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: 6c8ca91c731b7bf2b081694bda85b7dadc2b7aff
  34 diff --git a/generator.c b/generator.c
  35 --- a/generator.c
  36 +++ b/generator.c
  37 @@ -67,6 +67,7 @@ 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 @@ -618,7 +619,7 @@ int quick_check_ok(enum filetype ftype, const char *fn, struct file_struct *file
  46  {
  47         switch (ftype) {
  48           case FT_REG:
  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 @@ -191,7 +191,7 @@ 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 @@ -119,6 +119,7 @@ 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 @@ -179,6 +180,8 @@ 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 *early_input_file = NULL;
  84  char *rsync_path = RSYNC_PATH;
  85  char *backup_dir = NULL;
  86 @@ -689,6 +692,7 @@ static struct poptOption long_options[] = {
  87    {"chmod",            0,  POPT_ARG_STRING, 0, OPT_CHMOD, 0, 0 },
  88    {"ignore-times",    'I', POPT_ARG_NONE,   &ignore_times, 0, 0, 0 },
  89    {"size-only",        0,  POPT_ARG_NONE,   &size_only, 0, 0, 0 },
  90 +  {"times-only",       0,  POPT_ARG_NONE,   &times_only , 0, 0, 0 },
  91    {"one-file-system", 'x', POPT_ARG_NONE,   0, 'x', 0, 0 },
  92    {"no-one-file-system",0, POPT_ARG_VAL,    &one_file_system, 0, 0, 0 },
  93    {"no-x",             0,  POPT_ARG_VAL,    &one_file_system, 0, 0, 0 },
  94 @@ -829,6 +833,8 @@ static struct poptOption long_options[] = {
  95    {"early-input",      0,  POPT_ARG_STRING, &early_input_file, 0, 0, 0 },
  96    {"blocking-io",      0,  POPT_ARG_VAL,    &blocking_io, 1, 0, 0 },
  97    {"no-blocking-io",   0,  POPT_ARG_VAL,    &blocking_io, 0, 0, 0 },
  98 +  {"source-filter",    0,  POPT_ARG_STRING, &source_filter, 0, 0, 0 },
  99 +  {"dest-filter",      0,  POPT_ARG_STRING, &dest_filter, 0, 0, 0 },
 100    {"outbuf",           0,  POPT_ARG_STRING, &outbuf_mode, 0, 0, 0 },
 101    {"remote-option",   'M', POPT_ARG_STRING, 0, 'M', 0, 0 },
 102    {"protocol",         0,  POPT_ARG_INT,    &protocol_version, 0, 0, 0 },
 103 @@ -2444,6 +2450,16 @@ int parse_arguments(int *argc_p, const char ***argv_p)
 104                 }
 105         }
 106
 107 +       if (source_filter || dest_filter) {
 108 +               if (whole_file == 0) {
 109 +                       snprintf(err_buf, sizeof err_buf,
 110 +                                "--no-whole-file cannot be used with --%s-filter\n",
 111 +                                source_filter ? "source" : "dest");
 112 +                       return 0;
 113 +               }
 114 +               whole_file = 1;
 115 +       }
 116 +
 117         if (files_from) {
 118                 char *h, *p;
 119                 int q;
 120 @@ -2852,6 +2868,25 @@ void server_options(char **args, int *argc_p)
 121         else if (missing_args == 1 && !am_sender)
 122                 args[ac++] = "--ignore-missing-args";
 123
 124 +       if (times_only && am_sender)
 125 +               args[ac++] = "--times-only";
 126 +
 127 +       if (source_filter && !am_sender) {
 128 +               /* Need to single quote the arg to keep the remote shell
 129 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 130 +               if (asprintf(&arg, "--source-filter='%s'", source_filter) < 0)
 131 +                       goto oom;
 132 +               args[ac++] = arg;
 133 +       }
 134 +
 135 +       if (dest_filter && am_sender) {
 136 +               /* Need to single quote the arg to keep the remote shell
 137 +                * from splitting it.  FIXME: breaks if command has single quotes. */
 138 +               if (asprintf(&arg, "--dest-filter='%s'", dest_filter) < 0)
 139 +                       goto oom;
 140 +               args[ac++] = arg;
 141 +       }
 142 +
 143         if (modify_window_set && am_sender) {
 144                 char *fmt = modify_window < 0 ? "-@%d" : "--modify-window=%d";
 145                 if (asprintf(&arg, fmt, modify_window) < 0)
 146 diff --git a/pipe.c b/pipe.c
 147 --- a/pipe.c
 148 +++ b/pipe.c
 149 @@ -27,6 +27,7 @@ extern int am_server;
 150  extern int blocking_io;
 151  extern int filesfrom_fd;
 152  extern int munge_symlinks;
 153 +extern mode_t orig_umask;
 154  extern char *logfile_name;
 155  extern int remote_option_cnt;
 156  extern const char **remote_options;
 157 @@ -176,3 +177,77 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
 158
 159         return pid;
 160  }
 161 +
 162 +pid_t run_filter(char *command[], int out, int *pipe_to_filter)
 163 +{
 164 +       pid_t pid;
 165 +       int pipefds[2];
 166 +
 167 +       if (DEBUG_GTE(CMD, 1))
 168 +               print_child_argv("opening connection using:", command);
 169 +
 170 +       if (pipe(pipefds) < 0) {
 171 +               rsyserr(FERROR, errno, "pipe");
 172 +               exit_cleanup(RERR_IPC);
 173 +       }
 174 +
 175 +       pid = do_fork();
 176 +       if (pid == -1) {
 177 +               rsyserr(FERROR, errno, "fork");
 178 +               exit_cleanup(RERR_IPC);
 179 +       }
 180 +
 181 +       if (pid == 0) {
 182 +               if (dup2(pipefds[0], STDIN_FILENO) < 0
 183 +                || close(pipefds[1]) < 0
 184 +                || dup2(out, STDOUT_FILENO) < 0) {
 185 +                       rsyserr(FERROR, errno, "Failed dup/close");
 186 +                       exit_cleanup(RERR_IPC);
 187 +               }
 188 +               umask(orig_umask);
 189 +               set_blocking(STDIN_FILENO);
 190 +               if (blocking_io)
 191 +                       set_blocking(STDOUT_FILENO);
 192 +               execvp(command[0], command);
 193 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 194 +               exit_cleanup(RERR_IPC);
 195 +       }
 196 +
 197 +       if (close(pipefds[0]) < 0) {
 198 +               rsyserr(FERROR, errno, "Failed to close");
 199 +               exit_cleanup(RERR_IPC);
 200 +       }
 201 +
 202 +       *pipe_to_filter = pipefds[1];
 203 +
 204 +       return pid;
 205 +}
 206 +
 207 +pid_t run_filter_on_file(char *command[], int out, int in)
 208 +{
 209 +       pid_t pid;
 210 +
 211 +       if (DEBUG_GTE(CMD, 1))
 212 +               print_child_argv("opening connection using:", command);
 213 +
 214 +       pid = do_fork();
 215 +       if (pid == -1) {
 216 +               rsyserr(FERROR, errno, "fork");
 217 +               exit_cleanup(RERR_IPC);
 218 +       }
 219 +
 220 +       if (pid == 0) {
 221 +               if (dup2(in, STDIN_FILENO) < 0
 222 +                || dup2(out, STDOUT_FILENO) < 0) {
 223 +                       rsyserr(FERROR, errno, "Failed to dup2");
 224 +                       exit_cleanup(RERR_IPC);
 225 +               }
 226 +               if (blocking_io)
 227 +                       set_blocking(STDOUT_FILENO);
 228 +               execvp(command[0], command);
 229 +               rsyserr(FERROR, errno, "Failed to exec %s", command[0]);
 230 +               exit_cleanup(RERR_IPC);
 231 +       }
 232 +
 233 +       return pid;
 234 +}
 235 diff --git a/receiver.c b/receiver.c
 236 --- a/receiver.c
 237 +++ b/receiver.c
 238 @@ -60,6 +60,7 @@ extern BOOL want_progress_now;
 239  extern mode_t orig_umask;
 240  extern struct stats stats;
 241  extern char *tmpdir;
 242 +extern char *dest_filter;
 243  extern char *partial_dir;
 244  extern char *basis_dir[MAX_BASIS_DIRS+1];
 245  extern char sender_file_sum[MAX_DIGEST_LEN];
 246 @@ -528,6 +529,7 @@ int recv_files(int f_in, int f_out, char *local_name)
 247         char *fnametmp, fnametmpbuf[MAXPATHLEN];
 248         char *fnamecmp, *partialptr;
 249         char fnamecmpbuf[MAXPATHLEN];
 250 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 251         uchar fnamecmp_type;
 252         struct file_struct *file;
 253         int itemizing = am_server ? logfile_format_has_i : stdout_format_has_i;
 254 @@ -538,6 +540,7 @@ int recv_files(int f_in, int f_out, char *local_name)
 255         const char *parent_dirname = "";
 256  #endif
 257         int ndx, recv_ok, one_inplace;
 258 +       pid_t pid = 0;
 259
 260         if (DEBUG_GTE(RECV, 1))
 261                 rprintf(FINFO, "recv_files(%d) starting\n", cur_flist->used);
 262 @@ -548,6 +551,23 @@ int recv_files(int f_in, int f_out, char *local_name)
 263         if (whole_file < 0)
 264                 whole_file = 0;
 265
 266 +       if (dest_filter) {
 267 +               char *p;
 268 +               char *sep = " \t";
 269 +               int i;
 270 +               for (p = strtok(dest_filter, sep), i = 0;
 271 +                    p && i < MAX_FILTER_ARGS;
 272 +                    p = strtok(0, sep))
 273 +                       filter_argv[i++] = p;
 274 +               filter_argv[i] = NULL;
 275 +               if (p) {
 276 +                       rprintf(FERROR,
 277 +                               "Too many arguments to dest-filter (> %d)\n",
 278 +                               MAX_FILTER_ARGS);
 279 +                       exit_cleanup(RERR_SYNTAX);
 280 +               }
 281 +       }
 282 +
 283         progress_init();
 284
 285         while (1) {
 286 @@ -873,6 +893,9 @@ int recv_files(int f_in, int f_out, char *local_name)
 287                 else if (!am_server && INFO_GTE(NAME, 1) && INFO_EQ(PROGRESS, 1))
 288                         rprintf(FINFO, "%s\n", fname);
 289
 290 +               if (dest_filter)
 291 +                       pid = run_filter(filter_argv, fd2, &fd2);
 292 +
 293                 /* recv file data */
 294                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size, fname, fd2, file, inplace || one_inplace);
 295
 296 @@ -888,6 +911,16 @@ int recv_files(int f_in, int f_out, char *local_name)
 297                         exit_cleanup(RERR_FILEIO);
 298                 }
 299
 300 +               if (dest_filter) {
 301 +                       int status;
 302 +                       wait_process_with_flush(pid, &status);
 303 +                       if (status != 0) {
 304 +                               rprintf(FERROR, "filter %s exited code: %d\n",
 305 +                                       dest_filter, status);
 306 +                               continue;
 307 +                       }
 308 +               }
 309 +
 310                 if ((recv_ok && (!delay_updates || !partialptr)) || inplace) {
 311                         if (partialptr == fname)
 312                                 partialptr = NULL;
 313 diff --git a/rsync.1.md b/rsync.1.md
 314 --- a/rsync.1.md
 315 +++ b/rsync.1.md
 316 @@ -504,6 +504,7 @@ has its own detailed description later in this manpage.
 317  --contimeout=SECONDS     set daemon connection timeout in seconds
 318  --ignore-times, -I       don't skip files that match size and time
 319  --size-only              skip files that match in size
 320 +--times-only             skip files that match in mod-time
 321  --modify-window=NUM, -@  set the accuracy for mod-time comparisons
 322  --temp-dir=DIR, -T       create temporary files in directory DIR
 323  --fuzzy, -y              find similar file for basis if no dest file
 324 @@ -553,6 +554,8 @@ has its own detailed description later in this manpage.
 325  --write-batch=FILE       write a batched update to FILE
 326  --only-write-batch=FILE  like --write-batch but w/o updating dest
 327  --read-batch=FILE        read a batched update from FILE
 328 +--source-filter=COMMAND  filter file through COMMAND at source
 329 +--dest-filter=COMMAND    filter file through COMMAND at destination
 330  --protocol=NUM           force an older protocol version to be used
 331  --iconv=CONVERT_SPEC     request charset conversion of filenames
 332  --checksum-seed=NUM      set block/file checksum seed (advanced)
 333 @@ -3713,6 +3716,36 @@ expand it.
 334      [`--write-batch`](#opt).  If _FILE_ is `-`, the batch data will be read
 335      from standard input. See the "BATCH MODE" section for details.
 336
 337 +0.  `--source-filter=COMMAND`
 338 +
 339 +    This option allows the user to specify a filter program that will be
 340 +    applied to the contents of all transferred regular files before the data is
 341 +    sent to destination.  COMMAND will receive the data on its standard input
 342 +    and it should write the filtered data to standard output.  COMMAND should
 343 +    exit non-zero if it cannot process the data or if it encounters an error
 344 +    when writing the data to stdout.
 345 +
 346 +    Example: `--source-filter="gzip -9"` will cause remote files to be
 347 +    compressed.  Use of `--source-filter` automatically enables
 348 +    [`--whole-file`](#opt).  If your filter does not output the same number of
 349 +    bytes that it received on input, you should use `--times-only` to
 350 +    disable size and content checks on subsequent rsync runs.
 351 +
 352 +0.  `--dest-filter=COMMAND`
 353 +
 354 +    This option allows you to specify a filter program that will be applied to
 355 +    the contents of all transferred regular files before the data is written to
 356 +    disk.  COMMAND will receive the data on its standard input and it should
 357 +    write the filtered data to standard output.  COMMAND should exit non-zero
 358 +    if it cannot process the data or if it encounters an error when writing the
 359 +    data to stdout.
 360 +
 361 +    Example: --dest-filter="gzip -9" will cause remote files to be compressed.
 362 +    Use of --dest-filter automatically enables --whole-file.  If your filter
 363 +    does not output the same number of bytes that it received on input, you
 364 +    should use --times-only to disable size and content checks on subsequent
 365 +    rsync runs.
 366 +
 367  0.  `--protocol=NUM`
 368
 369      Force an older protocol version to be used.  This is useful for creating a
 370 diff --git a/rsync.h b/rsync.h
 371 --- a/rsync.h
 372 +++ b/rsync.h
 373 @@ -169,6 +169,7 @@
 374  #define IOERR_DEL_LIMIT (1<<2)
 375
 376  #define MAX_ARGS 1000
 377 +#define MAX_FILTER_ARGS 100
 378  #define MAX_BASIS_DIRS 20
 379  #define MAX_SERVER_ARGS (MAX_BASIS_DIRS*2 + 100)
 380
 381 diff --git a/sender.c b/sender.c
 382 --- a/sender.c
 383 +++ b/sender.c
 384 @@ -21,6 +21,7 @@
 385
 386  #include "rsync.h"
 387  #include "inums.h"
 388 +#include "ifuncs.h"
 389
 390  extern int do_xfers;
 391  extern int am_server;
 392 @@ -50,6 +51,7 @@ extern int batch_fd;
 393  extern int write_batch;
 394  extern int file_old_total;
 395  extern BOOL want_progress_now;
 396 +extern char *source_filter;
 397  extern struct stats stats;
 398  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 399  extern char num_dev_ino_buf[4 + 8 + 8];
 400 @@ -211,6 +213,26 @@ void send_files(int f_in, int f_out)
 401         int f_xfer = write_batch < 0 ? batch_fd : f_out;
 402         int save_io_error = io_error;
 403         int ndx, j;
 404 +       char *filter_argv[MAX_FILTER_ARGS + 1];
 405 +       char *tmp = 0;
 406 +       int unlink_tmp = 0;
 407 +
 408 +       if (source_filter) {
 409 +               char *p;
 410 +               char *sep = " \t";
 411 +               int i;
 412 +               for (p = strtok(source_filter, sep), i = 0;
 413 +                    p && i < MAX_FILTER_ARGS;
 414 +                    p = strtok(0, sep))
 415 +                       filter_argv[i++] = p;
 416 +               filter_argv[i] = NULL;
 417 +               if (p) {
 418 +                       rprintf(FERROR,
 419 +                               "Too many arguments to source-filter (> %d)\n",
 420 +                               MAX_FILTER_ARGS);
 421 +                       exit_cleanup(RERR_SYNTAX);
 422 +               }
 423 +       }
 424
 425         if (DEBUG_GTE(SEND, 1))
 426                 rprintf(FINFO, "send_files starting\n");
 427 @@ -348,6 +370,7 @@ void send_files(int f_in, int f_out)
 428                         exit_cleanup(RERR_PROTOCOL);
 429                 }
 430
 431 +               unlink_tmp = 0;
 432                 fd = do_open(fname, O_RDONLY, 0);
 433                 if (fd == -1) {
 434                         if (errno == ENOENT) {
 435 @@ -367,6 +390,33 @@ void send_files(int f_in, int f_out)
 436                         continue;
 437                 }
 438
 439 +               if (source_filter) {
 440 +                       int fd2;
 441 +                       char *tmpl = "/tmp/rsync-filtered_sourceXXXXXX";
 442 +
 443 +                       tmp = strdup(tmpl);
 444 +                       fd2 = mkstemp(tmp);
 445 +                       if (fd2 == -1) {
 446 +                               rprintf(FERROR, "mkstemp %s failed: %s\n",
 447 +                                       tmp, strerror(errno));
 448 +                       } else {
 449 +                               int status;
 450 +                               pid_t pid = run_filter_on_file(filter_argv, fd2, fd);
 451 +                               close(fd);
 452 +                               close(fd2);
 453 +                               wait_process_with_flush(pid, &status);
 454 +                               if (status != 0) {
 455 +                                       rprintf(FERROR,
 456 +                                           "bypassing source filter %s; exited with code: %d\n",
 457 +                                           source_filter, status);
 458 +                                       fd = do_open(fname, O_RDONLY, 0);
 459 +                               } else {
 460 +                                       fd = do_open(tmp, O_RDONLY, 0);
 461 +                                       unlink_tmp = 1;
 462 +                               }
 463 +                       }
 464 +               }
 465 +
 466                 /* map the local file */
 467                 if (do_fstat(fd, &st) != 0) {
 468                         io_error |= IOERR_GENERAL;
 469 @@ -437,6 +487,8 @@ void send_files(int f_in, int f_out)
 470                         }
 471                 }
 472                 close(fd);
 473 +               if (unlink_tmp)
 474 +                       unlink(tmp);
 475
 476                 free_sums(s);
 477