit_cleanup(RERR_IPC);
 217 +       }
 218 +
 219 +       *pipe_to_filter = pipefds[1];
 220 +
 221 +       return pid;
 222 +}
 223 +
 224 +pid_t run_filter_on_file(char *command[], int out, int in)
 225 +{
 226 +       pid_t pid;
 227 +
 228 +       if (DEBUG_GTE(CMD, 1))
 229 +               print_child_argv("opening connection using:", command);
 230 +
 231 +       pid = do_fork();
 232 +       if (pid == -1) {
 233 +               rsyserr(FERROR, errno, "fork");
 234 +               exit_cleanup(RERR_IPC);
 235 +       }
 236 +
 237 +       if (pid == 0) {
 238 +               if (dup2(in, STDIN_FILENO) < 0
 239 +                || dup2(out, STDOUT_FILENO) < 0) {
 240 +                       rsyserr(FERROR, errno, "Failed to dup2");
 241 +                       exit_cleanup(RERR_IPC);
 242 +               }
 243 +               if (blocking_io)
 244 +                       set_blocking(STDOUT_FILENO);
 245 +               execvp(command[0], command);
 246 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 247 +               exit_cleanup(RERR_IPC);
 248 +       }
 249 +
 250 +       return pid;
 251 +}
 252 diff --git a/receiver.c b/receiver.c
 253 --- a/receiver.c
 254 +++ b/receiver.c
 255 @@ -60,6 +60,7 @@ extern BOOL want_progress_now;
 256  extern mode_t orig_umask;
 257  extern struct stats stats;
 258  extern char *tmpdir;
 259 +extern char *dest_filter;
 260  extern char *partial_dir;
 261  extern char *basis_dir[MAX_BASIS_DIRS+1];
 262  extern char sender_file_sum[MAX_DIGEST_LEN];
 263 @@ -522,6 +523,7 @@ int recv_files(int f_in, int f_out, char *local_name)
 264         char *fnametmp, fnametmpbuf[MAXPATHLEN];
 265         char *fnamecmp, *partialptr;
 266         char fnamecmpbuf[MAXPATHLEN];
 267 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 268         uchar fnamecmp_type;
 269         struct file_struct *file;
 270         int itemizing = am_server ? logfile_format_has_i : stdout_format_has_i;
 271 @@ -532,6 +534,7 @@ int recv_files(int f_in, int f_out, char *local_name)
 272         const char *parent_dirname = "";
 273  #endif
 274         int ndx, recv_ok, one_inplace;
 275 +       pid_t pid = 0;
 276
 277         if (DEBUG_GTE(RECV, 1))
 278                 rprintf(FINFO, "recv_files(%d) starting\n", cur_flist->used);
 279 @@ -539,6 +542,23 @@ int recv_files(int f_in, int f_out, char *local_name)
 280         if (delay_updates)
 281                 delayed_bits = bitbag_create(cur_flist->used + 1);
 282
 283 +       if (dest_filter) {
 284 +               char *p;
 285 +               char *sep = " \t";
 286 +               int i;
 287 +               for (p = strtok(dest_filter, sep), i = 0;
 288 +                    p && i < MAX_FILTER_ARGS;
 289 +                    p = strtok(0, sep))
 290 +                       filter_argv[i++] = p;
 291 +               filter_argv[i] = NULL;
 292 +               if (p) {
 293 +                       rprintf(FERROR,
 294 +                               "Too many arguments to dest-filter (> %d)\n",
 295 +                               MAX_FILTER_ARGS);
 296 +                       exit_cleanup(RERR_SYNTAX);
 297 +               }
 298 +       }
 299 +
 300         progress_init();
 301
 302         while (1) {
 303 @@ -865,6 +885,9 @@ int recv_files(int f_in, int f_out, char *local_name)
 304                 else if (!am_server && INFO_GTE(NAME, 1) && INFO_EQ(PROGRESS, 1))
 305                         rprintf(FINFO, "%s\n", fname);
 306
 307 +               if (dest_filter)
 308 +                       pid = run_filter(filter_argv, fd2, &fd2);
 309 +
 310                 /* recv file data */
 311                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size, fname, fd2, file, inplace || one_inplace);
 312
 313 @@ -880,6 +903,16 @@ int recv_files(int f_in, int f_out, char *local_name)
 314                         exit_cleanup(RERR_FILEIO);
 315                 }
 316
 317 +               if (dest_filter) {
 318 +                       int status;
 319 +                       wait_process_with_flush(pid, &status);
 320 +                       if (status != 0) {
 321 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 322 +                                       dest_filter, status);
 323 +                               continue;
 324 +                       }
 325 +               }
 326 +
 327                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 328                         if (partialptr == fname)
 329                                 partialptr = NULL;
 330 diff --git a/rsync.1.md b/rsync.1.md
 331 --- a/rsync.1.md
 332 +++ b/rsync.1.md
 333 @@ -412,6 +412,7 @@ detailed description below for a complete description.
 334  --contimeout=SECONDS        set daemon connection timeout in seconds
 335  --ignore-times, -I          don't skip files that match size and time
 336  --size-only                 skip files that match in size
 337 +--times-only                skip files that match in mod-time
 338  --modify-window=NUM, -@     set the accuracy for mod-time comparisons
 339  --temp-dir=DIR, -T          create temporary files in directory DIR
 340  --fuzzy, -y                 find similar file for basis if no dest file
 341 @@ -454,6 +455,8 @@ detailed description below for a complete description.
 342  --write-batch=FILE          write a batched update to FILE
 343  --only-write-batch=FILE     like --write-batch but w/o updating dest
 344  --read-batch=FILE           read a batched update from FILE
 345 +--source-filter=COMMAND     filter file through COMMAND at source
 346 +--dest-filter=COMMAND       filter file through COMMAND at destination
 347  --protocol=NUM              force an older protocol version to be used
 348  --iconv=CONVERT_SPEC        request charset conversion of filenames
 349  --checksum-seed=NUM         set block/file checksum seed (advanced)
 350 @@ -3008,6 +3011,36 @@ your home directory (remove the '=' for that).
 351      `--write-batch`.  If _FILE_ is `-`, the batch data will be read from
 352      standard input. See the "BATCH MODE" section for details.
 353
 354 +0.  `--source-filter=COMMAND`
 355 +
 356 +    This option allows the user to specify a filter program that will be
 357 +    applied to the contents of all transferred regular files before the data is
 358 +    sent to destination.  COMMAND will receive the data on its standard input
 359 +    and it should write the filtered data to standard output.  COMMAND should
 360 +    exit non-zero if it cannot process the data or if it encounters an error
 361 +    when writing the data to stdout.
 362 +
 363 +    Example: `--source-filter="gzip -9"` will cause remote files to be
 364 +    compressed.  Use of `--source-filter` automatically enables `--whole-file`.
 365 +    If your filter does not output the same number of bytes that it received on
 366 +    input, you should use `--times-only` to disable size and content checks on
 367 +    subsequent rsync runs.
 368 +
 369 +0.  `--dest-filter=COMMAND`
 370 +
 371 +    This option allows you to specify a filter program that will be applied to
 372 +    the contents of all transferred regular files before the data is written to
 373 +    disk.  COMMAND will receive the data on its standard input and it should
 374 +    write the filtered data to standard output.  COMMAND should exit non-zero
 375 +    if it cannot process the data or if it encounters an error when writing the
 376 +    data to stdout.
 377 +
 378 +    Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 379 +    Use of --dest-filter automatically enables --whole-file.  If your filter
 380 +    does not output the same number of bytes that it received on input, you
 381 +    should use --times-only to disable size and content checks on subsequent
 382 +    rsync runs.
 383 +
 384  0. `--protocol=NUM`
 385
 386      Force an older protocol version to be used.  This is useful for creating a
 387 diff --git a/rsync.h b/rsync.h
 388 --- a/rsync.h
 389 +++ b/rsync.h
 390 @@ -158,6 +158,7 @@
 391  #define IOERR_DEL_LIMIT (1<<2)
 392
 393  #define MAX_ARGS 1000
 394 +#define MAX_FILTER_ARGS 100
 395  #define MAX_BASIS_DIRS 20
 396  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 397
 398 diff --git a/sender.c b/sender.c
 399 --- a/sender.c
 400 +++ b/sender.c
 401 @@ -47,6 +47,7 @@ extern int batch_fd;
 402  extern int write_batch;
 403  extern int file_old_total;
 404  extern BOOL want_progress_now;
 405 +extern char *source_filter;
 406  extern struct stats stats;
 407  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 408
 409 @@ -204,6 +205,26 @@ void send_files(int f_in, int f_out)
 410         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 411         int save_io_error = io_error;
 412         int ndx, j;
 413 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 414 +       char *tmp = 0;
 415 +       int unlink_tmp = 0;
 416 +
 417 +       if (source_filter) {
 418 +               char *p;
 419 +               char *sep = " \t";
 420 +               int i;
 421 +               for (p = strtok(source_filter, sep), i = 0;
 422 +                    p && i < MAX_FILTER_ARGS;
 423 +                    p = strtok(0, sep))
 424 +                       filter_argv[i++] = p;
 425 +               filter_argv[i] = NULL;
 426 +               if (p) {
 427 +                       rprintf(FERROR,
 428 +                               "Too many arguments to source-filter (> %d)\n",
 429 +                               MAX_FILTER_ARGS);
 430 +                       exit_cleanup(RERR_SYNTAX);
 431 +               }
 432 +       }
 433
 434         if (DEBUG_GTE(SEND, 1))
 435                 rprintf(FINFO, "send_files starting\n");
 436 @@ -340,6 +361,7 @@ void send_files(int f_in, int f_out)
 437                         exit_cleanup(RERR_PROTOCOL);
 438                 }
 439
 440 +               unlink_tmp = 0;
 441                 fd = do_open(fname, O_RDONLY, 0);
 442                 if (fd == -1) {
 443                         if (errno == ENOENT) {
 444 @@ -361,6 +383,33 @@ void send_files(int f_in, int f_out)
 445                         continue;
 446                 }
 447
 448 +               if (source_filter) {
 449 +                       int fd2;
 450 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 451 +
 452 +                       tmp = strdup(tmpl);
 453 +                       fd2 = mkstemp(tmp);
 454 +                       if (fd2 == -1) {
 455 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 456 +                                       tmp, strerror(errno));
 457 +                       } else {
 458 +                               int status;
 459 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 460 +                               close(fd);
 461 +                               close(fd2);
 462 +                               wait_process_with_flush(pid, &status);
 463 +                               if (status != 0) {
 464 +                                       rprintf(FERROR,
 465 +                                           "bypassing source filter %s; exited with code: %d\n",
 466 +                                           source_filter, status);
 467 +                                       fd = do_open(fname, O_RDONLY, 0);
 468 +                               } else {
 469 +                                       fd = do_open(tmp, O_RDONLY, 0);
 470 +                                       unlink_tmp = 1;
 471 +                               }
 472 +                       }
 473 +               }
 474 +
 475                 /* map the local file */
 476                 if (do_fstat(fd, &st) != 0) {
 477                         io_error |= IOERR_GENERAL;
 478 @@ -413,6 +462,8 @@ void send_files(int f_in, int f_out)
 479                         }
 480                 }
 481                 close(fd);
 482 +               if (unlink_tmp)
 483 +                       unlink(tmp);
 484
 485                 free_sums(s);
 486