exclude.c

   1 /*
   2  * The filter include/exclude routines.
   3  *
   4  * Copyright (C) 1996-2001 Andrew Tridgell <tridge@samba.org>
   5  * Copyright (C) 1996 Paul Mackerras
   6  * Copyright (C) 2002 Martin Pool
   7  * Copyright (C) 2003-2022 Wayne Davison
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 3 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License along
  20  * with this program; if not, visit the http://fsf.org website.
  21  */
  22
  23 #include "rsync.h"
  24 #include "ifuncs.h"
  25
  26 extern int am_server;
  27 extern int am_sender;
  28 extern int am_generator;
  29 extern int eol_nulls;
  30 extern int io_error;
  31 extern int xfer_dirs;
  32 extern int recurse;
  33 extern int local_server;
  34 extern int prune_empty_dirs;
  35 extern int ignore_perishable;
  36 extern int relative_paths;
  37 extern int delete_mode;
  38 extern int delete_excluded;
  39 extern int cvs_exclude;
  40 extern int sanitize_paths;
  41 extern int protocol_version;
  42 extern int trust_sender_args;
  43 extern int module_id;
  44
  45 extern char curr_dir[MAXPATHLEN];
  46 extern unsigned int curr_dir_len;
  47 extern unsigned int module_dirlen;
  48
  49 filter_rule_list filter_list = { .debug_type = "" };
  50 filter_rule_list cvs_filter_list = { .debug_type = " [global CVS]" };
  51 filter_rule_list daemon_filter_list = { .debug_type = " [daemon]" };
  52 filter_rule_list implied_filter_list = { .debug_type = " [implied]" };
  53
  54 int saw_xattr_filter = 0;
  55 int trust_sender_args = 0;
  56 int trust_sender_filter = 0;
  57
  58 /* Need room enough for ":MODS " prefix plus some room to grow. */
  59 #define MAX_RULE_PREFIX (16)
  60
  61 #define SLASH_WILD3_SUFFIX "/***"
  62
  63 /* The dirbuf is set by push_local_filters() to the current subdirectory
  64  * relative to curr_dir that is being processed.  The path always has a
  65  * trailing slash appended, and the variable dirbuf_len contains the length
  66  * of this path prefix.  The path is always absolute. */
  67 static char dirbuf[MAXPATHLEN+1];
  68 static unsigned int dirbuf_len = 0;
  69 static int dirbuf_depth;
  70
  71 /* This is True when we're scanning parent dirs for per-dir merge-files. */
  72 static BOOL parent_dirscan = False;
  73
  74 /* This array contains a list of all the currently active per-dir merge
  75  * files.  This makes it easier to save the appropriate values when we
  76  * "push" down into each subdirectory. */
  77 static filter_rule **mergelist_parents;
  78 static int mergelist_cnt = 0;
  79 static int mergelist_size = 0;
  80
  81 /* Each filter_list_struct describes a singly-linked list by keeping track
  82  * of both the head and tail pointers.  The list is slightly unusual in that
  83  * a parent-dir's content can be appended to the end of the local list in a
  84  * special way:  the last item in the local list has its "next" pointer set
  85  * to point to the inherited list, but the local list's tail pointer points
  86  * at the end of the local list.  Thus, if the local list is empty, the head
  87  * will be pointing at the inherited content but the tail will be NULL.  To
  88  * help you visualize this, here are the possible list arrangements:
  89  *
  90  * Completely Empty                     Local Content Only
  91  * ==================================   ====================================
  92  * head -> NULL                         head -> Local1 -> Local2 -> NULL
  93  * tail -> NULL                         tail -------------^
  94  *
  95  * Inherited Content Only               Both Local and Inherited Content
  96  * ==================================   ====================================
  97  * head -> Parent1 -> Parent2 -> NULL   head -> L1 -> L2 -> P1 -> P2 -> NULL
  98  * tail -> NULL                         tail ---------^
  99  *
 100  * This means that anyone wanting to traverse the whole list to use it just
 101  * needs to start at the head and use the "next" pointers until it goes
 102  * NULL.  To add new local content, we insert the item after the tail item
 103  * and update the tail (obviously, if "tail" was NULL, we insert it at the
 104  * head).  To clear the local list, WE MUST NOT FREE THE INHERITED CONTENT
 105  * because it is shared between the current list and our parent list(s).
 106  * The easiest way to handle this is to simply truncate the list after the
 107  * tail item and then free the local list from the head.  When inheriting
 108  * the list for a new local dir, we just save off the filter_list_struct
 109  * values (so we can pop back to them later) and set the tail to NULL.
 110  */
 111
 112 static void teardown_mergelist(filter_rule *ex)
 113 {
 114         int j;
 115
 116         if (!ex->u.mergelist)
 117                 return;
 118
 119         if (DEBUG_GTE(FILTER, 2)) {
 120                 rprintf(FINFO, "[%s] deactivating mergelist #%d%s\n",
 121                         who_am_i(), mergelist_cnt - 1,
 122                         ex->u.mergelist->debug_type);
 123         }
 124
 125         free(ex->u.mergelist->debug_type);
 126         free(ex->u.mergelist);
 127
 128         for (j = 0; j < mergelist_cnt; j++) {
 129                 if (mergelist_parents[j] == ex) {
 130                         mergelist_parents[j] = NULL;
 131                         break;
 132                 }
 133         }
 134         while (mergelist_cnt && mergelist_parents[mergelist_cnt-1] == NULL)
 135                 mergelist_cnt--;
 136 }
 137
 138 static void free_filter(filter_rule *ex)
 139 {
 140         if (ex->rflags & FILTRULE_PERDIR_MERGE)
 141                 teardown_mergelist(ex);
 142         free(ex->pattern);
 143         free(ex);
 144 }
 145
 146 static void free_filters(filter_rule *ent)
 147 {
 148         while (ent) {
 149                 filter_rule *next = ent->next;
 150                 free_filter(ent);
 151                 ent = next;
 152         }
 153 }
 154
 155 /* Build a filter structure given a filter pattern.  The value in "pat"
 156  * is not null-terminated.  "rule" is either held or freed, so the
 157  * caller should not free it. */
 158 static void add_rule(filter_rule_list *listp, const char *pat, unsigned int pat_len,
 159                      filter_rule *rule, int xflags)
 160 {
 161         const char *cp;
 162         unsigned int pre_len, suf_len, slash_cnt = 0;
 163         char *mention_rule_suffix;
 164
 165         if (DEBUG_GTE(FILTER, 1) && pat_len && (pat[pat_len-1] == ' ' || pat[pat_len-1] == '\t'))
 166                 mention_rule_suffix = " -- CAUTION: trailing whitespace!";
 167         else
 168                 mention_rule_suffix = DEBUG_GTE(FILTER, 2) ? "" : NULL;
 169         if (mention_rule_suffix) {
 170                 rprintf(FINFO, "[%s] add_rule(%s%.*s%s)%s%s\n",
 171                         who_am_i(), get_rule_prefix(rule, pat, 0, NULL),
 172                         (int)pat_len, pat, (rule->rflags & FILTRULE_DIRECTORY) ? "/" : "",
 173                         listp->debug_type, mention_rule_suffix);
 174         }
 175
 176         /* These flags also indicate that we're reading a list that
 177          * needs to be filtered now, not post-filtered later. */
 178         if (xflags & (XFLG_ANCHORED2ABS|XFLG_ABS_IF_SLASH)
 179                 && (rule->rflags & FILTRULES_SIDES)
 180                         == (am_sender ? FILTRULE_RECEIVER_SIDE : FILTRULE_SENDER_SIDE)) {
 181                 /* This filter applies only to the other side.  Drop it. */
 182                 free_filter(rule);
 183                 return;
 184         }
 185
 186         if (pat_len > 1 && pat[pat_len-1] == '/') {
 187                 pat_len--;
 188                 rule->rflags |= FILTRULE_DIRECTORY;
 189         }
 190
 191         for (cp = pat; cp < pat + pat_len; cp++) {
 192                 if (*cp == '/')
 193                         slash_cnt++;
 194         }
 195
 196         if (!(rule->rflags & (FILTRULE_ABS_PATH | FILTRULE_MERGE_FILE))
 197          && ((xflags & (XFLG_ANCHORED2ABS|XFLG_ABS_IF_SLASH) && *pat == '/')
 198           || (xflags & XFLG_ABS_IF_SLASH && slash_cnt))) {
 199                 rule->rflags |= FILTRULE_ABS_PATH;
 200                 if (*pat == '/')
 201                         pre_len = dirbuf_len - module_dirlen - 1;
 202                 else
 203                         pre_len = 0;
 204         } else
 205                 pre_len = 0;
 206
 207         /* The daemon wants dir-exclude rules to get an appended "/" + "***". */
 208         if (xflags & XFLG_DIR2WILD3
 209          && BITS_SETnUNSET(rule->rflags, FILTRULE_DIRECTORY, FILTRULE_INCLUDE)) {
 210                 rule->rflags &= ~FILTRULE_DIRECTORY;
 211                 suf_len = sizeof SLASH_WILD3_SUFFIX - 1;
 212         } else
 213                 suf_len = 0;
 214
 215         rule->pattern = new_array(char, pre_len + pat_len + suf_len + 1);
 216         if (pre_len) {
 217                 memcpy(rule->pattern, dirbuf + module_dirlen, pre_len);
 218                 for (cp = rule->pattern; cp < rule->pattern + pre_len; cp++) {
 219                         if (*cp == '/')
 220                                 slash_cnt++;
 221                 }
 222         }
 223         strlcpy(rule->pattern + pre_len, pat, pat_len + 1);
 224         pat_len += pre_len;
 225         if (suf_len) {
 226                 memcpy(rule->pattern + pat_len, SLASH_WILD3_SUFFIX, suf_len+1);
 227                 pat_len += suf_len;
 228                 slash_cnt++;
 229         }
 230
 231         if (strpbrk(rule->pattern, "*[?")) {
 232                 rule->rflags |= FILTRULE_WILD;
 233                 if ((cp = strstr(rule->pattern, "**")) != NULL) {
 234                         rule->rflags |= FILTRULE_WILD2;
 235                         /* If the pattern starts with **, note that. */
 236                         if (cp == rule->pattern)
 237                                 rule->rflags |= FILTRULE_WILD2_PREFIX;
 238                         /* If the pattern ends with ***, note that. */
 239                         if (pat_len >= 3
 240                          && rule->pattern[pat_len-3] == '*'
 241                          && rule->pattern[pat_len-2] == '*'
 242                          && rule->pattern[pat_len-1] == '*')
 243                                 rule->rflags |= FILTRULE_WILD3_SUFFIX;
 244                 }
 245         }
 246
 247         if (rule->rflags & FILTRULE_PERDIR_MERGE) {
 248                 filter_rule_list *lp;
 249                 unsigned int len;
 250                 int i;
 251
 252                 if ((cp = strrchr(rule->pattern, '/')) != NULL)
 253                         cp++;
 254                 else
 255                         cp = rule->pattern;
 256
 257                 /* If the local merge file was already mentioned, don't
 258                  * add it again. */
 259                 for (i = 0; i < mergelist_cnt; i++) {
 260                         filter_rule *ex = mergelist_parents[i];
 261                         const char *s;
 262                         if (!ex)
 263                                 continue;
 264                         s = strrchr(ex->pattern, '/');
 265                         if (s)
 266                                 s++;
 267                         else
 268                                 s = ex->pattern;
 269                         len = strlen(s);
 270                         if (len == pat_len - (cp - rule->pattern) && memcmp(s, cp, len) == 0) {
 271                                 free_filter(rule);
 272                                 return;
 273                         }
 274                 }
 275
 276                 lp = new_array0(filter_rule_list, 1);
 277                 if (asprintf(&lp->debug_type, " [per-dir %s]", cp) < 0)
 278                         out_of_memory("add_rule");
 279                 rule->u.mergelist = lp;
 280
 281                 if (mergelist_cnt == mergelist_size) {
 282                         mergelist_size += 5;
 283                         mergelist_parents = realloc_array(mergelist_parents, filter_rule *, mergelist_size);
 284                 }
 285                 if (DEBUG_GTE(FILTER, 2)) {
 286                         rprintf(FINFO, "[%s] activating mergelist #%d%s\n",
 287                                 who_am_i(), mergelist_cnt, lp->debug_type);
 288                 }
 289                 mergelist_parents[mergelist_cnt++] = rule;
 290         } else
 291                 rule->u.slash_cnt = slash_cnt;
 292
 293         if (!listp->tail) {
 294                 rule->next = listp->head;
 295                 listp->head = listp->tail = rule;
 296         } else {
 297                 rule->next = listp->tail->next;
 298                 listp->tail->next = rule;
 299                 listp->tail = rule;
 300         }
 301 }
 302
 303 /* If the wildcards failed, the remote shell might give us a file matching the literal
 304  * wildcards.  Since "*" & "?" already match themselves, this just needs to deal with
 305  * failed "[foo]" idioms.
 306  */
 307 static void maybe_add_literal_brackets_rule(filter_rule const *based_on, int arg_len)
 308 {
 309         filter_rule *rule;
 310         const char *arg = based_on->pattern, *cp;
 311         char *p;
 312         int cnt = 0;
 313
 314         if (arg_len < 0)
 315                 arg_len = strlen(arg);
 316
 317         for (cp = arg; *cp; cp++) {
 318                 if (*cp == '\\' && cp[1]) {
 319                         cp++;
 320                 } else if (*cp == '[')
 321                         cnt++;
 322         }
 323         if (!cnt)
 324                 return;
 325
 326         rule = new0(filter_rule);
 327         rule->rflags = based_on->rflags;
 328         rule->u.slash_cnt = based_on->u.slash_cnt;
 329         p = rule->pattern = new_array(char, arg_len + cnt + 1);
 330         for (cp = arg; *cp; ) {
 331                 if (*cp == '\\' && cp[1]) {
 332                         *p++ = *cp++;
 333                 } else if (*cp == '[')
 334                         *p++ = '\\';
 335                 *p++ = *cp++;
 336         }
 337         *p++ = '\0';
 338
 339         rule->next = implied_filter_list.head;
 340         implied_filter_list.head = rule;
 341         if (DEBUG_GTE(FILTER, 3)) {
 342                 rprintf(FINFO, "[%s] add_implied_include(%s%s)\n", who_am_i(), rule->pattern,
 343                         rule->rflags & FILTRULE_DIRECTORY ? "/" : "");
 344         }
 345 }
 346
 347 static char *partial_string_buf = NULL;
 348 static int partial_string_len = 0;
 349 void implied_include_partial_string(const char *s_start, const char *s_end)
 350 {
 351         partial_string_len = s_end - s_start;
 352         if (partial_string_len <= 0 || partial_string_len >= MAXPATHLEN) { /* too-large should be impossible... */
 353                 partial_string_len = 0;
 354                 return;
 355         }
 356         if (!partial_string_buf)
 357                 partial_string_buf = new_array(char, MAXPATHLEN);
 358         memcpy(partial_string_buf, s_start, partial_string_len);
 359 }
 360
 361 void free_implied_include_partial_string()
 362 {
 363         if (partial_string_buf) {
 364                 free(partial_string_buf);
 365                 partial_string_buf = NULL;
 366         }
 367         partial_string_len = 0; /* paranoia */
 368 }
 369
 370 /* Each arg the client sends to the remote sender turns into an implied include
 371  * that the receiver uses to validate the file list from the sender. */
 372 void add_implied_include(const char *arg, int skip_daemon_module)
 373 {
 374         filter_rule *rule;
 375         int arg_len, saw_wild = 0, saw_live_open_brkt = 0, backslash_cnt = 0;
 376         int slash_cnt = 1; /* We know we're adding a leading slash. */
 377         const char *cp;
 378         char *p;
 379         if (trust_sender_args)
 380                 return;
 381         if (partial_string_len) {
 382                 arg_len = strlen(arg);
 383                 if (partial_string_len + arg_len >= MAXPATHLEN) {
 384                         partial_string_len = 0;
 385                         return; /* Should be impossible... */
 386                 }
 387                 memcpy(partial_string_buf + partial_string_len, arg, arg_len + 1);
 388                 partial_string_len = 0;
 389                 arg = partial_string_buf;
 390         }
 391         if (skip_daemon_module) {
 392                 if ((cp = strchr(arg, '/')) != NULL)
 393                         arg = cp + 1;
 394                 else
 395                         arg = "";
 396         }
 397         if (relative_paths) {
 398                 if ((cp = strstr(arg, "/./")) != NULL)
 399                         arg = cp + 3;
 400         } else if ((cp = strrchr(arg, '/')) != NULL) {
 401                 arg = cp + 1;
 402         }
 403         if (*arg == '.' && arg[1] == '\0')
 404                 arg++;
 405         arg_len = strlen(arg);
 406         if (arg_len) {
 407                 if (strpbrk(arg, "*[?")) {
 408                         /* We need to add room to escape backslashes if wildcard chars are present. */
 409                         for (cp = arg; (cp = strchr(cp, '\\')) != NULL; cp++)
 410                                 arg_len++;
 411                         saw_wild = 1;
 412                 }
 413                 arg_len++; /* Leave room for the prefixed slash */
 414                 rule = new0(filter_rule);
 415                 if (!implied_filter_list.head)
 416                         implied_filter_list.head = implied_filter_list.tail = rule;
 417                 else {
 418                         rule->next = implied_filter_list.head;
 419                         implied_filter_list.head = rule;
 420                 }
 421                 rule->rflags = FILTRULE_INCLUDE + (saw_wild ? FILTRULE_WILD : 0);
 422                 p = rule->pattern = new_array(char, arg_len + 1);
 423                 *p++ = '/';
 424                 for (cp = arg; *cp; ) {
 425                         switch (*cp) {
 426                           case '\\':
 427                                 if (cp[1] == ']') {
 428                                         if (!saw_wild)
 429                                                 cp++; /* A \] in a non-wild filter causes a problem, so drop the \ . */
 430                                 } else if (!strchr("*[?", cp[1])) {
 431                                         backslash_cnt++;
 432                                         if (saw_wild)
 433                                                 *p++ = '\\';
 434                                 }
 435                                 *p++ = *cp++;
 436                                 break;
 437                           case '/':
 438                                 if (p[-1] == '/') { /* This is safe because of the initial slash. */
 439                                         cp++;
 440                                         break;
 441                                 }
 442                                 if (relative_paths) {
 443                                         filter_rule const *ent;
 444                                         int found = 0;
 445                                         *p = '\0';
 446                                         for (ent = implied_filter_list.head; ent; ent = ent->next) {
 447                                                 if (ent != rule && strcmp(ent->pattern, rule->pattern) == 0) {
 448                                                         found = 1;
 449                                                         break;
 450                                                 }
 451                                         }
 452                                         if (!found) {
 453                                                 filter_rule *R_rule = new0(filter_rule);
 454                                                 R_rule->rflags = FILTRULE_INCLUDE | FILTRULE_DIRECTORY;
 455                                                 /* Check if our sub-path has wildcards or escaped backslashes */
 456                                                 if (saw_wild && strpbrk(rule->pattern, "*[?\\"))
 457                                                         R_rule->rflags |= FILTRULE_WILD;
 458                                                 R_rule->pattern = strdup(rule->pattern);
 459                                                 R_rule->u.slash_cnt = slash_cnt;
 460                                                 R_rule->next = implied_filter_list.head;
 461                                                 implied_filter_list.head = R_rule;
 462                                                 if (DEBUG_GTE(FILTER, 3)) {
 463                                                         rprintf(FINFO, "[%s] add_implied_include(%s/)\n",
 464                                                                 who_am_i(), R_rule->pattern);
 465                                                 }
 466                                                 if (saw_live_open_brkt)
 467                                                         maybe_add_literal_brackets_rule(R_rule, -1);
 468                                         }
 469                                 }
 470                                 slash_cnt++;
 471                                 *p++ = *cp++;
 472                                 break;
 473                           case '[':
 474                                 saw_live_open_brkt = 1;
 475                                 *p++ = *cp++;
 476                                 break;
 477                           default:
 478                                 *p++ = *cp++;
 479                                 break;
 480                         }
 481                 }
 482                 *p = '\0';
 483                 rule->u.slash_cnt = slash_cnt;
 484                 arg = rule->pattern;
 485                 arg_len = p - arg; /* We recompute it due to backslash weirdness. */
 486                 if (DEBUG_GTE(FILTER, 3))
 487                         rprintf(FINFO, "[%s] add_implied_include(%s)\n", who_am_i(), rule->pattern);
 488                 if (saw_live_open_brkt)
 489                         maybe_add_literal_brackets_rule(rule, arg_len);
 490         }
 491
 492         if (recurse || xfer_dirs) {
 493                 /* Now create a rule with an added "/" & "**" or "*" at the end */
 494                 rule = new0(filter_rule);
 495                 rule->rflags = FILTRULE_INCLUDE | FILTRULE_WILD;
 496                 if (recurse)
 497                         rule->rflags |= FILTRULE_WILD2;
 498                 /* We must leave enough room for / * * \0. */
 499                 if (!saw_wild && backslash_cnt) {
 500                         /* We are appending a wildcard, so now the backslashes need to be escaped. */
 501                         p = rule->pattern = new_array(char, arg_len + backslash_cnt + 3 + 1);
 502                         for (cp = arg; *cp; ) {
 503                                 if (*cp == '\\')
 504                                         *p++ = '\\';
 505                                 *p++ = *cp++;
 506                         }
 507                 } else {
 508                         p = rule->pattern = new_array(char, arg_len + 3 + 1);
 509                         if (arg_len) {
 510                                 memcpy(p, arg, arg_len);
 511                                 p += arg_len;
 512                         }
 513                 }
 514                 if (p[-1] != '/')
 515                         *p++ = '/';
 516                 *p++ = '*';
 517                 if (recurse)
 518                         *p++ = '*';
 519                 *p = '\0';
 520                 rule->u.slash_cnt = slash_cnt + 1;
 521                 rule->next = implied_filter_list.head;
 522                 implied_filter_list.head = rule;
 523                 if (DEBUG_GTE(FILTER, 3))
 524                         rprintf(FINFO, "[%s] add_implied_include(%s)\n", who_am_i(), rule->pattern);
 525                 if (saw_live_open_brkt)
 526                         maybe_add_literal_brackets_rule(rule, p - rule->pattern);
 527         }
 528 }
 529
 530 /* This frees any non-inherited items, leaving just inherited items on the list. */
 531 static void pop_filter_list(filter_rule_list *listp)
 532 {
 533         filter_rule *inherited;
 534
 535         if (!listp->tail)
 536                 return;
 537
 538         inherited = listp->tail->next;
 539
 540         /* Truncate any inherited items from the local list. */
 541         listp->tail->next = NULL;
 542         /* Now free everything that is left. */
 543         free_filters(listp->head);
 544
 545         listp->head = inherited;
 546         listp->tail = NULL;
 547 }
 548
 549 /* This returns an expanded (absolute) filename for the merge-file name if
 550  * the name has any slashes in it OR if the parent_dirscan var is True;
 551  * otherwise it returns the original merge_file name.  If the len_ptr value
 552  * is non-NULL the merge_file name is limited by the referenced length
 553  * value and will be updated with the length of the resulting name.  We
 554  * always return a name that is null terminated, even if the merge_file
 555  * name was not. */
 556 static char *parse_merge_name(const char *merge_file, unsigned int *len_ptr,
 557                               unsigned int prefix_skip)
 558 {
 559         static char buf[MAXPATHLEN];
 560         char *fn, tmpbuf[MAXPATHLEN];
 561         unsigned int fn_len;
 562
 563         if (!parent_dirscan && *merge_file != '/') {
 564                 /* Return the name unchanged it doesn't have any slashes. */
 565                 if (len_ptr) {
 566                         const char *p = merge_file + *len_ptr;
 567                         while (--p > merge_file && *p != '/') {}
 568                         if (p == merge_file) {
 569                                 strlcpy(buf, merge_file, *len_ptr + 1);
 570                                 return buf;
 571                         }
 572                 } else if (strchr(merge_file, '/') == NULL)
 573                         return (char *)merge_file;
 574         }
 575
 576         fn = *merge_file == '/' ? buf : tmpbuf;
 577         if (sanitize_paths) {
 578                 const char *r = prefix_skip ? "/" : NULL;
 579                 /* null-terminate the name if it isn't already */
 580                 if (len_ptr && merge_file[*len_ptr]) {
 581                         char *to = fn == buf ? tmpbuf : buf;
 582                         strlcpy(to, merge_file, *len_ptr + 1);
 583                         merge_file = to;
 584                 }
 585                 if (!sanitize_path(fn, merge_file, r, dirbuf_depth, SP_DEFAULT)) {
 586                         rprintf(FERROR, "merge-file name overflows: %s\n",
 587                                 merge_file);
 588                         return NULL;
 589                 }
 590                 fn_len = strlen(fn);
 591         } else {
 592                 strlcpy(fn, merge_file, len_ptr ? *len_ptr + 1 : MAXPATHLEN);
 593                 fn_len = clean_fname(fn, CFN_COLLAPSE_DOT_DOT_DIRS);
 594         }
 595
 596         /* If the name isn't in buf yet, it wasn't absolute. */
 597         if (fn != buf) {
 598                 int d_len = dirbuf_len - prefix_skip;
 599                 if (d_len + fn_len >= MAXPATHLEN) {
 600                         rprintf(FERROR, "merge-file name overflows: %s\n", fn);
 601                         return NULL;
 602                 }
 603                 memcpy(buf, dirbuf + prefix_skip, d_len);
 604                 memcpy(buf + d_len, fn, fn_len + 1);
 605                 fn_len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
 606         }
 607
 608         if (len_ptr)
 609                 *len_ptr = fn_len;
 610         return buf;
 611 }
 612
 613 /* Sets the dirbuf and dirbuf_len values. */
 614 void set_filter_dir(const char *dir, unsigned int dirlen)
 615 {
 616         unsigned int len;
 617         if (*dir != '/') {
 618                 memcpy(dirbuf, curr_dir, curr_dir_len);
 619                 dirbuf[curr_dir_len] = '/';
 620                 len = curr_dir_len + 1;
 621                 if (len + dirlen >= MAXPATHLEN)
 622                         dirlen = 0;
 623         } else
 624                 len = 0;
 625         memcpy(dirbuf + len, dir, dirlen);
 626         dirbuf[dirlen + len] = '\0';
 627         dirbuf_len = clean_fname(dirbuf, CFN_COLLAPSE_DOT_DOT_DIRS);
 628         if (dirbuf_len > 1 && dirbuf[dirbuf_len-1] == '.'
 629             && dirbuf[dirbuf_len-2] == '/')
 630                 dirbuf_len -= 2;
 631         if (dirbuf_len != 1)
 632                 dirbuf[dirbuf_len++] = '/';
 633         dirbuf[dirbuf_len] = '\0';
 634         if (sanitize_paths)
 635                 dirbuf_depth = count_dir_elements(dirbuf + module_dirlen);
 636 }
 637
 638 /* This routine takes a per-dir merge-file entry and finishes its setup.
 639  * If the name has a path portion then we check to see if it refers to a
 640  * parent directory of the first transfer dir.  If it does, we scan all the
 641  * dirs from that point through the parent dir of the transfer dir looking
 642  * for the per-dir merge-file in each one. */
 643 static BOOL setup_merge_file(int mergelist_num, filter_rule *ex,
 644                              filter_rule_list *lp)
 645 {
 646         char buf[MAXPATHLEN];
 647         char *x, *y, *pat = ex->pattern;
 648         unsigned int len;
 649
 650         if (!(x = parse_merge_name(pat, NULL, 0)) || *x != '/')
 651                 return 0;
 652
 653         if (DEBUG_GTE(FILTER, 2)) {
 654                 rprintf(FINFO, "[%s] performing parent_dirscan for mergelist #%d%s\n",
 655                         who_am_i(), mergelist_num, lp->debug_type);
 656         }
 657         y = strrchr(x, '/');
 658         *y = '\0';
 659         ex->pattern = strdup(y+1);
 660         if (!*x)
 661                 x = "/";
 662         if (*x == '/')
 663                 strlcpy(buf, x, MAXPATHLEN);
 664         else
 665                 pathjoin(buf, MAXPATHLEN, dirbuf, x);
 666
 667         len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
 668         if (len != 1 && len < MAXPATHLEN-1) {
 669                 buf[len++] = '/';
 670                 buf[len] = '\0';
 671         }
 672         /* This ensures that the specified dir is a parent of the transfer. */
 673         for (x = buf, y = dirbuf; *x && *x == *y; x++, y++) {}
 674         if (*x)
 675                 y += strlen(y); /* nope -- skip the scan */
 676
 677         parent_dirscan = True;
 678         while (*y) {
 679                 char save[MAXPATHLEN];
 680                 strlcpy(save, y, MAXPATHLEN);
 681                 *y = '\0';
 682                 dirbuf_len = y - dirbuf;
 683                 strlcpy(x, ex->pattern, MAXPATHLEN - (x - buf));
 684                 parse_filter_file(lp, buf, ex, XFLG_ANCHORED2ABS);
 685                 if (ex->rflags & FILTRULE_NO_INHERIT) {
 686                         /* Free the undesired rules to clean up any per-dir
 687                          * mergelists they defined.  Otherwise pop_local_filters
 688                          * may crash trying to restore nonexistent state for
 689                          * those mergelists. */
 690                         free_filters(lp->head);
 691                         lp->head = NULL;
 692                 }
 693                 lp->tail = NULL;
 694                 strlcpy(y, save, MAXPATHLEN);
 695                 while ((*x++ = *y++) != '/') {}
 696         }
 697         parent_dirscan = False;
 698         if (DEBUG_GTE(FILTER, 2)) {
 699                 rprintf(FINFO, "[%s] completed parent_dirscan for mergelist #%d%s\n",
 700                         who_am_i(), mergelist_num, lp->debug_type);
 701         }
 702         free(pat);
 703         return 1;
 704 }
 705
 706 struct local_filter_state {
 707         int mergelist_cnt;
 708         filter_rule_list mergelists[1];
 709 };
 710
 711 /* Each time rsync changes to a new directory it call this function to
 712  * handle all the per-dir merge-files.  The "dir" value is the current path
 713  * relative to curr_dir (which might not be null-terminated).  We copy it
 714  * into dirbuf so that we can easily append a file name on the end. */
 715 void *push_local_filters(const char *dir, unsigned int dirlen)
 716 {
 717         struct local_filter_state *push;
 718         int i;
 719
 720         set_filter_dir(dir, dirlen);
 721         if (DEBUG_GTE(FILTER, 2)) {
 722                 rprintf(FINFO, "[%s] pushing local filters for %s\n",
 723                         who_am_i(), dirbuf);
 724         }
 725
 726         if (!mergelist_cnt) {
 727                 /* No old state to save and no new merge files to push. */
 728                 return NULL;
 729         }
 730
 731         push = (struct local_filter_state *)new_array(char,
 732                           sizeof (struct local_filter_state)
 733                         + (mergelist_cnt-1) * sizeof (filter_rule_list));
 734
 735         push->mergelist_cnt = mergelist_cnt;
 736         for (i = 0; i < mergelist_cnt; i++) {
 737                 filter_rule *ex = mergelist_parents[i];
 738                 if (!ex)
 739                         continue;
 740                 memcpy(&push->mergelists[i], ex->u.mergelist, sizeof (filter_rule_list));
 741         }
 742
 743         /* Note: parse_filter_file() might increase mergelist_cnt, so keep
 744          * this loop separate from the above loop. */
 745         for (i = 0; i < mergelist_cnt; i++) {
 746                 filter_rule *ex = mergelist_parents[i];
 747                 filter_rule_list *lp;
 748                 if (!ex)
 749                         continue;
 750                 lp = ex->u.mergelist;
 751
 752                 if (DEBUG_GTE(FILTER, 2)) {
 753                         rprintf(FINFO, "[%s] pushing mergelist #%d%s\n",
 754                                 who_am_i(), i, lp->debug_type);
 755                 }
 756
 757                 lp->tail = NULL; /* Switch any local rules to inherited. */
 758                 if (ex->rflags & FILTRULE_NO_INHERIT)
 759                         lp->head = NULL;
 760
 761                 if (ex->rflags & FILTRULE_FINISH_SETUP) {
 762                         ex->rflags &= ~FILTRULE_FINISH_SETUP;
 763                         if (setup_merge_file(i, ex, lp))
 764                                 set_filter_dir(dir, dirlen);
 765                 }
 766
 767                 if (strlcpy(dirbuf + dirbuf_len, ex->pattern,
 768                     MAXPATHLEN - dirbuf_len) < MAXPATHLEN - dirbuf_len) {
 769                         parse_filter_file(lp, dirbuf, ex,
 770                                           XFLG_ANCHORED2ABS);
 771                 } else {
 772                         io_error |= IOERR_GENERAL;
 773                         rprintf(FERROR,
 774                             "cannot add local filter rules in long-named directory: %s\n",
 775                             full_fname(dirbuf));
 776                 }
 777                 dirbuf[dirbuf_len] = '\0';
 778         }
 779
 780         return (void*)push;
 781 }
 782
 783 void pop_local_filters(void *mem)
 784 {
 785         struct local_filter_state *pop = (struct local_filter_state *)mem;
 786         int i;
 787         int old_mergelist_cnt = pop ? pop->mergelist_cnt : 0;
 788
 789         if (DEBUG_GTE(FILTER, 2))
 790                 rprintf(FINFO, "[%s] popping local filters\n", who_am_i());
 791
 792         for (i = mergelist_cnt; i-- > 0; ) {
 793                 filter_rule *ex = mergelist_parents[i];
 794                 filter_rule_list *lp;
 795                 if (!ex)
 796                         continue;
 797                 lp = ex->u.mergelist;
 798
 799                 if (DEBUG_GTE(FILTER, 2)) {
 800                         rprintf(FINFO, "[%s] popping mergelist #%d%s\n",
 801                                 who_am_i(), i, lp->debug_type);
 802                 }
 803
 804                 pop_filter_list(lp);
 805                 if (i >= old_mergelist_cnt && lp->head) {
 806                         /* This mergelist does not exist in the state to be restored, but it
 807                          * still has inherited rules.  This can sometimes happen if a per-dir
 808                          * merge file calls setup_merge_file() in push_local_filters() and that
 809                          * leaves some inherited rules that aren't in the pushed list state. */
 810                         if (DEBUG_GTE(FILTER, 2)) {
 811                                 rprintf(FINFO, "[%s] freeing parent_dirscan filters of mergelist #%d%s\n",
 812                                         who_am_i(), i, ex->u.mergelist->debug_type);
 813                         }
 814                         pop_filter_list(lp);
 815                 }
 816         }
 817
 818         if (!pop)
 819                 return; /* No state to restore. */
 820
 821         for (i = 0; i < old_mergelist_cnt; i++) {
 822                 filter_rule *ex = mergelist_parents[i];
 823                 if (!ex)
 824                         continue;
 825                 memcpy(ex->u.mergelist, &pop->mergelists[i], sizeof (filter_rule_list));
 826         }
 827
 828         free(pop);
 829 }
 830
 831 void change_local_filter_dir(const char *dname, int dlen, int dir_depth)
 832 {
 833         static int cur_depth = -1;
 834         static void *filt_array[MAXPATHLEN/2+1];
 835
 836         if (!dname) {
 837                 for ( ; cur_depth >= 0; cur_depth--) {
 838                         if (filt_array[cur_depth]) {
 839                                 pop_local_filters(filt_array[cur_depth]);
 840                                 filt_array[cur_depth] = NULL;
 841                         }
 842                 }
 843                 return;
 844         }
 845
 846         assert(dir_depth < MAXPATHLEN/2+1);
 847
 848         for ( ; cur_depth >= dir_depth; cur_depth--) {
 849                 if (filt_array[cur_depth]) {
 850                         pop_local_filters(filt_array[cur_depth]);
 851                         filt_array[cur_depth] = NULL;
 852                 }
 853         }
 854
 855         cur_depth = dir_depth;
 856         filt_array[cur_depth] = push_local_filters(dname, dlen);
 857 }
 858
 859 static int rule_matches(const char *fname, filter_rule *ex, int name_flags)
 860 {
 861         int slash_handling, str_cnt = 0, anchored_match = 0;
 862         int ret_match = ex->rflags & FILTRULE_NEGATE ? 0 : 1;
 863         char *p, *pattern = ex->pattern;
 864         const char *strings[16]; /* more than enough */
 865         const char *name = fname + (*fname == '/');
 866
 867         if (!*name)
 868                 return 0;
 869
 870         if (!(name_flags & NAME_IS_XATTR) ^ !(ex->rflags & FILTRULE_XATTR))
 871                 return 0;
 872
 873         if (!ex->u.slash_cnt && !(ex->rflags & FILTRULE_WILD2)) {
 874                 /* If the pattern does not have any slashes AND it does
 875                  * not have a "**" (which could match a slash), then we
 876                  * just match the name portion of the path. */
 877                 if ((p = strrchr(name,'/')) != NULL)
 878                         name = p+1;
 879         } else if (ex->rflags & FILTRULE_ABS_PATH && *fname != '/'
 880             && curr_dir_len > module_dirlen + 1) {
 881                 /* If we're matching against an absolute-path pattern,
 882                  * we need to prepend our full path info. */
 883                 strings[str_cnt++] = curr_dir + module_dirlen + 1;
 884                 strings[str_cnt++] = "/";
 885         } else if (ex->rflags & FILTRULE_WILD2_PREFIX && *fname != '/') {
 886                 /* Allow "**"+"/" to match at the start of the string. */
 887                 strings[str_cnt++] = "/";
 888         }
 889         strings[str_cnt++] = name;
 890         if (name_flags & NAME_IS_DIR) {
 891                 /* Allow a trailing "/"+"***" to match the directory. */
 892                 if (ex->rflags & FILTRULE_WILD3_SUFFIX)
 893                         strings[str_cnt++] = "/";
 894         } else if (ex->rflags & FILTRULE_DIRECTORY)
 895                 return !ret_match;
 896         strings[str_cnt] = NULL;
 897
 898         if (*pattern == '/') {
 899                 anchored_match = 1;
 900                 pattern++;
 901         }
 902
 903         if (!anchored_match && ex->u.slash_cnt
 904             && !(ex->rflags & FILTRULE_WILD2)) {
 905                 /* A non-anchored match with an infix slash and no "**"
 906                  * needs to match the last slash_cnt+1 name elements. */
 907                 slash_handling = ex->u.slash_cnt + 1;
 908         } else if (!anchored_match && !(ex->rflags & FILTRULE_WILD2_PREFIX)
 909                                    && ex->rflags & FILTRULE_WILD2) {
 910                 /* A non-anchored match with an infix or trailing "**" (but not
 911                  * a prefixed "**") needs to try matching after every slash. */
 912                 slash_handling = -1;
 913         } else {
 914                 /* The pattern matches only at the start of the path or name. */
 915                 slash_handling = 0;
 916         }
 917
 918         if (ex->rflags & FILTRULE_WILD) {
 919                 if (wildmatch_array(pattern, strings, slash_handling))
 920                         return ret_match;
 921         } else if (str_cnt > 1) {
 922                 if (litmatch_array(pattern, strings, slash_handling))
 923                         return ret_match;
 924         } else if (anchored_match) {
 925                 if (strcmp(name, pattern) == 0)
 926                         return ret_match;
 927         } else {
 928                 int l1 = strlen(name);
 929                 int l2 = strlen(pattern);
 930                 if (l2 <= l1 &&
 931                     strcmp(name+(l1-l2),pattern) == 0 &&
 932                     (l1==l2 || name[l1-(l2+1)] == '/')) {
 933                         return ret_match;
 934                 }
 935         }
 936
 937         return !ret_match;
 938 }
 939
 940 static void report_filter_result(enum logcode code, char const *name,
 941                                  filter_rule const *ent,
 942                                  int name_flags, const char *type)
 943 {
 944         int log_level = am_sender || am_generator ? 1 : 3;
 945
 946         /* If a trailing slash is present to match only directories,
 947          * then it is stripped out by add_rule().  So as a special
 948          * case we add it back in the log output. */
 949         if (DEBUG_GTE(FILTER, log_level)) {
 950                 static char *actions[2][2]
 951                     = { {"show", "hid"}, {"risk", "protect"} };
 952                 const char *w = who_am_i();
 953                 const char *t = name_flags & NAME_IS_XATTR ? "xattr"
 954                               : name_flags & NAME_IS_DIR ? "directory"
 955                               : "file";
 956                 rprintf(code, "[%s] %sing %s %s because of pattern %s%s%s\n",
 957                     w, actions[*w=='g'][!(ent->rflags & FILTRULE_INCLUDE)],
 958                     t, name, ent->pattern,
 959                     ent->rflags & FILTRULE_DIRECTORY ? "/" : "", type);
 960         }
 961 }
 962
 963 /* This function is used to check if a file should be included/excluded
 964  * from the list of files based on its name and type etc.  The value of
 965  * filter_level is set to either SERVER_FILTERS or ALL_FILTERS. */
 966 int name_is_excluded(const char *fname, int name_flags, int filter_level)
 967 {
 968         if (daemon_filter_list.head && check_filter(&daemon_filter_list, FLOG, fname, name_flags) < 0) {
 969                 if (!(name_flags & NAME_IS_XATTR))
 970                         errno = ENOENT;
 971                 return 1;
 972         }
 973
 974         if (filter_level != ALL_FILTERS)
 975                 return 0;
 976
 977         if (filter_list.head && check_filter(&filter_list, FINFO, fname, name_flags) < 0)
 978                 return 1;
 979
 980         return 0;
 981 }
 982
 983 /* Return -1 if file "name" is defined to be excluded by the specified
 984  * exclude list, 1 if it is included, and 0 if it was not matched. */
 985 int check_filter(filter_rule_list *listp, enum logcode code,
 986                  const char *name, int name_flags)
 987 {
 988         filter_rule *ent;
 989
 990         for (ent = listp->head; ent; ent = ent->next) {
 991                 if (ignore_perishable && ent->rflags & FILTRULE_PERISHABLE)
 992                         continue;
 993                 if (ent->rflags & FILTRULE_PERDIR_MERGE) {
 994                         int rc = check_filter(ent->u.mergelist, code, name, name_flags);
 995                         if (rc)
 996                                 return rc;
 997                         continue;
 998                 }
 999                 if (ent->rflags & FILTRULE_CVS_IGNORE) {
1000                         int rc = check_filter(&cvs_filter_list, code, name, name_flags);
1001                         if (rc)
1002                                 return rc;
1003                         continue;
1004                 }
1005                 if (rule_matches(name, ent, name_flags)) {
1006                         report_filter_result(code, name, ent, name_flags, listp->debug_type);
1007                         return ent->rflags & FILTRULE_INCLUDE ? 1 : -1;
1008                 }
1009         }
1010
1011         return 0;
1012 }
1013
1014 #define RULE_STRCMP(s,r) rule_strcmp((s), (r), sizeof (r) - 1)
1015
1016 static const uchar *rule_strcmp(const uchar *str, const char *rule, int rule_len)
1017 {
1018         if (strncmp((char*)str, rule, rule_len) != 0)
1019                 return NULL;
1020         if (isspace(str[rule_len]) || str[rule_len] == '_' || !str[rule_len])
1021                 return str + rule_len - 1;
1022         if (str[rule_len] == ',')
1023                 return str + rule_len;
1024         return NULL;
1025 }
1026
1027 #define FILTRULES_FROM_CONTAINER (FILTRULE_ABS_PATH | FILTRULE_INCLUDE \
1028                                 | FILTRULE_DIRECTORY | FILTRULE_NEGATE \
1029                                 | FILTRULE_PERISHABLE)
1030
1031 /* Gets the next include/exclude rule from *rulestr_ptr and advances
1032  * *rulestr_ptr to point beyond it.  Stores the pattern's start (within
1033  * *rulestr_ptr) and length in *pat_ptr and *pat_len_ptr, and returns a newly
1034  * allocated filter_rule containing the rest of the information.  Returns
1035  * NULL if there are no more rules in the input.
1036  *
1037  * The template provides defaults for the new rule to inherit, and the
1038  * template rflags and the xflags additionally affect parsing. */
1039 static filter_rule *parse_rule_tok(const char **rulestr_ptr,
1040                                    const filter_rule *template, int xflags,
1041                                    const char **pat_ptr, unsigned int *pat_len_ptr)
1042 {
1043         const uchar *s = (const uchar *)*rulestr_ptr;
1044         filter_rule *rule;
1045         unsigned int len;
1046
1047         if (template->rflags & FILTRULE_WORD_SPLIT) {
1048                 /* Skip over any initial whitespace. */
1049                 while (isspace(*s))
1050                         s++;
1051                 /* Update to point to real start of rule. */
1052                 *rulestr_ptr = (const char *)s;
1053         }
1054         if (!*s)
1055                 return NULL;
1056
1057         rule = new0(filter_rule);
1058
1059         /* Inherit from the template.  Don't inherit FILTRULES_SIDES; we check
1060          * that later. */
1061         rule->rflags = template->rflags & FILTRULES_FROM_CONTAINER;
1062
1063         /* Figure out what kind of a filter rule "s" is pointing at.  Note
1064          * that if FILTRULE_NO_PREFIXES is set, the rule is either an include
1065          * or an exclude based on the inheritance of the FILTRULE_INCLUDE
1066          * flag (above).  XFLG_OLD_PREFIXES indicates a compatibility mode
1067          * for old include/exclude patterns where just "+ " and "- " are
1068          * allowed as optional prefixes.  */
1069         if (template->rflags & FILTRULE_NO_PREFIXES) {
1070                 if (*s == '!' && template->rflags & FILTRULE_CVS_IGNORE)
1071                         rule->rflags |= FILTRULE_CLEAR_LIST; /* Tentative! */
1072         } else if (xflags & XFLG_OLD_PREFIXES) {
1073                 if (*s == '-' && s[1] == ' ') {
1074                         rule->rflags &= ~FILTRULE_INCLUDE;
1075                         s += 2;
1076                 } else if (*s == '+' && s[1] == ' ') {
1077                         rule->rflags |= FILTRULE_INCLUDE;
1078                         s += 2;
1079                 } else if (*s == '!')
1080                         rule->rflags |= FILTRULE_CLEAR_LIST; /* Tentative! */
1081         } else {
1082                 char ch = 0;
1083                 BOOL prefix_specifies_side = False;
1084                 switch (*s) {
1085                 case 'c':
1086                         if ((s = RULE_STRCMP(s, "clear")) != NULL)
1087                                 ch = '!';
1088                         break;
1089                 case 'd':
1090                         if ((s = RULE_STRCMP(s, "dir-merge")) != NULL)
1091                                 ch = ':';
1092                         break;
1093                 case 'e':
1094                         if ((s = RULE_STRCMP(s, "exclude")) != NULL)
1095                                 ch = '-';
1096                         break;
1097                 case 'h':
1098                         if ((s = RULE_STRCMP(s, "hide")) != NULL)
1099                                 ch = 'H';
1100                         break;
1101                 case 'i':
1102                         if ((s = RULE_STRCMP(s, "include")) != NULL)
1103                                 ch = '+';
1104                         break;
1105                 case 'm':
1106                         if ((s = RULE_STRCMP(s, "merge")) != NULL)
1107                                 ch = '.';
1108                         break;
1109                 case 'p':
1110                         if ((s = RULE_STRCMP(s, "protect")) != NULL)
1111                                 ch = 'P';
1112                         break;
1113                 case 'r':
1114                         if ((s = RULE_STRCMP(s, "risk")) != NULL)
1115                                 ch = 'R';
1116                         break;
1117                 case 's':
1118                         if ((s = RULE_STRCMP(s, "show")) != NULL)
1119                                 ch = 'S';
1120                         break;
1121                 default:
1122                         ch = *s;
1123                         if (s[1] == ',')
1124                                 s++;
1125                         break;
1126                 }
1127                 switch (ch) {
1128                 case ':':
1129                         trust_sender_filter = 1;
1130                         rule->rflags |= FILTRULE_PERDIR_MERGE
1131                                       | FILTRULE_FINISH_SETUP;
1132                         /* FALL THROUGH */
1133                 case '.':
1134                         rule->rflags |= FILTRULE_MERGE_FILE;
1135                         break;
1136                 case '+':
1137                         rule->rflags |= FILTRULE_INCLUDE;
1138                         break;
1139                 case '-':
1140                         break;
1141                 case 'S':
1142                         rule->rflags |= FILTRULE_INCLUDE;
1143                         /* FALL THROUGH */
1144                 case 'H':
1145                         rule->rflags |= FILTRULE_SENDER_SIDE;
1146                         prefix_specifies_side = True;
1147                         break;
1148                 case 'R':
1149                         rule->rflags |= FILTRULE_INCLUDE;
1150                         /* FALL THROUGH */
1151                 case 'P':
1152                         rule->rflags |= FILTRULE_RECEIVER_SIDE;
1153                         prefix_specifies_side = True;
1154                         break;
1155                 case '!':
1156                         rule->rflags |= FILTRULE_CLEAR_LIST;
1157                         break;
1158                 default:
1159                         rprintf(FERROR, "Unknown filter rule: `%s'\n", *rulestr_ptr);
1160                         exit_cleanup(RERR_SYNTAX);
1161                 }
1162                 while (ch != '!' && *++s && *s != ' ' && *s != '_') {
1163                         if (template->rflags & FILTRULE_WORD_SPLIT && isspace(*s)) {
1164                                 s--;
1165                                 break;
1166                         }
1167                         switch (*s) {
1168                         default:
1169                             invalid:
1170                                 rprintf(FERROR,
1171                                         "invalid modifier '%c' at position %d in filter rule: %s\n",
1172                                         *s, (int)(s - (const uchar *)*rulestr_ptr), *rulestr_ptr);
1173                                 exit_cleanup(RERR_SYNTAX);
1174                         case '-':
1175                                 if (!BITS_SETnUNSET(rule->rflags, FILTRULE_MERGE_FILE, FILTRULE_NO_PREFIXES))
1176                                         goto invalid;
1177                                 rule->rflags |= FILTRULE_NO_PREFIXES;
1178                                 break;
1179                         case '+':
1180                                 if (!BITS_SETnUNSET(rule->rflags, FILTRULE_MERGE_FILE, FILTRULE_NO_PREFIXES))
1181                                         goto invalid;
1182                                 rule->rflags |= FILTRULE_NO_PREFIXES
1183                                               | FILTRULE_INCLUDE;
1184                                 break;
1185                         case '/':
1186                                 rule->rflags |= FILTRULE_ABS_PATH;
1187                                 break;
1188                         case '!':
1189                                 /* Negation really goes with the pattern, so it
1190                                  * isn't useful as a merge-file default. */
1191                                 if (rule->rflags & FILTRULE_MERGE_FILE)
1192                                         goto invalid;
1193                                 rule->rflags |= FILTRULE_NEGATE;
1194                                 break;
1195                         case 'C':
1196                                 if (rule->rflags & FILTRULE_NO_PREFIXES || prefix_specifies_side)
1197                                         goto invalid;
1198                                 rule->rflags |= FILTRULE_NO_PREFIXES
1199                                               | FILTRULE_WORD_SPLIT
1200                                               | FILTRULE_NO_INHERIT
1201                                               | FILTRULE_CVS_IGNORE;
1202                                 break;
1203                         case 'e':
1204                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1205                                         goto invalid;
1206                                 rule->rflags |= FILTRULE_EXCLUDE_SELF;
1207                                 break;
1208                         case 'n':
1209                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1210                                         goto invalid;
1211                                 rule->rflags |= FILTRULE_NO_INHERIT;
1212                                 break;
1213                         case 'p':
1214                                 rule->rflags |= FILTRULE_PERISHABLE;
1215                                 break;
1216                         case 'r':
1217                                 if (prefix_specifies_side)
1218                                         goto invalid;
1219                                 rule->rflags |= FILTRULE_RECEIVER_SIDE;
1220                                 break;
1221                         case 's':
1222                                 if (prefix_specifies_side)
1223                                         goto invalid;
1224                                 rule->rflags |= FILTRULE_SENDER_SIDE;
1225                                 break;
1226                         case 'w':
1227                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1228                                         goto invalid;
1229                                 rule->rflags |= FILTRULE_WORD_SPLIT;
1230                                 break;
1231                         case 'x':
1232                                 rule->rflags |= FILTRULE_XATTR;
1233                                 saw_xattr_filter = 1;
1234                                 break;
1235                         }
1236                 }
1237                 if (*s)
1238                         s++;
1239         }
1240         if (template->rflags & FILTRULES_SIDES) {
1241                 if (rule->rflags & FILTRULES_SIDES) {
1242                         /* The filter and template both specify side(s).  This
1243                          * is dodgy (and won't work correctly if the template is
1244                          * a one-sided per-dir merge rule), so reject it. */
1245                         rprintf(FERROR,
1246                                 "specified-side merge file contains specified-side filter: %s\n",
1247                                 *rulestr_ptr);
1248                         exit_cleanup(RERR_SYNTAX);
1249                 }
1250                 rule->rflags |= template->rflags & FILTRULES_SIDES;
1251         }
1252
1253         if (template->rflags & FILTRULE_WORD_SPLIT) {
1254                 const uchar *cp = s;
1255                 /* Token ends at whitespace or the end of the string. */
1256                 while (!isspace(*cp) && *cp != '\0')
1257                         cp++;
1258                 len = cp - s;
1259         } else
1260                 len = strlen((char*)s);
1261
1262         if (rule->rflags & FILTRULE_CLEAR_LIST) {
1263                 if (!(rule->rflags & FILTRULE_NO_PREFIXES)
1264                  && !(xflags & XFLG_OLD_PREFIXES) && len) {
1265                         rprintf(FERROR,
1266                                 "'!' rule has trailing characters: %s\n", *rulestr_ptr);
1267                         exit_cleanup(RERR_SYNTAX);
1268                 }
1269                 if (len > 1)
1270                         rule->rflags &= ~FILTRULE_CLEAR_LIST;
1271         } else if (!len && !(rule->rflags & FILTRULE_CVS_IGNORE)) {
1272                 rprintf(FERROR, "unexpected end of filter rule: %s\n", *rulestr_ptr);
1273                 exit_cleanup(RERR_SYNTAX);
1274         }
1275
1276         /* --delete-excluded turns an un-modified include/exclude into a sender-side rule.  */
1277         if (delete_excluded
1278          && !(rule->rflags & (FILTRULES_SIDES|FILTRULE_MERGE_FILE|FILTRULE_PERDIR_MERGE)))
1279                 rule->rflags |= FILTRULE_SENDER_SIDE;
1280
1281         *pat_ptr = (const char *)s;
1282         *pat_len_ptr = len;
1283         *rulestr_ptr = *pat_ptr + len;
1284         return rule;
1285 }
1286
1287 static void get_cvs_excludes(uint32 rflags)
1288 {
1289         static int initialized = 0;
1290         char *p, fname[MAXPATHLEN];
1291
1292         if (initialized)
1293                 return;
1294         initialized = 1;
1295
1296         parse_filter_str(&cvs_filter_list, default_cvsignore(),
1297                          rule_template(rflags | (protocol_version >= 30 ? FILTRULE_PERISHABLE : 0)),
1298                          0);
1299
1300         p = module_id >= 0 && lp_use_chroot(module_id) ? "/" : getenv("HOME");
1301         if (p && pathjoin(fname, MAXPATHLEN, p, ".cvsignore") < MAXPATHLEN)
1302                 parse_filter_file(&cvs_filter_list, fname, rule_template(rflags), 0);
1303
1304         parse_filter_str(&cvs_filter_list, getenv("CVSIGNORE"), rule_template(rflags), 0);
1305 }
1306
1307 const filter_rule *rule_template(uint32 rflags)
1308 {
1309         static filter_rule template; /* zero-initialized */
1310         template.rflags = rflags;
1311         return &template;
1312 }
1313
1314 void parse_filter_str(filter_rule_list *listp, const char *rulestr,
1315                      const filter_rule *template, int xflags)
1316 {
1317         filter_rule *rule;
1318         const char *pat;
1319         unsigned int pat_len;
1320
1321         if (!rulestr)
1322                 return;
1323
1324         while (1) {
1325                 uint32 new_rflags;
1326
1327                 /* Remember that the returned string is NOT '\0' terminated! */
1328                 if (!(rule = parse_rule_tok(&rulestr, template, xflags, &pat, &pat_len)))
1329                         break;
1330
1331                 if (pat_len >= MAXPATHLEN) {
1332                         rprintf(FERROR, "discarding over-long filter: %.*s\n",
1333                                 (int)pat_len, pat);
1334                     free_continue:
1335                         free_filter(rule);
1336                         continue;
1337                 }
1338
1339                 new_rflags = rule->rflags;
1340                 if (new_rflags & FILTRULE_CLEAR_LIST) {
1341                         if (DEBUG_GTE(FILTER, 2)) {
1342                                 rprintf(FINFO,
1343                                         "[%s] clearing filter list%s\n",
1344                                         who_am_i(), listp->debug_type);
1345                         }
1346                         pop_filter_list(listp);
1347                         listp->head = NULL;
1348                         goto free_continue;
1349                 }
1350
1351                 if (new_rflags & FILTRULE_MERGE_FILE) {
1352                         if (!pat_len) {
1353                                 pat = ".cvsignore";
1354                                 pat_len = 10;
1355                         }
1356                         if (new_rflags & FILTRULE_EXCLUDE_SELF) {
1357                                 const char *name;
1358                                 filter_rule *excl_self;
1359
1360                                 excl_self = new0(filter_rule);
1361                                 /* Find the beginning of the basename and add an exclude for it. */
1362                                 for (name = pat + pat_len; name > pat && name[-1] != '/'; name--) {}
1363                                 add_rule(listp, name, (pat + pat_len) - name, excl_self, 0);
1364                                 rule->rflags &= ~FILTRULE_EXCLUDE_SELF;
1365                         }
1366                         if (new_rflags & FILTRULE_PERDIR_MERGE) {
1367                                 if (parent_dirscan) {
1368                                         const char *p;
1369                                         unsigned int len = pat_len;
1370                                         if ((p = parse_merge_name(pat, &len, module_dirlen)))
1371                                                 add_rule(listp, p, len, rule, 0);
1372                                         else
1373                                                 free_filter(rule);
1374                                         continue;
1375                                 }
1376                         } else {
1377                                 const char *p;
1378                                 unsigned int len = pat_len;
1379                                 if ((p = parse_merge_name(pat, &len, 0)))
1380                                         parse_filter_file(listp, p, rule, XFLG_FATAL_ERRORS);
1381                                 free_filter(rule);
1382                                 continue;
1383                         }
1384                 }
1385
1386                 add_rule(listp, pat, pat_len, rule, xflags);
1387
1388                 if (new_rflags & FILTRULE_CVS_IGNORE
1389                     && !(new_rflags & FILTRULE_MERGE_FILE))
1390                         get_cvs_excludes(new_rflags);
1391         }
1392 }
1393
1394 void parse_filter_file(filter_rule_list *listp, const char *fname, const filter_rule *template, int xflags)
1395 {
1396         FILE *fp;
1397         char line[BIGPATHBUFLEN];
1398         char *eob = line + sizeof line - 1;
1399         BOOL word_split = (template->rflags & FILTRULE_WORD_SPLIT) != 0;
1400
1401         if (!fname || !*fname)
1402                 return;
1403
1404         if (*fname != '-' || fname[1] || am_server) {
1405                 if (daemon_filter_list.head) {
1406                         strlcpy(line, fname, sizeof line);
1407                         clean_fname(line, CFN_COLLAPSE_DOT_DOT_DIRS);
1408                         if (check_filter(&daemon_filter_list, FLOG, line, 0) < 0)
1409                                 fp = NULL;
1410                         else
1411                                 fp = fopen(line, "rb");
1412                 } else
1413                         fp = fopen(fname, "rb");
1414         } else
1415                 fp = stdin;
1416
1417         if (DEBUG_GTE(FILTER, 2)) {
1418                 rprintf(FINFO, "[%s] parse_filter_file(%s,%x,%x)%s\n",
1419                         who_am_i(), fname, template->rflags, xflags,
1420                         fp ? "" : " [not found]");
1421         }
1422
1423         if (!fp) {
1424                 if (xflags & XFLG_FATAL_ERRORS) {
1425                         rsyserr(FERROR, errno,
1426                                 "failed to open %sclude file %s",
1427                                 template->rflags & FILTRULE_INCLUDE ? "in" : "ex",
1428                                 fname);
1429                         exit_cleanup(RERR_FILEIO);
1430                 }
1431                 return;
1432         }
1433         dirbuf[dirbuf_len] = '\0';
1434
1435         while (1) {
1436                 char *s = line;
1437                 int ch, overflow = 0;
1438                 while (1) {
1439                         if ((ch = getc(fp)) == EOF) {
1440                                 if (ferror(fp) && errno == EINTR) {
1441                                         clearerr(fp);
1442                                         continue;
1443                                 }
1444                                 break;
1445                         }
1446                         if (word_split && isspace(ch))
1447                                 break;
1448                         if (eol_nulls? !ch : (ch == '\n' || ch == '\r'))
1449                                 break;
1450                         if (s < eob)
1451                                 *s++ = ch;
1452                         else
1453                                 overflow = 1;
1454                 }
1455                 if (overflow) {
1456                         rprintf(FERROR, "discarding over-long filter: %s...\n", line);
1457                         s = line;
1458                 }
1459                 *s = '\0';
1460                 /* Skip an empty token and (when line parsing) comments. */
1461                 if (*line && (word_split || (*line != ';' && *line != '#')))
1462                         parse_filter_str(listp, line, template, xflags);
1463                 if (ch == EOF)
1464                         break;
1465         }
1466         fclose(fp);
1467 }
1468
1469 /* If the "for_xfer" flag is set, the prefix is made compatible with the
1470  * current protocol_version (if possible) or a NULL is returned (if not
1471  * possible). */
1472 char *get_rule_prefix(filter_rule *rule, const char *pat, int for_xfer,
1473                       unsigned int *plen_ptr)
1474 {
1475         static char buf[MAX_RULE_PREFIX+1];
1476         char *op = buf;
1477         int legal_len = for_xfer && protocol_version < 29 ? 1 : MAX_RULE_PREFIX-1;
1478
1479         if (rule->rflags & FILTRULE_PERDIR_MERGE) {
1480                 if (legal_len == 1)
1481                         return NULL;
1482                 *op++ = ':';
1483         } else if (rule->rflags & FILTRULE_INCLUDE)
1484                 *op++ = '+';
1485         else if (legal_len != 1
1486             || ((*pat == '-' || *pat == '+') && pat[1] == ' '))
1487                 *op++ = '-';
1488         else
1489                 legal_len = 0;
1490
1491         if (rule->rflags & FILTRULE_ABS_PATH)
1492                 *op++ = '/';
1493         if (rule->rflags & FILTRULE_NEGATE)
1494                 *op++ = '!';
1495         if (rule->rflags & FILTRULE_CVS_IGNORE)
1496                 *op++ = 'C';
1497         else {
1498                 if (rule->rflags & FILTRULE_NO_INHERIT)
1499                         *op++ = 'n';
1500                 if (rule->rflags & FILTRULE_WORD_SPLIT)
1501                         *op++ = 'w';
1502                 if (rule->rflags & FILTRULE_NO_PREFIXES) {
1503                         if (rule->rflags & FILTRULE_INCLUDE)
1504                                 *op++ = '+';
1505                         else
1506                                 *op++ = '-';
1507                 }
1508         }
1509         if (rule->rflags & FILTRULE_EXCLUDE_SELF)
1510                 *op++ = 'e';
1511         if (rule->rflags & FILTRULE_XATTR)
1512                 *op++ = 'x';
1513         if (rule->rflags & FILTRULE_SENDER_SIDE
1514             && (!for_xfer || protocol_version >= 29))
1515                 *op++ = 's';
1516         if (rule->rflags & FILTRULE_RECEIVER_SIDE
1517             && (!for_xfer || protocol_version >= 29
1518              || (delete_excluded && am_sender)))
1519                 *op++ = 'r';
1520         if (rule->rflags & FILTRULE_PERISHABLE) {
1521                 if (!for_xfer || protocol_version >= 30)
1522                         *op++ = 'p';
1523                 else if (am_sender)
1524                         return NULL;
1525         }
1526         if (op - buf > legal_len)
1527                 return NULL;
1528         if (legal_len)
1529                 *op++ = ' ';
1530         *op = '\0';
1531         if (plen_ptr)
1532                 *plen_ptr = op - buf;
1533         return buf;
1534 }
1535
1536 static void send_rules(int f_out, filter_rule_list *flp)
1537 {
1538         filter_rule *ent, *prev = NULL;
1539
1540         for (ent = flp->head; ent; ent = ent->next) {
1541                 unsigned int len, plen, dlen;
1542                 int elide = 0;
1543                 char *p;
1544
1545                 /* Note we need to check delete_excluded here in addition to
1546                  * the code in parse_rule_tok() because some rules may have
1547                  * been added before we found the --delete-excluded option.
1548                  * We must also elide any CVS merge-file rules to avoid a
1549                  * backward compatibility problem, and we elide any no-prefix
1550                  * merge files as an optimization (since they can only have
1551                  * include/exclude rules). */
1552                 if (ent->rflags & FILTRULE_SENDER_SIDE)
1553                         elide = am_sender ? 1 : -1;
1554                 if (ent->rflags & FILTRULE_RECEIVER_SIDE)
1555                         elide = elide ? 0 : am_sender ? -1 : 1;
1556                 else if (delete_excluded && !elide
1557                  && (!(ent->rflags & FILTRULE_PERDIR_MERGE)
1558                   || ent->rflags & FILTRULE_NO_PREFIXES))
1559                         elide = am_sender ? 1 : -1;
1560                 if (elide < 0) {
1561                         if (prev)
1562                                 prev->next = ent->next;
1563                         else
1564                                 flp->head = ent->next;
1565                 } else
1566                         prev = ent;
1567                 if (elide > 0)
1568                         continue;
1569                 if (ent->rflags & FILTRULE_CVS_IGNORE
1570                     && !(ent->rflags & FILTRULE_MERGE_FILE)) {
1571                         int f = am_sender || protocol_version < 29 ? f_out : -2;
1572                         send_rules(f, &cvs_filter_list);
1573                         if (f == f_out)
1574                                 continue;
1575                 }
1576                 p = get_rule_prefix(ent, ent->pattern, 1, &plen);
1577                 if (!p) {
1578                         rprintf(FERROR,
1579                                 "filter rules are too modern for remote rsync.\n");
1580                         exit_cleanup(RERR_PROTOCOL);
1581                 }
1582                 if (f_out < 0)
1583                         continue;
1584                 len = strlen(ent->pattern);
1585                 dlen = ent->rflags & FILTRULE_DIRECTORY ? 1 : 0;
1586                 if (!(plen + len + dlen))
1587                         continue;
1588                 write_int(f_out, plen + len + dlen);
1589                 if (plen)
1590                         write_buf(f_out, p, plen);
1591                 write_buf(f_out, ent->pattern, len);
1592                 if (dlen)
1593                         write_byte(f_out, '/');
1594         }
1595         flp->tail = prev;
1596 }
1597
1598 /* This is only called by the client. */
1599 void send_filter_list(int f_out)
1600 {
1601         int receiver_wants_list = prune_empty_dirs
1602             || (delete_mode && (!delete_excluded || protocol_version >= 29));
1603
1604         if (local_server || (am_sender && !receiver_wants_list))
1605                 f_out = -1;
1606         if (cvs_exclude && am_sender) {
1607                 if (protocol_version >= 29)
1608                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1609                 parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1610         }
1611
1612         send_rules(f_out, &filter_list);
1613
1614         if (f_out >= 0)
1615                 write_int(f_out, 0);
1616
1617         if (cvs_exclude) {
1618                 if (!am_sender || protocol_version < 29)
1619                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1620                 if (!am_sender)
1621                         parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1622         }
1623 }
1624
1625 /* This is only called by the server. */
1626 void recv_filter_list(int f_in)
1627 {
1628         char line[BIGPATHBUFLEN];
1629         int xflags = protocol_version >= 29 ? 0 : XFLG_OLD_PREFIXES;
1630         int receiver_wants_list = prune_empty_dirs
1631             || (delete_mode && (!delete_excluded || protocol_version >= 29));
1632         unsigned int len;
1633
1634         if (!local_server && (am_sender || receiver_wants_list)) {
1635                 while ((len = read_int(f_in)) != 0) {
1636                         if (len >= sizeof line)
1637                                 overflow_exit("recv_rules");
1638                         read_sbuf(f_in, line, len);
1639                         parse_filter_str(&filter_list, line, rule_template(0), xflags);
1640                 }
1641         }
1642
1643         if (cvs_exclude) {
1644                 if (local_server || am_sender || protocol_version < 29)
1645                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1646                 if (local_server || am_sender)
1647                         parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1648         }
1649
1650         if (local_server) /* filter out any rules that aren't for us. */
1651                 send_rules(-1, &filter_list);
1652 }