Commit | Line | Data |
---|---|---|
824abc86 | 1 | The changes to generator.c were greatly simplified, making the patch |
8c5b8235 | 2 | easier to maintain and fixing the failing test in the testsuite. |
0808daa5 | 3 | Lightly tested. |
241013b4 | 4 | |
824abc86 WD |
5 | Be sure to run "make proto" before "make". |
6 | ||
be73a66e | 7 | --- orig/generator.c 2005-01-23 07:55:40 |
a27b0830 | 8 | +++ generator.c 2005-01-19 18:39:15 |
be73a66e | 9 | @@ -45,6 +45,7 @@ extern int size_only; |
0808daa5 | 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[]; |
be73a66e | 17 | @@ -225,6 +226,88 @@ static void generate_and_send_sums(int f |
4370504a | 18 | unmap_file(mapbuf); |
58118c25 WD |
19 | } |
20 | ||
fc82f579 WD |
21 | +/* Try to find a filename in the same dir as "fname" with a similar name. |
22 | + * | |
23 | + * TODO: | |
4370504a | 24 | + * - We should be using a cache of names for the current dir, not |
fc82f579 | 25 | + * re-reading the destination directory for every file. |
4370504a WD |
26 | + * - We must not return an rsync tempfile from the current transfer. |
27 | + * - If the highest-rated name is not a normal file, we should fall- | |
fc82f579 | 28 | + * back to the next highest-rated file. |
4370504a | 29 | + * - We must not return a destination file that is being updated |
fc82f579 WD |
30 | + * during the current transfer, even if we already processed it |
31 | + * (since the receiver may not be done with it yet). | |
4370504a | 32 | + * - We must weed out any names that a daemon's config has excluded. |
fc82f579 | 33 | + */ |
09fb2223 | 34 | +static int find_fuzzy(const char *fname, char *buf, STRUCT_STAT *st_ptr) |
47dd7a31 WD |
35 | +{ |
36 | + DIR *d; | |
37 | + struct dirent *di; | |
0f626034 | 38 | + char *basename, *dirname, *slash; |
47dd7a31 | 39 | + char bestname[MAXPATHLEN]; |
4cd1daea | 40 | + int suf_len, basename_len; |
054f3f90 | 41 | + uint32 lowest_dist = 0x7FFFFFFF; |
4cd1daea | 42 | + const char *suf; |
47dd7a31 | 43 | + |
0f626034 WD |
44 | + strlcpy(buf, fname, MAXPATHLEN); |
45 | + if ((slash = strrchr(buf, '/')) != NULL) { | |
46 | + dirname = buf; | |
47 | + *slash = '\0'; | |
48 | + basename = slash + 1; | |
49 | + } else { | |
50 | + basename = buf; | |
51 | + dirname = "."; | |
52 | + } | |
054f3f90 | 53 | + basename_len = strlen(basename); |
47dd7a31 | 54 | + |
47dd7a31 WD |
55 | + if (!(d = opendir(dirname))) { |
56 | + rsyserr(FERROR, errno, "recv_generator opendir(%s)", dirname); | |
57 | + return -1; | |
58 | + } | |
e55625fb WD |
59 | + if (slash) |
60 | + *slash = '/'; | |
47dd7a31 | 61 | + |
a27b0830 | 62 | + suf = find_filename_suffix(basename, basename_len, &suf_len); |
47dd7a31 | 63 | + |
054f3f90 | 64 | + bestname[0] = '\0'; |
47dd7a31 | 65 | + while ((di = readdir(d)) != NULL) { |
4cd1daea | 66 | + const char *dname_suf, *dname = d_name(di); |
054f3f90 | 67 | + uint32 dist; |
4cd1daea | 68 | + int dname_len, dname_suf_len; |
47dd7a31 WD |
69 | + |
70 | + if (dname[0] == '.' && (dname[1] == '\0' | |
71 | + || (dname[1] == '.' && dname[2] == '\0'))) | |
72 | + continue; | |
73 | + | |
a27b0830 WD |
74 | + dname_len = strlen(dname); |
75 | + dname_suf = find_filename_suffix(dname, dname_len, &dname_suf_len); | |
054f3f90 WD |
76 | + |
77 | + dist = fuzzy_distance(dname, dname_len, basename, basename_len); | |
fc82f579 | 78 | + /* Add some extra weight to how well the suffixes match. */ |
4cd1daea WD |
79 | + dist += fuzzy_distance(dname_suf, dname_suf_len, suf, suf_len) * 10; |
80 | + if (verbose > 4) { | |
81 | + rprintf(FINFO, "fuzzy distance for %s = %d (%d)\n", | |
82 | + dname, (int)(dist>>16), (int)(dist&0xFFFF)); | |
47dd7a31 | 83 | + } |
fc82f579 | 84 | + if (dist <= lowest_dist) { |
8c5b8235 | 85 | + strlcpy(bestname, dname, sizeof bestname); |
054f3f90 | 86 | + lowest_dist = dist; |
47dd7a31 WD |
87 | + } |
88 | + } | |
89 | + closedir(d); | |
90 | + | |
91 | + /* Found a candidate. */ | |
054f3f90 | 92 | + if (bestname[0] != '\0') { |
e55625fb | 93 | + strlcpy(basename, bestname, MAXPATHLEN - (basename - buf)); |
054f3f90 WD |
94 | + if (verbose > 2) { |
95 | + rprintf(FINFO, "fuzzy match %s->%s\n", | |
96 | + safe_fname(fname), buf); | |
97 | + } | |
58118c25 | 98 | + return link_stat(buf, st_ptr, 0); |
47dd7a31 WD |
99 | + } |
100 | + return -1; | |
101 | +} | |
58118c25 WD |
102 | + |
103 | ||
104 | /* | |
105 | * Acts on file number @p i from @p flist, whose name is @p fname. | |
be73a66e | 106 | @@ -479,6 +562,15 @@ static void recv_generator(char *fname, |
f48a237e WD |
107 | } else |
108 | partialptr = NULL; | |
824abc86 | 109 | |
09fb2223 WD |
110 | + if (statret == -1 && fuzzy_basis) { |
111 | + if (find_fuzzy(fname, fnamecmpbuf, &st) == 0 | |
112 | + && S_ISREG(st.st_mode)) { | |
113 | + statret = 0; | |
114 | + fnamecmp = fnamecmpbuf; | |
0edc7d7f | 115 | + fnamecmp_type = FNAMECMP_FUZZY; |
09fb2223 | 116 | + } |
824abc86 WD |
117 | + } |
118 | + | |
09fb2223 WD |
119 | if (statret == -1) { |
120 | if (preserve_hard_links && hard_link_check(file, HL_SKIP)) | |
7628f156 | 121 | return; |
be73a66e | 122 | @@ -507,6 +599,8 @@ static void recv_generator(char *fname, |
241013b4 | 123 | |
9cf86680 | 124 | if (!compare_dest && fnamecmp_type <= FNAMECMP_BASIS_DIR_HIGH) |
0808daa5 WD |
125 | ; |
126 | + else if (fnamecmp_type == FNAMECMP_FUZZY) | |
127 | + ; | |
128 | else if (unchanged_file(fnamecmp, file, &st)) { | |
0edc7d7f | 129 | if (fnamecmp_type == FNAMECMP_FNAME) |
8c5b8235 | 130 | set_perms(fname, file, &st, PERMS_REPORT); |
be73a66e | 131 | @@ -581,8 +675,24 @@ notify_others: |
0edc7d7f | 132 | write_int(f_out, i); |
0f626034 WD |
133 | if (protocol_version >= 29 && inplace && !read_batch) |
134 | write_byte(f_out, fnamecmp_type); | |
0edc7d7f WD |
135 | - if (f_out_name >= 0) |
136 | + if (f_out_name >= 0) { | |
09fb2223 | 137 | write_byte(f_out_name, fnamecmp_type); |
0edc7d7f | 138 | + if (fnamecmp_type == FNAMECMP_FUZZY) { |
09fb2223 WD |
139 | + uchar lenbuf[3], *lb = lenbuf; |
140 | + int len = strlen(fnamecmpbuf); | |
141 | + if (len > 0x7F) { | |
142 | +#if MAXPATHLEN > 0x7FFF | |
143 | + *lb++ = len / 0x10000 + 0x80; | |
144 | + *lb++ = len / 0x100; | |
145 | +#else | |
146 | + *lb++ = len / 0x100 + 0x80; | |
147 | +#endif | |
148 | + } | |
149 | + *lb = len; | |
150 | + write_buf(f_out_name, lenbuf, lb - lenbuf + 1); | |
151 | + write_buf(f_out_name, fnamecmpbuf, len); | |
152 | + } | |
0edc7d7f | 153 | + } |
09fb2223 | 154 | |
0edc7d7f WD |
155 | if (dry_run || read_batch) |
156 | return; | |
be73a66e | 157 | --- orig/main.c 2005-01-23 01:48:52 |
3eabe3a3 | 158 | +++ main.c 2005-01-14 18:33:15 |
4370504a | 159 | @@ -49,6 +49,7 @@ extern int keep_dirlinks; |
495f1899 WD |
160 | extern int preserve_hard_links; |
161 | extern int protocol_version; | |
162 | extern int recurse; | |
09fb2223 | 163 | +extern int fuzzy_basis; |
495f1899 WD |
164 | extern int relative_paths; |
165 | extern int rsync_port; | |
166 | extern int whole_file; | |
4370504a | 167 | @@ -465,7 +466,8 @@ static int do_recv(int f_in,int f_out,st |
495f1899 WD |
168 | int pid; |
169 | int status = 0; | |
170 | int error_pipe[2], name_pipe[2]; | |
3eabe3a3 WD |
171 | - BOOL need_name_pipe = (basis_dir[0] || partial_dir) && !dry_run; |
172 | + BOOL need_name_pipe = (basis_dir[0] || partial_dir || fuzzy_basis) | |
173 | + && !dry_run; | |
495f1899 | 174 | |
d5753a22 WD |
175 | /* The receiving side mustn't obey this, or an existing symlink that |
176 | * points to an identical file won't be replaced by the referent. */ | |
be73a66e | 177 | --- orig/options.c 2005-01-23 01:53:35 |
9cf86680 | 178 | +++ options.c 2005-01-15 21:08:13 |
be73a66e | 179 | @@ -90,6 +90,7 @@ int copy_unsafe_links = 0; |
f6c3b300 | 180 | int size_only = 0; |
0808daa5 | 181 | int daemon_bwlimit = 0; |
f6c3b300 | 182 | int bwlimit = 0; |
09fb2223 | 183 | +int fuzzy_basis = 0; |
f6c3b300 | 184 | size_t bwlimit_writemax = 0; |
f6c3b300 | 185 | int only_existing = 0; |
be73a66e WD |
186 | int opt_ignore_existing = 0; |
187 | @@ -293,6 +294,7 @@ void usage(enum logcode F) | |
f0533c4c | 188 | rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n"); |
0808daa5 WD |
189 | rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n"); |
190 | rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n"); | |
dc800efa | 191 | + rprintf(F," --fuzzy find similar file for basis when no dest file\n"); |
f0533c4c WD |
192 | rprintf(F," -P equivalent to --partial --progress\n"); |
193 | rprintf(F," -z, --compress compress file data\n"); | |
194 | rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n"); | |
be73a66e | 195 | @@ -393,6 +395,7 @@ static struct poptOption long_options[] |
0808daa5 WD |
196 | {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 }, |
197 | {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 }, | |
198 | {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 }, | |
09fb2223 | 199 | + {"fuzzy", 0, POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 }, |
f0533c4c WD |
200 | /* TODO: Should this take an optional int giving the compression level? */ |
201 | {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 }, | |
5388f859 | 202 | {"stats", 0, POPT_ARG_NONE, &do_stats, 0, 0, 0 }, |
be73a66e | 203 | @@ -1251,6 +1254,9 @@ void server_options(char **args,int *arg |
4370504a WD |
204 | if (!implied_dirs && !am_sender) |
205 | args[ac++] = "--no-implied-dirs"; | |
7b675ff5 | 206 | |
09fb2223 | 207 | + if (fuzzy_basis && am_sender) |
241013b4 | 208 | + args[ac++] = "--fuzzy"; |
7b675ff5 | 209 | + |
241013b4 | 210 | *argc = ac; |
f74d2272 | 211 | return; |
7b675ff5 | 212 | |
be73a66e | 213 | --- orig/receiver.c 2005-01-23 07:55:58 |
9cf86680 | 214 | +++ receiver.c 2005-01-15 21:21:02 |
be73a66e | 215 | @@ -251,6 +251,27 @@ static int receive_data(int f_in, char * |
09fb2223 WD |
216 | } |
217 | ||
218 | ||
219 | +static void read_gen_name(int fd, char *buf) | |
220 | +{ | |
221 | + int len = read_byte(fd); | |
222 | + if (len & 0x80) { | |
223 | +#if MAXPATHLEN > 32767 | |
224 | + uchar lenbuf[2]; | |
225 | + read_buf(fd, (char *)lenbuf, 2); | |
226 | + len = (len & ~0x80) * 0x10000 + lenbuf[0] * 0x100 + lenbuf[1]; | |
227 | +#else | |
228 | + len = (len & ~0x80) * 0x100 + read_byte(fd); | |
229 | +#endif | |
230 | + } | |
231 | + if (len >= MAXPATHLEN) { | |
232 | + rprintf(FERROR, "bogus data on generator name pipe\n"); | |
233 | + exit_cleanup(RERR_PROTOCOL); | |
234 | + } | |
235 | + | |
236 | + read_sbuf(fd, buf, len); | |
237 | +} | |
238 | + | |
239 | + | |
240 | static void discard_receive_data(int f_in, OFF_T length) | |
241 | { | |
242 | receive_data(f_in, NULL, -1, 0, NULL, -1, length); | |
be73a66e | 243 | @@ -381,6 +402,10 @@ int recv_files(int f_in, struct file_lis |
0edc7d7f | 244 | case FNAMECMP_BACKUP: |
f48a237e | 245 | fnamecmp = get_backup_name(fname); |
09fb2223 | 246 | break; |
0edc7d7f | 247 | + case FNAMECMP_FUZZY: |
09fb2223 WD |
248 | + read_gen_name(f_in_name, fnamecmpbuf); |
249 | + fnamecmp = fnamecmpbuf; | |
250 | + break; | |
251 | default: | |
9cf86680 WD |
252 | if (j >= basis_dir_cnt) { |
253 | rprintf(FERROR, | |
be73a66e | 254 | --- orig/rsync.h 2005-01-21 00:35:26 |
a27b0830 | 255 | +++ rsync.h 2005-01-19 18:36:47 |
0f626034 | 256 | @@ -128,6 +128,7 @@ |
0edc7d7f WD |
257 | #define FNAMECMP_FNAME 0x80 |
258 | #define FNAMECMP_PARTIAL_DIR 0x81 | |
259 | #define FNAMECMP_BACKUP 0x82 | |
260 | +#define FNAMECMP_FUZZY 0x83 | |
09fb2223 | 261 | |
a27b0830 | 262 | /* For calling delete_file() */ |
be73a66e WD |
263 | #define DEL_DIR (1<<0) |
264 | --- orig/rsync.yo 2005-01-23 02:57:49 | |
637c560e | 265 | +++ rsync.yo 2005-01-15 21:48:52 |
be73a66e | 266 | @@ -360,6 +360,7 @@ verb( |
f0533c4c | 267 | --compare-dest=DIR also compare received files relative to DIR |
0808daa5 WD |
268 | --copy-dest=DIR ... and include copies of unchanged files |
269 | --link-dest=DIR hardlink to files in DIR when unchanged | |
dc800efa | 270 | + --fuzzy find similar file for basis when no dest |
f0533c4c WD |
271 | -P equivalent to --partial --progress |
272 | -z, --compress compress file data | |
273 | -C, --cvs-exclude auto ignore files in the same way CVS does | |
be73a66e | 274 | @@ -896,6 +897,14 @@ Note that rsync versions prior to 2.6.1 |
0f626034 WD |
275 | (or implied by -a). You can work-around this bug by avoiding the -o option |
276 | when sending to an old rsync. | |
637c560e WD |
277 | |
278 | +dit(bf(--fuzzy)) This option tells rsync that it should look around for a | |
279 | +basis file for any destination file that is missing. The current algorithm | |
280 | +looks for a similarly-named file in the same directory as the destination | |
fc82f579 WD |
281 | +file, and, if found, uses that to try to speed up the transfer. Note that |
282 | +the use of the --delete option might get rid of any potential fuzzy-match | |
283 | +files, so either use --delete-after or filename exclusions if you need to | |
284 | +prevent this. | |
637c560e WD |
285 | + |
286 | dit(bf(-z, --compress)) With this option, rsync compresses any data from | |
287 | the files that it sends to the destination machine. This | |
288 | option is useful on slow connections. The compression method used is the | |
be73a66e | 289 | --- orig/util.c 2005-01-20 23:05:34 |
a27b0830 WD |
290 | +++ util.c 2005-01-19 17:30:51 |
291 | @@ -1213,3 +1213,108 @@ void *_realloc_array(void *ptr, unsigned | |
054f3f90 WD |
292 | return malloc(size * num); |
293 | return realloc(ptr, size * num); | |
294 | } | |
295 | + | |
4cd1daea WD |
296 | +/* Take a filename and filename length and return the most significant |
297 | + * filename suffix we can find. This ignores suffixes such as "~", | |
298 | + * ".bak", ".orig", ".~1~", etc. */ | |
a27b0830 | 299 | +const char *find_filename_suffix(const char *fn, int fn_len, int *len_ptr) |
4cd1daea WD |
300 | +{ |
301 | + const char *suf, *s; | |
4cd1daea | 302 | + BOOL had_tilde; |
a27b0830 | 303 | + int s_len; |
4cd1daea WD |
304 | + |
305 | + /* One or more dots at the start aren't a suffix. */ | |
306 | + while (fn_len && *fn == '.') fn++, fn_len--; | |
307 | + | |
308 | + /* Ignore the ~ in a "foo~" filename. */ | |
309 | + if (fn_len > 1 && fn[fn_len-1] == '~') | |
310 | + fn_len--, had_tilde = True; | |
311 | + else | |
312 | + had_tilde = False; | |
313 | + | |
314 | + /* Assume we don't find an suffix. */ | |
315 | + suf = ""; | |
316 | + *len_ptr = 0; | |
317 | + | |
318 | + /* Find the last significant suffix. */ | |
319 | + for (s = fn + fn_len - 1; fn_len > 1; ) { | |
320 | + while (*s != '.' && s != fn) s--; | |
321 | + if (s == fn) | |
322 | + break; | |
323 | + s_len = fn_len - (s - fn); | |
324 | + fn_len = s - fn; | |
325 | + if (s_len == 3) { | |
326 | + if (strcmp(s+1, "bak") == 0 | |
327 | + || strcmp(s+1, "old") == 0) | |
328 | + continue; | |
329 | + } else if (s_len == 4) { | |
330 | + if (strcmp(s+1, "orig") == 0) | |
331 | + continue; | |
332 | + } else if (s_len > 2 && had_tilde | |
333 | + && s[1] == '~' && isdigit(s[2])) | |
334 | + continue; | |
335 | + *len_ptr = s_len; | |
336 | + suf = s; | |
337 | + /* Determine if the suffix is all digits. */ | |
338 | + for (s++, s_len--; s_len > 0; s++, s_len--) { | |
339 | + if (!isdigit(*s)) | |
340 | + return suf; | |
341 | + } | |
342 | + /* An all-digit suffix may not be that signficant. */ | |
343 | + continue; | |
344 | + } | |
345 | + | |
346 | + return suf; | |
347 | +} | |
348 | + | |
054f3f90 WD |
349 | +/* This is an implementation of the Levenshtein distance algorithm. It |
350 | + * was implemented to avoid needing a two-dimensional matrix (to save | |
351 | + * memory). It was also tweaked to try to factor in the ASCII distance | |
352 | + * between changed characters as a minor distance quantity. The normal | |
353 | + * Levenshtein units of distance (each signifying a single change between | |
354 | + * the two strings) are defined as a "UNIT". */ | |
355 | + | |
356 | +#define UNIT (1 << 16) | |
357 | + | |
358 | +uint32 fuzzy_distance(const char *s1, int len1, const char *s2, int len2) | |
359 | +{ | |
360 | + uint32 a[MAXPATHLEN], diag, above, left, diag_inc, above_inc, left_inc; | |
361 | + int32 cost; | |
362 | + int i1, i2; | |
363 | + | |
4cd1daea WD |
364 | + if (!len1 || !len2) { |
365 | + if (!len1) { | |
366 | + s1 = s2; | |
367 | + len1 = len2; | |
368 | + } | |
369 | + for (i1 = 0, cost = 0; i1 < len1; i1++) | |
370 | + cost += s1[i1]; | |
371 | + return (int32)len1 * UNIT + cost; | |
372 | + } | |
054f3f90 WD |
373 | + |
374 | + for (i2 = 0; i2 < len2; i2++) | |
375 | + a[i2] = (i2+1) * UNIT; | |
376 | + | |
377 | + for (i1 = 0; i1 < len1; i1++) { | |
378 | + diag = i1 * UNIT; | |
379 | + above = (i1+1) * UNIT; | |
380 | + for (i2 = 0; i2 < len2; i2++) { | |
381 | + left = a[i2]; | |
382 | + if ((cost = *((uchar*)s1+i1) - *((uchar*)s2+i2)) != 0) { | |
383 | + if (cost < 0) | |
384 | + cost = UNIT - cost; | |
385 | + else | |
386 | + cost = UNIT + cost; | |
387 | + } | |
388 | + diag_inc = diag + cost; | |
389 | + left_inc = left + UNIT + *((uchar*)s1+i1); | |
390 | + above_inc = above + UNIT + *((uchar*)s2+i2); | |
391 | + a[i2] = above = left < above | |
392 | + ? (left_inc < diag_inc ? left_inc : diag_inc) | |
393 | + : (above_inc < diag_inc ? above_inc : diag_inc); | |
394 | + diag = left; | |
395 | + } | |
396 | + } | |
397 | + | |
398 | + return a[len2-1]; | |
399 | +} |