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: ee51a745c163f3c422a30b22f4beda0e1ead7c20
  34 diff --git a/generator.c b/generator.c
  35 --- a/generator.c
  36 +++ b/generator.c
  37 @@ -62,6 +62,7 @@ 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 @@ -567,7 +568,7 @@ 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 @@ -154,7 +154,7 @@ pid_t wait_process(pid_t pid, int *status_ptr, int flags)
  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 @@ -106,6 +106,7 @@ 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 @@ -165,6 +166,8 @@ 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 @@ -754,6 +757,7 @@ 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     compare mod-times with reduced accuracy\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 @@ -796,6 +800,8 @@ 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 @@ -908,6 +914,7 @@ 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 @@ -1029,6 +1036,8 @@ 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 @@ -2279,6 +2288,16 @@ 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 @@ -2630,6 +2649,25 @@ 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) {
 161                 if (asprintf(&arg, "--modify-window=%d", modify_window) < 0)
 162                         goto oom;
 163 diff --git a/pipe.c b/pipe.c
 164 --- a/pipe.c
 165 +++ b/pipe.c
 166 @@ -178,3 +178,77 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
 167
 168         return pid;
 169  }
 170 +
 171 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 172 +{
 173 +       pid_t pid;
 174 +       int pipefds[2];
 175 +
 176 +       if (DEBUG_GTE(CMD, 1))
 177 +               print_child_argv("opening connection using:", command);
 178 +
 179 +       if (pipe(pipefds) < 0) {
 180 +               rsyserr(FERROR, errno, "pipe");
 181 +               exit_cleanup(RERR_IPC);
 182 +       }
 183 +
 184 +       pid = do_fork();
 185 +       if (pid == -1) {
 186 +               rsyserr(FERROR, errno, "fork");
 187 +               exit_cleanup(RERR_IPC);
 188 +       }
 189 +
 190 +       if (pid == 0) {
 191 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 192 +                || close(pipefds[1]) < 0
 193 +                || dup2(out, STDOUT_FILENO) < 0) {
 194 +                       rsyserr(FERROR, errno, "Failed dup/close");
 195 +                       exit_cleanup(RERR_IPC);
 196 +               }
 197 +               umask(orig_umask);
 198 +               set_blocking(STDIN_FILENO);
 199 +               if (blocking_io)
 200 +                       set_blocking(STDOUT_FILENO);
 201 +               execvp(command[0], command);
 202 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 203 +               exit_cleanup(RERR_IPC);
 204 +       }
 205 +
 206 +       if (close(pipefds[0]) < 0) {
 207 +               rsyserr(FERROR, errno, "Failed to close");
 208 +               exit_cleanup(RERR_IPC);
 209 +       }
 210 +
 211 +       *pipe_to_filter = pipefds[1];
 212 +
 213 +       return pid;
 214 +}
 215 +
 216 +pid_t run_filter_on_file(char *command[], int out, int in)
 217 +{
 218 +       pid_t pid;
 219 +
 220 +       if (DEBUG_GTE(CMD, 1))
 221 +               print_child_argv("opening connection using:", command);
 222 +
 223 +       pid = do_fork();
 224 +       if (pid == -1) {
 225 +               rsyserr(FERROR, errno, "fork");
 226 +               exit_cleanup(RERR_IPC);
 227 +       }
 228 +
 229 +       if (pid == 0) {
 230 +               if (dup2(in, STDIN_FILENO) < 0
 231 +                || dup2(out, STDOUT_FILENO) < 0) {
 232 +                       rsyserr(FERROR, errno, "Failed to dup2");
 233 +                       exit_cleanup(RERR_IPC);
 234 +               }
 235 +               if (blocking_io)
 236 +                       set_blocking(STDOUT_FILENO);
 237 +               execvp(command[0], command);
 238 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 239 +               exit_cleanup(RERR_IPC);
 240 +       }
 241 +
 242 +       return pid;
 243 +}
 244 diff --git a/receiver.c b/receiver.c
 245 --- a/receiver.c
 246 +++ b/receiver.c
 247 @@ -55,6 +55,7 @@ extern int delay_updates;
 248  extern mode_t orig_umask;
 249  extern struct stats stats;
 250  extern char *tmpdir;
 251 +extern char *dest_filter;
 252  extern char *partial_dir;
 253  extern char *basis_dir[MAX_BASIS_DIRS+1];
 254  extern char sender_file_sum[MAX_DIGEST_LEN];
 255 @@ -530,6 +531,8 @@ int recv_files(int f_in, int f_out, char *local_name)
 256         const char *parent_dirname = "";
 257  #endif
 258         int ndx, recv_ok;
 259 +       pid_t pid = 0;
 260 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 261
 262         if (DEBUG_GTE(RECV, 1))
 263                 rprintf(FINFO, "recv_files(%d) starting\n", cur_flist->used);
 264 @@ -537,6 +540,23 @@ int recv_files(int f_in, int f_out, char *local_name)
 265         if (delay_updates)
 266                 delayed_bits = bitbag_create(cur_flist->used + 1);
 267
 268 +       if (dest_filter) {
 269 +               char *p;
 270 +               char *sep = " \t";
 271 +               int i;
 272 +               for (p = strtok(dest_filter, sep), i = 0;
 273 +                    p && i < MAX_FILTER_ARGS;
 274 +                    p = strtok(0, sep))
 275 +                       filter_argv[i++] = p;
 276 +               filter_argv[i] = NULL;
 277 +               if (p) {
 278 +                       rprintf(FERROR,
 279 +                               "Too many arguments to dest-filter (> %d)\n",
 280 +                               MAX_FILTER_ARGS);
 281 +                       exit_cleanup(RERR_SYNTAX);
 282 +               }
 283 +       }
 284 +
 285         while (1) {
 286                 cleanup_disable();
 287
 288 @@ -841,6 +861,9 @@ int recv_files(int f_in, int f_out, char *local_name)
 289                 else if (!am_server && INFO_GTE(NAME, 1) && INFO_EQ(PROGRESS, 1))
 290                         rprintf(FINFO, "%s\n", fname);
 291
 292 +               if (dest_filter)
 293 +                       pid = run_filter(filter_argv, fd2, &fd2);
 294 +
 295                 /* recv file data */
 296                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 297                                        fname, fd2, F_LENGTH(file));
 298 @@ -855,6 +878,16 @@ int recv_files(int f_in, int f_out, char *local_name)
 299                         exit_cleanup(RERR_FILEIO);
 300                 }
 301
 302 +               if (dest_filter) {
 303 +                       int status;
 304 +                       wait_process_with_flush(pid, &status);
 305 +                       if (status != 0) {
 306 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 307 +                                       dest_filter, status);
 308 +                               continue;
 309 +                       }
 310 +               }
 311 +
 312                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 313                         if (partialptr == fname)
 314                                 partialptr = NULL;
 315 diff --git a/rsync.h b/rsync.h
 316 --- a/rsync.h
 317 +++ b/rsync.h
 318 @@ -148,6 +148,7 @@
 319  #define IOERR_DEL_LIMIT (1<<2)
 320
 321  #define MAX_ARGS 1000
 322 +#define MAX_FILTER_ARGS 100
 323  #define MAX_BASIS_DIRS 20
 324  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 325
 326 diff --git a/rsync.yo b/rsync.yo
 327 --- a/rsync.yo
 328 +++ b/rsync.yo
 329 @@ -410,6 +410,7 @@ to the detailed description below for a complete description.  verb(
 330       --contimeout=SECONDS    set daemon connection timeout in seconds
 331   -I, --ignore-times          don't skip files that match size and time
 332       --size-only             skip files that match in size
 333 +     --times-only            skip files that match in mod-time
 334       --modify-window=NUM     compare mod-times with reduced accuracy
 335   -T, --temp-dir=DIR          create temporary files in directory DIR
 336   -y, --fuzzy                 find similar file for basis if no dest file
 337 @@ -451,6 +452,8 @@ to the detailed description below for a complete description.  verb(
 338       --write-batch=FILE      write a batched update to FILE
 339       --only-write-batch=FILE like --write-batch but w/o updating dest
 340       --read-batch=FILE       read a batched update from FILE
 341 +     --source-filter=COMMAND filter file through COMMAND at source
 342 +     --dest-filter=COMMAND   filter file through COMMAND at destination
 343       --protocol=NUM          force an older protocol version to be used
 344       --iconv=CONVERT_SPEC    request charset conversion of filenames
 345       --checksum-seed=NUM     set block/file checksum seed (advanced)
 346 @@ -2495,6 +2498,33 @@ file previously generated by bf(--write-batch).
 347  If em(FILE) is bf(-), the batch data will be read from standard input.
 348  See the "BATCH MODE" section for details.
 349
 350 +dit(bf(--source-filter=COMMAND)) This option allows the user to specify a
 351 +filter program that will be applied to the contents of all transferred
 352 +regular files before the data is sent to destination.  COMMAND will receive
 353 +the data on its standard input and it should write the filtered data to
 354 +standard output.  COMMAND should exit non-zero if it cannot process the
 355 +data or if it encounters an error when writing the data to stdout.
 356 +
 357 +Example: --source-filter="gzip -9" will cause remote files to be
 358 +compressed.
 359 +Use of --source-filter automatically enables --whole-file.
 360 +If your filter does not output the same number of bytes that it received on
 361 +input, you should use --times-only to disable size and content checks on
 362 +subsequent rsync runs.
 363 +
 364 +dit(bf(--dest-filter=COMMAND)) This option allows you to specify a filter
 365 +program that will be applied to the contents of all transferred regular
 366 +files before the data is written to disk.  COMMAND will receive the data on
 367 +its standard input and it should write the filtered data to standard
 368 +output.  COMMAND should exit non-zero if it cannot process the data or if
 369 +it encounters an error when writing the data to stdout.
 370 +
 371 +Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 372 +Use of --dest-filter automatically enables --whole-file.
 373 +If your filter does not output the same number of bytes that it
 374 +received on input, you should use --times-only to disable size and
 375 +content checks on subsequent rsync runs.
 376 +
 377  dit(bf(--protocol=NUM)) Force an older protocol version to be used.  This
 378  is useful for creating a batch file that is compatible with an older
 379  version of rsync.  For instance, if rsync 2.6.4 is being used with the
 380 diff --git a/sender.c b/sender.c
 381 --- a/sender.c
 382 +++ b/sender.c
 383 @@ -43,6 +43,7 @@ extern int inplace;
 384  extern int batch_fd;
 385  extern int write_batch;
 386  extern int file_old_total;
 387 +extern char *source_filter;
 388  extern struct stats stats;
 389  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 390
 391 @@ -200,6 +201,26 @@ void send_files(int f_in, int f_out)
 392         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 393         int save_io_error = io_error;
 394         int ndx, j;
 395 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 396 +       char *tmp = 0;
 397 +       int unlink_tmp = 0;
 398 +
 399 +       if (source_filter) {
 400 +               char *p;
 401 +               char *sep = " \t";
 402 +               int i;
 403 +               for (p = strtok(source_filter, sep), i = 0;
 404 +                    p && i < MAX_FILTER_ARGS;
 405 +                    p = strtok(0, sep))
 406 +                       filter_argv[i++] = p;
 407 +               filter_argv[i] = NULL;
 408 +               if (p) {
 409 +                       rprintf(FERROR,
 410 +                               "Too many arguments to source-filter (> %d)\n",
 411 +                               MAX_FILTER_ARGS);
 412 +                       exit_cleanup(RERR_SYNTAX);
 413 +               }
 414 +       }
 415
 416         if (DEBUG_GTE(SEND, 1))
 417                 rprintf(FINFO, "send_files starting\n");
 418 @@ -334,6 +355,7 @@ void send_files(int f_in, int f_out)
 419                         exit_cleanup(RERR_PROTOCOL);
 420                 }
 421
 422 +               unlink_tmp = 0;
 423                 fd = do_open(fname, O_RDONLY, 0);
 424                 if (fd == -1) {
 425                         if (errno == ENOENT) {
 426 @@ -355,6 +377,33 @@ void send_files(int f_in, int f_out)
 427                         continue;
 428                 }
 429
 430 +               if (source_filter) {
 431 +                       int fd2;
 432 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 433 +
 434 +                       tmp = strdup(tmpl);
 435 +                       fd2 = mkstemp(tmp);
 436 +                       if (fd2 == -1) {
 437 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 438 +                                       tmp, strerror(errno));
 439 +                       } else {
 440 +                               int status;
 441 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 442 +                               close(fd);
 443 +                               close(fd2);
 444 +                               wait_process_with_flush(pid, &status);
 445 +                               if (status != 0) {
 446 +                                       rprintf(FERROR,
 447 +                                           "bypassing source filter %s; exited with code: %d\n",
 448 +                                           source_filter, status);
 449 +                                       fd = do_open(fname, O_RDONLY, 0);
 450 +                               } else {
 451 +                                       fd = do_open(tmp, O_RDONLY, 0);
 452 +                                       unlink_tmp = 1;
 453 +                               }
 454 +                       }
 455 +               }
 456 +
 457                 /* map the local file */
 458                 if (do_fstat(fd, &st) != 0) {
 459                         io_error |= IOERR_GENERAL;
 460 @@ -405,6 +454,8 @@ void send_files(int f_in, int f_out)
 461                         }
 462                 }
 463                 close(fd);
 464 +               if (unlink_tmp)
 465 +                       unlink(tmp);
 466
 467                 free_sums(s);
 468