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 @@ -61,6 +61,7 @@ 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 @@ -379,7 +380,7 @@ 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/options.c
  52 +++ new/options.c
  53 @@ -98,6 +98,7 @@ int keep_partial = 0;
  54  int safe_symlinks = 0;
  55  int copy_unsafe_links = 0;
  56  int size_only = 0;
  57 +int times_only = 0;
  58  int daemon_bwlimit = 0;
  59  int bwlimit = 0;
  60  int fuzzy_basis = 0;
  61 @@ -147,6 +148,8 @@ char *basis_dir[MAX_BASIS_DIRS+1];
  62  char *config_file = NULL;
  63  char *shell_cmd = NULL;
  64  char *log_format = NULL;
  65 +char *source_filter = NULL;
  66 +char *dest_filter = NULL;
  67  char *password_file = NULL;
  68  char *rsync_path = RSYNC_PATH;
  69  char *backup_dir = NULL;
  70 @@ -331,6 +334,7 @@ void usage(enum logcode F)
  71    rprintf(F,"     --timeout=TIME          set I/O timeout in seconds\n");
  72    rprintf(F," -I, --ignore-times          don't skip files that match in size and mod-time\n");
  73    rprintf(F,"     --size-only             skip files that match in size\n");
  74 +  rprintf(F,"     --times-only            skip files that match in mod-time\n");
  75    rprintf(F,"     --modify-window=NUM     compare mod-times with reduced accuracy\n");
  76    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
  77    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
  78 @@ -366,6 +370,8 @@ void usage(enum logcode F)
  79    rprintf(F,"     --write-batch=FILE      write a batched update to FILE\n");
  80    rprintf(F,"     --only-write-batch=FILE like --write-batch but w/o updating destination\n");
  81    rprintf(F,"     --read-batch=FILE       read a batched update from FILE\n");
  82 +  rprintf(F,"     --source-filter=COMMAND filter file through COMMAND at source\n");
  83 +  rprintf(F,"     --dest-filter=COMMAND   filter file through COMMAND at destination\n");
  84    rprintf(F,"     --protocol=NUM          force an older protocol version to be used\n");
  85  #ifdef INET6
  86    rprintf(F," -4, --ipv4                  prefer IPv4\n");
  87 @@ -443,6 +449,7 @@ static struct poptOption long_options[]
  88    {"chmod",            0,  POPT_ARG_STRING, 0, OPT_CHMOD, 0, 0 },
  89    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
  90    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
  91 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
  92    {"one-file-system", 'x', POPT_ARG_NONE,   0, 'x', 0, 0 },
  93    {"update",          'u', POPT_ARG_NONE,   &update_only, 0, 0, 0 },
  94    {"existing",         0,  POPT_ARG_NONE,   &ignore_non_existing, 0, 0, 0 },
  95 @@ -516,6 +523,8 @@ static struct poptOption long_options[]
  96    {"password-file",    0,  POPT_ARG_STRING, &password_file, 0, 0, 0 },
  97    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
  98    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
  99 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
 100 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
 101    {"protocol",         0,  POPT_ARG_INT,    &protocol_version, 0, 0, 0 },
 102    {"checksum-seed",    0,  POPT_ARG_INT,    &checksum_seed, 0, 0, 0 },
 103    {"server",           0,  POPT_ARG_NONE,   0, OPT_SERVER, 0, 0 },
 104 @@ -1380,6 +1389,16 @@ int parse_arguments(int *argc, const cha
 105                 }
 106         }
 107
 108 +       if (source_filter || dest_filter) {
 109 +               if (whole_file == 0) {
 110 +                       snprintf(err_buf, sizeof err_buf,
 111 +                                "--no-whole-file cannot be used with --%s-filter\n",
 112 +                                source_filter ? "source" : "dest");
 113 +                       return 0;
 114 +               }
 115 +               whole_file = 1;
 116 +       }
 117 +
 118         if (files_from) {
 119                 char *h, *p;
 120                 int q;
 121 @@ -1640,6 +1659,25 @@ void server_options(char **args,int *arg
 122                         args[ac++] = "--super";
 123         }
 124
 125 +       if (times_only && am_sender)
 126 +               args[ac++] = "--times-only";
 127 +
 128 +       if (source_filter && !am_sender) {
 129 +               /* Need to single quote the arg to keep the remote shell
 130 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 131 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 132 +                       goto oom;
 133 +               args[ac++] = arg;
 134 +       }
 135 +
 136 +       if (dest_filter && am_sender) {
 137 +               /* Need to single quote the arg to keep the remote shell
 138 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 139 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 140 +                       goto oom;
 141 +               args[ac++] = arg;
 142 +       }
 143 +
 144         if (size_only)
 145                 args[ac++] = "--size-only";
 146
 147 --- old/pipe.c
 148 +++ new/pipe.c
 149 @@ -157,3 +157,77 @@ pid_t local_child(int argc, char **argv,
 150
 151         return pid;
 152  }
 153 +
 154 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 155 +{
 156 +       pid_t pid;
 157 +       int pipefds[2];
 158 +
 159 +       if (verbose >= 2)
 160 +               print_child_argv(command);
 161 +
 162 +       if (pipe(pipefds) < 0) {
 163 +               rsyserr(FERROR, errno, "pipe");
 164 +               exit_cleanup(RERR_IPC);
 165 +       }
 166 +
 167 +       pid = do_fork();
 168 +       if (pid == -1) {
 169 +               rsyserr(FERROR, errno, "fork");
 170 +               exit_cleanup(RERR_IPC);
 171 +       }
 172 +
 173 +       if (pid == 0) {
 174 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 175 +                || close(pipefds[1]) < 0
 176 +                || dup2(out, STDOUT_FILENO) < 0) {
 177 +                       rsyserr(FERROR, errno, "Failed dup/close");
 178 +                       exit_cleanup(RERR_IPC);
 179 +               }
 180 +               umask(orig_umask);
 181 +               set_blocking(STDIN_FILENO);
 182 +               if (blocking_io)
 183 +                       set_blocking(STDOUT_FILENO);
 184 +               execvp(command[0], command);
 185 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 186 +               exit_cleanup(RERR_IPC);
 187 +       }
 188 +
 189 +       if (close(pipefds[0]) < 0) {
 190 +               rsyserr(FERROR, errno, "Failed to close");
 191 +               exit_cleanup(RERR_IPC);
 192 +       }
 193 +
 194 +       *pipe_to_filter = pipefds[1];
 195 +
 196 +       return pid;
 197 +}
 198 +
 199 +pid_t run_filter_on_file(char *command[], int out, int in)
 200 +{
 201 +       pid_t pid;
 202 +
 203 +       if (verbose >= 2)
 204 +               print_child_argv(command);
 205 +
 206 +       pid = do_fork();
 207 +       if (pid == -1) {
 208 +               rsyserr(FERROR, errno, "fork");
 209 +               exit_cleanup(RERR_IPC);
 210 +       }
 211 +
 212 +       if (pid == 0) {
 213 +               if (dup2(in, STDIN_FILENO) < 0
 214 +                || dup2(out, STDOUT_FILENO) < 0) {
 215 +                       rsyserr(FERROR, errno, "Failed to dup2");
 216 +                       exit_cleanup(RERR_IPC);
 217 +               }
 218 +               if (blocking_io)
 219 +                       set_blocking(STDOUT_FILENO);
 220 +               execvp(command[0], command);
 221 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 222 +               exit_cleanup(RERR_IPC);
 223 +       }
 224 +
 225 +       return pid;
 226 +}
 227 --- old/receiver.c
 228 +++ new/receiver.c
 229 @@ -53,6 +53,7 @@ extern int inplace;
 230  extern int delay_updates;
 231  extern struct stats stats;
 232  extern char *log_format;
 233 +extern char *dest_filter;
 234  extern char *tmpdir;
 235  extern char *partial_dir;
 236  extern char *basis_dir[];
 237 @@ -411,6 +412,8 @@ int recv_files(int f_in, struct file_lis
 238                       : !am_server && log_format_has_i;
 239         int max_phase = protocol_version >= 29 ? 2 : 1;
 240         int i, recv_ok;
 241 +       pid_t pid = 0;
 242 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 243
 244         if (verbose > 2)
 245                 rprintf(FINFO,"recv_files(%d) starting\n",flist->count);
 246 @@ -423,6 +426,23 @@ int recv_files(int f_in, struct file_lis
 247         if (delay_updates)
 248                 init_delayed_bits(flist->count);
 249
 250 +       if (dest_filter) {
 251 +               char *p;
 252 +               char *sep = " \t";
 253 +               int i;
 254 +               for (p = strtok(dest_filter, sep), i = 0;
 255 +                    p && i < MAX_FILTER_ARGS;
 256 +                    p = strtok(0, sep))
 257 +                       filter_argv[i++] = p;
 258 +               filter_argv[i] = NULL;
 259 +               if (p) {
 260 +                       rprintf(FERROR,
 261 +                               "Too many arguments to dest-filter (> %d)\n",
 262 +                               MAX_FILTER_ARGS);
 263 +                       exit_cleanup(RERR_SYNTAX);
 264 +               }
 265 +       }
 266 +
 267         while (1) {
 268                 cleanup_disable();
 269
 270 @@ -665,6 +685,9 @@ int recv_files(int f_in, struct file_lis
 271                 else if (!am_server && verbose && do_progress)
 272                         rprintf(FINFO, "%s\n", fname);
 273
 274 +               if (dest_filter)
 275 +                       pid = run_filter(filter_argv, fd2, &fd2);
 276 +
 277                 /* recv file data */
 278                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 279                                        fname, fd2, file->length);
 280 @@ -680,6 +703,16 @@ int recv_files(int f_in, struct file_lis
 281                         exit_cleanup(RERR_FILEIO);
 282                 }
 283
 284 +               if (dest_filter) {
 285 +                       int status;
 286 +                       wait_process(pid, &status);
 287 +                       if (status != 0) {
 288 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 289 +                                       dest_filter, status);
 290 +                               continue;
 291 +                       }
 292 +               }
 293 +
 294                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 295                         if (partialptr == fname || *partial_dir == '/')
 296                                 partialptr = NULL;
 297 --- old/rsync.h
 298 +++ new/rsync.h
 299 @@ -103,6 +103,7 @@
 300  #define IOERR_DEL_LIMIT (1<<2)
 301
 302  #define MAX_ARGS 1000
 303 +#define MAX_FILTER_ARGS 100
 304  #define MAX_BASIS_DIRS 20
 305  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 306
 307 --- old/rsync.yo
 308 +++ new/rsync.yo
 309 @@ -355,6 +355,7 @@ to the detailed description below for a
 310       --timeout=TIME          set I/O timeout in seconds
 311   -I, --ignore-times          don't skip files that match size and time
 312       --size-only             skip files that match in size
 313 +     --times-only            skip files that match in mod-time
 314       --modify-window=NUM     compare mod-times with reduced accuracy
 315   -T, --temp-dir=DIR          create temporary files in directory DIR
 316   -y, --fuzzy                 find similar file for basis if no dest file
 317 @@ -390,6 +391,8 @@ to the detailed description below for a
 318       --write-batch=FILE      write a batched update to FILE
 319       --only-write-batch=FILE like --write-batch but w/o updating dest
 320       --read-batch=FILE       read a batched update from FILE
 321 +     --source-filter=COMMAND filter file through COMMAND at source
 322 +     --dest-filter=COMMAND   filter file through COMMAND at destination
 323       --protocol=NUM          force an older protocol version to be used
 324       --checksum-seed=NUM     set block/file checksum seed (advanced)
 325   -4, --ipv4                  prefer IPv4
 326 @@ -1598,6 +1601,33 @@ file previously generated by bf(--write-
 327  If em(FILE) is bf(-), the batch data will be read from standard input.
 328  See the "BATCH MODE" section for details.
 329
 330 +dit(bf(--source-filter=COMMAND)) This option allows the user to specify a
 331 +filter program that will be applied to the contents of all transferred
 332 +regular files before the data is sent to destination.  COMMAND will receive
 333 +the data on its standard input and it should write the filtered data to
 334 +standard output.  COMMAND should exit non-zero if it cannot process the
 335 +data or if it encounters an error when writing the data to stdout.
 336 +
 337 +Example: --source-filter="gzip -9" will cause remote files to be
 338 +compressed.
 339 +Use of --source-filter automatically enables --whole-file.
 340 +If your filter does not output the same number of bytes that it received on
 341 +input, you should use --times-only to disable size and content checks on
 342 +subsequent rsync runs.
 343 +
 344 +dit(bf(--dest-filter=COMMAND)) This option allows you to specify a filter
 345 +program that will be applied to the contents of all transferred regular
 346 +files before the data is written to disk.  COMMAND will receive the data on
 347 +its standard input and it should write the filtered data to standard
 348 +output.  COMMAND should exit non-zero if it cannot process the data or if
 349 +it encounters an error when writing the data to stdout.
 350 +
 351 +Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 352 +Use of --dest-filter automatically enables --whole-file.
 353 +If your filter does not output the same number of bytes that it
 354 +received on input, you should use --times-only to disable size and
 355 +content checks on subsequent rsync runs.
 356 +
 357  dit(bf(--protocol=NUM)) Force an older protocol version to be used.  This
 358  is useful for creating a batch file that is compatible with an older
 359  version of rsync.  For instance, if rsync 2.6.4 is being used with the
 360 --- old/sender.c
 361 +++ new/sender.c
 362 @@ -41,6 +41,7 @@ extern int write_batch;
 363  extern struct stats stats;
 364  extern struct file_list *the_file_list;
 365  extern char *log_format;
 366 +extern char *source_filter;
 367
 368
 369  /**
 370 @@ -219,6 +220,26 @@ void send_files(struct file_list *flist,
 371                       : !am_server && log_format_has_i;
 372         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 373         int i, j;
 374 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 375 +       char *tmp = 0;
 376 +       int unlink_tmp = 0;
 377 +
 378 +       if (source_filter) {
 379 +               char *p;
 380 +               char *sep = " \t";
 381 +               int i;
 382 +               for (p = strtok(source_filter, sep), i = 0;
 383 +                    p && i < MAX_FILTER_ARGS;
 384 +                    p = strtok(0, sep))
 385 +                       filter_argv[i++] = p;
 386 +               filter_argv[i] = NULL;
 387 +               if (p) {
 388 +                       rprintf(FERROR,
 389 +                               "Too many arguments to source-filter (> %d)\n",
 390 +                               MAX_FILTER_ARGS);
 391 +                       exit_cleanup(RERR_SYNTAX);
 392 +               }
 393 +       }
 394
 395         if (verbose > 2)
 396                 rprintf(FINFO, "send_files starting\n");
 397 @@ -293,6 +314,7 @@ void send_files(struct file_list *flist,
 398                         return;
 399                 }
 400
 401 +               unlink_tmp = 0;
 402                 fd = do_open(fname, O_RDONLY, 0);
 403                 if (fd == -1) {
 404                         if (errno == ENOENT) {
 405 @@ -321,6 +343,33 @@ void send_files(struct file_list *flist,
 406                         return;
 407                 }
 408
 409 +               if (source_filter) {
 410 +                       int fd2;
 411 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 412 +
 413 +                       tmp = strdup(tmpl);
 414 +                       fd2 = mkstemp(tmp);
 415 +                       if (fd2 == -1) {
 416 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 417 +                                       tmp, strerror(errno));
 418 +                       } else {
 419 +                               int status;
 420 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 421 +                               close(fd);
 422 +                               close(fd2);
 423 +                               wait_process(pid, &status);
 424 +                               if (status != 0) {
 425 +                                       rprintf(FERROR,
 426 +                                           "bypassing source filter %s; exited with code: %d\n",
 427 +                                           source_filter, status);
 428 +                                       fd = do_open(fname, O_RDONLY, 0);
 429 +                               } else {
 430 +                                       fd = do_open(tmp, O_RDONLY, 0);
 431 +                                       unlink_tmp = 1;
 432 +                               }
 433 +                       }
 434 +               }
 435 +
 436                 if (st.st_size) {
 437                         int32 read_size = MAX(s->blength * 3, MAX_MAP_SIZE);
 438                         mbuf = map_file(fd, st.st_size, read_size, s->blength);
 439 @@ -363,6 +412,8 @@ void send_files(struct file_list *flist,
 440                         }
 441                 }
 442                 close(fd);
 443 +               if (unlink_tmp)
 444 +                       unlink(tmp);
 445
 446                 free_sums(s);
 447