1 The changes to generator.c were greatly simplified, making the patch
2 easier to maintain and fixing the failing test in the testsuite.
5 Be sure to run "make proto" before "make".
7 --- orig/generator.c 2005-01-01 21:11:00
8 +++ generator.c 2004-11-27 18:12:57
9 @@ -44,6 +44,7 @@ extern int size_only;
10 extern OFF_T max_size;
11 extern int io_timeout;
12 extern int protocol_version;
13 +extern int fuzzy_basis;
14 extern int always_checksum;
15 extern char *partial_dir;
16 extern char *basis_dir[];
17 @@ -239,6 +240,92 @@ static void generate_and_send_sums(int f
21 +static void split_names(char *fname, char **dirname, char **basename)
23 + char *slash = strrchr(fname, '/');
27 + *basename = slash+1;
35 +static unsigned int measure_name(const char *name, const char *basename,
38 + int namelen = strlen(name);
39 + int extlen = strlen(ext);
40 + unsigned int score = 0;
42 + /* Extensions must match */
43 + if (namelen <= extlen || strcmp(name + namelen - extlen, ext) != 0)
46 + /* Now score depends on similarity of prefix */
47 + for (; *name == *basename && *name; name++, basename++)
53 +static int find_fuzzy(const char *fname, char *buf, STRUCT_STAT *st_ptr)
57 + char *basename, *dirname;
58 + char mangled_name[MAXPATHLEN];
59 + char bestname[MAXPATHLEN];
60 + unsigned int bestscore = 0;
63 + strlcpy(mangled_name, fname, sizeof mangled_name);
65 + split_names(mangled_name, &dirname, &basename);
66 + if (!(d = opendir(dirname))) {
67 + rsyserr(FERROR, errno, "recv_generator opendir(%s)", dirname);
71 + /* Get final extension, eg. .gz; never full basename though. */
72 + if (!(ext = strrchr(basename + 1, '.')))
73 + ext = basename + strlen(basename); /* ext = "" */
75 + while ((di = readdir(d)) != NULL) {
76 + const char *dname = d_name(di);
79 + if (dname[0] == '.' && (dname[1] == '\0'
80 + || (dname[1] == '.' && dname[2] == '\0')))
83 + score = measure_name(dname, basename, ext);
85 + rprintf(FINFO, "[%s] fuzzy score for %s = %u\n",
86 + who_am_i(), dname, score);
88 + if (score > bestscore) {
89 + strlcpy(bestname, dname, sizeof bestname);
95 + /* Found a candidate. */
96 + if (bestscore != 0) {
97 + pathjoin(buf, MAXPATHLEN, dirname, bestname);
99 + rprintf(FINFO, "[%s] fuzzy match %s->%s\n",
100 + who_am_i(), fname, buf);
102 + return link_stat(buf, st_ptr, 0);
109 * Acts on file number @p i from @p flist, whose name is @p fname.
110 @@ -493,6 +580,15 @@ static void recv_generator(char *fname,
114 + if (statret == -1 && fuzzy_basis) {
115 + if (find_fuzzy(fname, fnamecmpbuf, &st) == 0
116 + && S_ISREG(st.st_mode)) {
118 + fnamecmp = fnamecmpbuf;
119 + fnamecmp_type = FNAMECMP_FUZZY;
124 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
126 @@ -521,6 +617,8 @@ static void recv_generator(char *fname,
128 if ((link_dest || copy_dest) && fnamecmp_type != FNAMECMP_FNAME)
130 + else if (fnamecmp_type == FNAMECMP_FUZZY)
132 else if (unchanged_file(fnamecmp, file, &st)) {
133 if (fnamecmp_type == FNAMECMP_FNAME)
134 set_perms(fname, file, &st, PERMS_REPORT);
135 @@ -592,8 +690,24 @@ prepare_to_open:
139 - if (f_out_name >= 0)
140 + if (f_out_name >= 0) {
141 write_byte(f_out_name, fnamecmp_type);
142 + if (fnamecmp_type == FNAMECMP_FUZZY) {
143 + uchar lenbuf[3], *lb = lenbuf;
144 + int len = strlen(fnamecmpbuf);
146 +#if MAXPATHLEN > 0x7FFF
147 + *lb++ = len / 0x10000 + 0x80;
148 + *lb++ = len / 0x100;
150 + *lb++ = len / 0x100 + 0x80;
154 + write_buf(f_out_name, lenbuf, lb - lenbuf + 1);
155 + write_buf(f_out_name, fnamecmpbuf, len);
159 if (dry_run || read_batch)
161 --- orig/main.c 2005-01-01 21:11:00
162 +++ main.c 2004-11-27 18:13:51
163 @@ -48,6 +48,7 @@ extern int keep_dirlinks;
164 extern int preserve_hard_links;
165 extern int protocol_version;
167 +extern int fuzzy_basis;
168 extern int relative_paths;
169 extern int rsync_port;
170 extern int whole_file;
171 @@ -463,7 +464,7 @@ static int do_recv(int f_in,int f_out,st
174 int error_pipe[2], name_pipe[2];
175 - BOOL need_name_pipe = basis_dir[0] && !dry_run;
176 + BOOL need_name_pipe = (basis_dir[0] || fuzzy_basis) && !dry_run;
178 /* The receiving side mustn't obey this, or an existing symlink that
179 * points to an identical file won't be replaced by the referent. */
180 --- orig/options.c 2005-01-01 21:11:00
181 +++ options.c 2004-11-29 01:36:48
182 @@ -86,6 +86,7 @@ int copy_unsafe_links = 0;
184 int daemon_bwlimit = 0;
186 +int fuzzy_basis = 0;
187 size_t bwlimit_writemax = 0;
188 int delete_after = 0;
189 int only_existing = 0;
190 @@ -288,6 +289,7 @@ void usage(enum logcode F)
191 rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n");
192 rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n");
193 rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n");
194 + rprintf(F," --fuzzy use similar file as basis if basis doesn't exist\n");
195 rprintf(F," -P equivalent to --partial --progress\n");
196 rprintf(F," -z, --compress compress file data\n");
197 rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n");
198 @@ -384,6 +386,7 @@ static struct poptOption long_options[]
199 {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
200 {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
201 {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 },
202 + {"fuzzy", 0, POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 },
203 /* TODO: Should this take an optional int giving the compression level? */
204 {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 },
205 {"stats", 0, POPT_ARG_NONE, &do_stats, 0, 0, 0 },
206 @@ -958,10 +961,10 @@ int parse_arguments(int *argc, const cha
207 am_server ? "server" : "client");
210 - if (compare_dest || copy_dest || link_dest) {
211 + if (dest_option || fuzzy_basis) {
212 snprintf(err_buf, sizeof err_buf,
213 "--inplace does not yet work with %s\n",
215 + dest_option ? dest_option : "--fuzzy");
219 @@ -1240,6 +1243,9 @@ void server_options(char **args,int *arg
220 args[ac++] = "--no-relative";
223 + if (fuzzy_basis && am_sender)
224 + args[ac++] = "--fuzzy";
229 --- orig/receiver.c 2005-01-01 21:11:00
230 +++ receiver.c 2004-11-27 18:15:01
231 @@ -323,6 +323,27 @@ static int receive_data(int f_in, char *
235 +static void read_gen_name(int fd, char *buf)
237 + int len = read_byte(fd);
239 +#if MAXPATHLEN > 32767
241 + read_buf(fd, (char *)lenbuf, 2);
242 + len = (len & ~0x80) * 0x10000 + lenbuf[0] * 0x100 + lenbuf[1];
244 + len = (len & ~0x80) * 0x100 + read_byte(fd);
247 + if (len >= MAXPATHLEN) {
248 + rprintf(FERROR, "bogus data on generator name pipe\n");
249 + exit_cleanup(RERR_PROTOCOL);
252 + read_sbuf(fd, buf, len);
256 static void discard_receive_data(int f_in, OFF_T length)
258 receive_data(f_in, NULL, -1, 0, NULL, -1, length);
259 @@ -453,6 +474,10 @@ int recv_files(int f_in, struct file_lis
260 case FNAMECMP_BACKUP:
261 fnamecmp = get_backup_name(fname);
263 + case FNAMECMP_FUZZY:
264 + read_gen_name(f_in_name, fnamecmpbuf);
265 + fnamecmp = fnamecmpbuf;
267 case FNAMECMP_BASIS_DIR:
269 pathjoin(fnamecmpbuf, sizeof fnamecmpbuf,
270 --- orig/rsync.h 2005-01-01 21:11:01
271 +++ rsync.h 2004-11-03 22:53:09
273 #define FNAMECMP_FNAME 0x80
274 #define FNAMECMP_PARTIAL_DIR 0x81
275 #define FNAMECMP_BACKUP 0x82
276 +#define FNAMECMP_FUZZY 0x83
279 /* Log-message categories. FLOG is only used on the daemon side to
280 --- orig/rsync.yo 2005-01-01 21:11:01
281 +++ rsync.yo 2004-11-27 18:15:22
282 @@ -358,6 +358,7 @@ verb(
283 --compare-dest=DIR also compare received files relative to DIR
284 --copy-dest=DIR ... and include copies of unchanged files
285 --link-dest=DIR hardlink to files in DIR when unchanged
286 + --fuzzy use similar file as basis if basis is gone
287 -P equivalent to --partial --progress
288 -z, --compress compress file data
289 -C, --cvs-exclude auto ignore files in the same way CVS does