Need to make sure that the destination file doesn't exist before we
[rsync/rsync-patches.git] / fuzzy.diff
CommitLineData
824abc86 1Depends-On-Patch: g2r-basis-filename.diff
8c5b8235 2
824abc86 3The changes to generator.c were greatly simplified, making the patch
8c5b8235 4easier to maintain and fixing the failing test in the testsuite.
58118c25 5Very lightly tested.
241013b4 6
824abc86
WD
7Be sure to run "make proto" before "make".
8
a6587818 9--- orig/generator.c 2004-09-20 19:57:58
dd39e657 10+++ generator.c 2004-09-23 15:33:08
f48a237e 11@@ -43,6 +43,7 @@ extern int ignore_times;
58118c25
WD
12 extern int size_only;
13 extern int io_timeout;
14 extern int protocol_version;
09fb2223 15+extern int fuzzy_basis;
58118c25 16 extern int always_checksum;
b952a177 17 extern char *partial_dir;
58118c25 18 extern char *compare_dest;
dd39e657 19@@ -244,6 +245,92 @@ static void generate_and_send_sums(int f
58118c25
WD
20 }
21
22
47dd7a31
WD
23+static void split_names(char *fname, char **dirname, char **basename)
24+{
25+ char *slash = strrchr(fname, '/');
26+ if (slash) {
27+ *dirname = fname;
28+ *slash = '\0';
29+ *basename = slash+1;
30+ } else {
31+ *basename = fname;
32+ *dirname = ".";
33+ }
34+}
35+
58118c25 36+
47dd7a31
WD
37+static unsigned int measure_name(const char *name, const char *basename,
38+ const char *ext)
39+{
40+ int namelen = strlen(name);
41+ int extlen = strlen(ext);
42+ unsigned int score = 0;
43+
44+ /* Extensions must match */
45+ if (namelen <= extlen || strcmp(name + namelen - extlen, ext) != 0)
46+ return 0;
47+
48+ /* Now score depends on similarity of prefix */
49+ for (; *name == *basename && *name; name++, basename++)
50+ score++;
51+ return score;
52+}
53+
58118c25 54+
09fb2223 55+static int find_fuzzy(const char *fname, char *buf, STRUCT_STAT *st_ptr)
47dd7a31
WD
56+{
57+ DIR *d;
58+ struct dirent *di;
59+ char *basename, *dirname;
60+ char mangled_name[MAXPATHLEN];
61+ char bestname[MAXPATHLEN];
62+ unsigned int bestscore = 0;
63+ const char *ext;
64+
09fb2223 65+ strlcpy(mangled_name, fname, sizeof mangled_name);
47dd7a31
WD
66+
67+ split_names(mangled_name, &dirname, &basename);
68+ if (!(d = opendir(dirname))) {
69+ rsyserr(FERROR, errno, "recv_generator opendir(%s)", dirname);
70+ return -1;
71+ }
72+
73+ /* Get final extension, eg. .gz; never full basename though. */
dd39e657 74+ if (!(ext = strrchr(basename + 1, '.')))
47dd7a31
WD
75+ ext = basename + strlen(basename); /* ext = "" */
76+
77+ while ((di = readdir(d)) != NULL) {
78+ const char *dname = d_name(di);
79+ unsigned int score;
80+
81+ if (dname[0] == '.' && (dname[1] == '\0'
82+ || (dname[1] == '.' && dname[2] == '\0')))
83+ continue;
84+
85+ score = measure_name(dname, basename, ext);
86+ if (verbose > 4) {
8c5b8235
WD
87+ rprintf(FINFO, "[%s] fuzzy score for %s = %u\n",
88+ who_am_i(), dname, score);
47dd7a31
WD
89+ }
90+ if (score > bestscore) {
8c5b8235 91+ strlcpy(bestname, dname, sizeof bestname);
47dd7a31
WD
92+ bestscore = score;
93+ }
94+ }
95+ closedir(d);
96+
97+ /* Found a candidate. */
98+ if (bestscore != 0) {
8c5b8235 99+ pathjoin(buf, MAXPATHLEN, dirname, bestname);
47dd7a31 100+ if (verbose > 2) {
8c5b8235 101+ rprintf(FINFO, "[%s] fuzzy match %s->%s\n",
09fb2223 102+ who_am_i(), fname, buf);
47dd7a31 103+ }
58118c25 104+ return link_stat(buf, st_ptr, 0);
47dd7a31
WD
105+ }
106+ return -1;
107+}
58118c25
WD
108+
109
110 /*
111 * Acts on file number @p i from @p flist, whose name is @p fname.
dd39e657
WD
112@@ -256,11 +343,11 @@ static void generate_and_send_sums(int f
113 static void recv_generator(char *fname, struct file_struct *file, int i,
114 int f_out, int f_out_name)
115 {
116- int fd = -1, f_copy;
117+ int fd = -1, f_copy = -1;
118 STRUCT_STAT st, partial_st;
119- struct file_struct *back_file;
120+ struct file_struct *back_file = NULL;
121 int statret, stat_errno;
122- char *fnamecmp, *partialptr, *backupptr;
123+ char *fnamecmp, *partialptr, *backupptr = NULL;
124 char fnamecmpbuf[MAXPATHLEN];
125 uchar fnamecmp_type;
126
127@@ -468,6 +555,15 @@ static void recv_generator(char *fname,
f48a237e
WD
128 } else
129 partialptr = NULL;
824abc86 130
09fb2223
WD
131+ if (statret == -1 && fuzzy_basis) {
132+ if (find_fuzzy(fname, fnamecmpbuf, &st) == 0
133+ && S_ISREG(st.st_mode)) {
134+ statret = 0;
135+ fnamecmp = fnamecmpbuf;
136+ fnamecmp_type = G2R_FUZZY;
137+ }
824abc86
WD
138+ }
139+
09fb2223
WD
140 if (statret == -1) {
141 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
7628f156 142 return;
dd39e657 143@@ -494,7 +590,7 @@ static void recv_generator(char *fname,
241013b4
MP
144 return;
145 }
146
ec8473c6 147- if (skip_file(fnamecmp, file, &st)) {
09fb2223
WD
148+ if (fnamecmp_type != G2R_FUZZY && skip_file(fnamecmp, file, &st)) {
149 if (fnamecmp_type == G2R_FNAME)
8c5b8235
WD
150 set_perms(fname, file, &st, PERMS_REPORT);
151 return;
dd39e657
WD
152@@ -553,10 +649,6 @@ prepare_to_open:
153 return;
154 }
155 fnamecmp_type = G2R_BACKUP;
156- } else {
157- backupptr = NULL;
158- back_file = NULL;
159- f_copy = -1;
160 }
161
162 if (verbose > 3) {
163@@ -570,6 +662,21 @@ prepare_to_open:
09fb2223
WD
164 notify_others:
165 if (f_out_name >= 0) {
166 write_byte(f_out_name, fnamecmp_type);
167+ if (fnamecmp_type == G2R_FUZZY) {
168+ uchar lenbuf[3], *lb = lenbuf;
169+ int len = strlen(fnamecmpbuf);
170+ if (len > 0x7F) {
171+#if MAXPATHLEN > 0x7FFF
172+ *lb++ = len / 0x10000 + 0x80;
173+ *lb++ = len / 0x100;
174+#else
175+ *lb++ = len / 0x100 + 0x80;
176+#endif
177+ }
178+ *lb = len;
179+ write_buf(f_out_name, lenbuf, lb - lenbuf + 1);
180+ write_buf(f_out_name, fnamecmpbuf, len);
181+ }
182 io_flush(NORMAL_FLUSH); /* XXX make this more efficient! */
183 }
184
2f5fa77e
WD
185--- orig/main.c 2004-07-22 00:10:43
186+++ main.c 2004-07-22 00:32:31
d5753a22 187@@ -48,6 +48,7 @@ extern int keep_dirlinks;
495f1899
WD
188 extern int preserve_hard_links;
189 extern int protocol_version;
190 extern int recurse;
09fb2223 191+extern int fuzzy_basis;
495f1899
WD
192 extern int relative_paths;
193 extern int rsync_port;
194 extern int whole_file;
f635ed27 195@@ -463,7 +464,7 @@ static int do_recv(int f_in,int f_out,st
495f1899
WD
196 int pid;
197 int status = 0;
198 int error_pipe[2], name_pipe[2];
c2530f70 199- BOOL need_name_pipe = compare_dest && !dry_run;
09fb2223 200+ BOOL need_name_pipe = (compare_dest || fuzzy_basis) && !dry_run;
495f1899 201
d5753a22
WD
202 /* The receiving side mustn't obey this, or an existing symlink that
203 * points to an identical file won't be replaced by the referent. */
dd39e657
WD
204--- orig/options.c 2004-09-22 08:47:31
205+++ options.c 2004-09-23 14:59:46
f635ed27 206@@ -85,6 +85,7 @@ int safe_symlinks = 0;
f6c3b300
WD
207 int copy_unsafe_links = 0;
208 int size_only = 0;
209 int bwlimit = 0;
09fb2223 210+int fuzzy_basis = 0;
f6c3b300
WD
211 size_t bwlimit_writemax = 0;
212 int delete_after = 0;
213 int only_existing = 0;
f635ed27 214@@ -279,6 +280,7 @@ void usage(enum logcode F)
c2530f70 215 rprintf(F," -T, --temp-dir=DIR create temporary files in directory DIR\n");
f0533c4c
WD
216 rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n");
217 rprintf(F," --link-dest=DIR create hardlinks to DIR for unchanged files\n");
218+ rprintf(F," --fuzzy use similar file as basis if basis doesn't exist\n");
219 rprintf(F," -P equivalent to --partial --progress\n");
220 rprintf(F," -z, --compress compress file data\n");
221 rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n");
f635ed27 222@@ -378,6 +380,7 @@ static struct poptOption long_options[]
f0533c4c
WD
223 {"temp-dir", 'T', POPT_ARG_STRING, &tmpdir, 0, 0, 0 },
224 {"compare-dest", 0, POPT_ARG_STRING, &compare_dest, 0, 0, 0 },
225 {"link-dest", 0, POPT_ARG_STRING, &compare_dest, OPT_LINK_DEST, 0, 0 },
09fb2223 226+ {"fuzzy", 0, POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 },
f0533c4c
WD
227 /* TODO: Should this take an optional int giving the compression level? */
228 {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 },
229 {"daemon", 0, POPT_ARG_NONE, &daemon_opt, 0, 0, 0 },
dd39e657
WD
230@@ -833,6 +836,11 @@ int parse_arguments(int *argc, const cha
231 link_dest ? "--link-dest" : "--compare-dest");
232 return 0;
233 }
234+ if (compare_dest) {
235+ snprintf(err_buf, sizeof err_buf,
236+ "--inplace does not yet work with --fuzzy\n");
237+ return 0;
238+ }
239 } else {
240 if (keep_partial && !partial_dir)
241 partial_dir = getenv("RSYNC_PARTIAL_DIR");
242@@ -1099,6 +1107,9 @@ void server_options(char **args,int *arg
f74d2272 243 }
241013b4 244 }
7b675ff5 245
09fb2223 246+ if (fuzzy_basis && am_sender)
241013b4 247+ args[ac++] = "--fuzzy";
7b675ff5 248+
241013b4 249 *argc = ac;
f74d2272 250 return;
7b675ff5 251
f48a237e 252--- orig/receiver.c 2004-09-07 21:57:20
09fb2223 253+++ receiver.c 2004-07-30 18:21:38
a6587818 254@@ -319,6 +319,27 @@ static int receive_data(int f_in, char *
09fb2223
WD
255 }
256
257
258+static void read_gen_name(int fd, char *buf)
259+{
260+ int len = read_byte(fd);
261+ if (len & 0x80) {
262+#if MAXPATHLEN > 32767
263+ uchar lenbuf[2];
264+ read_buf(fd, (char *)lenbuf, 2);
265+ len = (len & ~0x80) * 0x10000 + lenbuf[0] * 0x100 + lenbuf[1];
266+#else
267+ len = (len & ~0x80) * 0x100 + read_byte(fd);
268+#endif
269+ }
270+ if (len >= MAXPATHLEN) {
271+ rprintf(FERROR, "bogus data on generator name pipe\n");
272+ exit_cleanup(RERR_PROTOCOL);
273+ }
274+
275+ read_sbuf(fd, buf, len);
276+}
277+
278+
279 static void discard_receive_data(int f_in, OFF_T length)
280 {
281 receive_data(f_in, NULL, -1, 0, NULL, -1, length);
a6587818 282@@ -448,6 +469,10 @@ int recv_files(int f_in, struct file_lis
f48a237e
WD
283 case G2R_BACKUP:
284 fnamecmp = get_backup_name(fname);
09fb2223
WD
285 break;
286+ case G2R_FUZZY:
287+ read_gen_name(f_in_name, fnamecmpbuf);
288+ fnamecmp = fnamecmpbuf;
289+ break;
290 default:
291 pathjoin(fnamecmpbuf, sizeof fnamecmpbuf,
292 compare_dest, fname);
f48a237e
WD
293--- orig/rsync.h 2004-09-07 21:52:22
294+++ rsync.h 2004-09-07 22:02:37
dd39e657 295@@ -122,6 +122,7 @@
09fb2223
WD
296 #define G2R_FNAME 0x80
297 #define G2R_PARTIAL_DIR 0x81
f48a237e
WD
298 #define G2R_BACKUP 0x82
299+#define G2R_FUZZY 0x83
09fb2223
WD
300
301
302 /* Log-message categories. FLOG is only used on the daemon side to
dd39e657 303--- orig/rsync.yo 2004-09-22 08:47:31
824abc86 304+++ rsync.yo 2004-07-03 19:27:25
67c08134 305@@ -356,6 +356,7 @@ verb(
f0533c4c
WD
306 -T --temp-dir=DIR create temporary files in directory DIR
307 --compare-dest=DIR also compare received files relative to DIR
308 --link-dest=DIR create hardlinks to DIR for unchanged files
309+ --fuzzy use similar file as basis if basis is gone
310 -P equivalent to --partial --progress
311 -z, --compress compress file data
312 -C, --cvs-exclude auto ignore files in the same way CVS does