Fixed a failing hunk.
[rsync/rsync-patches.git] / fuzzy.diff
CommitLineData
824abc86 1The changes to generator.c were greatly simplified, making the patch
8c5b8235 2easier to maintain and fixing the failing test in the testsuite.
0808daa5 3Lightly tested.
241013b4 4
824abc86
WD
5Be sure to run "make proto" before "make".
6
3eabe3a3 7--- orig/generator.c 2005-01-14 18:30:18
0808daa5
WD
8+++ generator.c 2004-11-27 18:12:57
9@@ -44,6 +44,7 @@ extern int size_only;
10 extern OFF_T max_size;
58118c25
WD
11 extern int io_timeout;
12 extern int protocol_version;
09fb2223 13+extern int fuzzy_basis;
58118c25 14 extern int always_checksum;
b952a177 15 extern char *partial_dir;
0808daa5 16 extern char *basis_dir[];
37da98ae 17@@ -239,6 +240,92 @@ static void generate_and_send_sums(int f
58118c25
WD
18 }
19
20
47dd7a31
WD
21+static void split_names(char *fname, char **dirname, char **basename)
22+{
23+ char *slash = strrchr(fname, '/');
24+ if (slash) {
25+ *dirname = fname;
26+ *slash = '\0';
27+ *basename = slash+1;
28+ } else {
29+ *basename = fname;
30+ *dirname = ".";
31+ }
32+}
33+
58118c25 34+
47dd7a31
WD
35+static unsigned int measure_name(const char *name, const char *basename,
36+ const char *ext)
37+{
38+ int namelen = strlen(name);
39+ int extlen = strlen(ext);
40+ unsigned int score = 0;
41+
42+ /* Extensions must match */
43+ if (namelen <= extlen || strcmp(name + namelen - extlen, ext) != 0)
44+ return 0;
45+
46+ /* Now score depends on similarity of prefix */
47+ for (; *name == *basename && *name; name++, basename++)
48+ score++;
49+ return score;
50+}
51+
58118c25 52+
09fb2223 53+static int find_fuzzy(const char *fname, char *buf, STRUCT_STAT *st_ptr)
47dd7a31
WD
54+{
55+ DIR *d;
56+ struct dirent *di;
57+ char *basename, *dirname;
58+ char mangled_name[MAXPATHLEN];
59+ char bestname[MAXPATHLEN];
60+ unsigned int bestscore = 0;
61+ const char *ext;
62+
09fb2223 63+ strlcpy(mangled_name, fname, sizeof mangled_name);
47dd7a31
WD
64+
65+ split_names(mangled_name, &dirname, &basename);
66+ if (!(d = opendir(dirname))) {
67+ rsyserr(FERROR, errno, "recv_generator opendir(%s)", dirname);
68+ return -1;
69+ }
70+
71+ /* Get final extension, eg. .gz; never full basename though. */
dd39e657 72+ if (!(ext = strrchr(basename + 1, '.')))
47dd7a31
WD
73+ ext = basename + strlen(basename); /* ext = "" */
74+
75+ while ((di = readdir(d)) != NULL) {
76+ const char *dname = d_name(di);
77+ unsigned int score;
78+
79+ if (dname[0] == '.' && (dname[1] == '\0'
80+ || (dname[1] == '.' && dname[2] == '\0')))
81+ continue;
82+
83+ score = measure_name(dname, basename, ext);
84+ if (verbose > 4) {
8c5b8235
WD
85+ rprintf(FINFO, "[%s] fuzzy score for %s = %u\n",
86+ who_am_i(), dname, score);
47dd7a31
WD
87+ }
88+ if (score > bestscore) {
8c5b8235 89+ strlcpy(bestname, dname, sizeof bestname);
47dd7a31
WD
90+ bestscore = score;
91+ }
92+ }
93+ closedir(d);
94+
95+ /* Found a candidate. */
96+ if (bestscore != 0) {
8c5b8235 97+ pathjoin(buf, MAXPATHLEN, dirname, bestname);
47dd7a31 98+ if (verbose > 2) {
8c5b8235 99+ rprintf(FINFO, "[%s] fuzzy match %s->%s\n",
09fb2223 100+ who_am_i(), fname, buf);
47dd7a31 101+ }
58118c25 102+ return link_stat(buf, st_ptr, 0);
47dd7a31
WD
103+ }
104+ return -1;
105+}
58118c25
WD
106+
107
108 /*
109 * Acts on file number @p i from @p flist, whose name is @p fname.
37da98ae 110@@ -493,6 +580,15 @@ static void recv_generator(char *fname,
f48a237e
WD
111 } else
112 partialptr = NULL;
824abc86 113
09fb2223
WD
114+ if (statret == -1 && fuzzy_basis) {
115+ if (find_fuzzy(fname, fnamecmpbuf, &st) == 0
116+ && S_ISREG(st.st_mode)) {
117+ statret = 0;
118+ fnamecmp = fnamecmpbuf;
0edc7d7f 119+ fnamecmp_type = FNAMECMP_FUZZY;
09fb2223 120+ }
824abc86
WD
121+ }
122+
09fb2223
WD
123 if (statret == -1) {
124 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
7628f156 125 return;
37da98ae 126@@ -521,6 +617,8 @@ static void recv_generator(char *fname,
241013b4 127
0808daa5
WD
128 if ((link_dest || copy_dest) && fnamecmp_type != FNAMECMP_FNAME)
129 ;
130+ else if (fnamecmp_type == FNAMECMP_FUZZY)
131+ ;
132 else if (unchanged_file(fnamecmp, file, &st)) {
0edc7d7f 133 if (fnamecmp_type == FNAMECMP_FNAME)
8c5b8235 134 set_perms(fname, file, &st, PERMS_REPORT);
3eabe3a3 135@@ -593,8 +691,24 @@ prepare_to_open:
0edc7d7f 136
09fb2223 137 notify_others:
0edc7d7f
WD
138 write_int(f_out, i);
139- if (f_out_name >= 0)
140+ if (f_out_name >= 0) {
09fb2223 141 write_byte(f_out_name, fnamecmp_type);
0edc7d7f 142+ if (fnamecmp_type == FNAMECMP_FUZZY) {
09fb2223
WD
143+ uchar lenbuf[3], *lb = lenbuf;
144+ int len = strlen(fnamecmpbuf);
145+ if (len > 0x7F) {
146+#if MAXPATHLEN > 0x7FFF
147+ *lb++ = len / 0x10000 + 0x80;
148+ *lb++ = len / 0x100;
149+#else
150+ *lb++ = len / 0x100 + 0x80;
151+#endif
152+ }
153+ *lb = len;
154+ write_buf(f_out_name, lenbuf, lb - lenbuf + 1);
155+ write_buf(f_out_name, fnamecmpbuf, len);
156+ }
0edc7d7f 157+ }
09fb2223 158
0edc7d7f
WD
159 if (dry_run || read_batch)
160 return;
3eabe3a3
WD
161--- orig/main.c 2005-01-14 18:30:18
162+++ main.c 2005-01-14 18:33:15
d5753a22 163@@ -48,6 +48,7 @@ extern int keep_dirlinks;
495f1899
WD
164 extern int preserve_hard_links;
165 extern int protocol_version;
166 extern int recurse;
09fb2223 167+extern int fuzzy_basis;
495f1899
WD
168 extern int relative_paths;
169 extern int rsync_port;
170 extern int whole_file;
3eabe3a3 171@@ -464,7 +465,8 @@ static int do_recv(int f_in,int f_out,st
495f1899
WD
172 int pid;
173 int status = 0;
174 int error_pipe[2], name_pipe[2];
3eabe3a3
WD
175- BOOL need_name_pipe = (basis_dir[0] || partial_dir) && !dry_run;
176+ BOOL need_name_pipe = (basis_dir[0] || partial_dir || fuzzy_basis)
177+ && !dry_run;
495f1899 178
d5753a22
WD
179 /* The receiving side mustn't obey this, or an existing symlink that
180 * points to an identical file won't be replaced by the referent. */
37da98ae 181--- orig/options.c 2005-01-01 21:11:00
89072d7e 182+++ options.c 2004-11-29 01:36:48
0808daa5 183@@ -86,6 +86,7 @@ int copy_unsafe_links = 0;
f6c3b300 184 int size_only = 0;
0808daa5 185 int daemon_bwlimit = 0;
f6c3b300 186 int bwlimit = 0;
09fb2223 187+int fuzzy_basis = 0;
f6c3b300
WD
188 size_t bwlimit_writemax = 0;
189 int delete_after = 0;
190 int only_existing = 0;
0808daa5 191@@ -288,6 +289,7 @@ void usage(enum logcode F)
f0533c4c 192 rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n");
0808daa5
WD
193 rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n");
194 rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n");
f0533c4c
WD
195+ rprintf(F," --fuzzy use similar file as basis if basis doesn't exist\n");
196 rprintf(F," -P equivalent to --partial --progress\n");
197 rprintf(F," -z, --compress compress file data\n");
198 rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n");
37da98ae 199@@ -384,6 +386,7 @@ static struct poptOption long_options[]
0808daa5
WD
200 {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
201 {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
202 {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 },
09fb2223 203+ {"fuzzy", 0, POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 },
f0533c4c
WD
204 /* TODO: Should this take an optional int giving the compression level? */
205 {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 },
5388f859 206 {"stats", 0, POPT_ARG_NONE, &do_stats, 0, 0, 0 },
37da98ae 207@@ -958,10 +961,10 @@ int parse_arguments(int *argc, const cha
89072d7e
WD
208 am_server ? "server" : "client");
209 return 0;
210 #endif
211- if (compare_dest || copy_dest || link_dest) {
212+ if (dest_option || fuzzy_basis) {
213 snprintf(err_buf, sizeof err_buf,
214 "--inplace does not yet work with %s\n",
215- dest_option);
216+ dest_option ? dest_option : "--fuzzy");
dd39e657
WD
217 return 0;
218 }
dd39e657 219 } else {
37da98ae
WD
220@@ -1240,6 +1243,9 @@ void server_options(char **args,int *arg
221 args[ac++] = "--no-relative";
241013b4 222 }
7b675ff5 223
09fb2223 224+ if (fuzzy_basis && am_sender)
241013b4 225+ args[ac++] = "--fuzzy";
7b675ff5 226+
241013b4 227 *argc = ac;
f74d2272 228 return;
7b675ff5 229
3eabe3a3 230--- orig/receiver.c 2005-01-10 09:46:01
0808daa5 231+++ receiver.c 2004-11-27 18:15:01
37da98ae 232@@ -323,6 +323,27 @@ static int receive_data(int f_in, char *
09fb2223
WD
233 }
234
235
236+static void read_gen_name(int fd, char *buf)
237+{
238+ int len = read_byte(fd);
239+ if (len & 0x80) {
240+#if MAXPATHLEN > 32767
241+ uchar lenbuf[2];
242+ read_buf(fd, (char *)lenbuf, 2);
243+ len = (len & ~0x80) * 0x10000 + lenbuf[0] * 0x100 + lenbuf[1];
244+#else
245+ len = (len & ~0x80) * 0x100 + read_byte(fd);
246+#endif
247+ }
248+ if (len >= MAXPATHLEN) {
249+ rprintf(FERROR, "bogus data on generator name pipe\n");
250+ exit_cleanup(RERR_PROTOCOL);
251+ }
252+
253+ read_sbuf(fd, buf, len);
254+}
255+
256+
257 static void discard_receive_data(int f_in, OFF_T length)
258 {
259 receive_data(f_in, NULL, -1, 0, NULL, -1, length);
37da98ae 260@@ -453,6 +474,10 @@ int recv_files(int f_in, struct file_lis
0edc7d7f 261 case FNAMECMP_BACKUP:
f48a237e 262 fnamecmp = get_backup_name(fname);
09fb2223 263 break;
0edc7d7f 264+ case FNAMECMP_FUZZY:
09fb2223
WD
265+ read_gen_name(f_in_name, fnamecmpbuf);
266+ fnamecmp = fnamecmpbuf;
267+ break;
0808daa5 268 case FNAMECMP_BASIS_DIR:
09fb2223
WD
269 default:
270 pathjoin(fnamecmpbuf, sizeof fnamecmpbuf,
3eabe3a3 271--- orig/rsync.h 2005-01-10 00:21:12
0edc7d7f 272+++ rsync.h 2004-11-03 22:53:09
0808daa5 273@@ -125,6 +125,7 @@
0edc7d7f
WD
274 #define FNAMECMP_FNAME 0x80
275 #define FNAMECMP_PARTIAL_DIR 0x81
276 #define FNAMECMP_BACKUP 0x82
277+#define FNAMECMP_FUZZY 0x83
09fb2223
WD
278
279
280 /* Log-message categories. FLOG is only used on the daemon side to
37da98ae 281--- orig/rsync.yo 2005-01-01 21:11:01
0808daa5
WD
282+++ rsync.yo 2004-11-27 18:15:22
283@@ -358,6 +358,7 @@ verb(
f0533c4c 284 --compare-dest=DIR also compare received files relative to DIR
0808daa5
WD
285 --copy-dest=DIR ... and include copies of unchanged files
286 --link-dest=DIR hardlink to files in DIR when unchanged
f0533c4c
WD
287+ --fuzzy use similar file as basis if basis is gone
288 -P equivalent to --partial --progress
289 -z, --compress compress file data
290 -C, --cvs-exclude auto ignore files in the same way CVS does