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                 if (partial_string_len)
 365                         add_implied_include("", 0);
 366                 free(partial_string_buf);
 367                 partial_string_buf = NULL;
 368         }
 369         partial_string_len = 0; /* paranoia */
 370 }
 371
 372 /* Each arg the client sends to the remote sender turns into an implied include
 373  * that the receiver uses to validate the file list from the sender. */
 374 void add_implied_include(const char *arg, int skip_daemon_module)
 375 {
 376         int arg_len, saw_wild = 0, saw_live_open_brkt = 0, backslash_cnt = 0;
 377         int slash_cnt = 0;
 378         const char *cp;
 379         char *p;
 380         if (trust_sender_args)
 381                 return;
 382         if (partial_string_len) {
 383                 arg_len = strlen(arg);
 384                 if (partial_string_len + arg_len >= MAXPATHLEN) {
 385                         partial_string_len = 0;
 386                         return; /* Should be impossible... */
 387                 }
 388                 memcpy(partial_string_buf + partial_string_len, arg, arg_len + 1);
 389                 partial_string_len = 0;
 390                 arg = partial_string_buf;
 391         }
 392         if (skip_daemon_module) {
 393                 if ((cp = strchr(arg, '/')) != NULL)
 394                         arg = cp + 1;
 395                 else
 396                         arg = "";
 397         }
 398         if (relative_paths) {
 399                 if ((cp = strstr(arg, "/./")) != NULL)
 400                         arg = cp + 3;
 401         } else if ((cp = strrchr(arg, '/')) != NULL) {
 402                 arg = cp + 1;
 403         }
 404         if (*arg == '.' && arg[1] == '\0')
 405                 arg++;
 406         arg_len = strlen(arg);
 407         if (arg_len) {
 408                 char *new_pat;
 409                 if (strpbrk(arg, "*[?")) {
 410                         /* We need to add room to escape backslashes if wildcard chars are present. */
 411                         for (cp = arg; (cp = strchr(cp, '\\')) != NULL; cp++)
 412                                 arg_len++;
 413                         saw_wild = 1;
 414                 }
 415                 arg_len++; /* Leave room for the prefixed slash */
 416                 p = new_pat = new_array(char, arg_len + 1);
 417                 *p++ = '/';
 418                 slash_cnt++;
 419                 for (cp = arg; *cp; ) {
 420                         switch (*cp) {
 421                           case '\\':
 422                                 if (cp[1] == ']') {
 423                                         if (!saw_wild)
 424                                                 cp++; /* A \] in a non-wild filter causes a problem, so drop the \ . */
 425                                 } else if (!strchr("*[?", cp[1])) {
 426                                         backslash_cnt++;
 427                                         if (saw_wild)
 428                                                 *p++ = '\\';
 429                                 }
 430                                 *p++ = *cp++;
 431                                 break;
 432                           case '/':
 433                                 if (p[-1] == '/') { /* This is safe because of the initial slash. */
 434                                         if (*++cp == '\0') {
 435                                                 slash_cnt--;
 436                                                 p--;
 437                                         }
 438                                 } else if (cp[1] == '\0') {
 439                                         cp++;
 440                                 } else {
 441                                         slash_cnt++;
 442                                         *p++ = *cp++;
 443                                 }
 444                                 break;
 445                           case '.':
 446                                 if (p[-1] == '/') {
 447                                         if (cp[1] == '/') {
 448                                                 cp += 2;
 449                                                 if (!*cp) {
 450                                                         slash_cnt--;
 451                                                         p--;
 452                                                 }
 453                                         } else if (cp[1] == '\0') {
 454                                                 cp++;
 455                                                 slash_cnt--;
 456                                                 p--;
 457                                                 break;
 458                                         }
 459                                 } else
 460                                         *p++ = *cp++;
 461                                 break;
 462                           case '[':
 463                                 saw_live_open_brkt = 1;
 464                                 *p++ = *cp++;
 465                                 break;
 466                           default:
 467                                 *p++ = *cp++;
 468                                 break;
 469                         }
 470                 }
 471                 *p = '\0';
 472                 arg_len = p - new_pat;
 473                 if (!arg_len)
 474                         free(new_pat);
 475                 else {
 476                         filter_rule *rule = new0(filter_rule);
 477                         rule->rflags = FILTRULE_INCLUDE + (saw_wild ? FILTRULE_WILD : 0);
 478                         rule->u.slash_cnt = slash_cnt;
 479                         arg = rule->pattern = new_pat;
 480                         if (!implied_filter_list.head)
 481                                 implied_filter_list.head = implied_filter_list.tail = rule;
 482                         else {
 483                                 rule->next = implied_filter_list.head;
 484                                 implied_filter_list.head = rule;
 485                         }
 486                         if (DEBUG_GTE(FILTER, 3))
 487                                 rprintf(FINFO, "[%s] add_IMPlied_include(%s)\n", who_am_i(), arg);
 488                         if (saw_live_open_brkt)
 489                                 maybe_add_literal_brackets_rule(rule, arg_len);
 490                         if (relative_paths && slash_cnt) {
 491                                 filter_rule const *ent;
 492                                 int found = 0;
 493                                 slash_cnt = 1;
 494                                 for (p = new_pat + 1; (p = strchr(p, '/')) != NULL; p++) {
 495                                         *p = '\0';
 496                                         for (ent = implied_filter_list.head; ent; ent = ent->next) {
 497                                                 if (ent != rule && strcmp(ent->pattern, new_pat) == 0) {
 498                                                         found = 1;
 499                                                         break;
 500                                                 }
 501                                         }
 502                                         if (!found) {
 503                                                 filter_rule *R_rule = new0(filter_rule);
 504                                                 R_rule->rflags = FILTRULE_INCLUDE | FILTRULE_DIRECTORY;
 505                                                 /* Check if our sub-path has wildcards or escaped backslashes */
 506                                                 if (saw_wild && strpbrk(rule->pattern, "*[?\\"))
 507                                                         R_rule->rflags |= FILTRULE_WILD;
 508                                                 R_rule->pattern = strdup(new_pat);
 509                                                 R_rule->u.slash_cnt = slash_cnt;
 510                                                 R_rule->next = implied_filter_list.head;
 511                                                 implied_filter_list.head = R_rule;
 512                                                 if (DEBUG_GTE(FILTER, 3)) {
 513                                                         rprintf(FINFO, "[%s] add_implied_include(%s/)\n",
 514                                                                 who_am_i(), R_rule->pattern);
 515                                                 }
 516                                                 if (saw_live_open_brkt)
 517                                                         maybe_add_literal_brackets_rule(R_rule, -1);
 518                                         }
 519                                         *p = '/';
 520                                         slash_cnt++;
 521                                 }
 522                         }
 523                 }
 524         }
 525
 526         if (recurse || xfer_dirs) {
 527                 /* Now create a rule with an added "/" & "**" or "*" at the end */
 528                 filter_rule *rule = new0(filter_rule);
 529                 rule->rflags = FILTRULE_INCLUDE | FILTRULE_WILD;
 530                 if (recurse)
 531                         rule->rflags |= FILTRULE_WILD2;
 532                 /* We must leave enough room for / * * \0. */
 533                 if (!saw_wild && backslash_cnt) {
 534                         /* We are appending a wildcard, so now the backslashes need to be escaped. */
 535                         p = rule->pattern = new_array(char, arg_len + backslash_cnt + 3 + 1);
 536                         for (cp = arg; *cp; ) { /* Note that arg_len != 0 because backslash_cnt > 0 */
 537                                 if (*cp == '\\')
 538                                         *p++ = '\\';
 539                                 *p++ = *cp++;
 540                         }
 541                 } else {
 542                         p = rule->pattern = new_array(char, arg_len + 3 + 1);
 543                         if (arg_len) {
 544                                 memcpy(p, arg, arg_len);
 545                                 p += arg_len;
 546                         }
 547                 }
 548                 if (p[-1] != '/') {
 549                         *p++ = '/';
 550                         slash_cnt++;
 551                 }
 552                 *p++ = '*';
 553                 if (recurse)
 554                         *p++ = '*';
 555                 *p = '\0';
 556                 rule->u.slash_cnt = slash_cnt;
 557                 rule->next = implied_filter_list.head;
 558                 implied_filter_list.head = rule;
 559                 if (DEBUG_GTE(FILTER, 3))
 560                         rprintf(FINFO, "[%s] add_implied_include(%s)\n", who_am_i(), rule->pattern);
 561                 if (saw_live_open_brkt)
 562                         maybe_add_literal_brackets_rule(rule, p - rule->pattern);
 563         }
 564 }
 565
 566 /* This frees any non-inherited items, leaving just inherited items on the list. */
 567 static void pop_filter_list(filter_rule_list *listp)
 568 {
 569         filter_rule *inherited;
 570
 571         if (!listp->tail)
 572                 return;
 573
 574         inherited = listp->tail->next;
 575
 576         /* Truncate any inherited items from the local list. */
 577         listp->tail->next = NULL;
 578         /* Now free everything that is left. */
 579         free_filters(listp->head);
 580
 581         listp->head = inherited;
 582         listp->tail = NULL;
 583 }
 584
 585 /* This returns an expanded (absolute) filename for the merge-file name if
 586  * the name has any slashes in it OR if the parent_dirscan var is True;
 587  * otherwise it returns the original merge_file name.  If the len_ptr value
 588  * is non-NULL the merge_file name is limited by the referenced length
 589  * value and will be updated with the length of the resulting name.  We
 590  * always return a name that is null terminated, even if the merge_file
 591  * name was not. */
 592 static char *parse_merge_name(const char *merge_file, unsigned int *len_ptr,
 593                               unsigned int prefix_skip)
 594 {
 595         static char buf[MAXPATHLEN];
 596         char *fn, tmpbuf[MAXPATHLEN];
 597         unsigned int fn_len;
 598
 599         if (!parent_dirscan && *merge_file != '/') {
 600                 /* Return the name unchanged it doesn't have any slashes. */
 601                 if (len_ptr) {
 602                         const char *p = merge_file + *len_ptr;
 603                         while (--p > merge_file && *p != '/') {}
 604                         if (p == merge_file) {
 605                                 strlcpy(buf, merge_file, *len_ptr + 1);
 606                                 return buf;
 607                         }
 608                 } else if (strchr(merge_file, '/') == NULL)
 609                         return (char *)merge_file;
 610         }
 611
 612         fn = *merge_file == '/' ? buf : tmpbuf;
 613         if (sanitize_paths) {
 614                 const char *r = prefix_skip ? "/" : NULL;
 615                 /* null-terminate the name if it isn't already */
 616                 if (len_ptr && merge_file[*len_ptr]) {
 617                         char *to = fn == buf ? tmpbuf : buf;
 618                         strlcpy(to, merge_file, *len_ptr + 1);
 619                         merge_file = to;
 620                 }
 621                 if (!sanitize_path(fn, merge_file, r, dirbuf_depth, SP_DEFAULT)) {
 622                         rprintf(FERROR, "merge-file name overflows: %s\n",
 623                                 merge_file);
 624                         return NULL;
 625                 }
 626                 fn_len = strlen(fn);
 627         } else {
 628                 strlcpy(fn, merge_file, len_ptr ? *len_ptr + 1 : MAXPATHLEN);
 629                 fn_len = clean_fname(fn, CFN_COLLAPSE_DOT_DOT_DIRS);
 630         }
 631
 632         /* If the name isn't in buf yet, it wasn't absolute. */
 633         if (fn != buf) {
 634                 int d_len = dirbuf_len - prefix_skip;
 635                 if (d_len + fn_len >= MAXPATHLEN) {
 636                         rprintf(FERROR, "merge-file name overflows: %s\n", fn);
 637                         return NULL;
 638                 }
 639                 memcpy(buf, dirbuf + prefix_skip, d_len);
 640                 memcpy(buf + d_len, fn, fn_len + 1);
 641                 fn_len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
 642         }
 643
 644         if (len_ptr)
 645                 *len_ptr = fn_len;
 646         return buf;
 647 }
 648
 649 /* Sets the dirbuf and dirbuf_len values. */
 650 void set_filter_dir(const char *dir, unsigned int dirlen)
 651 {
 652         unsigned int len;
 653         if (*dir != '/') {
 654                 memcpy(dirbuf, curr_dir, curr_dir_len);
 655                 dirbuf[curr_dir_len] = '/';
 656                 len = curr_dir_len + 1;
 657                 if (len + dirlen >= MAXPATHLEN)
 658                         dirlen = 0;
 659         } else
 660                 len = 0;
 661         memcpy(dirbuf + len, dir, dirlen);
 662         dirbuf[dirlen + len] = '\0';
 663         dirbuf_len = clean_fname(dirbuf, CFN_COLLAPSE_DOT_DOT_DIRS);
 664         if (dirbuf_len > 1 && dirbuf[dirbuf_len-1] == '.'
 665             && dirbuf[dirbuf_len-2] == '/')
 666                 dirbuf_len -= 2;
 667         if (dirbuf_len != 1)
 668                 dirbuf[dirbuf_len++] = '/';
 669         dirbuf[dirbuf_len] = '\0';
 670         if (sanitize_paths)
 671                 dirbuf_depth = count_dir_elements(dirbuf + module_dirlen);
 672 }
 673
 674 /* This routine takes a per-dir merge-file entry and finishes its setup.
 675  * If the name has a path portion then we check to see if it refers to a
 676  * parent directory of the first transfer dir.  If it does, we scan all the
 677  * dirs from that point through the parent dir of the transfer dir looking
 678  * for the per-dir merge-file in each one. */
 679 static BOOL setup_merge_file(int mergelist_num, filter_rule *ex,
 680                              filter_rule_list *lp)
 681 {
 682         char buf[MAXPATHLEN];
 683         char *x, *y, *pat = ex->pattern;
 684         unsigned int len;
 685
 686         if (!(x = parse_merge_name(pat, NULL, 0)) || *x != '/')
 687                 return 0;
 688
 689         if (DEBUG_GTE(FILTER, 2)) {
 690                 rprintf(FINFO, "[%s] performing parent_dirscan for mergelist #%d%s\n",
 691                         who_am_i(), mergelist_num, lp->debug_type);
 692         }
 693         y = strrchr(x, '/');
 694         *y = '\0';
 695         ex->pattern = strdup(y+1);
 696         if (!*x)
 697                 x = "/";
 698         if (*x == '/')
 699                 strlcpy(buf, x, MAXPATHLEN);
 700         else
 701                 pathjoin(buf, MAXPATHLEN, dirbuf, x);
 702
 703         len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
 704         if (len != 1 && len < MAXPATHLEN-1) {
 705                 buf[len++] = '/';
 706                 buf[len] = '\0';
 707         }
 708         /* This ensures that the specified dir is a parent of the transfer. */
 709         for (x = buf, y = dirbuf; *x && *x == *y; x++, y++) {}
 710         if (*x)
 711                 y += strlen(y); /* nope -- skip the scan */
 712
 713         parent_dirscan = True;
 714         while (*y) {
 715                 char save[MAXPATHLEN];
 716                 strlcpy(save, y, MAXPATHLEN);
 717                 *y = '\0';
 718                 dirbuf_len = y - dirbuf;
 719                 strlcpy(x, ex->pattern, MAXPATHLEN - (x - buf));
 720                 parse_filter_file(lp, buf, ex, XFLG_ANCHORED2ABS);
 721                 if (ex->rflags & FILTRULE_NO_INHERIT) {
 722                         /* Free the undesired rules to clean up any per-dir
 723                          * mergelists they defined.  Otherwise pop_local_filters
 724                          * may crash trying to restore nonexistent state for
 725                          * those mergelists. */
 726                         free_filters(lp->head);
 727                         lp->head = NULL;
 728                 }
 729                 lp->tail = NULL;
 730                 strlcpy(y, save, MAXPATHLEN);
 731                 while ((*x++ = *y++) != '/') {}
 732         }
 733         parent_dirscan = False;
 734         if (DEBUG_GTE(FILTER, 2)) {
 735                 rprintf(FINFO, "[%s] completed parent_dirscan for mergelist #%d%s\n",
 736                         who_am_i(), mergelist_num, lp->debug_type);
 737         }
 738         free(pat);
 739         return 1;
 740 }
 741
 742 struct local_filter_state {
 743         int mergelist_cnt;
 744         filter_rule_list mergelists[1];
 745 };
 746
 747 /* Each time rsync changes to a new directory it call this function to
 748  * handle all the per-dir merge-files.  The "dir" value is the current path
 749  * relative to curr_dir (which might not be null-terminated).  We copy it
 750  * into dirbuf so that we can easily append a file name on the end. */
 751 void *push_local_filters(const char *dir, unsigned int dirlen)
 752 {
 753         struct local_filter_state *push;
 754         int i;
 755
 756         set_filter_dir(dir, dirlen);
 757         if (DEBUG_GTE(FILTER, 2)) {
 758                 rprintf(FINFO, "[%s] pushing local filters for %s\n",
 759                         who_am_i(), dirbuf);
 760         }
 761
 762         if (!mergelist_cnt) {
 763                 /* No old state to save and no new merge files to push. */
 764                 return NULL;
 765         }
 766
 767         push = (struct local_filter_state *)new_array(char,
 768                           sizeof (struct local_filter_state)
 769                         + (mergelist_cnt-1) * sizeof (filter_rule_list));
 770
 771         push->mergelist_cnt = mergelist_cnt;
 772         for (i = 0; i < mergelist_cnt; i++) {
 773                 filter_rule *ex = mergelist_parents[i];
 774                 if (!ex)
 775                         continue;
 776                 memcpy(&push->mergelists[i], ex->u.mergelist, sizeof (filter_rule_list));
 777         }
 778
 779         /* Note: parse_filter_file() might increase mergelist_cnt, so keep
 780          * this loop separate from the above loop. */
 781         for (i = 0; i < mergelist_cnt; i++) {
 782                 filter_rule *ex = mergelist_parents[i];
 783                 filter_rule_list *lp;
 784                 if (!ex)
 785                         continue;
 786                 lp = ex->u.mergelist;
 787
 788                 if (DEBUG_GTE(FILTER, 2)) {
 789                         rprintf(FINFO, "[%s] pushing mergelist #%d%s\n",
 790                                 who_am_i(), i, lp->debug_type);
 791                 }
 792
 793                 lp->tail = NULL; /* Switch any local rules to inherited. */
 794                 if (ex->rflags & FILTRULE_NO_INHERIT)
 795                         lp->head = NULL;
 796
 797                 if (ex->rflags & FILTRULE_FINISH_SETUP) {
 798                         ex->rflags &= ~FILTRULE_FINISH_SETUP;
 799                         if (setup_merge_file(i, ex, lp))
 800                                 set_filter_dir(dir, dirlen);
 801                 }
 802
 803                 if (strlcpy(dirbuf + dirbuf_len, ex->pattern,
 804                     MAXPATHLEN - dirbuf_len) < MAXPATHLEN - dirbuf_len) {
 805                         parse_filter_file(lp, dirbuf, ex,
 806                                           XFLG_ANCHORED2ABS);
 807                 } else {
 808                         io_error |= IOERR_GENERAL;
 809                         rprintf(FERROR,
 810                             "cannot add local filter rules in long-named directory: %s\n",
 811                             full_fname(dirbuf));
 812                 }
 813                 dirbuf[dirbuf_len] = '\0';
 814         }
 815
 816         return (void*)push;
 817 }
 818
 819 void pop_local_filters(void *mem)
 820 {
 821         struct local_filter_state *pop = (struct local_filter_state *)mem;
 822         int i;
 823         int old_mergelist_cnt = pop ? pop->mergelist_cnt : 0;
 824
 825         if (DEBUG_GTE(FILTER, 2))
 826                 rprintf(FINFO, "[%s] popping local filters\n", who_am_i());
 827
 828         for (i = mergelist_cnt; i-- > 0; ) {
 829                 filter_rule *ex = mergelist_parents[i];
 830                 filter_rule_list *lp;
 831                 if (!ex)
 832                         continue;
 833                 lp = ex->u.mergelist;
 834
 835                 if (DEBUG_GTE(FILTER, 2)) {
 836                         rprintf(FINFO, "[%s] popping mergelist #%d%s\n",
 837                                 who_am_i(), i, lp->debug_type);
 838                 }
 839
 840                 pop_filter_list(lp);
 841                 if (i >= old_mergelist_cnt && lp->head) {
 842                         /* This mergelist does not exist in the state to be restored, but it
 843                          * still has inherited rules.  This can sometimes happen if a per-dir
 844                          * merge file calls setup_merge_file() in push_local_filters() and that
 845                          * leaves some inherited rules that aren't in the pushed list state. */
 846                         if (DEBUG_GTE(FILTER, 2)) {
 847                                 rprintf(FINFO, "[%s] freeing parent_dirscan filters of mergelist #%d%s\n",
 848                                         who_am_i(), i, ex->u.mergelist->debug_type);
 849                         }
 850                         pop_filter_list(lp);
 851                 }
 852         }
 853
 854         if (!pop)
 855                 return; /* No state to restore. */
 856
 857         for (i = 0; i < old_mergelist_cnt; i++) {
 858                 filter_rule *ex = mergelist_parents[i];
 859                 if (!ex)
 860                         continue;
 861                 memcpy(ex->u.mergelist, &pop->mergelists[i], sizeof (filter_rule_list));
 862         }
 863
 864         free(pop);
 865 }
 866
 867 void change_local_filter_dir(const char *dname, int dlen, int dir_depth)
 868 {
 869         static int cur_depth = -1;
 870         static void *filt_array[MAXPATHLEN/2+1];
 871
 872         if (!dname) {
 873                 for ( ; cur_depth >= 0; cur_depth--) {
 874                         if (filt_array[cur_depth]) {
 875                                 pop_local_filters(filt_array[cur_depth]);
 876                                 filt_array[cur_depth] = NULL;
 877                         }
 878                 }
 879                 return;
 880         }
 881
 882         assert(dir_depth < MAXPATHLEN/2+1);
 883
 884         for ( ; cur_depth >= dir_depth; cur_depth--) {
 885                 if (filt_array[cur_depth]) {
 886                         pop_local_filters(filt_array[cur_depth]);
 887                         filt_array[cur_depth] = NULL;
 888                 }
 889         }
 890
 891         cur_depth = dir_depth;
 892         filt_array[cur_depth] = push_local_filters(dname, dlen);
 893 }
 894
 895 static int rule_matches(const char *fname, filter_rule *ex, int name_flags)
 896 {
 897         int slash_handling, str_cnt = 0, anchored_match = 0;
 898         int ret_match = ex->rflags & FILTRULE_NEGATE ? 0 : 1;
 899         char *p, *pattern = ex->pattern;
 900         const char *strings[16]; /* more than enough */
 901         const char *name = fname + (*fname == '/');
 902
 903         if (!*name)
 904                 return 0;
 905
 906         if (!(name_flags & NAME_IS_XATTR) ^ !(ex->rflags & FILTRULE_XATTR))
 907                 return 0;
 908
 909         if (!ex->u.slash_cnt && !(ex->rflags & FILTRULE_WILD2)) {
 910                 /* If the pattern does not have any slashes AND it does
 911                  * not have a "**" (which could match a slash), then we
 912                  * just match the name portion of the path. */
 913                 if ((p = strrchr(name,'/')) != NULL)
 914                         name = p+1;
 915         } else if (ex->rflags & FILTRULE_ABS_PATH && *fname != '/'
 916             && curr_dir_len > module_dirlen + 1) {
 917                 /* If we're matching against an absolute-path pattern,
 918                  * we need to prepend our full path info. */
 919                 strings[str_cnt++] = curr_dir + module_dirlen + 1;
 920                 strings[str_cnt++] = "/";
 921         } else if (ex->rflags & FILTRULE_WILD2_PREFIX && *fname != '/') {
 922                 /* Allow "**"+"/" to match at the start of the string. */
 923                 strings[str_cnt++] = "/";
 924         }
 925         strings[str_cnt++] = name;
 926         if (name_flags & NAME_IS_DIR) {
 927                 /* Allow a trailing "/"+"***" to match the directory. */
 928                 if (ex->rflags & FILTRULE_WILD3_SUFFIX)
 929                         strings[str_cnt++] = "/";
 930         } else if (ex->rflags & FILTRULE_DIRECTORY)
 931                 return !ret_match;
 932         strings[str_cnt] = NULL;
 933
 934         if (*pattern == '/') {
 935                 anchored_match = 1;
 936                 pattern++;
 937         }
 938
 939         if (!anchored_match && ex->u.slash_cnt
 940             && !(ex->rflags & FILTRULE_WILD2)) {
 941                 /* A non-anchored match with an infix slash and no "**"
 942                  * needs to match the last slash_cnt+1 name elements. */
 943                 slash_handling = ex->u.slash_cnt + 1;
 944         } else if (!anchored_match && !(ex->rflags & FILTRULE_WILD2_PREFIX)
 945                                    && ex->rflags & FILTRULE_WILD2) {
 946                 /* A non-anchored match with an infix or trailing "**" (but not
 947                  * a prefixed "**") needs to try matching after every slash. */
 948                 slash_handling = -1;
 949         } else {
 950                 /* The pattern matches only at the start of the path or name. */
 951                 slash_handling = 0;
 952         }
 953
 954         if (ex->rflags & FILTRULE_WILD) {
 955                 if (wildmatch_array(pattern, strings, slash_handling))
 956                         return ret_match;
 957         } else if (str_cnt > 1) {
 958                 if (litmatch_array(pattern, strings, slash_handling))
 959                         return ret_match;
 960         } else if (anchored_match) {
 961                 if (strcmp(name, pattern) == 0)
 962                         return ret_match;
 963         } else {
 964                 int l1 = strlen(name);
 965                 int l2 = strlen(pattern);
 966                 if (l2 <= l1 &&
 967                     strcmp(name+(l1-l2),pattern) == 0 &&
 968                     (l1==l2 || name[l1-(l2+1)] == '/')) {
 969                         return ret_match;
 970                 }
 971         }
 972
 973         return !ret_match;
 974 }
 975
 976 static void report_filter_result(enum logcode code, char const *name,
 977                                  filter_rule const *ent,
 978                                  int name_flags, const char *type)
 979 {
 980         int log_level = am_sender || am_generator ? 1 : 3;
 981
 982         /* If a trailing slash is present to match only directories,
 983          * then it is stripped out by add_rule().  So as a special
 984          * case we add it back in the log output. */
 985         if (DEBUG_GTE(FILTER, log_level)) {
 986                 static char *actions[2][2]
 987                     = { {"show", "hid"}, {"risk", "protect"} };
 988                 const char *w = who_am_i();
 989                 const char *t = name_flags & NAME_IS_XATTR ? "xattr"
 990                               : name_flags & NAME_IS_DIR ? "directory"
 991                               : "file";
 992                 rprintf(code, "[%s] %sing %s %s because of pattern %s%s%s\n",
 993                     w, actions[*w=='g'][!(ent->rflags & FILTRULE_INCLUDE)],
 994                     t, name, ent->pattern,
 995                     ent->rflags & FILTRULE_DIRECTORY ? "/" : "", type);
 996         }
 997 }
 998
 999 /* This function is used to check if a file should be included/excluded
1000  * from the list of files based on its name and type etc.  The value of
1001  * filter_level is set to either SERVER_FILTERS or ALL_FILTERS. */
1002 int name_is_excluded(const char *fname, int name_flags, int filter_level)
1003 {
1004         if (daemon_filter_list.head && check_filter(&daemon_filter_list, FLOG, fname, name_flags) < 0) {
1005                 if (!(name_flags & NAME_IS_XATTR))
1006                         errno = ENOENT;
1007                 return 1;
1008         }
1009
1010         if (filter_level != ALL_FILTERS)
1011                 return 0;
1012
1013         if (filter_list.head && check_filter(&filter_list, FINFO, fname, name_flags) < 0)
1014                 return 1;
1015
1016         return 0;
1017 }
1018
1019 /* Return -1 if file "name" is defined to be excluded by the specified
1020  * exclude list, 1 if it is included, and 0 if it was not matched. */
1021 int check_filter(filter_rule_list *listp, enum logcode code,
1022                  const char *name, int name_flags)
1023 {
1024         filter_rule *ent;
1025
1026         for (ent = listp->head; ent; ent = ent->next) {
1027                 if (ignore_perishable && ent->rflags & FILTRULE_PERISHABLE)
1028                         continue;
1029                 if (ent->rflags & FILTRULE_PERDIR_MERGE) {
1030                         int rc = check_filter(ent->u.mergelist, code, name, name_flags);
1031                         if (rc)
1032                                 return rc;
1033                         continue;
1034                 }
1035                 if (ent->rflags & FILTRULE_CVS_IGNORE) {
1036                         int rc = check_filter(&cvs_filter_list, code, name, name_flags);
1037                         if (rc)
1038                                 return rc;
1039                         continue;
1040                 }
1041                 if (rule_matches(name, ent, name_flags)) {
1042                         report_filter_result(code, name, ent, name_flags, listp->debug_type);
1043                         return ent->rflags & FILTRULE_INCLUDE ? 1 : -1;
1044                 }
1045         }
1046
1047         return 0;
1048 }
1049
1050 #define RULE_STRCMP(s,r) rule_strcmp((s), (r), sizeof (r) - 1)
1051
1052 static const uchar *rule_strcmp(const uchar *str, const char *rule, int rule_len)
1053 {
1054         if (strncmp((char*)str, rule, rule_len) != 0)
1055                 return NULL;
1056         if (isspace(str[rule_len]) || str[rule_len] == '_' || !str[rule_len])
1057                 return str + rule_len - 1;
1058         if (str[rule_len] == ',')
1059                 return str + rule_len;
1060         return NULL;
1061 }
1062
1063 #define FILTRULES_FROM_CONTAINER (FILTRULE_ABS_PATH | FILTRULE_INCLUDE \
1064                                 | FILTRULE_DIRECTORY | FILTRULE_NEGATE \
1065                                 | FILTRULE_PERISHABLE)
1066
1067 /* Gets the next include/exclude rule from *rulestr_ptr and advances
1068  * *rulestr_ptr to point beyond it.  Stores the pattern's start (within
1069  * *rulestr_ptr) and length in *pat_ptr and *pat_len_ptr, and returns a newly
1070  * allocated filter_rule containing the rest of the information.  Returns
1071  * NULL if there are no more rules in the input.
1072  *
1073  * The template provides defaults for the new rule to inherit, and the
1074  * template rflags and the xflags additionally affect parsing. */
1075 static filter_rule *parse_rule_tok(const char **rulestr_ptr,
1076                                    const filter_rule *template, int xflags,
1077                                    const char **pat_ptr, unsigned int *pat_len_ptr)
1078 {
1079         const uchar *s = (const uchar *)*rulestr_ptr;
1080         filter_rule *rule;
1081         unsigned int len;
1082
1083         if (template->rflags & FILTRULE_WORD_SPLIT) {
1084                 /* Skip over any initial whitespace. */
1085                 while (isspace(*s))
1086                         s++;
1087                 /* Update to point to real start of rule. */
1088                 *rulestr_ptr = (const char *)s;
1089         }
1090         if (!*s)
1091                 return NULL;
1092
1093         rule = new0(filter_rule);
1094
1095         /* Inherit from the template.  Don't inherit FILTRULES_SIDES; we check
1096          * that later. */
1097         rule->rflags = template->rflags & FILTRULES_FROM_CONTAINER;
1098
1099         /* Figure out what kind of a filter rule "s" is pointing at.  Note
1100          * that if FILTRULE_NO_PREFIXES is set, the rule is either an include
1101          * or an exclude based on the inheritance of the FILTRULE_INCLUDE
1102          * flag (above).  XFLG_OLD_PREFIXES indicates a compatibility mode
1103          * for old include/exclude patterns where just "+ " and "- " are
1104          * allowed as optional prefixes.  */
1105         if (template->rflags & FILTRULE_NO_PREFIXES) {
1106                 if (*s == '!' && template->rflags & FILTRULE_CVS_IGNORE)
1107                         rule->rflags |= FILTRULE_CLEAR_LIST; /* Tentative! */
1108         } else if (xflags & XFLG_OLD_PREFIXES) {
1109                 if (*s == '-' && s[1] == ' ') {
1110                         rule->rflags &= ~FILTRULE_INCLUDE;
1111                         s += 2;
1112                 } else if (*s == '+' && s[1] == ' ') {
1113                         rule->rflags |= FILTRULE_INCLUDE;
1114                         s += 2;
1115                 } else if (*s == '!')
1116                         rule->rflags |= FILTRULE_CLEAR_LIST; /* Tentative! */
1117         } else {
1118                 char ch = 0;
1119                 BOOL prefix_specifies_side = False;
1120                 switch (*s) {
1121                 case 'c':
1122                         if ((s = RULE_STRCMP(s, "clear")) != NULL)
1123                                 ch = '!';
1124                         break;
1125                 case 'd':
1126                         if ((s = RULE_STRCMP(s, "dir-merge")) != NULL)
1127                                 ch = ':';
1128                         break;
1129                 case 'e':
1130                         if ((s = RULE_STRCMP(s, "exclude")) != NULL)
1131                                 ch = '-';
1132                         break;
1133                 case 'h':
1134                         if ((s = RULE_STRCMP(s, "hide")) != NULL)
1135                                 ch = 'H';
1136                         break;
1137                 case 'i':
1138                         if ((s = RULE_STRCMP(s, "include")) != NULL)
1139                                 ch = '+';
1140                         break;
1141                 case 'm':
1142                         if ((s = RULE_STRCMP(s, "merge")) != NULL)
1143                                 ch = '.';
1144                         break;
1145                 case 'p':
1146                         if ((s = RULE_STRCMP(s, "protect")) != NULL)
1147                                 ch = 'P';
1148                         break;
1149                 case 'r':
1150                         if ((s = RULE_STRCMP(s, "risk")) != NULL)
1151                                 ch = 'R';
1152                         break;
1153                 case 's':
1154                         if ((s = RULE_STRCMP(s, "show")) != NULL)
1155                                 ch = 'S';
1156                         break;
1157                 default:
1158                         ch = *s;
1159                         if (s[1] == ',')
1160                                 s++;
1161                         break;
1162                 }
1163                 switch (ch) {
1164                 case ':':
1165                         trust_sender_filter = 1;
1166                         rule->rflags |= FILTRULE_PERDIR_MERGE
1167                                       | FILTRULE_FINISH_SETUP;
1168                         /* FALL THROUGH */
1169                 case '.':
1170                         rule->rflags |= FILTRULE_MERGE_FILE;
1171                         break;
1172                 case '+':
1173                         rule->rflags |= FILTRULE_INCLUDE;
1174                         break;
1175                 case '-':
1176                         break;
1177                 case 'S':
1178                         rule->rflags |= FILTRULE_INCLUDE;
1179                         /* FALL THROUGH */
1180                 case 'H':
1181                         rule->rflags |= FILTRULE_SENDER_SIDE;
1182                         prefix_specifies_side = True;
1183                         break;
1184                 case 'R':
1185                         rule->rflags |= FILTRULE_INCLUDE;
1186                         /* FALL THROUGH */
1187                 case 'P':
1188                         rule->rflags |= FILTRULE_RECEIVER_SIDE;
1189                         prefix_specifies_side = True;
1190                         break;
1191                 case '!':
1192                         rule->rflags |= FILTRULE_CLEAR_LIST;
1193                         break;
1194                 default:
1195                         rprintf(FERROR, "Unknown filter rule: `%s'\n", *rulestr_ptr);
1196                         exit_cleanup(RERR_SYNTAX);
1197                 }
1198                 while (ch != '!' && *++s && *s != ' ' && *s != '_') {
1199                         if (template->rflags & FILTRULE_WORD_SPLIT && isspace(*s)) {
1200                                 s--;
1201                                 break;
1202                         }
1203                         switch (*s) {
1204                         default:
1205                             invalid:
1206                                 rprintf(FERROR,
1207                                         "invalid modifier '%c' at position %d in filter rule: %s\n",
1208                                         *s, (int)(s - (const uchar *)*rulestr_ptr), *rulestr_ptr);
1209                                 exit_cleanup(RERR_SYNTAX);
1210                         case '-':
1211                                 if (!BITS_SETnUNSET(rule->rflags, FILTRULE_MERGE_FILE, FILTRULE_NO_PREFIXES))
1212                                         goto invalid;
1213                                 rule->rflags |= FILTRULE_NO_PREFIXES;
1214                                 break;
1215                         case '+':
1216                                 if (!BITS_SETnUNSET(rule->rflags, FILTRULE_MERGE_FILE, FILTRULE_NO_PREFIXES))
1217                                         goto invalid;
1218                                 rule->rflags |= FILTRULE_NO_PREFIXES
1219                                               | FILTRULE_INCLUDE;
1220                                 break;
1221                         case '/':
1222                                 rule->rflags |= FILTRULE_ABS_PATH;
1223                                 break;
1224                         case '!':
1225                                 /* Negation really goes with the pattern, so it
1226                                  * isn't useful as a merge-file default. */
1227                                 if (rule->rflags & FILTRULE_MERGE_FILE)
1228                                         goto invalid;
1229                                 rule->rflags |= FILTRULE_NEGATE;
1230                                 break;
1231                         case 'C':
1232                                 if (rule->rflags & FILTRULE_NO_PREFIXES || prefix_specifies_side)
1233                                         goto invalid;
1234                                 rule->rflags |= FILTRULE_NO_PREFIXES
1235                                               | FILTRULE_WORD_SPLIT
1236                                               | FILTRULE_NO_INHERIT
1237                                               | FILTRULE_CVS_IGNORE;
1238                                 break;
1239                         case 'e':
1240                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1241                                         goto invalid;
1242                                 rule->rflags |= FILTRULE_EXCLUDE_SELF;
1243                                 break;
1244                         case 'n':
1245                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1246                                         goto invalid;
1247                                 rule->rflags |= FILTRULE_NO_INHERIT;
1248                                 break;
1249                         case 'p':
1250                                 rule->rflags |= FILTRULE_PERISHABLE;
1251                                 break;
1252                         case 'r':
1253                                 if (prefix_specifies_side)
1254                                         goto invalid;
1255                                 rule->rflags |= FILTRULE_RECEIVER_SIDE;
1256                                 break;
1257                         case 's':
1258                                 if (prefix_specifies_side)
1259                                         goto invalid;
1260                                 rule->rflags |= FILTRULE_SENDER_SIDE;
1261                                 break;
1262                         case 'w':
1263                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1264                                         goto invalid;
1265                                 rule->rflags |= FILTRULE_WORD_SPLIT;
1266                                 break;
1267                         case 'x':
1268                                 rule->rflags |= FILTRULE_XATTR;
1269                                 saw_xattr_filter = 1;
1270                                 break;
1271                         }
1272                 }
1273                 if (*s)
1274                         s++;
1275         }
1276         if (template->rflags & FILTRULES_SIDES) {
1277                 if (rule->rflags & FILTRULES_SIDES) {
1278                         /* The filter and template both specify side(s).  This
1279                          * is dodgy (and won't work correctly if the template is
1280                          * a one-sided per-dir merge rule), so reject it. */
1281                         rprintf(FERROR,
1282                                 "specified-side merge file contains specified-side filter: %s\n",
1283                                 *rulestr_ptr);
1284                         exit_cleanup(RERR_SYNTAX);
1285                 }
1286                 rule->rflags |= template->rflags & FILTRULES_SIDES;
1287         }
1288
1289         if (template->rflags & FILTRULE_WORD_SPLIT) {
1290                 const uchar *cp = s;
1291                 /* Token ends at whitespace or the end of the string. */
1292                 while (!isspace(*cp) && *cp != '\0')
1293                         cp++;
1294                 len = cp - s;
1295         } else
1296                 len = strlen((char*)s);
1297
1298         if (rule->rflags & FILTRULE_CLEAR_LIST) {
1299                 if (!(rule->rflags & FILTRULE_NO_PREFIXES)
1300                  && !(xflags & XFLG_OLD_PREFIXES) && len) {
1301                         rprintf(FERROR,
1302                                 "'!' rule has trailing characters: %s\n", *rulestr_ptr);
1303                         exit_cleanup(RERR_SYNTAX);
1304                 }
1305                 if (len > 1)
1306                         rule->rflags &= ~FILTRULE_CLEAR_LIST;
1307         } else if (!len && !(rule->rflags & FILTRULE_CVS_IGNORE)) {
1308                 rprintf(FERROR, "unexpected end of filter rule: %s\n", *rulestr_ptr);
1309                 exit_cleanup(RERR_SYNTAX);
1310         }
1311
1312         /* --delete-excluded turns an un-modified include/exclude into a sender-side rule.  */
1313         if (delete_excluded
1314          && !(rule->rflags & (FILTRULES_SIDES|FILTRULE_MERGE_FILE|FILTRULE_PERDIR_MERGE)))
1315                 rule->rflags |= FILTRULE_SENDER_SIDE;
1316
1317         *pat_ptr = (const char *)s;
1318         *pat_len_ptr = len;
1319         *rulestr_ptr = *pat_ptr + len;
1320         return rule;
1321 }
1322
1323 static void get_cvs_excludes(uint32 rflags)
1324 {
1325         static int initialized = 0;
1326         char *p, fname[MAXPATHLEN];
1327
1328         if (initialized)
1329                 return;
1330         initialized = 1;
1331
1332         parse_filter_str(&cvs_filter_list, default_cvsignore(),
1333                          rule_template(rflags | (protocol_version >= 30 ? FILTRULE_PERISHABLE : 0)),
1334                          0);
1335
1336         p = module_id >= 0 && lp_use_chroot(module_id) ? "/" : getenv("HOME");
1337         if (p && pathjoin(fname, MAXPATHLEN, p, ".cvsignore") < MAXPATHLEN)
1338                 parse_filter_file(&cvs_filter_list, fname, rule_template(rflags), 0);
1339
1340         parse_filter_str(&cvs_filter_list, getenv("CVSIGNORE"), rule_template(rflags), 0);
1341 }
1342
1343 const filter_rule *rule_template(uint32 rflags)
1344 {
1345         static filter_rule template; /* zero-initialized */
1346         template.rflags = rflags;
1347         return &template;
1348 }
1349
1350 void parse_filter_str(filter_rule_list *listp, const char *rulestr,
1351                      const filter_rule *template, int xflags)
1352 {
1353         filter_rule *rule;
1354         const char *pat;
1355         unsigned int pat_len;
1356
1357         if (!rulestr)
1358                 return;
1359
1360         while (1) {
1361                 uint32 new_rflags;
1362
1363                 /* Remember that the returned string is NOT '\0' terminated! */
1364                 if (!(rule = parse_rule_tok(&rulestr, template, xflags, &pat, &pat_len)))
1365                         break;
1366
1367                 if (pat_len >= MAXPATHLEN) {
1368                         rprintf(FERROR, "discarding over-long filter: %.*s\n",
1369                                 (int)pat_len, pat);
1370                     free_continue:
1371                         free_filter(rule);
1372                         continue;
1373                 }
1374
1375                 new_rflags = rule->rflags;
1376                 if (new_rflags & FILTRULE_CLEAR_LIST) {
1377                         if (DEBUG_GTE(FILTER, 2)) {
1378                                 rprintf(FINFO,
1379                                         "[%s] clearing filter list%s\n",
1380                                         who_am_i(), listp->debug_type);
1381                         }
1382                         pop_filter_list(listp);
1383                         listp->head = NULL;
1384                         goto free_continue;
1385                 }
1386
1387                 if (new_rflags & FILTRULE_MERGE_FILE) {
1388                         if (!pat_len) {
1389                                 pat = ".cvsignore";
1390                                 pat_len = 10;
1391                         }
1392                         if (new_rflags & FILTRULE_EXCLUDE_SELF) {
1393                                 const char *name;
1394                                 filter_rule *excl_self;
1395
1396                                 excl_self = new0(filter_rule);
1397                                 /* Find the beginning of the basename and add an exclude for it. */
1398                                 for (name = pat + pat_len; name > pat && name[-1] != '/'; name--) {}
1399                                 add_rule(listp, name, (pat + pat_len) - name, excl_self, 0);
1400                                 rule->rflags &= ~FILTRULE_EXCLUDE_SELF;
1401                         }
1402                         if (new_rflags & FILTRULE_PERDIR_MERGE) {
1403                                 if (parent_dirscan) {
1404                                         const char *p;
1405                                         unsigned int len = pat_len;
1406                                         if ((p = parse_merge_name(pat, &len, module_dirlen)))
1407                                                 add_rule(listp, p, len, rule, 0);
1408                                         else
1409                                                 free_filter(rule);
1410                                         continue;
1411                                 }
1412                         } else {
1413                                 const char *p;
1414                                 unsigned int len = pat_len;
1415                                 if ((p = parse_merge_name(pat, &len, 0)))
1416                                         parse_filter_file(listp, p, rule, XFLG_FATAL_ERRORS);
1417                                 free_filter(rule);
1418                                 continue;
1419                         }
1420                 }
1421
1422                 add_rule(listp, pat, pat_len, rule, xflags);
1423
1424                 if (new_rflags & FILTRULE_CVS_IGNORE
1425                     && !(new_rflags & FILTRULE_MERGE_FILE))
1426                         get_cvs_excludes(new_rflags);
1427         }
1428 }
1429
1430 void parse_filter_file(filter_rule_list *listp, const char *fname, const filter_rule *template, int xflags)
1431 {
1432         FILE *fp;
1433         char line[BIGPATHBUFLEN];
1434         char *eob = line + sizeof line - 1;
1435         BOOL word_split = (template->rflags & FILTRULE_WORD_SPLIT) != 0;
1436
1437         if (!fname || !*fname)
1438                 return;
1439
1440         if (*fname != '-' || fname[1] || am_server) {
1441                 if (daemon_filter_list.head) {
1442                         strlcpy(line, fname, sizeof line);
1443                         clean_fname(line, CFN_COLLAPSE_DOT_DOT_DIRS);
1444                         if (check_filter(&daemon_filter_list, FLOG, line, 0) < 0)
1445                                 fp = NULL;
1446                         else
1447                                 fp = fopen(line, "rb");
1448                 } else
1449                         fp = fopen(fname, "rb");
1450         } else
1451                 fp = stdin;
1452
1453         if (DEBUG_GTE(FILTER, 2)) {
1454                 rprintf(FINFO, "[%s] parse_filter_file(%s,%x,%x)%s\n",
1455                         who_am_i(), fname, template->rflags, xflags,
1456                         fp ? "" : " [not found]");
1457         }
1458
1459         if (!fp) {
1460                 if (xflags & XFLG_FATAL_ERRORS) {
1461                         rsyserr(FERROR, errno,
1462                                 "failed to open %sclude file %s",
1463                                 template->rflags & FILTRULE_INCLUDE ? "in" : "ex",
1464                                 fname);
1465                         exit_cleanup(RERR_FILEIO);
1466                 }
1467                 return;
1468         }
1469         dirbuf[dirbuf_len] = '\0';
1470
1471         while (1) {
1472                 char *s = line;
1473                 int ch, overflow = 0;
1474                 while (1) {
1475                         if ((ch = getc(fp)) == EOF) {
1476                                 if (ferror(fp) && errno == EINTR) {
1477                                         clearerr(fp);
1478                                         continue;
1479                                 }
1480                                 break;
1481                         }
1482                         if (word_split && isspace(ch))
1483                                 break;
1484                         if (eol_nulls? !ch : (ch == '\n' || ch == '\r'))
1485                                 break;
1486                         if (s < eob)
1487                                 *s++ = ch;
1488                         else
1489                                 overflow = 1;
1490                 }
1491                 if (overflow) {
1492                         rprintf(FERROR, "discarding over-long filter: %s...\n", line);
1493                         s = line;
1494                 }
1495                 *s = '\0';
1496                 /* Skip an empty token and (when line parsing) comments. */
1497                 if (*line && (word_split || (*line != ';' && *line != '#')))
1498                         parse_filter_str(listp, line, template, xflags);
1499                 if (ch == EOF)
1500                         break;
1501         }
1502         fclose(fp);
1503 }
1504
1505 /* If the "for_xfer" flag is set, the prefix is made compatible with the
1506  * current protocol_version (if possible) or a NULL is returned (if not
1507  * possible). */
1508 char *get_rule_prefix(filter_rule *rule, const char *pat, int for_xfer,
1509                       unsigned int *plen_ptr)
1510 {
1511         static char buf[MAX_RULE_PREFIX+1];
1512         char *op = buf;
1513         int legal_len = for_xfer && protocol_version < 29 ? 1 : MAX_RULE_PREFIX-1;
1514
1515         if (rule->rflags & FILTRULE_PERDIR_MERGE) {
1516                 if (legal_len == 1)
1517                         return NULL;
1518                 *op++ = ':';
1519         } else if (rule->rflags & FILTRULE_INCLUDE)
1520                 *op++ = '+';
1521         else if (legal_len != 1
1522             || ((*pat == '-' || *pat == '+') && pat[1] == ' '))
1523                 *op++ = '-';
1524         else
1525                 legal_len = 0;
1526
1527         if (rule->rflags & FILTRULE_ABS_PATH)
1528                 *op++ = '/';
1529         if (rule->rflags & FILTRULE_NEGATE)
1530                 *op++ = '!';
1531         if (rule->rflags & FILTRULE_CVS_IGNORE)
1532                 *op++ = 'C';
1533         else {
1534                 if (rule->rflags & FILTRULE_NO_INHERIT)
1535                         *op++ = 'n';
1536                 if (rule->rflags & FILTRULE_WORD_SPLIT)
1537                         *op++ = 'w';
1538                 if (rule->rflags & FILTRULE_NO_PREFIXES) {
1539                         if (rule->rflags & FILTRULE_INCLUDE)
1540                                 *op++ = '+';
1541                         else
1542                                 *op++ = '-';
1543                 }
1544         }
1545         if (rule->rflags & FILTRULE_EXCLUDE_SELF)
1546                 *op++ = 'e';
1547         if (rule->rflags & FILTRULE_XATTR)
1548                 *op++ = 'x';
1549         if (rule->rflags & FILTRULE_SENDER_SIDE
1550             && (!for_xfer || protocol_version >= 29))
1551                 *op++ = 's';
1552         if (rule->rflags & FILTRULE_RECEIVER_SIDE
1553             && (!for_xfer || protocol_version >= 29
1554              || (delete_excluded && am_sender)))
1555                 *op++ = 'r';
1556         if (rule->rflags & FILTRULE_PERISHABLE) {
1557                 if (!for_xfer || protocol_version >= 30)
1558                         *op++ = 'p';
1559                 else if (am_sender)
1560                         return NULL;
1561         }
1562         if (op - buf > legal_len)
1563                 return NULL;
1564         if (legal_len)
1565                 *op++ = ' ';
1566         *op = '\0';
1567         if (plen_ptr)
1568                 *plen_ptr = op - buf;
1569         return buf;
1570 }
1571
1572 static void send_rules(int f_out, filter_rule_list *flp)
1573 {
1574         filter_rule *ent, *prev = NULL;
1575
1576         for (ent = flp->head; ent; ent = ent->next) {
1577                 unsigned int len, plen, dlen;
1578                 int elide = 0;
1579                 char *p;
1580
1581                 /* Note we need to check delete_excluded here in addition to
1582                  * the code in parse_rule_tok() because some rules may have
1583                  * been added before we found the --delete-excluded option.
1584                  * We must also elide any CVS merge-file rules to avoid a
1585                  * backward compatibility problem, and we elide any no-prefix
1586                  * merge files as an optimization (since they can only have
1587                  * include/exclude rules). */
1588                 if (ent->rflags & FILTRULE_SENDER_SIDE)
1589                         elide = am_sender ? 1 : -1;
1590                 if (ent->rflags & FILTRULE_RECEIVER_SIDE)
1591                         elide = elide ? 0 : am_sender ? -1 : 1;
1592                 else if (delete_excluded && !elide
1593                  && (!(ent->rflags & FILTRULE_PERDIR_MERGE)
1594                   || ent->rflags & FILTRULE_NO_PREFIXES))
1595                         elide = am_sender ? 1 : -1;
1596                 if (elide < 0) {
1597                         if (prev)
1598                                 prev->next = ent->next;
1599                         else
1600                                 flp->head = ent->next;
1601                 } else
1602                         prev = ent;
1603                 if (elide > 0)
1604                         continue;
1605                 if (ent->rflags & FILTRULE_CVS_IGNORE
1606                     && !(ent->rflags & FILTRULE_MERGE_FILE)) {
1607                         int f = am_sender || protocol_version < 29 ? f_out : -2;
1608                         send_rules(f, &cvs_filter_list);
1609                         if (f == f_out)
1610                                 continue;
1611                 }
1612                 p = get_rule_prefix(ent, ent->pattern, 1, &plen);
1613                 if (!p) {
1614                         rprintf(FERROR,
1615                                 "filter rules are too modern for remote rsync.\n");
1616                         exit_cleanup(RERR_PROTOCOL);
1617                 }
1618                 if (f_out < 0)
1619                         continue;
1620                 len = strlen(ent->pattern);
1621                 dlen = ent->rflags & FILTRULE_DIRECTORY ? 1 : 0;
1622                 if (!(plen + len + dlen))
1623                         continue;
1624                 write_int(f_out, plen + len + dlen);
1625                 if (plen)
1626                         write_buf(f_out, p, plen);
1627                 write_buf(f_out, ent->pattern, len);
1628                 if (dlen)
1629                         write_byte(f_out, '/');
1630         }
1631         flp->tail = prev;
1632 }
1633
1634 /* This is only called by the client. */
1635 void send_filter_list(int f_out)
1636 {
1637         int receiver_wants_list = prune_empty_dirs
1638             || (delete_mode && (!delete_excluded || protocol_version >= 29));
1639
1640         if (local_server || (am_sender && !receiver_wants_list))
1641                 f_out = -1;
1642         if (cvs_exclude && am_sender) {
1643                 if (protocol_version >= 29)
1644                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1645                 parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1646         }
1647
1648         send_rules(f_out, &filter_list);
1649
1650         if (f_out >= 0)
1651                 write_int(f_out, 0);
1652
1653         if (cvs_exclude) {
1654                 if (!am_sender || protocol_version < 29)
1655                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1656                 if (!am_sender)
1657                         parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1658         }
1659 }
1660
1661 /* This is only called by the server. */
1662 void recv_filter_list(int f_in)
1663 {
1664         char line[BIGPATHBUFLEN];
1665         int xflags = protocol_version >= 29 ? 0 : XFLG_OLD_PREFIXES;
1666         int receiver_wants_list = prune_empty_dirs
1667             || (delete_mode && (!delete_excluded || protocol_version >= 29));
1668         unsigned int len;
1669
1670         if (!local_server && (am_sender || receiver_wants_list)) {
1671                 while ((len = read_int(f_in)) != 0) {
1672                         if (len >= sizeof line)
1673                                 overflow_exit("recv_rules");
1674                         read_sbuf(f_in, line, len);
1675                         parse_filter_str(&filter_list, line, rule_template(0), xflags);
1676                 }
1677         }
1678
1679         if (cvs_exclude) {
1680                 if (local_server || am_sender || protocol_version < 29)
1681                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1682                 if (local_server || am_sender)
1683                         parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1684         }
1685
1686         if (local_server) /* filter out any rules that aren't for us. */
1687                 send_rules(-1, &filter_list);
1688 }