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 diff --git a/generator.c b/generator.c
  34 index 12007a1..88bd5e7 100644
  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 @@ -524,7 +525,7 @@ 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 index 2ef2f47..e7b4a05 100644
  56 --- a/main.c
  57 +++ b/main.c
  58 @@ -140,7 +140,7 @@ pid_t wait_process(pid_t pid, int *status_ptr, int flags)
  59  }
  60
  61  /* Wait for a process to exit, calling io_flush while waiting. */
  62 -static void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  63 +void wait_process_with_flush(pid_t pid, int *exit_code_ptr)
  64  {
  65         pid_t waited_pid;
  66         int status;
  67 diff --git a/options.c b/options.c
  68 index e7c6c61..059bcbf 100644
  69 --- a/options.c
  70 +++ b/options.c
  71 @@ -105,6 +105,7 @@ int safe_symlinks = 0;
  72  int copy_unsafe_links = 0;
  73  int munge_symlinks = 0;
  74  int size_only = 0;
  75 +int times_only = 0;
  76  int daemon_bwlimit = 0;
  77  int bwlimit = 0;
  78  int fuzzy_basis = 0;
  79 @@ -164,6 +165,8 @@ char *logfile_name = NULL;
  80  char *logfile_format = NULL;
  81  char *stdout_format = NULL;
  82  char *password_file = NULL;
  83 +char *source_filter = NULL;
  84 +char *dest_filter = NULL;
  85  char *rsync_path = RSYNC_PATH;
  86  char *backup_dir = NULL;
  87  char backup_dir_buf[MAXPATHLEN];
  88 @@ -739,6 +742,7 @@ void usage(enum logcode F)
  89    rprintf(F," -I, --ignore-times          don't skip files that match in size and mod-time\n");
  90    rprintf(F," -M, --remote-option=OPTION  send OPTION to the remote side only\n");
  91    rprintf(F,"     --size-only             skip files that match in size\n");
  92 +  rprintf(F,"     --times-only            skip files that match in mod-time\n");
  93    rprintf(F,"     --modify-window=NUM     compare mod-times with reduced accuracy\n");
  94    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
  95    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
  96 @@ -778,6 +782,8 @@ void usage(enum logcode F)
  97    rprintf(F,"     --write-batch=FILE      write a batched update to FILE\n");
  98    rprintf(F,"     --only-write-batch=FILE like --write-batch but w/o updating destination\n");
  99    rprintf(F,"     --read-batch=FILE       read a batched update from FILE\n");
 100 +  rprintf(F,"     --source-filter=COMMAND filter file through COMMAND at source\n");
 101 +  rprintf(F,"     --dest-filter=COMMAND   filter file through COMMAND at destination\n");
 102    rprintf(F,"     --protocol=NUM          force an older protocol version to be used\n");
 103  #ifdef ICONV_OPTION
 104    rprintf(F,"     --iconv=CONVERT_SPEC    request charset conversion of filenames\n");
 105 @@ -887,6 +893,7 @@ static struct poptOption long_options[] = {
 106    {"chmod",            0,  POPT_ARG_STRING, 0, OPT_CHMOD, 0, 0 },
 107    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
 108    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
 109 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
 110    {"one-file-system", 'x', POPT_ARG_NONE,   0, 'x', 0, 0 },
 111    {"no-one-file-system",'x',POPT_ARG_VAL,   &one_file_system, 0, 0, 0 },
 112    {"no-x",            'x', POPT_ARG_VAL,    &one_file_system, 0, 0, 0 },
 113 @@ -1007,6 +1014,8 @@ static struct poptOption long_options[] = {
 114    {"password-file",    0,  POPT_ARG_STRING, &password_file, 0, 0, 0 },
 115    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
 116    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
 117 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
 118 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
 119    {"remote-option",   'M', POPT_ARG_STRING, 0, 'M', 0, 0 },
 120    {"protocol",         0,  POPT_ARG_INT,    &protocol_version, 0, 0, 0 },
 121    {"checksum-seed",    0,  POPT_ARG_INT,    &checksum_seed, 0, 0, 0 },
 122 @@ -2149,6 +2158,16 @@ int parse_arguments(int *argc_p, const char ***argv_p)
 123                 }
 124         }
 125
 126 +       if (source_filter || dest_filter) {
 127 +               if (whole_file == 0) {
 128 +                       snprintf(err_buf, sizeof err_buf,
 129 +                                "--no-whole-file cannot be used with --%s-filter\n",
 130 +                                source_filter ? "source" : "dest");
 131 +                       return 0;
 132 +               }
 133 +               whole_file = 1;
 134 +       }
 135 +
 136         if (files_from) {
 137                 char *h, *p;
 138                 int q;
 139 @@ -2493,6 +2512,25 @@ void server_options(char **args, int *argc_p)
 140         else if (missing_args == 1 && !am_sender)
 141                 args[ac++] = "--ignore-missing-args";
 142
 143 +       if (times_only && am_sender)
 144 +               args[ac++] = "--times-only";
 145 +
 146 +       if (source_filter && !am_sender) {
 147 +               /* Need to single quote the arg to keep the remote shell
 148 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 149 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 150 +                       goto oom;
 151 +               args[ac++] = arg;
 152 +       }
 153 +
 154 +       if (dest_filter && am_sender) {
 155 +               /* Need to single quote the arg to keep the remote shell
 156 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 157 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 158 +                       goto oom;
 159 +               args[ac++] = arg;
 160 +       }
 161 +
 162         if (modify_window_set) {
 163                 if (asprintf(&arg, "--modify-window=%d", modify_window) < 0)
 164                         goto oom;
 165 diff --git a/pipe.c b/pipe.c
 166 index a33117c..43eea31 100644
 167 --- a/pipe.c
 168 +++ b/pipe.c
 169 @@ -180,3 +180,77 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
 170
 171         return pid;
 172  }
 173 +
 174 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 175 +{
 176 +       pid_t pid;
 177 +       int pipefds[2];
 178 +
 179 +       if (DEBUG_GTE(CMD, 1))
 180 +               print_child_argv("opening connection using:", command);
 181 +
 182 +       if (pipe(pipefds) < 0) {
 183 +               rsyserr(FERROR, errno, "pipe");
 184 +               exit_cleanup(RERR_IPC);
 185 +       }
 186 +
 187 +       pid = do_fork();
 188 +       if (pid == -1) {
 189 +               rsyserr(FERROR, errno, "fork");
 190 +               exit_cleanup(RERR_IPC);
 191 +       }
 192 +
 193 +       if (pid == 0) {
 194 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 195 +                || close(pipefds[1]) < 0
 196 +                || dup2(out, STDOUT_FILENO) < 0) {
 197 +                       rsyserr(FERROR, errno, "Failed dup/close");
 198 +                       exit_cleanup(RERR_IPC);
 199 +               }
 200 +               umask(orig_umask);
 201 +               set_blocking(STDIN_FILENO);
 202 +               if (blocking_io)
 203 +                       set_blocking(STDOUT_FILENO);
 204 +               execvp(command[0], command);
 205 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 206 +               exit_cleanup(RERR_IPC);
 207 +       }
 208 +
 209 +       if (close(pipefds[0]) < 0) {
 210 +               rsyserr(FERROR, errno, "Failed to close");
 211 +               exit_cleanup(RERR_IPC);
 212 +       }
 213 +
 214 +       *pipe_to_filter = pipefds[1];
 215 +
 216 +       return pid;
 217 +}
 218 +
 219 +pid_t run_filter_on_file(char *command[], int out, int in)
 220 +{
 221 +       pid_t pid;
 222 +
 223 +       if (DEBUG_GTE(CMD, 1))
 224 +               print_child_argv("opening connection using:", command);
 225 +
 226 +       pid = do_fork();
 227 +       if (pid == -1) {
 228 +               rsyserr(FERROR, errno, "fork");
 229 +               exit_cleanup(RERR_IPC);
 230 +       }
 231 +
 232 +       if (pid == 0) {
 233 +               if (dup2(in, STDIN_FILENO) < 0
 234 +                || dup2(out, STDOUT_FILENO) < 0) {
 235 +                       rsyserr(FERROR, errno, "Failed to dup2");
 236 +                       exit_cleanup(RERR_IPC);
 237 +               }
 238 +               if (blocking_io)
 239 +                       set_blocking(STDOUT_FILENO);
 240 +               execvp(command[0], command);
 241 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 242 +               exit_cleanup(RERR_IPC);
 243 +       }
 244 +
 245 +       return pid;
 246 +}
 247 diff --git a/receiver.c b/receiver.c
 248 index 4325e30..e5ede73 100644
 249 --- a/receiver.c
 250 +++ b/receiver.c
 251 @@ -52,6 +52,7 @@ extern int delay_updates;
 252  extern mode_t orig_umask;
 253  extern struct stats stats;
 254  extern char *tmpdir;
 255 +extern char *dest_filter;
 256  extern char *partial_dir;
 257  extern char *basis_dir[MAX_BASIS_DIRS+1];
 258  extern char sender_file_sum[MAX_DIGEST_LEN];
 259 @@ -441,6 +442,8 @@ int recv_files(int f_in, char *local_name)
 260         const char *parent_dirname = "";
 261  #endif
 262         int ndx, recv_ok;
 263 +       pid_t pid = 0;
 264 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 265
 266         if (DEBUG_GTE(RECV, 1))
 267                 rprintf(FINFO, "recv_files(%d) starting\n", cur_flist->used);
 268 @@ -448,6 +451,23 @@ int recv_files(int f_in, char *local_name)
 269         if (delay_updates)
 270                 delayed_bits = bitbag_create(cur_flist->used + 1);
 271
 272 +       if (dest_filter) {
 273 +               char *p;
 274 +               char *sep = " \t";
 275 +               int i;
 276 +               for (p = strtok(dest_filter, sep), i = 0;
 277 +                    p && i < MAX_FILTER_ARGS;
 278 +                    p = strtok(0, sep))
 279 +                       filter_argv[i++] = p;
 280 +               filter_argv[i] = NULL;
 281 +               if (p) {
 282 +                       rprintf(FERROR,
 283 +                               "Too many arguments to dest-filter (> %d)\n",
 284 +                               MAX_FILTER_ARGS);
 285 +                       exit_cleanup(RERR_SYNTAX);
 286 +               }
 287 +       }
 288 +
 289         while (1) {
 290                 cleanup_disable();
 291
 292 @@ -742,6 +762,9 @@ int recv_files(int f_in, char *local_name)
 293                 else if (!am_server && INFO_GTE(NAME, 1) && INFO_EQ(PROGRESS, 1))
 294                         rprintf(FINFO, "%s\n", fname);
 295
 296 +               if (dest_filter)
 297 +                       pid = run_filter(filter_argv, fd2, &fd2);
 298 +
 299                 /* recv file data */
 300                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 301                                        fname, fd2, F_LENGTH(file));
 302 @@ -756,6 +779,16 @@ int recv_files(int f_in, char *local_name)
 303                         exit_cleanup(RERR_FILEIO);
 304                 }
 305
 306 +               if (dest_filter) {
 307 +                       int status;
 308 +                       wait_process_with_flush(pid, &status);
 309 +                       if (status != 0) {
 310 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 311 +                                       dest_filter, status);
 312 +                               continue;
 313 +                       }
 314 +               }
 315 +
 316                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 317                         if (partialptr == fname)
 318                                 partialptr = NULL;
 319 diff --git a/rsync.h b/rsync.h
 320 index be7cf8a..8d78818 100644
 321 --- a/rsync.h
 322 +++ b/rsync.h
 323 @@ -138,6 +138,7 @@
 324  #define IOERR_DEL_LIMIT (1<<2)
 325
 326  #define MAX_ARGS 1000
 327 +#define MAX_FILTER_ARGS 100
 328  #define MAX_BASIS_DIRS 20
 329  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 330
 331 diff --git a/rsync.yo b/rsync.yo
 332 index 941f7a5..edfad5e 100644
 333 --- a/rsync.yo
 334 +++ b/rsync.yo
 335 @@ -394,6 +394,7 @@ to the detailed description below for a complete description.  verb(
 336       --contimeout=SECONDS    set daemon connection timeout in seconds
 337   -I, --ignore-times          don't skip files that match size and time
 338       --size-only             skip files that match in size
 339 +     --times-only            skip files that match in mod-time
 340       --modify-window=NUM     compare mod-times with reduced accuracy
 341   -T, --temp-dir=DIR          create temporary files in directory DIR
 342   -y, --fuzzy                 find similar file for basis if no dest file
 343 @@ -434,6 +435,8 @@ to the detailed description below for a complete description.  verb(
 344       --write-batch=FILE      write a batched update to FILE
 345       --only-write-batch=FILE like --write-batch but w/o updating dest
 346       --read-batch=FILE       read a batched update from FILE
 347 +     --source-filter=COMMAND filter file through COMMAND at source
 348 +     --dest-filter=COMMAND   filter file through COMMAND at destination
 349       --protocol=NUM          force an older protocol version to be used
 350       --iconv=CONVERT_SPEC    request charset conversion of filenames
 351       --checksum-seed=NUM     set block/file checksum seed (advanced)
 352 @@ -2256,6 +2259,33 @@ file previously generated by bf(--write-batch).
 353  If em(FILE) is bf(-), the batch data will be read from standard input.
 354  See the "BATCH MODE" section for details.
 355
 356 +dit(bf(--source-filter=COMMAND)) This option allows the user to specify a
 357 +filter program that will be applied to the contents of all transferred
 358 +regular files before the data is sent to destination.  COMMAND will receive
 359 +the data on its standard input and it should write the filtered data to
 360 +standard output.  COMMAND should exit non-zero if it cannot process the
 361 +data or if it encounters an error when writing the data to stdout.
 362 +
 363 +Example: --source-filter="gzip -9" will cause remote files to be
 364 +compressed.
 365 +Use of --source-filter automatically enables --whole-file.
 366 +If your filter does not output the same number of bytes that it received on
 367 +input, you should use --times-only to disable size and content checks on
 368 +subsequent rsync runs.
 369 +
 370 +dit(bf(--dest-filter=COMMAND)) This option allows you to specify a filter
 371 +program that will be applied to the contents of all transferred regular
 372 +files before the data is written to disk.  COMMAND will receive the data on
 373 +its standard input and it should write the filtered data to standard
 374 +output.  COMMAND should exit non-zero if it cannot process the data or if
 375 +it encounters an error when writing the data to stdout.
 376 +
 377 +Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 378 +Use of --dest-filter automatically enables --whole-file.
 379 +If your filter does not output the same number of bytes that it
 380 +received on input, you should use --times-only to disable size and
 381 +content checks on subsequent rsync runs.
 382 +
 383  dit(bf(--protocol=NUM)) Force an older protocol version to be used.  This
 384  is useful for creating a batch file that is compatible with an older
 385  version of rsync.  For instance, if rsync 2.6.4 is being used with the
 386 diff --git a/sender.c b/sender.c
 387 index bf8221d..f315f80 100644
 388 --- a/sender.c
 389 +++ b/sender.c
 390 @@ -42,6 +42,7 @@ extern int make_backups;
 391  extern int inplace;
 392  extern int batch_fd;
 393  extern int write_batch;
 394 +extern char *source_filter;
 395  extern struct stats stats;
 396  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 397
 398 @@ -174,6 +175,26 @@ void send_files(int f_in, int f_out)
 399         enum logcode log_code = log_before_transfer ? FLOG : FINFO;
 400         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 401         int ndx, j;
 402 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 403 +       char *tmp = 0;
 404 +       int unlink_tmp = 0;
 405 +
 406 +       if (source_filter) {
 407 +               char *p;
 408 +               char *sep = " \t";
 409 +               int i;
 410 +               for (p = strtok(source_filter, sep), i = 0;
 411 +                    p && i < MAX_FILTER_ARGS;
 412 +                    p = strtok(0, sep))
 413 +                       filter_argv[i++] = p;
 414 +               filter_argv[i] = NULL;
 415 +               if (p) {
 416 +                       rprintf(FERROR,
 417 +                               "Too many arguments to source-filter (> %d)\n",
 418 +                               MAX_FILTER_ARGS);
 419 +                       exit_cleanup(RERR_SYNTAX);
 420 +               }
 421 +       }
 422
 423         if (DEBUG_GTE(SEND, 1))
 424                 rprintf(FINFO, "send_files starting\n");
 425 @@ -299,6 +320,7 @@ void send_files(int f_in, int f_out)
 426                         exit_cleanup(RERR_PROTOCOL);
 427                 }
 428
 429 +               unlink_tmp = 0;
 430                 fd = do_open(fname, O_RDONLY, 0);
 431                 if (fd == -1) {
 432                         if (errno == ENOENT) {
 433 @@ -320,6 +342,33 @@ void send_files(int f_in, int f_out)
 434                         continue;
 435                 }
 436
 437 +               if (source_filter) {
 438 +                       int fd2;
 439 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 440 +
 441 +                       tmp = strdup(tmpl);
 442 +                       fd2 = mkstemp(tmp);
 443 +                       if (fd2 == -1) {
 444 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 445 +                                       tmp, strerror(errno));
 446 +                       } else {
 447 +                               int status;
 448 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 449 +                               close(fd);
 450 +                               close(fd2);
 451 +                               wait_process_with_flush(pid, &status);
 452 +                               if (status != 0) {
 453 +                                       rprintf(FERROR,
 454 +                                           "bypassing source filter %s; exited with code: %d\n",
 455 +                                           source_filter, status);
 456 +                                       fd = do_open(fname, O_RDONLY, 0);
 457 +                               } else {
 458 +                                       fd = do_open(tmp, O_RDONLY, 0);
 459 +                                       unlink_tmp = 1;
 460 +                               }
 461 +                       }
 462 +               }
 463 +
 464                 /* map the local file */
 465                 if (do_fstat(fd, &st) != 0) {
 466                         io_error |= IOERR_GENERAL;
 467 @@ -370,6 +419,8 @@ void send_files(int f_in, int f_out)
 468                         }
 469                 }
 470                 close(fd);
 471 +               if (unlink_tmp)
 472 +                       unlink(tmp);
 473
 474                 free_sums(s);
 475