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) 2003-2024 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 #define LOCAL_RULE   1
  82 #define REMOTE_RULE  2
  83 static uchar cur_elide_value = REMOTE_RULE;
  84
  85 /* Each filter_list_struct describes a singly-linked list by keeping track
  86  * of both the head and tail pointers.  The list is slightly unusual in that
  87  * a parent-dir's content can be appended to the end of the local list in a
  88  * special way:  the last item in the local list has its "next" pointer set
  89  * to point to the inherited list, but the local list's tail pointer points
  90  * at the end of the local list.  Thus, if the local list is empty, the head
  91  * will be pointing at the inherited content but the tail will be NULL.  To
  92  * help you visualize this, here are the possible list arrangements:
  93  *
  94  * Completely Empty                     Local Content Only
  95  * ==================================   ====================================
  96  * head -> NULL                         head -> Local1 -> Local2 -> NULL
  97  * tail -> NULL                         tail -------------^
  98  *
  99  * Inherited Content Only               Both Local and Inherited Content
 100  * ==================================   ====================================
 101  * head -> Parent1 -> Parent2 -> NULL   head -> L1 -> L2 -> P1 -> P2 -> NULL
 102  * tail -> NULL                         tail ---------^
 103  *
 104  * This means that anyone wanting to traverse the whole list to use it just
 105  * needs to start at the head and use the "next" pointers until it goes
 106  * NULL.  To add new local content, we insert the item after the tail item
 107  * and update the tail (obviously, if "tail" was NULL, we insert it at the
 108  * head).  To clear the local list, WE MUST NOT FREE THE INHERITED CONTENT
 109  * because it is shared between the current list and our parent list(s).
 110  * The easiest way to handle this is to simply truncate the list after the
 111  * tail item and then free the local list from the head.  When inheriting
 112  * the list for a new local dir, we just save off the filter_list_struct
 113  * values (so we can pop back to them later) and set the tail to NULL.
 114  */
 115
 116 static void teardown_mergelist(filter_rule *ex)
 117 {
 118         int j;
 119
 120         if (!ex->u.mergelist)
 121                 return;
 122
 123         if (DEBUG_GTE(FILTER, 2)) {
 124                 rprintf(FINFO, "[%s] deactivating mergelist #%d%s\n",
 125                         who_am_i(), mergelist_cnt - 1,
 126                         ex->u.mergelist->debug_type);
 127         }
 128
 129         free(ex->u.mergelist->debug_type);
 130         free(ex->u.mergelist);
 131
 132         for (j = 0; j < mergelist_cnt; j++) {
 133                 if (mergelist_parents[j] == ex) {
 134                         mergelist_parents[j] = NULL;
 135                         break;
 136                 }
 137         }
 138         while (mergelist_cnt && mergelist_parents[mergelist_cnt-1] == NULL)
 139                 mergelist_cnt--;
 140 }
 141
 142 static void free_filter(filter_rule *ex)
 143 {
 144         if (ex->rflags & FILTRULE_PERDIR_MERGE)
 145                 teardown_mergelist(ex);
 146         free(ex->pattern);
 147         free(ex);
 148 }
 149
 150 static void free_filters(filter_rule *ent)
 151 {
 152         while (ent) {
 153                 filter_rule *next = ent->next;
 154                 free_filter(ent);
 155                 ent = next;
 156         }
 157 }
 158
 159 /* Build a filter structure given a filter pattern.  The value in "pat"
 160  * is not null-terminated.  "rule" is either held or freed, so the
 161  * caller should not free it. */
 162 static void add_rule(filter_rule_list *listp, const char *pat, unsigned int pat_len,
 163                      filter_rule *rule, int xflags)
 164 {
 165         const char *cp;
 166         unsigned int pre_len, suf_len, slash_cnt = 0;
 167         char *mention_rule_suffix;
 168
 169         if (DEBUG_GTE(FILTER, 1) && pat_len && (pat[pat_len-1] == ' ' || pat[pat_len-1] == '\t'))
 170                 mention_rule_suffix = " -- CAUTION: trailing whitespace!";
 171         else
 172                 mention_rule_suffix = DEBUG_GTE(FILTER, 2) ? "" : NULL;
 173         if (mention_rule_suffix) {
 174                 rprintf(FINFO, "[%s] add_rule(%s%.*s%s)%s%s\n",
 175                         who_am_i(), get_rule_prefix(rule, pat, 0, NULL),
 176                         (int)pat_len, pat, (rule->rflags & FILTRULE_DIRECTORY) ? "/" : "",
 177                         listp->debug_type, mention_rule_suffix);
 178         }
 179
 180         /* These flags also indicate that we're reading a list that
 181          * needs to be filtered now, not post-filtered later. */
 182         if (xflags & (XFLG_ANCHORED2ABS|XFLG_ABS_IF_SLASH)
 183                 && (rule->rflags & FILTRULES_SIDES)
 184                         == (am_sender ? FILTRULE_RECEIVER_SIDE : FILTRULE_SENDER_SIDE)) {
 185                 /* This filter applies only to the other side.  Drop it. */
 186                 free_filter(rule);
 187                 return;
 188         }
 189
 190         if (pat_len > 1 && pat[pat_len-1] == '/') {
 191                 pat_len--;
 192                 rule->rflags |= FILTRULE_DIRECTORY;
 193         }
 194
 195         for (cp = pat; cp < pat + pat_len; cp++) {
 196                 if (*cp == '/')
 197                         slash_cnt++;
 198         }
 199
 200         if (!(rule->rflags & (FILTRULE_ABS_PATH | FILTRULE_MERGE_FILE))
 201          && ((xflags & (XFLG_ANCHORED2ABS|XFLG_ABS_IF_SLASH) && *pat == '/')
 202           || (xflags & XFLG_ABS_IF_SLASH && slash_cnt))) {
 203                 rule->rflags |= FILTRULE_ABS_PATH;
 204                 if (*pat == '/')
 205                         pre_len = dirbuf_len - module_dirlen - 1;
 206                 else
 207                         pre_len = 0;
 208         } else
 209                 pre_len = 0;
 210
 211         /* The daemon wants dir-exclude rules to get an appended "/" + "***". */
 212         if (xflags & XFLG_DIR2WILD3
 213          && BITS_SETnUNSET(rule->rflags, FILTRULE_DIRECTORY, FILTRULE_INCLUDE)) {
 214                 rule->rflags &= ~FILTRULE_DIRECTORY;
 215                 suf_len = sizeof SLASH_WILD3_SUFFIX - 1;
 216         } else
 217                 suf_len = 0;
 218
 219         rule->pattern = new_array(char, pre_len + pat_len + suf_len + 1);
 220         if (pre_len) {
 221                 memcpy(rule->pattern, dirbuf + module_dirlen, pre_len);
 222                 for (cp = rule->pattern; cp < rule->pattern + pre_len; cp++) {
 223                         if (*cp == '/')
 224                                 slash_cnt++;
 225                 }
 226         }
 227         rule->elide = 0;
 228         strlcpy(rule->pattern + pre_len, pat, pat_len + 1);
 229         pat_len += pre_len;
 230         if (suf_len) {
 231                 memcpy(rule->pattern + pat_len, SLASH_WILD3_SUFFIX, suf_len+1);
 232                 pat_len += suf_len;
 233                 slash_cnt++;
 234         }
 235
 236         if (strpbrk(rule->pattern, "*[?")) {
 237                 rule->rflags |= FILTRULE_WILD;
 238                 if ((cp = strstr(rule->pattern, "**")) != NULL) {
 239                         rule->rflags |= FILTRULE_WILD2;
 240                         /* If the pattern starts with **, note that. */
 241                         if (cp == rule->pattern)
 242                                 rule->rflags |= FILTRULE_WILD2_PREFIX;
 243                         /* If the pattern ends with ***, note that. */
 244                         if (pat_len >= 3
 245                          && rule->pattern[pat_len-3] == '*'
 246                          && rule->pattern[pat_len-2] == '*'
 247                          && rule->pattern[pat_len-1] == '*')
 248                                 rule->rflags |= FILTRULE_WILD3_SUFFIX;
 249                 }
 250         }
 251
 252         if (rule->rflags & FILTRULE_PERDIR_MERGE) {
 253                 filter_rule_list *lp;
 254                 unsigned int len;
 255                 int i;
 256
 257                 if ((cp = strrchr(rule->pattern, '/')) != NULL)
 258                         cp++;
 259                 else
 260                         cp = rule->pattern;
 261
 262                 /* If the local merge file was already mentioned, don't
 263                  * add it again. */
 264                 for (i = 0; i < mergelist_cnt; i++) {
 265                         filter_rule *ex = mergelist_parents[i];
 266                         const char *s;
 267                         if (!ex)
 268                                 continue;
 269                         s = strrchr(ex->pattern, '/');
 270                         if (s)
 271                                 s++;
 272                         else
 273                                 s = ex->pattern;
 274                         len = strlen(s);
 275                         if (len == pat_len - (cp - rule->pattern) && memcmp(s, cp, len) == 0) {
 276                                 free_filter(rule);
 277                                 return;
 278                         }
 279                 }
 280
 281                 lp = new_array0(filter_rule_list, 1);
 282                 if (asprintf(&lp->debug_type, " [per-dir %s]", cp) < 0)
 283                         out_of_memory("add_rule");
 284                 rule->u.mergelist = lp;
 285
 286                 if (mergelist_cnt == mergelist_size) {
 287                         mergelist_size += 5;
 288                         mergelist_parents = realloc_array(mergelist_parents, filter_rule *, mergelist_size);
 289                 }
 290                 if (DEBUG_GTE(FILTER, 2)) {
 291                         rprintf(FINFO, "[%s] activating mergelist #%d%s\n",
 292                                 who_am_i(), mergelist_cnt, lp->debug_type);
 293                 }
 294                 mergelist_parents[mergelist_cnt++] = rule;
 295         } else
 296                 rule->u.slash_cnt = slash_cnt;
 297
 298         if (!listp->tail) {
 299                 rule->next = listp->head;
 300                 listp->head = listp->tail = rule;
 301         } else {
 302                 rule->next = listp->tail->next;
 303                 listp->tail->next = rule;
 304                 listp->tail = rule;
 305         }
 306 }
 307
 308 /* If the wildcards failed, the remote shell might give us a file matching the literal
 309  * wildcards.  Since "*" & "?" already match themselves, this just needs to deal with
 310  * failed "[foo]" idioms.
 311  */
 312 static void maybe_add_literal_brackets_rule(filter_rule const *based_on, int arg_len)
 313 {
 314         filter_rule *rule;
 315         const char *arg = based_on->pattern, *cp;
 316         char *p;
 317         int cnt = 0;
 318
 319         if (arg_len < 0)
 320                 arg_len = strlen(arg);
 321
 322         for (cp = arg; *cp; cp++) {
 323                 if (*cp == '\\' && cp[1]) {
 324                         cp++;
 325                 } else if (*cp == '[')
 326                         cnt++;
 327         }
 328         if (!cnt)
 329                 return;
 330
 331         rule = new0(filter_rule);
 332         rule->rflags = based_on->rflags;
 333         rule->u.slash_cnt = based_on->u.slash_cnt;
 334         p = rule->pattern = new_array(char, arg_len + cnt + 1);
 335         for (cp = arg; *cp; ) {
 336                 if (*cp == '\\' && cp[1]) {
 337                         *p++ = *cp++;
 338                 } else if (*cp == '[')
 339                         *p++ = '\\';
 340                 *p++ = *cp++;
 341         }
 342         *p++ = '\0';
 343
 344         rule->next = implied_filter_list.head;
 345         implied_filter_list.head = rule;
 346         if (DEBUG_GTE(FILTER, 3)) {
 347                 rprintf(FINFO, "[%s] add_implied_include(%s%s)\n", who_am_i(), rule->pattern,
 348                         rule->rflags & FILTRULE_DIRECTORY ? "/" : "");
 349         }
 350 }
 351
 352 static char *partial_string_buf = NULL;
 353 static int partial_string_len = 0;
 354 void implied_include_partial_string(const char *s_start, const char *s_end)
 355 {
 356         partial_string_len = s_end - s_start;
 357         if (partial_string_len <= 0 || partial_string_len >= MAXPATHLEN) { /* too-large should be impossible... */
 358                 partial_string_len = 0;
 359                 return;
 360         }
 361         if (!partial_string_buf)
 362                 partial_string_buf = new_array(char, MAXPATHLEN);
 363         memcpy(partial_string_buf, s_start, partial_string_len);
 364 }
 365
 366 void free_implied_include_partial_string()
 367 {
 368         if (partial_string_buf) {
 369                 if (partial_string_len)
 370                         add_implied_include("", 0);
 371                 free(partial_string_buf);
 372                 partial_string_buf = NULL;
 373         }
 374         partial_string_len = 0; /* paranoia */
 375 }
 376
 377 /* Each arg the client sends to the remote sender turns into an implied include
 378  * that the receiver uses to validate the file list from the sender. */
 379 void add_implied_include(const char *arg, int skip_daemon_module)
 380 {
 381         int arg_len, saw_wild = 0, saw_live_open_brkt = 0, backslash_cnt = 0;
 382         int slash_cnt = 0;
 383         const char *cp;
 384         char *p;
 385         if (trust_sender_args)
 386                 return;
 387         if (partial_string_len) {
 388                 arg_len = strlen(arg);
 389                 if (partial_string_len + arg_len >= MAXPATHLEN) {
 390                         partial_string_len = 0;
 391                         return; /* Should be impossible... */
 392                 }
 393                 memcpy(partial_string_buf + partial_string_len, arg, arg_len + 1);
 394                 partial_string_len = 0;
 395                 arg = partial_string_buf;
 396         }
 397         if (skip_daemon_module) {
 398                 if ((cp = strchr(arg, '/')) != NULL)
 399                         arg = cp + 1;
 400                 else
 401                         arg = "";
 402         }
 403         if (relative_paths) {
 404                 if ((cp = strstr(arg, "/./")) != NULL)
 405                         arg = cp + 3;
 406         } else if ((cp = strrchr(arg, '/')) != NULL) {
 407                 arg = cp + 1;
 408         }
 409         if (*arg == '.' && arg[1] == '\0')
 410                 arg++;
 411         arg_len = strlen(arg);
 412         if (arg_len) {
 413                 char *new_pat;
 414                 if (strpbrk(arg, "*[?")) {
 415                         /* We need to add room to escape backslashes if wildcard chars are present. */
 416                         for (cp = arg; (cp = strchr(cp, '\\')) != NULL; cp++)
 417                                 arg_len++;
 418                         saw_wild = 1;
 419                 }
 420                 arg_len++; /* Leave room for the prefixed slash */
 421                 p = new_pat = new_array(char, arg_len + 1);
 422                 *p++ = '/';
 423                 slash_cnt++;
 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                                         if (*++cp == '\0') {
 440                                                 slash_cnt--;
 441                                                 p--;
 442                                         }
 443                                 } else if (cp[1] == '\0') {
 444                                         cp++;
 445                                 } else {
 446                                         slash_cnt++;
 447                                         *p++ = *cp++;
 448                                 }
 449                                 break;
 450                           case '.':
 451                                 if (p[-1] == '/') {
 452                                         if (cp[1] == '/') {
 453                                                 cp += 2;
 454                                                 if (!*cp) {
 455                                                         slash_cnt--;
 456                                                         p--;
 457                                                 }
 458                                         } else if (cp[1] == '\0') {
 459                                                 cp++;
 460                                                 slash_cnt--;
 461                                                 p--;
 462                                         } else
 463                                                 *p++ = *cp++;
 464                                 } else
 465                                         *p++ = *cp++;
 466                                 break;
 467                           case '[':
 468                                 saw_live_open_brkt = 1;
 469                                 *p++ = *cp++;
 470                                 break;
 471                           default:
 472                                 *p++ = *cp++;
 473                                 break;
 474                         }
 475                 }
 476                 *p = '\0';
 477                 arg_len = p - new_pat;
 478                 if (!arg_len)
 479                         free(new_pat);
 480                 else {
 481                         filter_rule *rule = new0(filter_rule);
 482                         rule->rflags = FILTRULE_INCLUDE + (saw_wild ? FILTRULE_WILD : 0);
 483                         rule->u.slash_cnt = slash_cnt;
 484                         arg = rule->pattern = new_pat;
 485                         if (!implied_filter_list.head)
 486                                 implied_filter_list.head = implied_filter_list.tail = rule;
 487                         else {
 488                                 rule->next = implied_filter_list.head;
 489                                 implied_filter_list.head = rule;
 490                         }
 491                         if (DEBUG_GTE(FILTER, 3))
 492                                 rprintf(FINFO, "[%s] add_implied_include(%s)\n", who_am_i(), arg);
 493                         if (saw_live_open_brkt)
 494                                 maybe_add_literal_brackets_rule(rule, arg_len);
 495                         if (relative_paths && slash_cnt) {
 496                                 int sub_slash_cnt = slash_cnt;
 497                                 while ((p = strrchr(new_pat, '/')) != NULL && p != new_pat) {
 498                                         filter_rule const *ent;
 499                                         filter_rule *R_rule;
 500                                         int found = 0;
 501                                         *p = '\0';
 502                                         for (ent = implied_filter_list.head; ent; ent = ent->next) {
 503                                                 if (ent != rule && strcmp(ent->pattern, new_pat) == 0) {
 504                                                         found = 1;
 505                                                         break;
 506                                                 }
 507                                         }
 508                                         if (found) {
 509                                                 *p = '/';
 510                                                 break; /* We added all parent dirs already */
 511                                         }
 512                                         R_rule = new0(filter_rule);
 513                                         R_rule->rflags = FILTRULE_INCLUDE | FILTRULE_DIRECTORY;
 514                                         /* Check if our sub-path has wildcards or escaped backslashes */
 515                                         if (saw_wild && strpbrk(new_pat, "*[?\\"))
 516                                                 R_rule->rflags |= FILTRULE_WILD;
 517                                         R_rule->pattern = strdup(new_pat);
 518                                         R_rule->u.slash_cnt = --sub_slash_cnt;
 519                                         R_rule->next = implied_filter_list.head;
 520                                         implied_filter_list.head = R_rule;
 521                                         if (DEBUG_GTE(FILTER, 3)) {
 522                                                 rprintf(FINFO, "[%s] add_implied_include(%s/)\n",
 523                                                         who_am_i(), R_rule->pattern);
 524                                         }
 525                                         if (saw_live_open_brkt)
 526                                                 maybe_add_literal_brackets_rule(R_rule, -1);
 527                                 }
 528                                 for (p = new_pat; sub_slash_cnt < slash_cnt; sub_slash_cnt++) {
 529                                         p += strlen(p);
 530                                         *p = '/';
 531                                 }
 532                         }
 533                 }
 534         }
 535
 536         if (recurse || xfer_dirs) {
 537                 /* Now create a rule with an added "/" & "**" or "*" at the end */
 538                 filter_rule *rule = new0(filter_rule);
 539                 rule->rflags = FILTRULE_INCLUDE | FILTRULE_WILD;
 540                 if (recurse)
 541                         rule->rflags |= FILTRULE_WILD2;
 542                 /* We must leave enough room for / * * \0. */
 543                 if (!saw_wild && backslash_cnt) {
 544                         /* We are appending a wildcard, so now the backslashes need to be escaped. */
 545                         p = rule->pattern = new_array(char, arg_len + backslash_cnt + 3 + 1);
 546                         for (cp = arg; *cp; ) { /* Note that arg_len != 0 because backslash_cnt > 0 */
 547                                 if (*cp == '\\')
 548                                         *p++ = '\\';
 549                                 *p++ = *cp++;
 550                         }
 551                 } else {
 552                         p = rule->pattern = new_array(char, arg_len + 3 + 1);
 553                         if (arg_len) {
 554                                 memcpy(p, arg, arg_len);
 555                                 p += arg_len;
 556                         }
 557                 }
 558                 *p++ = '/';
 559                 *p++ = '*';
 560                 if (recurse)
 561                         *p++ = '*';
 562                 *p = '\0';
 563                 rule->u.slash_cnt = slash_cnt + 1;
 564                 rule->next = implied_filter_list.head;
 565                 implied_filter_list.head = rule;
 566                 if (DEBUG_GTE(FILTER, 3))
 567                         rprintf(FINFO, "[%s] add_implied_include(%s)\n", who_am_i(), rule->pattern);
 568                 if (saw_live_open_brkt)
 569                         maybe_add_literal_brackets_rule(rule, p - rule->pattern);
 570         }
 571 }
 572
 573 /* This frees any non-inherited items, leaving just inherited items on the list. */
 574 static void pop_filter_list(filter_rule_list *listp)
 575 {
 576         filter_rule *inherited;
 577
 578         if (!listp->tail)
 579                 return;
 580
 581         inherited = listp->tail->next;
 582
 583         /* Truncate any inherited items from the local list. */
 584         listp->tail->next = NULL;
 585         /* Now free everything that is left. */
 586         free_filters(listp->head);
 587
 588         listp->head = inherited;
 589         listp->tail = NULL;
 590 }
 591
 592 /* This returns an expanded (absolute) filename for the merge-file name if
 593  * the name has any slashes in it OR if the parent_dirscan var is True;
 594  * otherwise it returns the original merge_file name.  If the len_ptr value
 595  * is non-NULL the merge_file name is limited by the referenced length
 596  * value and will be updated with the length of the resulting name.  We
 597  * always return a name that is null terminated, even if the merge_file
 598  * name was not. */
 599 static char *parse_merge_name(const char *merge_file, unsigned int *len_ptr,
 600                               unsigned int prefix_skip)
 601 {
 602         static char buf[MAXPATHLEN];
 603         char *fn, tmpbuf[MAXPATHLEN];
 604         unsigned int fn_len;
 605
 606         if (!parent_dirscan && *merge_file != '/') {
 607                 /* Return the name unchanged it doesn't have any slashes. */
 608                 if (len_ptr) {
 609                         const char *p = merge_file + *len_ptr;
 610                         while (--p > merge_file && *p != '/') {}
 611                         if (p == merge_file) {
 612                                 strlcpy(buf, merge_file, *len_ptr + 1);
 613                                 return buf;
 614                         }
 615                 } else if (strchr(merge_file, '/') == NULL)
 616                         return (char *)merge_file;
 617         }
 618
 619         fn = *merge_file == '/' ? buf : tmpbuf;
 620         if (sanitize_paths) {
 621                 const char *r = prefix_skip ? "/" : NULL;
 622                 /* null-terminate the name if it isn't already */
 623                 if (len_ptr && merge_file[*len_ptr]) {
 624                         char *to = fn == buf ? tmpbuf : buf;
 625                         strlcpy(to, merge_file, *len_ptr + 1);
 626                         merge_file = to;
 627                 }
 628                 if (!sanitize_path(fn, merge_file, r, dirbuf_depth, SP_DEFAULT)) {
 629                         rprintf(FERROR, "merge-file name overflows: %s\n",
 630                                 merge_file);
 631                         return NULL;
 632                 }
 633                 fn_len = strlen(fn);
 634         } else {
 635                 strlcpy(fn, merge_file, len_ptr ? *len_ptr + 1 : MAXPATHLEN);
 636                 fn_len = clean_fname(fn, CFN_COLLAPSE_DOT_DOT_DIRS);
 637         }
 638
 639         /* If the name isn't in buf yet, it wasn't absolute. */
 640         if (fn != buf) {
 641                 int d_len = dirbuf_len - prefix_skip;
 642                 if (d_len + fn_len >= MAXPATHLEN) {
 643                         rprintf(FERROR, "merge-file name overflows: %s\n", fn);
 644                         return NULL;
 645                 }
 646                 memcpy(buf, dirbuf + prefix_skip, d_len);
 647                 memcpy(buf + d_len, fn, fn_len + 1);
 648                 fn_len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
 649         }
 650
 651         if (len_ptr)
 652                 *len_ptr = fn_len;
 653         return buf;
 654 }
 655
 656 /* Sets the dirbuf and dirbuf_len values. */
 657 void set_filter_dir(const char *dir, unsigned int dirlen)
 658 {
 659         unsigned int len;
 660         if (*dir != '/') {
 661                 memcpy(dirbuf, curr_dir, curr_dir_len);
 662                 dirbuf[curr_dir_len] = '/';
 663                 len = curr_dir_len + 1;
 664                 if (len + dirlen >= MAXPATHLEN)
 665                         dirlen = 0;
 666         } else
 667                 len = 0;
 668         memcpy(dirbuf + len, dir, dirlen);
 669         dirbuf[dirlen + len] = '\0';
 670         dirbuf_len = clean_fname(dirbuf, CFN_COLLAPSE_DOT_DOT_DIRS);
 671         if (dirbuf_len > 1 && dirbuf[dirbuf_len-1] == '.'
 672             && dirbuf[dirbuf_len-2] == '/')
 673                 dirbuf_len -= 2;
 674         if (dirbuf_len != 1)
 675                 dirbuf[dirbuf_len++] = '/';
 676         dirbuf[dirbuf_len] = '\0';
 677         if (sanitize_paths)
 678                 dirbuf_depth = count_dir_elements(dirbuf + module_dirlen);
 679 }
 680
 681 /* This routine takes a per-dir merge-file entry and finishes its setup.
 682  * If the name has a path portion then we check to see if it refers to a
 683  * parent directory of the first transfer dir.  If it does, we scan all the
 684  * dirs from that point through the parent dir of the transfer dir looking
 685  * for the per-dir merge-file in each one. */
 686 static BOOL setup_merge_file(int mergelist_num, filter_rule *ex,
 687                              filter_rule_list *lp)
 688 {
 689         char buf[MAXPATHLEN];
 690         char *x, *y, *pat = ex->pattern;
 691         unsigned int len;
 692
 693         if (!(x = parse_merge_name(pat, NULL, 0)) || *x != '/')
 694                 return 0;
 695
 696         if (DEBUG_GTE(FILTER, 2)) {
 697                 rprintf(FINFO, "[%s] performing parent_dirscan for mergelist #%d%s\n",
 698                         who_am_i(), mergelist_num, lp->debug_type);
 699         }
 700         y = strrchr(x, '/');
 701         *y = '\0';
 702         ex->pattern = strdup(y+1);
 703         if (!*x)
 704                 x = "/";
 705         if (*x == '/')
 706                 strlcpy(buf, x, MAXPATHLEN);
 707         else
 708                 pathjoin(buf, MAXPATHLEN, dirbuf, x);
 709
 710         len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
 711         if (len != 1 && len < MAXPATHLEN-1) {
 712                 buf[len++] = '/';
 713                 buf[len] = '\0';
 714         }
 715         /* This ensures that the specified dir is a parent of the transfer. */
 716         for (x = buf, y = dirbuf; *x && *x == *y; x++, y++) {}
 717         if (*x)
 718                 y += strlen(y); /* nope -- skip the scan */
 719
 720         parent_dirscan = True;
 721         while (*y) {
 722                 char save[MAXPATHLEN];
 723                 /* copylen is strlen(y) which is < MAXPATHLEN. +1 for \0 */
 724                 size_t copylen = strlcpy(save, y, MAXPATHLEN) + 1;
 725                 *y = '\0';
 726                 dirbuf_len = y - dirbuf;
 727                 strlcpy(x, ex->pattern, MAXPATHLEN - (x - buf));
 728                 parse_filter_file(lp, buf, ex, XFLG_ANCHORED2ABS);
 729                 if (ex->rflags & FILTRULE_NO_INHERIT) {
 730                         /* Free the undesired rules to clean up any per-dir
 731                          * mergelists they defined.  Otherwise pop_local_filters
 732                          * may crash trying to restore nonexistent state for
 733                          * those mergelists. */
 734                         free_filters(lp->head);
 735                         lp->head = NULL;
 736                 }
 737                 lp->tail = NULL;
 738                 strlcpy(y, save, copylen);
 739                 while ((*x++ = *y++) != '/') {}
 740         }
 741         parent_dirscan = False;
 742         if (DEBUG_GTE(FILTER, 2)) {
 743                 rprintf(FINFO, "[%s] completed parent_dirscan for mergelist #%d%s\n",
 744                         who_am_i(), mergelist_num, lp->debug_type);
 745         }
 746         free(pat);
 747         return 1;
 748 }
 749
 750 struct local_filter_state {
 751         int mergelist_cnt;
 752         filter_rule_list mergelists[1];
 753 };
 754
 755 /* Each time rsync changes to a new directory it call this function to
 756  * handle all the per-dir merge-files.  The "dir" value is the current path
 757  * relative to curr_dir (which might not be null-terminated).  We copy it
 758  * into dirbuf so that we can easily append a file name on the end. */
 759 void *push_local_filters(const char *dir, unsigned int dirlen)
 760 {
 761         struct local_filter_state *push;
 762         int i;
 763
 764         set_filter_dir(dir, dirlen);
 765         if (DEBUG_GTE(FILTER, 2)) {
 766                 rprintf(FINFO, "[%s] pushing local filters for %s\n",
 767                         who_am_i(), dirbuf);
 768         }
 769
 770         if (!mergelist_cnt) {
 771                 /* No old state to save and no new merge files to push. */
 772                 return NULL;
 773         }
 774
 775         push = (struct local_filter_state *)new_array(char,
 776                           sizeof (struct local_filter_state)
 777                         + (mergelist_cnt-1) * sizeof (filter_rule_list));
 778
 779         push->mergelist_cnt = mergelist_cnt;
 780         for (i = 0; i < mergelist_cnt; i++) {
 781                 filter_rule *ex = mergelist_parents[i];
 782                 if (!ex)
 783                         continue;
 784                 memcpy(&push->mergelists[i], ex->u.mergelist, sizeof (filter_rule_list));
 785         }
 786
 787         /* Note: parse_filter_file() might increase mergelist_cnt, so keep
 788          * this loop separate from the above loop. */
 789         for (i = 0; i < mergelist_cnt; i++) {
 790                 filter_rule *ex = mergelist_parents[i];
 791                 filter_rule_list *lp;
 792                 if (!ex)
 793                         continue;
 794                 lp = ex->u.mergelist;
 795
 796                 if (DEBUG_GTE(FILTER, 2)) {
 797                         rprintf(FINFO, "[%s] pushing mergelist #%d%s\n",
 798                                 who_am_i(), i, lp->debug_type);
 799                 }
 800
 801                 lp->tail = NULL; /* Switch any local rules to inherited. */
 802                 if (ex->rflags & FILTRULE_NO_INHERIT)
 803                         lp->head = NULL;
 804
 805                 if (ex->rflags & FILTRULE_FINISH_SETUP) {
 806                         ex->rflags &= ~FILTRULE_FINISH_SETUP;
 807                         if (setup_merge_file(i, ex, lp))
 808                                 set_filter_dir(dir, dirlen);
 809                 }
 810
 811                 if (strlcpy(dirbuf + dirbuf_len, ex->pattern,
 812                     MAXPATHLEN - dirbuf_len) < MAXPATHLEN - dirbuf_len) {
 813                         parse_filter_file(lp, dirbuf, ex,
 814                                           XFLG_ANCHORED2ABS);
 815                 } else {
 816                         io_error |= IOERR_GENERAL;
 817                         rprintf(FERROR,
 818                             "cannot add local filter rules in long-named directory: %s\n",
 819                             full_fname(dirbuf));
 820                 }
 821                 dirbuf[dirbuf_len] = '\0';
 822         }
 823
 824         return (void*)push;
 825 }
 826
 827 void pop_local_filters(void *mem)
 828 {
 829         struct local_filter_state *pop = (struct local_filter_state *)mem;
 830         int i;
 831         int old_mergelist_cnt = pop ? pop->mergelist_cnt : 0;
 832
 833         if (DEBUG_GTE(FILTER, 2))
 834                 rprintf(FINFO, "[%s] popping local filters\n", who_am_i());
 835
 836         for (i = mergelist_cnt; i-- > 0; ) {
 837                 filter_rule *ex = mergelist_parents[i];
 838                 filter_rule_list *lp;
 839                 if (!ex)
 840                         continue;
 841                 lp = ex->u.mergelist;
 842
 843                 if (DEBUG_GTE(FILTER, 2)) {
 844                         rprintf(FINFO, "[%s] popping mergelist #%d%s\n",
 845                                 who_am_i(), i, lp->debug_type);
 846                 }
 847
 848                 pop_filter_list(lp);
 849                 if (i >= old_mergelist_cnt && lp->head) {
 850                         /* This mergelist does not exist in the state to be restored, but it
 851                          * still has inherited rules.  This can sometimes happen if a per-dir
 852                          * merge file calls setup_merge_file() in push_local_filters() and that
 853                          * leaves some inherited rules that aren't in the pushed list state. */
 854                         if (DEBUG_GTE(FILTER, 2)) {
 855                                 rprintf(FINFO, "[%s] freeing parent_dirscan filters of mergelist #%d%s\n",
 856                                         who_am_i(), i, ex->u.mergelist->debug_type);
 857                         }
 858                         pop_filter_list(lp);
 859                 }
 860         }
 861
 862         if (!pop)
 863                 return; /* No state to restore. */
 864
 865         for (i = 0; i < old_mergelist_cnt; i++) {
 866                 filter_rule *ex = mergelist_parents[i];
 867                 if (!ex)
 868                         continue;
 869                 memcpy(ex->u.mergelist, &pop->mergelists[i], sizeof (filter_rule_list));
 870         }
 871
 872         free(pop);
 873 }
 874
 875 void change_local_filter_dir(const char *dname, int dlen, int dir_depth)
 876 {
 877         static int cur_depth = -1;
 878         static void *filt_array[MAXPATHLEN/2+1];
 879
 880         if (!dname) {
 881                 for ( ; cur_depth >= 0; cur_depth--) {
 882                         if (filt_array[cur_depth]) {
 883                                 pop_local_filters(filt_array[cur_depth]);
 884                                 filt_array[cur_depth] = NULL;
 885                         }
 886                 }
 887                 return;
 888         }
 889
 890         assert(dir_depth < MAXPATHLEN/2+1);
 891
 892         for ( ; cur_depth >= dir_depth; cur_depth--) {
 893                 if (filt_array[cur_depth]) {
 894                         pop_local_filters(filt_array[cur_depth]);
 895                         filt_array[cur_depth] = NULL;
 896                 }
 897         }
 898
 899         cur_depth = dir_depth;
 900         filt_array[cur_depth] = push_local_filters(dname, dlen);
 901 }
 902
 903 static int rule_matches(const char *fname, filter_rule *ex, int name_flags)
 904 {
 905         int slash_handling, str_cnt = 0, anchored_match = 0;
 906         int ret_match = ex->rflags & FILTRULE_NEGATE ? 0 : 1;
 907         char *p, *pattern = ex->pattern;
 908         const char *strings[16]; /* more than enough */
 909         const char *name = fname + (*fname == '/');
 910
 911         if (!*name || ex->elide == cur_elide_value)
 912                 return 0;
 913
 914         if (!(name_flags & NAME_IS_XATTR) ^ !(ex->rflags & FILTRULE_XATTR))
 915                 return 0;
 916
 917         if (!ex->u.slash_cnt && !(ex->rflags & FILTRULE_WILD2)) {
 918                 /* If the pattern does not have any slashes AND it does
 919                  * not have a "**" (which could match a slash), then we
 920                  * just match the name portion of the path. */
 921                 if ((p = strrchr(name,'/')) != NULL)
 922                         name = p+1;
 923         } else if (ex->rflags & FILTRULE_ABS_PATH && *fname != '/'
 924             && curr_dir_len > module_dirlen + 1) {
 925                 /* If we're matching against an absolute-path pattern,
 926                  * we need to prepend our full path info. */
 927                 strings[str_cnt++] = curr_dir + module_dirlen + 1;
 928                 strings[str_cnt++] = "/";
 929         } else if (ex->rflags & FILTRULE_WILD2_PREFIX && *fname != '/') {
 930                 /* Allow "**"+"/" to match at the start of the string. */
 931                 strings[str_cnt++] = "/";
 932         }
 933         strings[str_cnt++] = name;
 934         if (name_flags & NAME_IS_DIR) {
 935                 /* Allow a trailing "/"+"***" to match the directory. */
 936                 if (ex->rflags & FILTRULE_WILD3_SUFFIX)
 937                         strings[str_cnt++] = "/";
 938         } else if (ex->rflags & FILTRULE_DIRECTORY)
 939                 return !ret_match;
 940         strings[str_cnt] = NULL;
 941
 942         if (*pattern == '/') {
 943                 anchored_match = 1;
 944                 pattern++;
 945         }
 946
 947         if (!anchored_match && ex->u.slash_cnt
 948             && !(ex->rflags & FILTRULE_WILD2)) {
 949                 /* A non-anchored match with an infix slash and no "**"
 950                  * needs to match the last slash_cnt+1 name elements. */
 951                 slash_handling = ex->u.slash_cnt + 1;
 952         } else if (!anchored_match && !(ex->rflags & FILTRULE_WILD2_PREFIX)
 953                                    && ex->rflags & FILTRULE_WILD2) {
 954                 /* A non-anchored match with an infix or trailing "**" (but not
 955                  * a prefixed "**") needs to try matching after every slash. */
 956                 slash_handling = -1;
 957         } else {
 958                 /* The pattern matches only at the start of the path or name. */
 959                 slash_handling = 0;
 960         }
 961
 962         if (ex->rflags & FILTRULE_WILD) {
 963                 if (wildmatch_array(pattern, strings, slash_handling))
 964                         return ret_match;
 965         } else if (str_cnt > 1) {
 966                 if (litmatch_array(pattern, strings, slash_handling))
 967                         return ret_match;
 968         } else if (anchored_match) {
 969                 if (strcmp(name, pattern) == 0)
 970                         return ret_match;
 971         } else {
 972                 int l1 = strlen(name);
 973                 int l2 = strlen(pattern);
 974                 if (l2 <= l1 &&
 975                     strcmp(name+(l1-l2),pattern) == 0 &&
 976                     (l1==l2 || name[l1-(l2+1)] == '/')) {
 977                         return ret_match;
 978                 }
 979         }
 980
 981         return !ret_match;
 982 }
 983
 984 static void report_filter_result(enum logcode code, char const *name,
 985                                  filter_rule const *ent,
 986                                  int name_flags, const char *type)
 987 {
 988         int log_level = am_sender || am_generator ? 1 : 3;
 989
 990         /* If a trailing slash is present to match only directories,
 991          * then it is stripped out by add_rule().  So as a special
 992          * case we add it back in the log output. */
 993         if (DEBUG_GTE(FILTER, log_level)) {
 994                 static char *actions[2][2]
 995                     = { {"show", "hid"}, {"risk", "protect"} };
 996                 const char *w = who_am_i();
 997                 const char *t = name_flags & NAME_IS_XATTR ? "xattr"
 998                               : name_flags & NAME_IS_DIR ? "directory"
 999                               : "file";
1000                 rprintf(code, "[%s] %sing %s %s because of pattern %s%s%s\n",
1001                     w, actions[*w=='g'][!(ent->rflags & FILTRULE_INCLUDE)],
1002                     t, name, ent->pattern,
1003                     ent->rflags & FILTRULE_DIRECTORY ? "/" : "", type);
1004         }
1005 }
1006
1007 /* This function is used to check if a file should be included/excluded
1008  * from the list of files based on its name and type etc.  The value of
1009  * filter_level is set to either SERVER_FILTERS or ALL_FILTERS. */
1010 int name_is_excluded(const char *fname, int name_flags, int filter_level)
1011 {
1012         if (daemon_filter_list.head && check_filter(&daemon_filter_list, FLOG, fname, name_flags) < 0) {
1013                 if (!(name_flags & NAME_IS_XATTR))
1014                         errno = ENOENT;
1015                 return 1;
1016         }
1017
1018         if (filter_level != ALL_FILTERS)
1019                 return 0;
1020
1021         if (filter_list.head && check_filter(&filter_list, FINFO, fname, name_flags) < 0)
1022                 return 1;
1023
1024         return 0;
1025 }
1026
1027 int check_server_filter(filter_rule_list *listp, enum logcode code, const char *name, int name_flags)
1028 {
1029         int ret;
1030         cur_elide_value = LOCAL_RULE;
1031         ret = check_filter(listp, code, name, name_flags);
1032         cur_elide_value = REMOTE_RULE;
1033         return ret;
1034 }
1035
1036 /* Return -1 if file "name" is defined to be excluded by the specified
1037  * exclude list, 1 if it is included, and 0 if it was not matched. */
1038 int check_filter(filter_rule_list *listp, enum logcode code,
1039                  const char *name, int name_flags)
1040 {
1041         filter_rule *ent;
1042
1043         for (ent = listp->head; ent; ent = ent->next) {
1044                 if (ignore_perishable && ent->rflags & FILTRULE_PERISHABLE)
1045                         continue;
1046                 if (ent->rflags & FILTRULE_PERDIR_MERGE) {
1047                         int rc = check_filter(ent->u.mergelist, code, name, name_flags);
1048                         if (rc)
1049                                 return rc;
1050                         continue;
1051                 }
1052                 if (ent->rflags & FILTRULE_CVS_IGNORE) {
1053                         int rc = check_filter(&cvs_filter_list, code, name, name_flags);
1054                         if (rc)
1055                                 return rc;
1056                         continue;
1057                 }
1058                 if (rule_matches(name, ent, name_flags)) {
1059                         report_filter_result(code, name, ent, name_flags, listp->debug_type);
1060                         return ent->rflags & FILTRULE_INCLUDE ? 1 : -1;
1061                 }
1062         }
1063
1064         return 0;
1065 }
1066
1067 #define RULE_STRCMP(s,r) rule_strcmp((s), (r), sizeof (r) - 1)
1068
1069 static const uchar *rule_strcmp(const uchar *str, const char *rule, int rule_len)
1070 {
1071         if (strncmp((char*)str, rule, rule_len) != 0)
1072                 return NULL;
1073         if (isspace(str[rule_len]) || str[rule_len] == '_' || !str[rule_len])
1074                 return str + rule_len - 1;
1075         if (str[rule_len] == ',')
1076                 return str + rule_len;
1077         return NULL;
1078 }
1079
1080 #define FILTRULES_FROM_CONTAINER (FILTRULE_ABS_PATH | FILTRULE_INCLUDE \
1081                                 | FILTRULE_DIRECTORY | FILTRULE_NEGATE \
1082                                 | FILTRULE_PERISHABLE)
1083
1084 /* Gets the next include/exclude rule from *rulestr_ptr and advances
1085  * *rulestr_ptr to point beyond it.  Stores the pattern's start (within
1086  * *rulestr_ptr) and length in *pat_ptr and *pat_len_ptr, and returns a newly
1087  * allocated filter_rule containing the rest of the information.  Returns
1088  * NULL if there are no more rules in the input.
1089  *
1090  * The template provides defaults for the new rule to inherit, and the
1091  * template rflags and the xflags additionally affect parsing. */
1092 static filter_rule *parse_rule_tok(const char **rulestr_ptr,
1093                                    const filter_rule *template, int xflags,
1094                                    const char **pat_ptr, unsigned int *pat_len_ptr)
1095 {
1096         const uchar *s = (const uchar *)*rulestr_ptr;
1097         filter_rule *rule;
1098         unsigned int len;
1099
1100         if (template->rflags & FILTRULE_WORD_SPLIT) {
1101                 /* Skip over any initial whitespace. */
1102                 while (isspace(*s))
1103                         s++;
1104                 /* Update to point to real start of rule. */
1105                 *rulestr_ptr = (const char *)s;
1106         }
1107         if (!*s)
1108                 return NULL;
1109
1110         rule = new0(filter_rule);
1111
1112         /* Inherit from the template.  Don't inherit FILTRULES_SIDES; we check
1113          * that later. */
1114         rule->rflags = template->rflags & FILTRULES_FROM_CONTAINER;
1115
1116         /* Figure out what kind of a filter rule "s" is pointing at.  Note
1117          * that if FILTRULE_NO_PREFIXES is set, the rule is either an include
1118          * or an exclude based on the inheritance of the FILTRULE_INCLUDE
1119          * flag (above).  XFLG_OLD_PREFIXES indicates a compatibility mode
1120          * for old include/exclude patterns where just "+ " and "- " are
1121          * allowed as optional prefixes.  */
1122         if (template->rflags & FILTRULE_NO_PREFIXES) {
1123                 if (*s == '!' && template->rflags & FILTRULE_CVS_IGNORE)
1124                         rule->rflags |= FILTRULE_CLEAR_LIST; /* Tentative! */
1125         } else if (xflags & XFLG_OLD_PREFIXES) {
1126                 if (*s == '-' && s[1] == ' ') {
1127                         rule->rflags &= ~FILTRULE_INCLUDE;
1128                         s += 2;
1129                 } else if (*s == '+' && s[1] == ' ') {
1130                         rule->rflags |= FILTRULE_INCLUDE;
1131                         s += 2;
1132                 } else if (*s == '!')
1133                         rule->rflags |= FILTRULE_CLEAR_LIST; /* Tentative! */
1134         } else {
1135                 char ch = 0;
1136                 BOOL prefix_specifies_side = False;
1137                 switch (*s) {
1138                 case 'c':
1139                         if ((s = RULE_STRCMP(s, "clear")) != NULL)
1140                                 ch = '!';
1141                         break;
1142                 case 'd':
1143                         if ((s = RULE_STRCMP(s, "dir-merge")) != NULL)
1144                                 ch = ':';
1145                         break;
1146                 case 'e':
1147                         if ((s = RULE_STRCMP(s, "exclude")) != NULL)
1148                                 ch = '-';
1149                         break;
1150                 case 'h':
1151                         if ((s = RULE_STRCMP(s, "hide")) != NULL)
1152                                 ch = 'H';
1153                         break;
1154                 case 'i':
1155                         if ((s = RULE_STRCMP(s, "include")) != NULL)
1156                                 ch = '+';
1157                         break;
1158                 case 'm':
1159                         if ((s = RULE_STRCMP(s, "merge")) != NULL)
1160                                 ch = '.';
1161                         break;
1162                 case 'p':
1163                         if ((s = RULE_STRCMP(s, "protect")) != NULL)
1164                                 ch = 'P';
1165                         break;
1166                 case 'r':
1167                         if ((s = RULE_STRCMP(s, "risk")) != NULL)
1168                                 ch = 'R';
1169                         break;
1170                 case 's':
1171                         if ((s = RULE_STRCMP(s, "show")) != NULL)
1172                                 ch = 'S';
1173                         break;
1174                 default:
1175                         ch = *s;
1176                         if (s[1] == ',')
1177                                 s++;
1178                         break;
1179                 }
1180                 switch (ch) {
1181                 case ':':
1182                         trust_sender_filter = 1;
1183                         rule->rflags |= FILTRULE_PERDIR_MERGE
1184                                       | FILTRULE_FINISH_SETUP;
1185                         /* FALL THROUGH */
1186                 case '.':
1187                         rule->rflags |= FILTRULE_MERGE_FILE;
1188                         break;
1189                 case '+':
1190                         rule->rflags |= FILTRULE_INCLUDE;
1191                         break;
1192                 case '-':
1193                         break;
1194                 case 'S':
1195                         rule->rflags |= FILTRULE_INCLUDE;
1196                         /* FALL THROUGH */
1197                 case 'H':
1198                         rule->rflags |= FILTRULE_SENDER_SIDE;
1199                         prefix_specifies_side = True;
1200                         break;
1201                 case 'R':
1202                         rule->rflags |= FILTRULE_INCLUDE;
1203                         /* FALL THROUGH */
1204                 case 'P':
1205                         rule->rflags |= FILTRULE_RECEIVER_SIDE;
1206                         prefix_specifies_side = True;
1207                         break;
1208                 case '!':
1209                         rule->rflags |= FILTRULE_CLEAR_LIST;
1210                         break;
1211                 default:
1212                         rprintf(FERROR, "Unknown filter rule: `%s'\n", *rulestr_ptr);
1213                         exit_cleanup(RERR_SYNTAX);
1214                 }
1215                 while (ch != '!' && *++s && *s != ' ' && *s != '_') {
1216                         if (template->rflags & FILTRULE_WORD_SPLIT && isspace(*s)) {
1217                                 s--;
1218                                 break;
1219                         }
1220                         switch (*s) {
1221                         default:
1222                             invalid:
1223                                 rprintf(FERROR,
1224                                         "invalid modifier '%c' at position %d in filter rule: %s\n",
1225                                         *s, (int)(s - (const uchar *)*rulestr_ptr), *rulestr_ptr);
1226                                 exit_cleanup(RERR_SYNTAX);
1227                         case '-':
1228                                 if (!BITS_SETnUNSET(rule->rflags, FILTRULE_MERGE_FILE, FILTRULE_NO_PREFIXES))
1229                                         goto invalid;
1230                                 rule->rflags |= FILTRULE_NO_PREFIXES;
1231                                 break;
1232                         case '+':
1233                                 if (!BITS_SETnUNSET(rule->rflags, FILTRULE_MERGE_FILE, FILTRULE_NO_PREFIXES))
1234                                         goto invalid;
1235                                 rule->rflags |= FILTRULE_NO_PREFIXES
1236                                               | FILTRULE_INCLUDE;
1237                                 break;
1238                         case '/':
1239                                 rule->rflags |= FILTRULE_ABS_PATH;
1240                                 break;
1241                         case '!':
1242                                 /* Negation really goes with the pattern, so it
1243                                  * isn't useful as a merge-file default. */
1244                                 if (rule->rflags & FILTRULE_MERGE_FILE)
1245                                         goto invalid;
1246                                 rule->rflags |= FILTRULE_NEGATE;
1247                                 break;
1248                         case 'C':
1249                                 if (rule->rflags & FILTRULE_NO_PREFIXES || prefix_specifies_side)
1250                                         goto invalid;
1251                                 rule->rflags |= FILTRULE_NO_PREFIXES
1252                                               | FILTRULE_WORD_SPLIT
1253                                               | FILTRULE_NO_INHERIT
1254                                               | FILTRULE_CVS_IGNORE;
1255                                 break;
1256                         case 'e':
1257                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1258                                         goto invalid;
1259                                 rule->rflags |= FILTRULE_EXCLUDE_SELF;
1260                                 break;
1261                         case 'n':
1262                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1263                                         goto invalid;
1264                                 rule->rflags |= FILTRULE_NO_INHERIT;
1265                                 break;
1266                         case 'p':
1267                                 rule->rflags |= FILTRULE_PERISHABLE;
1268                                 break;
1269                         case 'r':
1270                                 if (prefix_specifies_side)
1271                                         goto invalid;
1272                                 rule->rflags |= FILTRULE_RECEIVER_SIDE;
1273                                 break;
1274                         case 's':
1275                                 if (prefix_specifies_side)
1276                                         goto invalid;
1277                                 rule->rflags |= FILTRULE_SENDER_SIDE;
1278                                 break;
1279                         case 'w':
1280                                 if (!(rule->rflags & FILTRULE_MERGE_FILE))
1281                                         goto invalid;
1282                                 rule->rflags |= FILTRULE_WORD_SPLIT;
1283                                 break;
1284                         case 'x':
1285                                 rule->rflags |= FILTRULE_XATTR;
1286                                 saw_xattr_filter = 1;
1287                                 break;
1288                         }
1289                 }
1290                 if (*s)
1291                         s++;
1292         }
1293         if (template->rflags & FILTRULES_SIDES) {
1294                 if (rule->rflags & FILTRULES_SIDES) {
1295                         /* The filter and template both specify side(s).  This
1296                          * is dodgy (and won't work correctly if the template is
1297                          * a one-sided per-dir merge rule), so reject it. */
1298                         rprintf(FERROR,
1299                                 "specified-side merge file contains specified-side filter: %s\n",
1300                                 *rulestr_ptr);
1301                         exit_cleanup(RERR_SYNTAX);
1302                 }
1303                 rule->rflags |= template->rflags & FILTRULES_SIDES;
1304         }
1305
1306         if (template->rflags & FILTRULE_WORD_SPLIT) {
1307                 const uchar *cp = s;
1308                 /* Token ends at whitespace or the end of the string. */
1309                 while (!isspace(*cp) && *cp != '\0')
1310                         cp++;
1311                 len = cp - s;
1312         } else
1313                 len = strlen((char*)s);
1314
1315         if (rule->rflags & FILTRULE_CLEAR_LIST) {
1316                 if (!(rule->rflags & FILTRULE_NO_PREFIXES)
1317                  && !(xflags & XFLG_OLD_PREFIXES) && len) {
1318                         rprintf(FERROR,
1319                                 "'!' rule has trailing characters: %s\n", *rulestr_ptr);
1320                         exit_cleanup(RERR_SYNTAX);
1321                 }
1322                 if (len > 1)
1323                         rule->rflags &= ~FILTRULE_CLEAR_LIST;
1324         } else if (!len && !(rule->rflags & FILTRULE_CVS_IGNORE)) {
1325                 rprintf(FERROR, "unexpected end of filter rule: %s\n", *rulestr_ptr);
1326                 exit_cleanup(RERR_SYNTAX);
1327         }
1328
1329         /* --delete-excluded turns an un-modified include/exclude into a sender-side rule.  */
1330         if (delete_excluded
1331          && !(rule->rflags & (FILTRULES_SIDES|FILTRULE_MERGE_FILE|FILTRULE_PERDIR_MERGE)))
1332                 rule->rflags |= FILTRULE_SENDER_SIDE;
1333
1334         *pat_ptr = (const char *)s;
1335         *pat_len_ptr = len;
1336         *rulestr_ptr = *pat_ptr + len;
1337         return rule;
1338 }
1339
1340 static void get_cvs_excludes(uint32 rflags)
1341 {
1342         static int initialized = 0;
1343         char *p, fname[MAXPATHLEN];
1344
1345         if (initialized)
1346                 return;
1347         initialized = 1;
1348
1349         parse_filter_str(&cvs_filter_list, default_cvsignore(),
1350                          rule_template(rflags | (protocol_version >= 30 ? FILTRULE_PERISHABLE : 0)),
1351                          0);
1352
1353         p = module_id >= 0 && lp_use_chroot(module_id) ? "/" : getenv("HOME");
1354         if (p && pathjoin(fname, MAXPATHLEN, p, ".cvsignore") < MAXPATHLEN)
1355                 parse_filter_file(&cvs_filter_list, fname, rule_template(rflags), 0);
1356
1357         parse_filter_str(&cvs_filter_list, getenv("CVSIGNORE"), rule_template(rflags), 0);
1358 }
1359
1360 const filter_rule *rule_template(uint32 rflags)
1361 {
1362         static filter_rule template; /* zero-initialized */
1363         template.rflags = rflags;
1364         return &template;
1365 }
1366
1367 void parse_filter_str(filter_rule_list *listp, const char *rulestr,
1368                      const filter_rule *template, int xflags)
1369 {
1370         filter_rule *rule;
1371         const char *pat;
1372         unsigned int pat_len;
1373
1374         if (!rulestr)
1375                 return;
1376
1377         while (1) {
1378                 uint32 new_rflags;
1379
1380                 /* Remember that the returned string is NOT '\0' terminated! */
1381                 if (!(rule = parse_rule_tok(&rulestr, template, xflags, &pat, &pat_len)))
1382                         break;
1383
1384                 if (pat_len >= MAXPATHLEN) {
1385                         rprintf(FERROR, "discarding over-long filter: %.*s\n",
1386                                 (int)pat_len, pat);
1387                     free_continue:
1388                         free_filter(rule);
1389                         continue;
1390                 }
1391
1392                 new_rflags = rule->rflags;
1393                 if (new_rflags & FILTRULE_CLEAR_LIST) {
1394                         if (DEBUG_GTE(FILTER, 2)) {
1395                                 rprintf(FINFO,
1396                                         "[%s] clearing filter list%s\n",
1397                                         who_am_i(), listp->debug_type);
1398                         }
1399                         pop_filter_list(listp);
1400                         listp->head = NULL;
1401                         goto free_continue;
1402                 }
1403
1404                 if (new_rflags & FILTRULE_MERGE_FILE) {
1405                         if (!pat_len) {
1406                                 pat = ".cvsignore";
1407                                 pat_len = 10;
1408                         }
1409                         if (new_rflags & FILTRULE_EXCLUDE_SELF) {
1410                                 const char *name;
1411                                 filter_rule *excl_self;
1412
1413                                 excl_self = new0(filter_rule);
1414                                 /* Find the beginning of the basename and add an exclude for it. */
1415                                 for (name = pat + pat_len; name > pat && name[-1] != '/'; name--) {}
1416                                 add_rule(listp, name, (pat + pat_len) - name, excl_self, 0);
1417                                 rule->rflags &= ~FILTRULE_EXCLUDE_SELF;
1418                         }
1419                         if (new_rflags & FILTRULE_PERDIR_MERGE) {
1420                                 if (parent_dirscan) {
1421                                         const char *p;
1422                                         unsigned int len = pat_len;
1423                                         if ((p = parse_merge_name(pat, &len, module_dirlen)))
1424                                                 add_rule(listp, p, len, rule, 0);
1425                                         else
1426                                                 free_filter(rule);
1427                                         continue;
1428                                 }
1429                         } else {
1430                                 const char *p;
1431                                 unsigned int len = pat_len;
1432                                 if ((p = parse_merge_name(pat, &len, 0)))
1433                                         parse_filter_file(listp, p, rule, XFLG_FATAL_ERRORS);
1434                                 free_filter(rule);
1435                                 continue;
1436                         }
1437                 }
1438
1439                 add_rule(listp, pat, pat_len, rule, xflags);
1440
1441                 if (new_rflags & FILTRULE_CVS_IGNORE
1442                     && !(new_rflags & FILTRULE_MERGE_FILE))
1443                         get_cvs_excludes(new_rflags);
1444         }
1445 }
1446
1447 void parse_filter_file(filter_rule_list *listp, const char *fname, const filter_rule *template, int xflags)
1448 {
1449         FILE *fp;
1450         char line[BIGPATHBUFLEN];
1451         char *eob = line + sizeof line - 1;
1452         BOOL word_split = (template->rflags & FILTRULE_WORD_SPLIT) != 0;
1453
1454         if (!fname || !*fname)
1455                 return;
1456
1457         if (*fname != '-' || fname[1] || am_server) {
1458                 if (daemon_filter_list.head) {
1459                         strlcpy(line, fname, sizeof line);
1460                         clean_fname(line, CFN_COLLAPSE_DOT_DOT_DIRS);
1461                         if (check_filter(&daemon_filter_list, FLOG, line, 0) < 0)
1462                                 fp = NULL;
1463                         else
1464                                 fp = fopen(line, "rb");
1465                 } else
1466                         fp = fopen(fname, "rb");
1467         } else
1468                 fp = stdin;
1469
1470         if (DEBUG_GTE(FILTER, 2)) {
1471                 rprintf(FINFO, "[%s] parse_filter_file(%s,%x,%x)%s\n",
1472                         who_am_i(), fname, template->rflags, xflags,
1473                         fp ? "" : " [not found]");
1474         }
1475
1476         if (!fp) {
1477                 if (xflags & XFLG_FATAL_ERRORS) {
1478                         rsyserr(FERROR, errno,
1479                                 "failed to open %sclude file %s",
1480                                 template->rflags & FILTRULE_INCLUDE ? "in" : "ex",
1481                                 fname);
1482                         exit_cleanup(RERR_FILEIO);
1483                 }
1484                 return;
1485         }
1486         dirbuf[dirbuf_len] = '\0';
1487
1488         while (1) {
1489                 char *s = line;
1490                 int ch, overflow = 0;
1491                 while (1) {
1492                         if ((ch = getc(fp)) == EOF) {
1493                                 if (ferror(fp) && errno == EINTR) {
1494                                         clearerr(fp);
1495                                         continue;
1496                                 }
1497                                 break;
1498                         }
1499                         if (word_split && isspace(ch))
1500                                 break;
1501                         if (eol_nulls? !ch : (ch == '\n' || ch == '\r'))
1502                                 break;
1503                         if (s < eob)
1504                                 *s++ = ch;
1505                         else
1506                                 overflow = 1;
1507                 }
1508                 if (overflow) {
1509                         rprintf(FERROR, "discarding over-long filter: %s...\n", line);
1510                         s = line;
1511                 }
1512                 *s = '\0';
1513                 /* Skip an empty token and (when line parsing) comments. */
1514                 if (*line && (word_split || (*line != ';' && *line != '#')))
1515                         parse_filter_str(listp, line, template, xflags);
1516                 if (ch == EOF)
1517                         break;
1518         }
1519         fclose(fp);
1520 }
1521
1522 /* If the "for_xfer" flag is set, the prefix is made compatible with the
1523  * current protocol_version (if possible) or a NULL is returned (if not
1524  * possible). */
1525 char *get_rule_prefix(filter_rule *rule, const char *pat, int for_xfer,
1526                       unsigned int *plen_ptr)
1527 {
1528         static char buf[MAX_RULE_PREFIX+1];
1529         char *op = buf;
1530         int legal_len = for_xfer && protocol_version < 29 ? 1 : MAX_RULE_PREFIX-1;
1531
1532         if (rule->rflags & FILTRULE_PERDIR_MERGE) {
1533                 if (legal_len == 1)
1534                         return NULL;
1535                 *op++ = ':';
1536         } else if (rule->rflags & FILTRULE_INCLUDE)
1537                 *op++ = '+';
1538         else if (legal_len != 1
1539             || ((*pat == '-' || *pat == '+') && pat[1] == ' '))
1540                 *op++ = '-';
1541         else
1542                 legal_len = 0;
1543
1544         if (rule->rflags & FILTRULE_ABS_PATH)
1545                 *op++ = '/';
1546         if (rule->rflags & FILTRULE_NEGATE)
1547                 *op++ = '!';
1548         if (rule->rflags & FILTRULE_CVS_IGNORE)
1549                 *op++ = 'C';
1550         else {
1551                 if (rule->rflags & FILTRULE_NO_INHERIT)
1552                         *op++ = 'n';
1553                 if (rule->rflags & FILTRULE_WORD_SPLIT)
1554                         *op++ = 'w';
1555                 if (rule->rflags & FILTRULE_NO_PREFIXES) {
1556                         if (rule->rflags & FILTRULE_INCLUDE)
1557                                 *op++ = '+';
1558                         else
1559                                 *op++ = '-';
1560                 }
1561         }
1562         if (rule->rflags & FILTRULE_EXCLUDE_SELF)
1563                 *op++ = 'e';
1564         if (rule->rflags & FILTRULE_XATTR)
1565                 *op++ = 'x';
1566         if (rule->rflags & FILTRULE_SENDER_SIDE
1567             && (!for_xfer || protocol_version >= 29))
1568                 *op++ = 's';
1569         if (rule->rflags & FILTRULE_RECEIVER_SIDE
1570             && (!for_xfer || protocol_version >= 29
1571              || (delete_excluded && am_sender)))
1572                 *op++ = 'r';
1573         if (rule->rflags & FILTRULE_PERISHABLE) {
1574                 if (!for_xfer || protocol_version >= 30)
1575                         *op++ = 'p';
1576                 else if (am_sender)
1577                         return NULL;
1578         }
1579         if (op - buf > legal_len)
1580                 return NULL;
1581         if (legal_len)
1582                 *op++ = ' ';
1583         *op = '\0';
1584         if (plen_ptr)
1585                 *plen_ptr = op - buf;
1586         return buf;
1587 }
1588
1589 static void send_rules(int f_out, filter_rule_list *flp)
1590 {
1591         filter_rule *ent;
1592
1593         for (ent = flp->head; ent; ent = ent->next) {
1594                 unsigned int len, plen, dlen;
1595                 int elide = 0;
1596                 char *p;
1597
1598                 /* Note we need to check delete_excluded here in addition to
1599                  * the code in parse_rule_tok() because some rules may have
1600                  * been added before we found the --delete-excluded option.
1601                  * We must also elide any CVS merge-file rules to avoid a
1602                  * backward compatibility problem, and we elide any no-prefix
1603                  * merge files as an optimization (since they can only have
1604                  * include/exclude rules). */
1605                 if (ent->rflags & FILTRULE_SENDER_SIDE)
1606                         elide = am_sender ? LOCAL_RULE : REMOTE_RULE;
1607                 if (ent->rflags & FILTRULE_RECEIVER_SIDE)
1608                         elide = elide ? 0 : am_sender ? REMOTE_RULE : LOCAL_RULE;
1609                 else if (delete_excluded && !elide
1610                  && (!(ent->rflags & FILTRULE_PERDIR_MERGE)
1611                   || ent->rflags & FILTRULE_NO_PREFIXES))
1612                         elide = am_sender ? LOCAL_RULE : REMOTE_RULE;
1613                 ent->elide = elide;
1614                 if (elide == LOCAL_RULE)
1615                         continue;
1616                 if (ent->rflags & FILTRULE_CVS_IGNORE
1617                     && !(ent->rflags & FILTRULE_MERGE_FILE)) {
1618                         int f = am_sender || protocol_version < 29 ? f_out : -2;
1619                         send_rules(f, &cvs_filter_list);
1620                         if (f == f_out)
1621                                 continue;
1622                 }
1623                 p = get_rule_prefix(ent, ent->pattern, 1, &plen);
1624                 if (!p) {
1625                         rprintf(FERROR,
1626                                 "filter rules are too modern for remote rsync.\n");
1627                         exit_cleanup(RERR_PROTOCOL);
1628                 }
1629                 if (f_out < 0)
1630                         continue;
1631                 len = strlen(ent->pattern);
1632                 dlen = ent->rflags & FILTRULE_DIRECTORY ? 1 : 0;
1633                 if (!(plen + len + dlen))
1634                         continue;
1635                 write_int(f_out, plen + len + dlen);
1636                 if (plen)
1637                         write_buf(f_out, p, plen);
1638                 write_buf(f_out, ent->pattern, len);
1639                 if (dlen)
1640                         write_byte(f_out, '/');
1641         }
1642 }
1643
1644 /* This is only called by the client. */
1645 void send_filter_list(int f_out)
1646 {
1647         int receiver_wants_list = prune_empty_dirs
1648             || (delete_mode && (!delete_excluded || protocol_version >= 29));
1649
1650         if (local_server || (am_sender && !receiver_wants_list))
1651                 f_out = -1;
1652         if (cvs_exclude && am_sender) {
1653                 if (protocol_version >= 29)
1654                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1655                 parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1656         }
1657
1658         send_rules(f_out, &filter_list);
1659
1660         if (f_out >= 0)
1661                 write_int(f_out, 0);
1662
1663         if (cvs_exclude) {
1664                 if (!am_sender || protocol_version < 29)
1665                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1666                 if (!am_sender)
1667                         parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1668         }
1669 }
1670
1671 /* This is only called by the server. */
1672 void recv_filter_list(int f_in)
1673 {
1674         char line[BIGPATHBUFLEN];
1675         int xflags = protocol_version >= 29 ? 0 : XFLG_OLD_PREFIXES;
1676         int receiver_wants_list = prune_empty_dirs
1677             || (delete_mode && (!delete_excluded || protocol_version >= 29));
1678         unsigned int len;
1679
1680         if (!local_server && (am_sender || receiver_wants_list)) {
1681                 while ((len = read_int(f_in)) != 0) {
1682                         if (len >= sizeof line)
1683                                 overflow_exit("recv_rules");
1684                         read_sbuf(f_in, line, len);
1685                         parse_filter_str(&filter_list, line, rule_template(0), xflags);
1686                 }
1687         }
1688
1689         if (cvs_exclude) {
1690                 if (local_server || am_sender || protocol_version < 29)
1691                         parse_filter_str(&filter_list, ":C", rule_template(0), 0);
1692                 if (local_server || am_sender)
1693                         parse_filter_str(&filter_list, "-C", rule_template(0), 0);
1694         }
1695
1696         if (local_server) /* filter out any rules that aren't for us. */
1697                 send_rules(-1, &filter_list);
1698 }