Refactorings to the filter code, most notably:
[rsync/rsync.git] / exclude.c
CommitLineData
0f78b815
WD
1/*
2 * The filter include/exclude routines.
0f2ac855 3 *
0f78b815
WD
4 * Copyright (C) 1996-2001 Andrew Tridgell <tridge@samba.org>
5 * Copyright (C) 1996 Paul Mackerras
6 * Copyright (C) 2002 Martin Pool
b3bf9b9d 7 * Copyright (C) 2003-2009 Wayne Davison
0f2ac855 8 *
07a874fd 9 * This program is free software; you can redistribute it and/or modify
8e41b68e
WD
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.
0f2ac855 13 *
07a874fd
MP
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.
0f2ac855 18 *
e7c67065 19 * You should have received a copy of the GNU General Public License along
4fd842f9 20 * with this program; if not, visit the http://fsf.org website.
07a874fd 21 */
c627d613 22
c627d613
AT
23#include "rsync.h"
24
5a016db9 25extern int am_server;
fdc79501 26extern int am_sender;
40d38dc0 27extern int eol_nulls;
6dfd07d0 28extern int io_error;
1412da7c 29extern int local_server;
d5782b52 30extern int prune_empty_dirs;
ccdb23bb 31extern int ignore_perishable;
fdc79501
WD
32extern int delete_mode;
33extern int delete_excluded;
34extern int cvs_exclude;
6dfd07d0
WD
35extern int sanitize_paths;
36extern int protocol_version;
bf39270e 37extern int module_id;
40d38dc0 38
b791d680 39extern char curr_dir[MAXPATHLEN];
6dfd07d0
WD
40extern unsigned int curr_dir_len;
41extern unsigned int module_dirlen;
c627d613 42
7634fc8e
WD
43struct filter_list_struct filter_list = { .debug_type = "" };
44struct filter_list_struct cvs_filter_list = { .debug_type = " [global CVS]" };
45struct filter_list_struct daemon_filter_list = { .debug_type = " [daemon]" };
c627d613 46
6dfd07d0 47/* Need room enough for ":MODS " prefix plus some room to grow. */
7842418b 48#define MAX_RULE_PREFIX (16)
6dfd07d0 49
d727f0ff 50#define MODIFIERS_MERGE_FILE "-+Cenw"
ccdb23bb
WD
51#define MODIFIERS_INCL_EXCL "/!Crsp"
52#define MODIFIERS_HIDE_PROTECT "/!p"
b6f06b8e 53
f5aeb6ff
WD
54#define SLASH_WILD3_SUFFIX "/***"
55
7842418b 56/* The dirbuf is set by push_local_filters() to the current subdirectory
6dfd07d0
WD
57 * relative to curr_dir that is being processed. The path always has a
58 * trailing slash appended, and the variable dirbuf_len contains the length
59 * of this path prefix. The path is always absolute. */
60static char dirbuf[MAXPATHLEN+1];
61static unsigned int dirbuf_len = 0;
62static int dirbuf_depth;
63
64/* This is True when we're scanning parent dirs for per-dir merge-files. */
65static BOOL parent_dirscan = False;
66
67/* This array contains a list of all the currently active per-dir merge
68 * files. This makes it easier to save the appropriate values when we
69 * "push" down into each subdirectory. */
7842418b 70static struct filter_struct **mergelist_parents;
6dfd07d0
WD
71static int mergelist_cnt = 0;
72static int mergelist_size = 0;
73
7842418b 74/* Each filter_list_struct describes a singly-linked list by keeping track
6dfd07d0
WD
75 * of both the head and tail pointers. The list is slightly unusual in that
76 * a parent-dir's content can be appended to the end of the local list in a
77 * special way: the last item in the local list has its "next" pointer set
78 * to point to the inherited list, but the local list's tail pointer points
79 * at the end of the local list. Thus, if the local list is empty, the head
80 * will be pointing at the inherited content but the tail will be NULL. To
81 * help you visualize this, here are the possible list arrangements:
82 *
83 * Completely Empty Local Content Only
84 * ================================== ====================================
85 * head -> NULL head -> Local1 -> Local2 -> NULL
86 * tail -> NULL tail -------------^
87 *
88 * Inherited Content Only Both Local and Inherited Content
89 * ================================== ====================================
90 * head -> Parent1 -> Parent2 -> NULL head -> L1 -> L2 -> P1 -> P2 -> NULL
91 * tail -> NULL tail ---------^
92 *
93 * This means that anyone wanting to traverse the whole list to use it just
94 * needs to start at the head and use the "next" pointers until it goes
95 * NULL. To add new local content, we insert the item after the tail item
96 * and update the tail (obviously, if "tail" was NULL, we insert it at the
97 * head). To clear the local list, WE MUST NOT FREE THE INHERITED CONTENT
98 * because it is shared between the current list and our parent list(s).
99 * The easiest way to handle this is to simply truncate the list after the
100 * tail item and then free the local list from the head. When inheriting
7842418b 101 * the list for a new local dir, we just save off the filter_list_struct
6dfd07d0
WD
102 * values (so we can pop back to them later) and set the tail to NULL.
103 */
104
daa8d920
MM
105static void teardown_mergelist(struct filter_struct *ex)
106{
7634fc8e 107 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
108 rprintf(FINFO, "[%s] deactivating mergelist #%d%s\n",
109 who_am_i(), mergelist_cnt - 1,
110 ex->u.mergelist->debug_type);
7634fc8e 111 }
daa8d920
MM
112
113 /* We should deactivate mergelists in LIFO order. */
114 assert(mergelist_cnt > 0);
115 assert(ex == mergelist_parents[mergelist_cnt - 1]);
116
117 /* The parent_dirscan filters should have been freed. */
118 assert(ex->u.mergelist->parent_dirscan_head == NULL);
119
120 free(ex->u.mergelist->debug_type);
121 free(ex->u.mergelist);
122 mergelist_cnt--;
123}
124
7842418b 125static void free_filter(struct filter_struct *ex)
6dfd07d0 126{
6dfd07d0
WD
127 free(ex->pattern);
128 free(ex);
129}
130
daa8d920
MM
131static void free_filters(struct filter_struct *head)
132{
133 struct filter_struct *rev_head = NULL;
134
135 /* Reverse the list so we deactivate mergelists in the proper LIFO
136 * order. */
137 while (head) {
138 struct filter_struct *next = head->next;
139 head->next = rev_head;
140 rev_head = head;
141 head = next;
142 }
143
144 while (rev_head) {
145 struct filter_struct *prev = rev_head->next;
ebaa4296
MM
146 /* Tear down mergelists here, not in free_filter, so that we
147 * affect only real filter lists and not temporarily allocated
148 * filters. */
149 if (rev_head->flags & MATCHFLG_PERDIR_MERGE)
150 teardown_mergelist(rev_head);
daa8d920
MM
151 free_filter(rev_head);
152 rev_head = prev;
153 }
154}
155
7842418b 156/* Build a filter structure given a filter pattern. The value in "pat"
ebaa4296
MM
157 * is not null-terminated. "filter" is either held or freed, so the
158 * caller should not free it. */
159static void add_rule(struct filter_list_struct *listp,
160 const char *pat, unsigned int pat_len,
161 struct filter_struct *filter, int xflags)
c627d613 162{
f8f72644 163 const char *cp;
f5aeb6ff 164 unsigned int pre_len, suf_len, slash_cnt = 0;
c627d613 165
951e826b 166 if (DEBUG_GTE(FILTER, 2)) {
fdc79501 167 rprintf(FINFO, "[%s] add_rule(%s%.*s%s)%s\n",
ebaa4296 168 who_am_i(), get_rule_prefix(filter, pat, 0, NULL),
fdc79501 169 (int)pat_len, pat,
ebaa4296 170 (filter->flags & MATCHFLG_DIRECTORY) ? "/" : "",
6dfd07d0
WD
171 listp->debug_type);
172 }
173
4fc8140a 174 /* These flags also indicate that we're reading a list that
6d7b3d52 175 * needs to be filtered now, not post-filtered later. */
ebaa4296
MM
176 if (xflags & (XFLG_ANCHORED2ABS|XFLG_ABS_IF_SLASH)
177 && (filter->flags & MATCHFLGS_SIDES)
178 == (am_sender ? MATCHFLG_RECEIVER_SIDE : MATCHFLG_SENDER_SIDE)) {
179 /* This filter applies only to the other side. Drop it. */
180 free_filter(filter);
181 return;
6d7b3d52
WD
182 }
183
4c74d44d
WD
184 if (pat_len > 1 && pat[pat_len-1] == '/') {
185 pat_len--;
ebaa4296 186 filter->flags |= MATCHFLG_DIRECTORY;
4c74d44d
WD
187 }
188
189 for (cp = pat; cp < pat + pat_len; cp++) {
190 if (*cp == '/')
191 slash_cnt++;
192 }
193
ebaa4296 194 if (!(filter->flags & (MATCHFLG_ABS_PATH | MATCHFLG_MERGE_FILE))
4fc8140a 195 && ((xflags & (XFLG_ANCHORED2ABS|XFLG_ABS_IF_SLASH) && *pat == '/')
4c74d44d 196 || (xflags & XFLG_ABS_IF_SLASH && slash_cnt))) {
ebaa4296 197 filter->flags |= MATCHFLG_ABS_PATH;
4fc8140a 198 if (*pat == '/')
4a86fbcd 199 pre_len = dirbuf_len - module_dirlen - 1;
4fc8140a 200 else
4a86fbcd 201 pre_len = 0;
6dfd07d0 202 } else
4a86fbcd 203 pre_len = 0;
4c74d44d 204
f5aeb6ff
WD
205 /* The daemon wants dir-exclude rules to get an appended "/" + "***". */
206 if (xflags & XFLG_DIR2WILD3
ebaa4296
MM
207 && BITS_SETnUNSET(filter->flags, MATCHFLG_DIRECTORY, MATCHFLG_INCLUDE)) {
208 filter->flags &= ~MATCHFLG_DIRECTORY;
f5aeb6ff
WD
209 suf_len = sizeof SLASH_WILD3_SUFFIX - 1;
210 } else
211 suf_len = 0;
212
ebaa4296 213 if (!(filter->pattern = new_array(char, pre_len + pat_len + suf_len + 1)))
fdc79501 214 out_of_memory("add_rule");
4c74d44d 215 if (pre_len) {
ebaa4296
MM
216 memcpy(filter->pattern, dirbuf + module_dirlen, pre_len);
217 for (cp = filter->pattern; cp < filter->pattern + pre_len; cp++) {
4c74d44d
WD
218 if (*cp == '/')
219 slash_cnt++;
220 }
221 }
ebaa4296 222 strlcpy(filter->pattern + pre_len, pat, pat_len + 1);
4a86fbcd 223 pat_len += pre_len;
f5aeb6ff 224 if (suf_len) {
ebaa4296 225 memcpy(filter->pattern + pat_len, SLASH_WILD3_SUFFIX, suf_len+1);
f5aeb6ff
WD
226 pat_len += suf_len;
227 slash_cnt++;
228 }
f8f72644 229
ebaa4296
MM
230 if (strpbrk(filter->pattern, "*[?")) {
231 filter->flags |= MATCHFLG_WILD;
232 if ((cp = strstr(filter->pattern, "**")) != NULL) {
233 filter->flags |= MATCHFLG_WILD2;
170381c0 234 /* If the pattern starts with **, note that. */
ebaa4296
MM
235 if (cp == filter->pattern)
236 filter->flags |= MATCHFLG_WILD2_PREFIX;
e5daa273
WD
237 /* If the pattern ends with ***, note that. */
238 if (pat_len >= 3
ebaa4296
MM
239 && filter->pattern[pat_len-3] == '*'
240 && filter->pattern[pat_len-2] == '*'
241 && filter->pattern[pat_len-1] == '*')
242 filter->flags |= MATCHFLG_WILD3_SUFFIX;
0f2ac855 243 }
2bca43f6 244 }
c627d613 245
ebaa4296 246 if (filter->flags & MATCHFLG_PERDIR_MERGE) {
7842418b 247 struct filter_list_struct *lp;
6dfd07d0
WD
248 unsigned int len;
249 int i;
250
ebaa4296 251 if ((cp = strrchr(filter->pattern, '/')) != NULL)
6dfd07d0
WD
252 cp++;
253 else
ebaa4296 254 cp = filter->pattern;
6dfd07d0
WD
255
256 /* If the local merge file was already mentioned, don't
257 * add it again. */
258 for (i = 0; i < mergelist_cnt; i++) {
7842418b 259 struct filter_struct *ex = mergelist_parents[i];
6dfd07d0
WD
260 const char *s = strrchr(ex->pattern, '/');
261 if (s)
bf39270e 262 s++;
6dfd07d0 263 else
bf39270e 264 s = ex->pattern;
6dfd07d0 265 len = strlen(s);
ebaa4296 266 if (len == pat_len - (cp - filter->pattern)
6dfd07d0 267 && memcmp(s, cp, len) == 0) {
ebaa4296 268 free_filter(filter);
6dfd07d0
WD
269 return;
270 }
271 }
272
7842418b 273 if (!(lp = new_array(struct filter_list_struct, 1)))
fdc79501 274 out_of_memory("add_rule");
daa8d920 275 lp->head = lp->tail = lp->parent_dirscan_head = NULL;
fdc79501
WD
276 if (asprintf(&lp->debug_type, " [per-dir %s]", cp) < 0)
277 out_of_memory("add_rule");
ebaa4296 278 filter->u.mergelist = lp;
6dfd07d0
WD
279
280 if (mergelist_cnt == mergelist_size) {
281 mergelist_size += 5;
282 mergelist_parents = realloc_array(mergelist_parents,
7842418b 283 struct filter_struct *,
6dfd07d0
WD
284 mergelist_size);
285 if (!mergelist_parents)
fdc79501 286 out_of_memory("add_rule");
6dfd07d0 287 }
7634fc8e 288 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
289 rprintf(FINFO, "[%s] activating mergelist #%d%s\n",
290 who_am_i(), mergelist_cnt, lp->debug_type);
7634fc8e 291 }
ebaa4296 292 mergelist_parents[mergelist_cnt++] = filter;
4c74d44d 293 } else
ebaa4296 294 filter->u.slash_cnt = slash_cnt;
c1b29492 295
6dfd07d0 296 if (!listp->tail) {
ebaa4296
MM
297 filter->next = listp->head;
298 listp->head = listp->tail = filter;
6dfd07d0 299 } else {
ebaa4296
MM
300 filter->next = listp->tail->next;
301 listp->tail->next = filter;
302 listp->tail = filter;
b2aa573b 303 }
2b6b4d53
AT
304}
305
7842418b 306static void clear_filter_list(struct filter_list_struct *listp)
2b6b4d53 307{
6dfd07d0 308 if (listp->tail) {
6dfd07d0
WD
309 /* Truncate any inherited items from the local list. */
310 listp->tail->next = NULL;
311 /* Now free everything that is left. */
daa8d920 312 free_filters(listp->head);
6dfd07d0
WD
313 }
314
315 listp->head = listp->tail = NULL;
2b6b4d53 316}
c627d613 317
6dfd07d0
WD
318/* This returns an expanded (absolute) filename for the merge-file name if
319 * the name has any slashes in it OR if the parent_dirscan var is True;
320 * otherwise it returns the original merge_file name. If the len_ptr value
321 * is non-NULL the merge_file name is limited by the referenced length
322 * value and will be updated with the length of the resulting name. We
323 * always return a name that is null terminated, even if the merge_file
324 * name was not. */
325static char *parse_merge_name(const char *merge_file, unsigned int *len_ptr,
326 unsigned int prefix_skip)
5be7fa93 327{
6dfd07d0
WD
328 static char buf[MAXPATHLEN];
329 char *fn, tmpbuf[MAXPATHLEN];
330 unsigned int fn_len;
331
332 if (!parent_dirscan && *merge_file != '/') {
333 /* Return the name unchanged it doesn't have any slashes. */
334 if (len_ptr) {
335 const char *p = merge_file + *len_ptr;
336 while (--p > merge_file && *p != '/') {}
337 if (p == merge_file) {
338 strlcpy(buf, merge_file, *len_ptr + 1);
339 return buf;
340 }
341 } else if (strchr(merge_file, '/') == NULL)
342 return (char *)merge_file;
343 }
5be7fa93 344
6dfd07d0
WD
345 fn = *merge_file == '/' ? buf : tmpbuf;
346 if (sanitize_paths) {
347 const char *r = prefix_skip ? "/" : NULL;
348 /* null-terminate the name if it isn't already */
349 if (len_ptr && merge_file[*len_ptr]) {
350 char *to = fn == buf ? tmpbuf : buf;
351 strlcpy(to, merge_file, *len_ptr + 1);
352 merge_file = to;
353 }
d48810ba 354 if (!sanitize_path(fn, merge_file, r, dirbuf_depth, SP_DEFAULT)) {
6dfd07d0 355 rprintf(FERROR, "merge-file name overflows: %s\n",
45c49b52 356 merge_file);
6dfd07d0
WD
357 return NULL;
358 }
798cde47 359 fn_len = strlen(fn);
6dfd07d0
WD
360 } else {
361 strlcpy(fn, merge_file, len_ptr ? *len_ptr + 1 : MAXPATHLEN);
b58bfb2f 362 fn_len = clean_fname(fn, CFN_COLLAPSE_DOT_DOT_DIRS);
b2aa573b 363 }
0f78b815 364
798cde47
WD
365 /* If the name isn't in buf yet, it's wasn't absolute. */
366 if (fn != buf) {
9793bbb3
WD
367 int d_len = dirbuf_len - prefix_skip;
368 if (d_len + fn_len >= MAXPATHLEN) {
798cde47
WD
369 rprintf(FERROR, "merge-file name overflows: %s\n", fn);
370 return NULL;
371 }
9793bbb3
WD
372 memcpy(buf, dirbuf + prefix_skip, d_len);
373 memcpy(buf + d_len, fn, fn_len + 1);
b58bfb2f 374 fn_len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
6dfd07d0 375 }
6dfd07d0 376
6dfd07d0
WD
377 if (len_ptr)
378 *len_ptr = fn_len;
379 return buf;
380}
5be7fa93 381
6dfd07d0 382/* Sets the dirbuf and dirbuf_len values. */
7842418b 383void set_filter_dir(const char *dir, unsigned int dirlen)
6dfd07d0
WD
384{
385 unsigned int len;
386 if (*dir != '/') {
387 memcpy(dirbuf, curr_dir, curr_dir_len);
388 dirbuf[curr_dir_len] = '/';
389 len = curr_dir_len + 1;
390 if (len + dirlen >= MAXPATHLEN)
391 dirlen = 0;
392 } else
393 len = 0;
394 memcpy(dirbuf + len, dir, dirlen);
395 dirbuf[dirlen + len] = '\0';
b58bfb2f 396 dirbuf_len = clean_fname(dirbuf, CFN_COLLAPSE_DOT_DOT_DIRS);
6dfd07d0
WD
397 if (dirbuf_len > 1 && dirbuf[dirbuf_len-1] == '.'
398 && dirbuf[dirbuf_len-2] == '/')
399 dirbuf_len -= 2;
400 if (dirbuf_len != 1)
401 dirbuf[dirbuf_len++] = '/';
402 dirbuf[dirbuf_len] = '\0';
403 if (sanitize_paths)
404 dirbuf_depth = count_dir_elements(dirbuf + module_dirlen);
405}
406
407/* This routine takes a per-dir merge-file entry and finishes its setup.
408 * If the name has a path portion then we check to see if it refers to a
409 * parent directory of the first transfer dir. If it does, we scan all the
410 * dirs from that point through the parent dir of the transfer dir looking
411 * for the per-dir merge-file in each one. */
daa8d920 412static BOOL setup_merge_file(int mergelist_num, struct filter_struct *ex,
b6f06b8e 413 struct filter_list_struct *lp)
6dfd07d0
WD
414{
415 char buf[MAXPATHLEN];
416 char *x, *y, *pat = ex->pattern;
417 unsigned int len;
418
419 if (!(x = parse_merge_name(pat, NULL, 0)) || *x != '/')
420 return 0;
421
7634fc8e 422 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
423 rprintf(FINFO, "[%s] performing parent_dirscan for mergelist #%d%s\n",
424 who_am_i(), mergelist_num, lp->debug_type);
7634fc8e 425 }
6dfd07d0
WD
426 y = strrchr(x, '/');
427 *y = '\0';
428 ex->pattern = strdup(y+1);
429 if (!*x)
430 x = "/";
431 if (*x == '/')
432 strlcpy(buf, x, MAXPATHLEN);
433 else
434 pathjoin(buf, MAXPATHLEN, dirbuf, x);
435
b58bfb2f 436 len = clean_fname(buf, CFN_COLLAPSE_DOT_DOT_DIRS);
6dfd07d0
WD
437 if (len != 1 && len < MAXPATHLEN-1) {
438 buf[len++] = '/';
439 buf[len] = '\0';
440 }
441 /* This ensures that the specified dir is a parent of the transfer. */
442 for (x = buf, y = dirbuf; *x && *x == *y; x++, y++) {}
443 if (*x)
444 y += strlen(y); /* nope -- skip the scan */
445
446 parent_dirscan = True;
447 while (*y) {
448 char save[MAXPATHLEN];
449 strlcpy(save, y, MAXPATHLEN);
450 *y = '\0';
451 dirbuf_len = y - dirbuf;
452 strlcpy(x, ex->pattern, MAXPATHLEN - (x - buf));
ebaa4296
MM
453 load_filter_file(lp, buf, ex, XFLG_ANCHORED2ABS);
454 if (ex->flags & MATCHFLG_NO_INHERIT) {
daa8d920
MM
455 /* Free the undesired rules to clean up any per-dir
456 * mergelists they defined. Otherwise pop_local_filters
457 * may crash trying to restore nonexistent state for
458 * those mergelists. */
459 free_filters(lp->head);
6dfd07d0 460 lp->head = NULL;
daa8d920 461 }
6dfd07d0
WD
462 lp->tail = NULL;
463 strlcpy(y, save, MAXPATHLEN);
464 while ((*x++ = *y++) != '/') {}
465 }
daa8d920
MM
466 /* Save current head for freeing when the mergelist becomes inactive. */
467 lp->parent_dirscan_head = lp->head;
6dfd07d0 468 parent_dirscan = False;
7634fc8e 469 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
470 rprintf(FINFO, "[%s] completed parent_dirscan for mergelist #%d%s\n",
471 who_am_i(), mergelist_num, lp->debug_type);
7634fc8e 472 }
6dfd07d0
WD
473 free(pat);
474 return 1;
475}
476
daa8d920
MM
477struct local_filter_state {
478 int mergelist_cnt;
2a147e9f 479 struct filter_list_struct mergelists[1];
daa8d920
MM
480};
481
6dfd07d0
WD
482/* Each time rsync changes to a new directory it call this function to
483 * handle all the per-dir merge-files. The "dir" value is the current path
484 * relative to curr_dir (which might not be null-terminated). We copy it
485 * into dirbuf so that we can easily append a file name on the end. */
7842418b 486void *push_local_filters(const char *dir, unsigned int dirlen)
6dfd07d0 487{
daa8d920 488 struct local_filter_state *push;
6dfd07d0
WD
489 int i;
490
7842418b 491 set_filter_dir(dir, dirlen);
7634fc8e 492 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
493 rprintf(FINFO, "[%s] pushing local filters for %s\n",
494 who_am_i(), dirbuf);
7634fc8e 495 }
6dfd07d0 496
7634fc8e 497 if (!mergelist_cnt) {
daa8d920 498 /* No old state to save and no new merge files to push. */
a2b371cd 499 return NULL;
7634fc8e 500 }
a2b371cd 501
7634fc8e
WD
502 push = (struct local_filter_state *)new_array(char,
503 sizeof (struct local_filter_state)
2a147e9f 504 + (mergelist_cnt-1) * sizeof (struct filter_list_struct));
6dfd07d0 505 if (!push)
7842418b 506 out_of_memory("push_local_filters");
6dfd07d0 507
daa8d920
MM
508 push->mergelist_cnt = mergelist_cnt;
509 for (i = 0; i < mergelist_cnt; i++) {
510 memcpy(&push->mergelists[i], mergelist_parents[i]->u.mergelist,
7842418b 511 sizeof (struct filter_list_struct));
6dfd07d0
WD
512 }
513
ebaa4296 514 /* Note: load_filter_file() might increase mergelist_cnt, so keep
6dfd07d0
WD
515 * this loop separate from the above loop. */
516 for (i = 0; i < mergelist_cnt; i++) {
7842418b
WD
517 struct filter_struct *ex = mergelist_parents[i];
518 struct filter_list_struct *lp = ex->u.mergelist;
6dfd07d0 519
951e826b 520 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
521 rprintf(FINFO, "[%s] pushing mergelist #%d%s\n",
522 who_am_i(), i, lp->debug_type);
6dfd07d0
WD
523 }
524
525 lp->tail = NULL; /* Switch any local rules to inherited. */
ebaa4296 526 if (ex->flags & MATCHFLG_NO_INHERIT)
6dfd07d0 527 lp->head = NULL;
6dfd07d0 528
ebaa4296
MM
529 if (ex->flags & MATCHFLG_FINISH_SETUP) {
530 ex->flags &= ~MATCHFLG_FINISH_SETUP;
daa8d920 531 if (setup_merge_file(i, ex, lp))
7842418b 532 set_filter_dir(dir, dirlen);
6dfd07d0
WD
533 }
534
535 if (strlcpy(dirbuf + dirbuf_len, ex->pattern,
b6f06b8e 536 MAXPATHLEN - dirbuf_len) < MAXPATHLEN - dirbuf_len) {
ebaa4296 537 load_filter_file(lp, dirbuf, ex,
fdc79501 538 XFLG_ANCHORED2ABS);
b6f06b8e 539 } else {
6dfd07d0 540 io_error |= IOERR_GENERAL;
342bfb5e 541 rprintf(FERROR,
7842418b 542 "cannot add local filter rules in long-named directory: %s\n",
6dfd07d0
WD
543 full_fname(dirbuf));
544 }
545 dirbuf[dirbuf_len] = '\0';
546 }
547
548 return (void*)push;
549}
550
7842418b 551void pop_local_filters(void *mem)
6dfd07d0 552{
daa8d920 553 struct local_filter_state *pop = (struct local_filter_state *)mem;
6dfd07d0 554 int i;
daa8d920
MM
555 int old_mergelist_cnt = pop ? pop->mergelist_cnt : 0;
556
557 if (DEBUG_GTE(FILTER, 2))
558 rprintf(FINFO, "[%s] popping local filters\n", who_am_i());
6dfd07d0
WD
559
560 for (i = mergelist_cnt; i-- > 0; ) {
7842418b
WD
561 struct filter_struct *ex = mergelist_parents[i];
562 struct filter_list_struct *lp = ex->u.mergelist;
6dfd07d0 563
951e826b 564 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
565 rprintf(FINFO, "[%s] popping mergelist #%d%s\n",
566 who_am_i(), i, lp->debug_type);
6dfd07d0
WD
567 }
568
7842418b 569 clear_filter_list(lp);
daa8d920
MM
570
571 if (i >= old_mergelist_cnt) {
572 /* This mergelist does not exist in the state to be
573 * restored. Free its parent_dirscan list to clean up
574 * any per-dir mergelists defined there so we don't
575 * crash trying to restore nonexistent state for them
576 * below. (Counterpart to setup_merge_file call in
577 * push_local_filters. Must be done here, not in
578 * free_filter, for LIFO order.) */
7634fc8e 579 if (DEBUG_GTE(FILTER, 2)) {
daa8d920
MM
580 rprintf(FINFO, "[%s] freeing parent_dirscan filters of mergelist #%d%s\n",
581 who_am_i(), i, ex->u.mergelist->debug_type);
7634fc8e 582 }
daa8d920
MM
583 free_filters(lp->parent_dirscan_head);
584 lp->parent_dirscan_head = NULL;
585 }
6dfd07d0
WD
586 }
587
daa8d920
MM
588 /* If we cleaned things up properly, the only still-active mergelists
589 * should be those with a state to be restored. */
590 assert(mergelist_cnt == old_mergelist_cnt);
591
7634fc8e 592 if (!pop) {
daa8d920 593 /* No state to restore. */
a2b371cd 594 return;
7634fc8e 595 }
a2b371cd 596
daa8d920
MM
597 for (i = 0; i < mergelist_cnt; i++) {
598 memcpy(mergelist_parents[i]->u.mergelist, &pop->mergelists[i],
7842418b 599 sizeof (struct filter_list_struct));
6dfd07d0
WD
600 }
601
602 free(pop);
5be7fa93
WD
603}
604
fe04532a
WD
605void change_local_filter_dir(const char *dname, int dlen, int dir_depth)
606{
7fa60281 607 static int cur_depth = -1;
fe04532a
WD
608 static void *filt_array[MAXPATHLEN/2+1];
609
610 if (!dname) {
7fa60281
WD
611 for ( ; cur_depth >= 0; cur_depth--) {
612 if (filt_array[cur_depth]) {
613 pop_local_filters(filt_array[cur_depth]);
614 filt_array[cur_depth] = NULL;
615 }
616 }
fe04532a
WD
617 return;
618 }
619
620 assert(dir_depth < MAXPATHLEN/2+1);
621
7fa60281
WD
622 for ( ; cur_depth >= dir_depth; cur_depth--) {
623 if (filt_array[cur_depth]) {
624 pop_local_filters(filt_array[cur_depth]);
625 filt_array[cur_depth] = NULL;
626 }
627 }
fe04532a 628
7fa60281 629 cur_depth = dir_depth;
fe04532a
WD
630 filt_array[cur_depth] = push_local_filters(dname, dlen);
631}
632
ffe8feb2 633static int rule_matches(const char *fname, struct filter_struct *ex, int name_is_dir)
2b6b4d53 634{
e5daa273 635 int slash_handling, str_cnt = 0, anchored_match = 0;
ebaa4296 636 int ret_match = ex->flags & MATCHFLG_NEGATE ? 0 : 1;
e5daa273
WD
637 char *p, *pattern = ex->pattern;
638 const char *strings[16]; /* more than enough */
ffe8feb2 639 const char *name = fname + (*fname == '/');
2b6b4d53 640
9f186578
WD
641 if (!*name)
642 return 0;
643
ebaa4296 644 if (!ex->u.slash_cnt && !(ex->flags & MATCHFLG_WILD2)) {
770de899
WD
645 /* If the pattern does not have any slashes AND it does
646 * not have a "**" (which could match a slash), then we
647 * just match the name portion of the path. */
5be7fa93
WD
648 if ((p = strrchr(name,'/')) != NULL)
649 name = p+1;
ebaa4296 650 } else if (ex->flags & MATCHFLG_ABS_PATH && *fname != '/'
6dfd07d0 651 && curr_dir_len > module_dirlen + 1) {
770de899
WD
652 /* If we're matching against an absolute-path pattern,
653 * we need to prepend our full path info. */
e5daa273
WD
654 strings[str_cnt++] = curr_dir + module_dirlen + 1;
655 strings[str_cnt++] = "/";
ebaa4296 656 } else if (ex->flags & MATCHFLG_WILD2_PREFIX && *fname != '/') {
e5daa273
WD
657 /* Allow "**"+"/" to match at the start of the string. */
658 strings[str_cnt++] = "/";
5be7fa93 659 }
e5daa273
WD
660 strings[str_cnt++] = name;
661 if (name_is_dir) {
662 /* Allow a trailing "/"+"***" to match the directory. */
ebaa4296 663 if (ex->flags & MATCHFLG_WILD3_SUFFIX)
e5daa273 664 strings[str_cnt++] = "/";
ebaa4296 665 } else if (ex->flags & MATCHFLG_DIRECTORY)
f2ae9e85 666 return !ret_match;
e5daa273 667 strings[str_cnt] = NULL;
2b6b4d53 668
170381c0 669 if (*pattern == '/') {
770de899 670 anchored_match = 1;
2b6b4d53
AT
671 pattern++;
672 }
673
e5daa273 674 if (!anchored_match && ex->u.slash_cnt
ebaa4296 675 && !(ex->flags & MATCHFLG_WILD2)) {
170381c0
WD
676 /* A non-anchored match with an infix slash and no "**"
677 * needs to match the last slash_cnt+1 name elements. */
e5daa273 678 slash_handling = ex->u.slash_cnt + 1;
ebaa4296
MM
679 } else if (!anchored_match && !(ex->flags & MATCHFLG_WILD2_PREFIX)
680 && ex->flags & MATCHFLG_WILD2) {
e5daa273
WD
681 /* A non-anchored match with an infix or trailing "**" (but not
682 * a prefixed "**") needs to try matching after every slash. */
683 slash_handling = -1;
684 } else {
685 /* The pattern matches only at the start of the path or name. */
686 slash_handling = 0;
687 }
688
ebaa4296 689 if (ex->flags & MATCHFLG_WILD) {
e5daa273
WD
690 if (wildmatch_array(pattern, strings, slash_handling))
691 return ret_match;
692 } else if (str_cnt > 1) {
693 if (litmatch_array(pattern, strings, slash_handling))
f2ae9e85 694 return ret_match;
770de899 695 } else if (anchored_match) {
48ecccce 696 if (strcmp(name, pattern) == 0)
f2ae9e85 697 return ret_match;
2b6b4d53
AT
698 } else {
699 int l1 = strlen(name);
ea2111d1 700 int l2 = strlen(pattern);
0f2ac855 701 if (l2 <= l1 &&
ea2111d1 702 strcmp(name+(l1-l2),pattern) == 0 &&
170381c0 703 (l1==l2 || name[l1-(l2+1)] == '/')) {
f2ae9e85 704 return ret_match;
c36cd317 705 }
2b6b4d53
AT
706 }
707
f2ae9e85 708 return !ret_match;
c627d613
AT
709}
710
711
1df02d13 712static void report_filter_result(enum logcode code, char const *name,
7634fc8e
WD
713 struct filter_struct const *ent,
714 int name_is_dir, const char *type)
d567322f 715{
0f2ac855 716 /* If a trailing slash is present to match only directories,
fdc79501 717 * then it is stripped out by add_rule(). So as a special
0f2ac855
WD
718 * case we add it back in here. */
719
951e826b 720 if (DEBUG_GTE(FILTER, 1)) {
d5782b52
WD
721 static char *actions[2][2]
722 = { {"show", "hid"}, {"risk", "protect"} };
723 const char *w = who_am_i();
1df02d13 724 rprintf(code, "[%s] %sing %s %s because of pattern %s%s%s\n",
ebaa4296 725 w, actions[*w!='s'][!(ent->flags&MATCHFLG_INCLUDE)],
d5782b52 726 name_is_dir ? "directory" : "file", name, ent->pattern,
ebaa4296 727 ent->flags & MATCHFLG_DIRECTORY ? "/" : "", type);
ea847c62 728 }
d567322f
MP
729}
730
731
732/*
a6536635
WD
733 * Return -1 if file "name" is defined to be excluded by the specified
734 * exclude list, 1 if it is included, and 0 if it was not matched.
d567322f 735 */
1df02d13
WD
736int check_filter(struct filter_list_struct *listp, enum logcode code,
737 const char *name, int name_is_dir)
c627d613 738{
7842418b 739 struct filter_struct *ent;
c627d613 740
b2aa573b 741 for (ent = listp->head; ent; ent = ent->next) {
ebaa4296 742 if (ignore_perishable && ent->flags & MATCHFLG_PERISHABLE)
ccdb23bb 743 continue;
ebaa4296 744 if (ent->flags & MATCHFLG_PERDIR_MERGE) {
1df02d13 745 int rc = check_filter(ent->u.mergelist, code, name,
7842418b 746 name_is_dir);
6dfd07d0
WD
747 if (rc)
748 return rc;
749 continue;
750 }
ebaa4296 751 if (ent->flags & MATCHFLG_CVS_IGNORE) {
1df02d13 752 int rc = check_filter(&cvs_filter_list, code, name,
fdc79501
WD
753 name_is_dir);
754 if (rc)
755 return rc;
756 continue;
757 }
7842418b 758 if (rule_matches(name, ent, name_is_dir)) {
1df02d13
WD
759 report_filter_result(code, name, ent, name_is_dir,
760 listp->debug_type);
ebaa4296 761 return ent->flags & MATCHFLG_INCLUDE ? 1 : -1;
0f2ac855 762 }
2b6b4d53 763 }
c627d613 764
2b6b4d53 765 return 0;
c627d613
AT
766}
767
345e0988 768#define RULE_STRCMP(s,r) rule_strcmp((s), (r), sizeof (r) - 1)
a1ac8edd 769
4adbb5f2 770static const uchar *rule_strcmp(const uchar *str, const char *rule, int rule_len)
a1ac8edd 771{
4adbb5f2 772 if (strncmp((char*)str, rule, rule_len) != 0)
a1ac8edd
WD
773 return NULL;
774 if (isspace(str[rule_len]) || str[rule_len] == '_' || !str[rule_len])
775 return str + rule_len - 1;
776 if (str[rule_len] == ',')
777 return str + rule_len;
778 return NULL;
779}
c627d613 780
ebaa4296
MM
781#define MATCHFLGS_FROM_CONTAINER (MATCHFLG_ABS_PATH | MATCHFLG_INCLUDE \
782 | MATCHFLG_DIRECTORY | MATCHFLG_NEGATE \
783 | MATCHFLG_PERISHABLE)
784
785/* Gets the next include/exclude rule from *rulestr_ptr and advances
786 * *rulestr_ptr to point beyond it. Stores the pattern's start (within
787 * *rulestr_ptr) and length in *pat_ptr and *pat_len_ptr, and returns a newly
788 * allocated filter_struct containing the rest of the information. Returns
789 * NULL if there are no more rules in the input.
790 *
791 * The template provides defaults for the new rule to inherit, and the
792 * template mflags and the xflags additionally affect parsing. */
793static struct filter_struct *parse_rule_tok(const char **rulestr_ptr,
794 struct filter_struct *template, int xflags,
795 const char **pat_ptr, unsigned int *pat_len_ptr)
f8f72644 796{
ebaa4296
MM
797 const uchar *s = (const uchar *)*rulestr_ptr;
798 struct filter_struct *filter;
fdc79501 799 unsigned int len;
f8f72644 800
ebaa4296 801 if (template->flags & MATCHFLG_WORD_SPLIT) {
96d3590a
WD
802 /* Skip over any initial whitespace. */
803 while (isspace(*s))
f8f72644 804 s++;
6dfd07d0 805 /* Update to point to real start of rule. */
ebaa4296 806 *rulestr_ptr = (const char *)s;
f8f72644 807 }
6dfd07d0
WD
808 if (!*s)
809 return NULL;
810
ebaa4296
MM
811 if (!(filter = new0(struct filter_struct)))
812 out_of_memory("parse_rule_tok");
813
814 /* Inherit from the template. Don't inherit MATCHFLGS_SIDES; we check
815 * that later. */
816 filter->flags = template->flags & MATCHFLGS_FROM_CONTAINER;
b6f06b8e
WD
817
818 /* Figure out what kind of a filter rule "s" is pointing at. Note
819 * that if MATCHFLG_NO_PREFIXES is set, the rule is either an include
820 * or an exclude based on the inheritance of the MATCHFLG_INCLUDE
821 * flag (above). XFLG_OLD_PREFIXES indicates a compatibility mode
822 * for old include/exclude patterns where just "+ " and "- " are
823 * allowed as optional prefixes. */
ebaa4296
MM
824 if (template->flags & MATCHFLG_NO_PREFIXES) {
825 if (*s == '!' && template->flags & MATCHFLG_CVS_IGNORE)
826 filter->flags |= MATCHFLG_CLEAR_LIST; /* Tentative! */
b6f06b8e
WD
827 } else if (xflags & XFLG_OLD_PREFIXES) {
828 if (*s == '-' && s[1] == ' ') {
ebaa4296 829 filter->flags &= ~MATCHFLG_INCLUDE;
b6f06b8e
WD
830 s += 2;
831 } else if (*s == '+' && s[1] == ' ') {
ebaa4296 832 filter->flags |= MATCHFLG_INCLUDE;
b6f06b8e 833 s += 2;
664cf327 834 } else if (*s == '!')
ebaa4296 835 filter->flags |= MATCHFLG_CLEAR_LIST; /* Tentative! */
b6f06b8e 836 } else {
ebaa4296
MM
837 char ch = 0;
838 BOOL prefix_specifies_side = False;
6dfd07d0 839 switch (*s) {
a1ac8edd 840 case 'c':
345e0988 841 if ((s = RULE_STRCMP(s, "clear")) != NULL)
a1ac8edd
WD
842 ch = '!';
843 break;
844 case 'd':
345e0988 845 if ((s = RULE_STRCMP(s, "dir-merge")) != NULL)
a1ac8edd
WD
846 ch = ':';
847 break;
848 case 'e':
345e0988 849 if ((s = RULE_STRCMP(s, "exclude")) != NULL)
a1ac8edd
WD
850 ch = '-';
851 break;
ed243f8c
WD
852 case 'h':
853 if ((s = RULE_STRCMP(s, "hide")) != NULL)
854 ch = 'H';
855 break;
a1ac8edd 856 case 'i':
345e0988 857 if ((s = RULE_STRCMP(s, "include")) != NULL)
a1ac8edd
WD
858 ch = '+';
859 break;
860 case 'm':
345e0988 861 if ((s = RULE_STRCMP(s, "merge")) != NULL)
a1ac8edd
WD
862 ch = '.';
863 break;
ed243f8c
WD
864 case 'p':
865 if ((s = RULE_STRCMP(s, "protect")) != NULL)
866 ch = 'P';
867 break;
868 case 'r':
869 if ((s = RULE_STRCMP(s, "risk")) != NULL)
870 ch = 'R';
871 break;
872 case 's':
873 if ((s = RULE_STRCMP(s, "show")) != NULL)
874 ch = 'S';
875 break;
a1ac8edd
WD
876 default:
877 ch = *s;
64b761c1
WD
878 if (s[1] == ',')
879 s++;
a1ac8edd
WD
880 break;
881 }
882 switch (ch) {
6dfd07d0 883 case ':':
ebaa4296
MM
884 filter->flags |= MATCHFLG_PERDIR_MERGE
885 | MATCHFLG_FINISH_SETUP;
6dfd07d0
WD
886 /* FALL THROUGH */
887 case '.':
ebaa4296 888 filter->flags |= MATCHFLG_MERGE_FILE;
6dfd07d0
WD
889 break;
890 case '+':
ebaa4296 891 filter->flags |= MATCHFLG_INCLUDE;
bf39270e 892 /* FALL THROUGH */
6dfd07d0
WD
893 case '-':
894 break;
ed243f8c 895 case 'S':
ebaa4296 896 filter->flags |= MATCHFLG_INCLUDE;
ed243f8c
WD
897 /* FALL THROUGH */
898 case 'H':
ebaa4296
MM
899 filter->flags |= MATCHFLG_SENDER_SIDE;
900 prefix_specifies_side = True;
ed243f8c
WD
901 break;
902 case 'R':
ebaa4296 903 filter->flags |= MATCHFLG_INCLUDE;
ed243f8c
WD
904 /* FALL THROUGH */
905 case 'P':
ebaa4296
MM
906 filter->flags |= MATCHFLG_RECEIVER_SIDE;
907 prefix_specifies_side = True;
ed243f8c 908 break;
6dfd07d0 909 case '!':
ebaa4296 910 filter->flags |= MATCHFLG_CLEAR_LIST;
6dfd07d0
WD
911 break;
912 default:
ebaa4296 913 rprintf(FERROR, "Unknown filter rule: `%s'\n", *rulestr_ptr);
6dfd07d0
WD
914 exit_cleanup(RERR_SYNTAX);
915 }
ebaa4296
MM
916 while (ch != '!' && *++s && *s != ' ' && *s != '_') {
917 if (template->flags & MATCHFLG_WORD_SPLIT && isspace(*s)) {
918 s--;
919 break;
920 }
921 switch (*s) {
922 default:
b6f06b8e 923 invalid:
6dfd07d0 924 rprintf(FERROR,
ebaa4296
MM
925 "invalid modifier '%c' at position %d in filter rule: %s\n",
926 *s, s - (const uchar *)*rulestr_ptr, *rulestr_ptr);
6dfd07d0 927 exit_cleanup(RERR_SYNTAX);
6dfd07d0 928 case '-':
ebaa4296
MM
929 if (!BITS_SETnUNSET(filter->flags, MATCHFLG_MERGE_FILE, MATCHFLG_NO_PREFIXES))
930 goto invalid;
931 filter->flags |= MATCHFLG_NO_PREFIXES;
6dfd07d0
WD
932 break;
933 case '+':
ebaa4296
MM
934 if (!BITS_SETnUNSET(filter->flags, MATCHFLG_MERGE_FILE, MATCHFLG_NO_PREFIXES))
935 goto invalid;
936 filter->flags |= MATCHFLG_NO_PREFIXES
937 | MATCHFLG_INCLUDE;
6dfd07d0 938 break;
bf39270e 939 case '/':
ebaa4296 940 filter->flags |= MATCHFLG_ABS_PATH;
bf39270e 941 break;
f2ae9e85 942 case '!':
ebaa4296
MM
943 /* Negation really goes with the pattern, so it
944 * isn't useful as a merge-file default. */
945 if (filter->flags & MATCHFLG_MERGE_FILE)
946 goto invalid;
947 filter->flags |= MATCHFLG_NEGATE;
f2ae9e85 948 break;
6dfd07d0 949 case 'C':
ebaa4296
MM
950 if (filter->flags & MATCHFLG_NO_PREFIXES || prefix_specifies_side)
951 goto invalid;
952 filter->flags |= MATCHFLG_NO_PREFIXES
953 | MATCHFLG_WORD_SPLIT
954 | MATCHFLG_NO_INHERIT
955 | MATCHFLG_CVS_IGNORE;
6dfd07d0
WD
956 break;
957 case 'e':
ebaa4296
MM
958 if (!(filter->flags & MATCHFLG_MERGE_FILE))
959 goto invalid;
960 filter->flags |= MATCHFLG_EXCLUDE_SELF;
6dfd07d0
WD
961 break;
962 case 'n':
ebaa4296
MM
963 if (!(filter->flags & MATCHFLG_MERGE_FILE))
964 goto invalid;
965 filter->flags |= MATCHFLG_NO_INHERIT;
6dfd07d0 966 break;
ccdb23bb 967 case 'p':
ebaa4296 968 filter->flags |= MATCHFLG_PERISHABLE;
ccdb23bb 969 break;
ed243f8c 970 case 'r':
ebaa4296
MM
971 if (prefix_specifies_side)
972 goto invalid;
973 filter->flags |= MATCHFLG_RECEIVER_SIDE;
ed243f8c
WD
974 break;
975 case 's':
ebaa4296
MM
976 if (prefix_specifies_side)
977 goto invalid;
978 filter->flags |= MATCHFLG_SENDER_SIDE;
ed243f8c 979 break;
0b2901b7 980 case 'w':
ebaa4296
MM
981 if (!(filter->flags & MATCHFLG_MERGE_FILE))
982 goto invalid;
983 filter->flags |= MATCHFLG_WORD_SPLIT;
6dfd07d0
WD
984 break;
985 }
986 }
987 if (*s)
988 s++;
6dfd07d0 989 }
ebaa4296
MM
990 if (template->flags & MATCHFLGS_SIDES) {
991 if (filter->flags & MATCHFLGS_SIDES) {
992 /* The filter and template both specify side(s). This
993 * is dodgy (and won't work correctly if the template is
994 * a one-sided per-dir merge rule), so reject it. */
995 rprintf(FERROR,
996 "specified-side merge file contains specified-side filter: %s\n",
997 *rulestr_ptr);
998 exit_cleanup(RERR_SYNTAX);
999 }
1000 filter->flags |= template->flags & MATCHFLGS_SIDES;
1001 }
6dfd07d0 1002
ebaa4296 1003 if (template->flags & MATCHFLG_WORD_SPLIT) {
fdc79501 1004 const uchar *cp = s;
96d3590a
WD
1005 /* Token ends at whitespace or the end of the string. */
1006 while (!isspace(*cp) && *cp != '\0')
1007 cp++;
1008 len = cp - s;
1009 } else
0eeb1cf8 1010 len = strlen((char*)s);
96d3590a 1011
ebaa4296
MM
1012 if (filter->flags & MATCHFLG_CLEAR_LIST) {
1013 if (!(filter->flags & MATCHFLG_NO_PREFIXES)
35a388b1 1014 && !(xflags & XFLG_OLD_PREFIXES) && len) {
6dfd07d0 1015 rprintf(FERROR,
ebaa4296 1016 "'!' rule has trailing characters: %s\n", *rulestr_ptr);
6dfd07d0
WD
1017 exit_cleanup(RERR_SYNTAX);
1018 }
1019 if (len > 1)
ebaa4296
MM
1020 filter->flags &= ~MATCHFLG_CLEAR_LIST;
1021 } else if (!len && !(filter->flags & MATCHFLG_CVS_IGNORE)) {
1022 rprintf(FERROR, "unexpected end of filter rule: %s\n", *rulestr_ptr);
6dfd07d0
WD
1023 exit_cleanup(RERR_SYNTAX);
1024 }
1025
f1678790 1026 /* --delete-excluded turns an un-modified include/exclude into a
48ecccce
WD
1027 * sender-side rule. We also affect per-dir merge files that take
1028 * no prefixes as a simple optimization. */
f1678790 1029 if (delete_excluded
ebaa4296
MM
1030 && !(filter->flags & MATCHFLGS_SIDES)
1031 && (!(filter->flags & MATCHFLG_PERDIR_MERGE)
1032 || filter->flags & MATCHFLG_NO_PREFIXES))
1033 filter->flags |= MATCHFLG_SENDER_SIDE;
1034
1035 *pat_ptr = (const char *)s;
1036 *pat_len_ptr = len;
1037 *rulestr_ptr = *pat_ptr + len;
1038 return filter;
f8f72644
WD
1039}
1040
1041
0f78b815 1042static char default_cvsignore[] =
46db1850
WD
1043 /* These default ignored items come from the CVS manual. */
1044 "RCS SCCS CVS CVS.adm RCSLOG cvslog.* tags TAGS"
1045 " .make.state .nse_depinfo *~ #* .#* ,* _$* *$"
1046 " *.old *.bak *.BAK *.orig *.rej .del-*"
1047 " *.a *.olb *.o *.obj *.so *.exe"
1048 " *.Z *.elc *.ln core"
1049 /* The rest we added to suit ourself. */
0485b451 1050 " .svn/ .git/ .bzr/";
46db1850 1051
ebaa4296 1052static void get_cvs_excludes(struct filter_struct *template)
46db1850 1053{
46db1850 1054 static int initialized = 0;
c2f0e4d9 1055 char *p, fname[MAXPATHLEN];
ebaa4296 1056 uint32 save_mflags;
46db1850
WD
1057
1058 if (initialized)
1059 return;
1060 initialized = 1;
1061
ebaa4296
MM
1062 save_mflags = template->flags;
1063 if (protocol_version >= 30)
1064 template->flags |= MATCHFLG_PERISHABLE;
1065 load_filter_str(&cvs_filter_list, default_cvsignore, template, 0);
1066 template->flags = save_mflags;
46db1850
WD
1067
1068 p = module_id >= 0 && lp_use_chroot(module_id) ? "/" : getenv("HOME");
1069 if (p && pathjoin(fname, MAXPATHLEN, p, ".cvsignore") < MAXPATHLEN)
ebaa4296
MM
1070 load_filter_file(&cvs_filter_list, fname, template, 0);
1071
1072 load_filter_str(&cvs_filter_list, getenv("CVSIGNORE"), template, 0);
1073}
1074
46db1850 1075
ebaa4296
MM
1076struct filter_struct *filter_template(uint32 mflags)
1077{
1078 static struct filter_struct template; /* zero-initialized */
1079 template.flags = mflags;
1080 return &template;
46db1850
WD
1081}
1082
1083
ebaa4296
MM
1084void load_filter_str(struct filter_list_struct *listp, const char *rulestr,
1085 struct filter_struct *template, int xflags)
c627d613 1086{
ebaa4296
MM
1087 struct filter_struct *filter;
1088 const char *pat;
fdc79501 1089 unsigned int pat_len;
5be7fa93 1090
ebaa4296 1091 if (!rulestr)
5e7dbaca 1092 return;
f8f72644 1093
b2aa573b 1094 while (1) {
6dfd07d0 1095 /* Remember that the returned string is NOT '\0' terminated! */
ebaa4296
MM
1096 filter = parse_rule_tok(&rulestr, template, xflags,
1097 &pat, &pat_len);
1098 if (!filter)
b2aa573b 1099 break;
969f7ed5 1100
6dfd07d0 1101 if (pat_len >= MAXPATHLEN) {
969f7ed5 1102 rprintf(FERROR, "discarding over-long filter: %.*s\n",
ebaa4296
MM
1103 (int)pat_len, pat);
1104 free_continue:
1105 free_filter(filter);
6dfd07d0
WD
1106 continue;
1107 }
5e972dcf 1108
ebaa4296 1109 if (filter->flags & MATCHFLG_CLEAR_LIST) {
951e826b 1110 if (DEBUG_GTE(FILTER, 2)) {
de91e757 1111 rprintf(FINFO,
7842418b 1112 "[%s] clearing filter list%s\n",
de91e757
WD
1113 who_am_i(), listp->debug_type);
1114 }
7842418b 1115 clear_filter_list(listp);
ebaa4296 1116 goto free_continue;
5e972dcf 1117 }
b2aa573b 1118
ebaa4296 1119 if (filter->flags & MATCHFLG_MERGE_FILE) {
fdc79501 1120 if (!pat_len) {
ebaa4296 1121 pat = ".cvsignore";
fdc79501
WD
1122 pat_len = 10;
1123 }
ebaa4296
MM
1124 if (filter->flags & MATCHFLG_EXCLUDE_SELF) {
1125 const char *name;
1126 struct filter_struct *excl_self;
1127
1128 if (!(excl_self = new0(struct filter_struct)))
1129 out_of_memory("load_filter_str");
1130 /* Find the beginning of the basename. */
1131 for (name = pat + pat_len; name > pat && name[-1] != '/'; name--)
1132 ;
1133 add_rule(listp, name, (pat + pat_len) - name, excl_self, 0);
1134 filter->flags &= ~MATCHFLG_EXCLUDE_SELF;
6dfd07d0 1135 }
ebaa4296 1136 if (filter->flags & MATCHFLG_PERDIR_MERGE) {
6dfd07d0 1137 if (parent_dirscan) {
ebaa4296
MM
1138 const char *p;
1139 unsigned int len = pat_len;
1140 if ((p = parse_merge_name(pat, &len,
b6f06b8e 1141 module_dirlen)))
ebaa4296
MM
1142 add_rule(listp, p, len, filter, 0);
1143 else
1144 free_filter(filter);
6dfd07d0
WD
1145 continue;
1146 }
1147 } else {
ebaa4296
MM
1148 const char *p;
1149 unsigned int len = pat_len;
1150 if ((p = parse_merge_name(pat, &len, 0))) {
1151 load_filter_file(listp, p, filter,
1152 XFLG_FATAL_ERRORS);
1153 }
1154 free_filter(filter);
6dfd07d0
WD
1155 continue;
1156 }
f8f72644 1157 }
6dfd07d0 1158
ebaa4296
MM
1159 if (filter->flags & MATCHFLG_CVS_IGNORE
1160 && !(filter->flags & MATCHFLG_MERGE_FILE))
1161 get_cvs_excludes(filter);
fdc79501 1162
ebaa4296 1163 add_rule(listp, pat, pat_len, filter, xflags);
8c35542d 1164 }
c627d613
AT
1165}
1166
c627d613 1167
ebaa4296
MM
1168void load_filter_file(struct filter_list_struct *listp, const char *fname,
1169 struct filter_struct *template, int xflags)
c627d613 1170{
5e7dbaca 1171 FILE *fp;
3fac51e2 1172 char line[BIGPATHBUFLEN];
7cd72c79 1173 char *eob = line + sizeof line - 1;
ebaa4296 1174 BOOL word_split = (template->flags & MATCHFLG_WORD_SPLIT) != 0;
ccdff3eb 1175
5be7fa93
WD
1176 if (!fname || !*fname)
1177 return;
1178
5a016db9 1179 if (*fname != '-' || fname[1] || am_server) {
819bfe45 1180 if (daemon_filter_list.head) {
5a016db9 1181 strlcpy(line, fname, sizeof line);
b58bfb2f 1182 clean_fname(line, CFN_COLLAPSE_DOT_DOT_DIRS);
1df02d13 1183 if (check_filter(&daemon_filter_list, FLOG, line, 0) < 0)
5a016db9
WD
1184 fp = NULL;
1185 else
1186 fp = fopen(line, "rb");
1187 } else
1188 fp = fopen(fname, "rb");
1189 } else
5e7dbaca 1190 fp = stdin;
bf39270e 1191
951e826b 1192 if (DEBUG_GTE(FILTER, 2)) {
ebaa4296
MM
1193 rprintf(FINFO, "[%s] load_filter_file(%s,%x,%x)%s\n",
1194 who_am_i(), fname, template->flags, xflags,
bf39270e
WD
1195 fp ? "" : " [not found]");
1196 }
1197
5e7dbaca 1198 if (!fp) {
f8f72644 1199 if (xflags & XFLG_FATAL_ERRORS) {
a039749b 1200 rsyserr(FERROR, errno,
6dfd07d0 1201 "failed to open %sclude file %s",
ebaa4296 1202 template->flags & MATCHFLG_INCLUDE ? "in" : "ex",
45c49b52 1203 fname);
65417579 1204 exit_cleanup(RERR_FILEIO);
2b6b4d53 1205 }
5be7fa93 1206 return;
2b6b4d53 1207 }
6dfd07d0
WD
1208 dirbuf[dirbuf_len] = '\0';
1209
ccdff3eb 1210 while (1) {
5e7dbaca 1211 char *s = line;
619d21ff 1212 int ch, overflow = 0;
ccdff3eb 1213 while (1) {
5e7dbaca 1214 if ((ch = getc(fp)) == EOF) {
61e16468
WD
1215 if (ferror(fp) && errno == EINTR) {
1216 clearerr(fp);
ccdff3eb 1217 continue;
61e16468 1218 }
ccdff3eb
WD
1219 break;
1220 }
40d38dc0
WD
1221 if (word_split && isspace(ch))
1222 break;
ccdff3eb
WD
1223 if (eol_nulls? !ch : (ch == '\n' || ch == '\r'))
1224 break;
1225 if (s < eob)
1226 *s++ = ch;
619d21ff
WD
1227 else
1228 overflow = 1;
1229 }
1230 if (overflow) {
7842418b 1231 rprintf(FERROR, "discarding over-long filter: %s...\n", line);
619d21ff 1232 s = line;
ccdff3eb
WD
1233 }
1234 *s = '\0';
7f0feb4d
WD
1235 /* Skip an empty token and (when line parsing) comments. */
1236 if (*line && (word_split || (*line != ';' && *line != '#')))
ebaa4296 1237 load_filter_str(listp, line, template, xflags);
5e7dbaca 1238 if (ch == EOF)
ccdff3eb 1239 break;
2b6b4d53 1240 }
5e7dbaca 1241 fclose(fp);
c627d613
AT
1242}
1243
a261103c 1244/* If the "for_xfer" flag is set, the prefix is made compatible with the
fdc79501
WD
1245 * current protocol_version (if possible) or a NULL is returned (if not
1246 * possible). */
ebaa4296 1247char *get_rule_prefix(struct filter_struct *filter, const char *pat, int for_xfer,
fdc79501 1248 unsigned int *plen_ptr)
417b5999
WD
1249{
1250 static char buf[MAX_RULE_PREFIX+1];
1251 char *op = buf;
2217b30a 1252 int legal_len = for_xfer && protocol_version < 29 ? 1 : MAX_RULE_PREFIX-1;
417b5999 1253
ebaa4296 1254 if (filter->flags & MATCHFLG_PERDIR_MERGE) {
fdc79501
WD
1255 if (legal_len == 1)
1256 return NULL;
417b5999 1257 *op++ = ':';
ebaa4296 1258 } else if (filter->flags & MATCHFLG_INCLUDE)
fdc79501
WD
1259 *op++ = '+';
1260 else if (legal_len != 1
1261 || ((*pat == '-' || *pat == '+') && pat[1] == ' '))
1262 *op++ = '-';
1263 else
1264 legal_len = 0;
1265
ebaa4296
MM
1266 if (filter->flags & MATCHFLG_ABS_PATH)
1267 *op++ = '/';
1268 if (filter->flags & MATCHFLG_NEGATE)
2a28dd32 1269 *op++ = '!';
ebaa4296 1270 if (filter->flags & MATCHFLG_CVS_IGNORE)
fdc79501
WD
1271 *op++ = 'C';
1272 else {
ebaa4296 1273 if (filter->flags & MATCHFLG_NO_INHERIT)
417b5999 1274 *op++ = 'n';
ebaa4296 1275 if (filter->flags & MATCHFLG_WORD_SPLIT)
d09e800a 1276 *op++ = 'w';
ebaa4296
MM
1277 if (filter->flags & MATCHFLG_NO_PREFIXES) {
1278 if (filter->flags & MATCHFLG_INCLUDE)
417b5999
WD
1279 *op++ = '+';
1280 else
1281 *op++ = '-';
1282 }
417b5999 1283 }
ebaa4296 1284 if (filter->flags & MATCHFLG_EXCLUDE_SELF)
46db1850 1285 *op++ = 'e';
ebaa4296 1286 if (filter->flags & MATCHFLG_SENDER_SIDE
ed243f8c
WD
1287 && (!for_xfer || protocol_version >= 29))
1288 *op++ = 's';
ebaa4296 1289 if (filter->flags & MATCHFLG_RECEIVER_SIDE
ed243f8c
WD
1290 && (!for_xfer || protocol_version >= 29
1291 || (delete_excluded && am_sender)))
1292 *op++ = 'r';
ebaa4296 1293 if (filter->flags & MATCHFLG_PERISHABLE) {
ccdb23bb
WD
1294 if (!for_xfer || protocol_version >= 30)
1295 *op++ = 'p';
1296 else if (am_sender)
1297 return NULL;
1298 }
448797a1
WD
1299 if (op - buf > legal_len)
1300 return NULL;
2217b30a
WD
1301 if (legal_len)
1302 *op++ = ' ';
417b5999
WD
1303 *op = '\0';
1304 if (plen_ptr)
1305 *plen_ptr = op - buf;
417b5999
WD
1306 return buf;
1307}
c627d613 1308
fdc79501 1309static void send_rules(int f_out, struct filter_list_struct *flp)
c627d613 1310{
ed243f8c 1311 struct filter_struct *ent, *prev = NULL;
25cf8893 1312
fdc79501 1313 for (ent = flp->head; ent; ent = ent->next) {
417b5999 1314 unsigned int len, plen, dlen;
ed243f8c 1315 int elide = 0;
417b5999 1316 char *p;
2fb139c1 1317
f1678790
WD
1318 /* Note we need to check delete_excluded here in addition to
1319 * the code in parse_rule_tok() because some rules may have
48ecccce
WD
1320 * been added before we found the --delete-excluded option.
1321 * We must also elide any CVS merge-file rules to avoid a
1322 * backward compatibility problem, and we elide any no-prefix
1323 * merge files as an optimization (since they can only have
1324 * include/exclude rules). */
ebaa4296 1325 if (ent->flags & MATCHFLG_SENDER_SIDE)
ed243f8c 1326 elide = am_sender ? 1 : -1;
ebaa4296 1327 if (ent->flags & MATCHFLG_RECEIVER_SIDE)
ed243f8c 1328 elide = elide ? 0 : am_sender ? -1 : 1;
97bf86f8 1329 else if (delete_excluded && !elide
ebaa4296
MM
1330 && (!(ent->flags & MATCHFLG_PERDIR_MERGE)
1331 || ent->flags & MATCHFLG_NO_PREFIXES))
ed243f8c
WD
1332 elide = am_sender ? 1 : -1;
1333 if (elide < 0) {
1334 if (prev)
1335 prev->next = ent->next;
1336 else
1337 flp->head = ent->next;
1338 } else
1339 prev = ent;
1340 if (elide > 0)
1341 continue;
ebaa4296
MM
1342 if (ent->flags & MATCHFLG_CVS_IGNORE
1343 && !(ent->flags & MATCHFLG_MERGE_FILE)) {
98e47414 1344 int f = am_sender || protocol_version < 29 ? f_out : -2;
ed243f8c 1345 send_rules(f, &cvs_filter_list);
98e47414 1346 if (f == f_out)
417b5999 1347 continue;
fdc79501 1348 }
ebaa4296 1349 p = get_rule_prefix(ent, ent->pattern, 1, &plen);
fdc79501 1350 if (!p) {
417b5999 1351 rprintf(FERROR,
fdc79501 1352 "filter rules are too modern for remote rsync.\n");
caf8299e 1353 exit_cleanup(RERR_PROTOCOL);
5f5be796 1354 }
ed243f8c
WD
1355 if (f_out < 0)
1356 continue;
fdc79501 1357 len = strlen(ent->pattern);
ebaa4296 1358 dlen = ent->flags & MATCHFLG_DIRECTORY ? 1 : 0;
fdc79501
WD
1359 if (!(plen + len + dlen))
1360 continue;
1361 write_int(f_out, plen + len + dlen);
417b5999 1362 if (plen)
fdc79501
WD
1363 write_buf(f_out, p, plen);
1364 write_buf(f_out, ent->pattern, len);
417b5999 1365 if (dlen)
fdc79501 1366 write_byte(f_out, '/');
0f2ac855 1367 }
ed243f8c 1368 flp->tail = prev;
c627d613
AT
1369}
1370
fdc79501
WD
1371/* This is only called by the client. */
1372void send_filter_list(int f_out)
1373{
d5782b52
WD
1374 int receiver_wants_list = prune_empty_dirs
1375 || (delete_mode && (!delete_excluded || protocol_version >= 29));
448797a1
WD
1376
1377 if (local_server || (am_sender && !receiver_wants_list))
fdc79501
WD
1378 f_out = -1;
1379 if (cvs_exclude && am_sender) {
1380 if (protocol_version >= 29)
ebaa4296
MM
1381 load_filter_str(&filter_list, ":C", filter_template(0), 0);
1382 load_filter_str(&filter_list, "-C", filter_template(0), 0);
fdc79501 1383 }
c627d613 1384
ed243f8c
WD
1385 send_rules(f_out, &filter_list);
1386
1387 if (f_out >= 0)
fdc79501 1388 write_int(f_out, 0);
fdc79501
WD
1389
1390 if (cvs_exclude) {
1391 if (!am_sender || protocol_version < 29)
ebaa4296 1392 load_filter_str(&filter_list, ":C", filter_template(0), 0);
fdc79501 1393 if (!am_sender)
ebaa4296 1394 load_filter_str(&filter_list, "-C", filter_template(0), 0);
fdc79501
WD
1395 }
1396}
1397
1398/* This is only called by the server. */
1399void recv_filter_list(int f_in)
c627d613 1400{
3fac51e2 1401 char line[BIGPATHBUFLEN];
fdc79501 1402 int xflags = protocol_version >= 29 ? 0 : XFLG_OLD_PREFIXES;
d5782b52 1403 int receiver_wants_list = prune_empty_dirs
e257c6c2
WD
1404 || (delete_mode
1405 && (!delete_excluded || protocol_version >= 29));
fdc79501
WD
1406 unsigned int len;
1407
448797a1 1408 if (!local_server && (am_sender || receiver_wants_list)) {
fdc79501
WD
1409 while ((len = read_int(f_in)) != 0) {
1410 if (len >= sizeof line)
a1f99493 1411 overflow_exit("recv_rules");
fdc79501 1412 read_sbuf(f_in, line, len);
ebaa4296 1413 load_filter_str(&filter_list, line, filter_template(0), xflags);
fdc79501
WD
1414 }
1415 }
9dd891bb 1416
fdc79501 1417 if (cvs_exclude) {
1412da7c 1418 if (local_server || am_sender || protocol_version < 29)
ebaa4296 1419 load_filter_str(&filter_list, ":C", filter_template(0), 0);
1412da7c 1420 if (local_server || am_sender)
ebaa4296 1421 load_filter_str(&filter_list, "-C", filter_template(0), 0);
651443a7 1422 }
ed243f8c
WD
1423
1424 if (local_server) /* filter out any rules that aren't for us. */
1425 send_rules(-1, &filter_list);
651443a7 1426}