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 | ||
0f626034 | 7 | --- orig/generator.c 2005-01-17 23:11:45 |
4cd1daea | 8 | +++ generator.c 2005-01-18 19:25:55 |
0808daa5 WD |
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[]; |
4cd1daea | 17 | @@ -242,6 +243,76 @@ static void generate_and_send_sums(int f |
58118c25 WD |
18 | } |
19 | ||
20 | ||
09fb2223 | 21 | +static int find_fuzzy(const char *fname, char *buf, STRUCT_STAT *st_ptr) |
47dd7a31 WD |
22 | +{ |
23 | + DIR *d; | |
24 | + struct dirent *di; | |
0f626034 | 25 | + char *basename, *dirname, *slash; |
47dd7a31 | 26 | + char bestname[MAXPATHLEN]; |
4cd1daea | 27 | + int suf_len, basename_len; |
054f3f90 | 28 | + uint32 lowest_dist = 0x7FFFFFFF; |
4cd1daea | 29 | + const char *suf; |
47dd7a31 | 30 | + |
0f626034 WD |
31 | + strlcpy(buf, fname, MAXPATHLEN); |
32 | + if ((slash = strrchr(buf, '/')) != NULL) { | |
33 | + dirname = buf; | |
34 | + *slash = '\0'; | |
35 | + basename = slash + 1; | |
36 | + } else { | |
37 | + basename = buf; | |
38 | + dirname = "."; | |
39 | + } | |
054f3f90 | 40 | + basename_len = strlen(basename); |
47dd7a31 | 41 | + |
47dd7a31 WD |
42 | + if (!(d = opendir(dirname))) { |
43 | + rsyserr(FERROR, errno, "recv_generator opendir(%s)", dirname); | |
44 | + return -1; | |
45 | + } | |
e55625fb WD |
46 | + if (slash) |
47 | + *slash = '/'; | |
47dd7a31 | 48 | + |
4cd1daea WD |
49 | + suf_len = basename_len; |
50 | + suf = find_filename_suffix(basename, &suf_len); | |
47dd7a31 | 51 | + |
054f3f90 | 52 | + bestname[0] = '\0'; |
47dd7a31 | 53 | + while ((di = readdir(d)) != NULL) { |
4cd1daea | 54 | + const char *dname_suf, *dname = d_name(di); |
054f3f90 | 55 | + uint32 dist; |
4cd1daea | 56 | + int dname_len, dname_suf_len; |
47dd7a31 WD |
57 | + |
58 | + if (dname[0] == '.' && (dname[1] == '\0' | |
59 | + || (dname[1] == '.' && dname[2] == '\0'))) | |
60 | + continue; | |
61 | + | |
4cd1daea WD |
62 | + dname_len = dname_suf_len = strlen(dname); |
63 | + dname_suf = find_filename_suffix(dname, &dname_suf_len); | |
054f3f90 WD |
64 | + |
65 | + dist = fuzzy_distance(dname, dname_len, basename, basename_len); | |
4cd1daea WD |
66 | + /* Add some extra weight to how well the suffixes matched. */ |
67 | + dist += fuzzy_distance(dname_suf, dname_suf_len, suf, suf_len) * 10; | |
68 | + if (verbose > 4) { | |
69 | + rprintf(FINFO, "fuzzy distance for %s = %d (%d)\n", | |
70 | + dname, (int)(dist>>16), (int)(dist&0xFFFF)); | |
47dd7a31 | 71 | + } |
054f3f90 | 72 | + if (dist < lowest_dist) { |
8c5b8235 | 73 | + strlcpy(bestname, dname, sizeof bestname); |
054f3f90 | 74 | + lowest_dist = dist; |
47dd7a31 WD |
75 | + } |
76 | + } | |
77 | + closedir(d); | |
78 | + | |
79 | + /* Found a candidate. */ | |
054f3f90 | 80 | + if (bestname[0] != '\0') { |
e55625fb | 81 | + strlcpy(basename, bestname, MAXPATHLEN - (basename - buf)); |
054f3f90 WD |
82 | + if (verbose > 2) { |
83 | + rprintf(FINFO, "fuzzy match %s->%s\n", | |
84 | + safe_fname(fname), buf); | |
85 | + } | |
58118c25 | 86 | + return link_stat(buf, st_ptr, 0); |
47dd7a31 WD |
87 | + } |
88 | + return -1; | |
89 | +} | |
58118c25 WD |
90 | + |
91 | ||
92 | /* | |
93 | * Acts on file number @p i from @p flist, whose name is @p fname. | |
4cd1daea | 94 | @@ -496,6 +567,15 @@ static void recv_generator(char *fname, |
f48a237e WD |
95 | } else |
96 | partialptr = NULL; | |
824abc86 | 97 | |
09fb2223 WD |
98 | + if (statret == -1 && fuzzy_basis) { |
99 | + if (find_fuzzy(fname, fnamecmpbuf, &st) == 0 | |
100 | + && S_ISREG(st.st_mode)) { | |
101 | + statret = 0; | |
102 | + fnamecmp = fnamecmpbuf; | |
0edc7d7f | 103 | + fnamecmp_type = FNAMECMP_FUZZY; |
09fb2223 | 104 | + } |
824abc86 WD |
105 | + } |
106 | + | |
09fb2223 WD |
107 | if (statret == -1) { |
108 | if (preserve_hard_links && hard_link_check(file, HL_SKIP)) | |
7628f156 | 109 | return; |
4cd1daea | 110 | @@ -524,6 +604,8 @@ static void recv_generator(char *fname, |
241013b4 | 111 | |
9cf86680 | 112 | if (!compare_dest && fnamecmp_type <= FNAMECMP_BASIS_DIR_HIGH) |
0808daa5 WD |
113 | ; |
114 | + else if (fnamecmp_type == FNAMECMP_FUZZY) | |
115 | + ; | |
116 | else if (unchanged_file(fnamecmp, file, &st)) { | |
0edc7d7f | 117 | if (fnamecmp_type == FNAMECMP_FNAME) |
8c5b8235 | 118 | set_perms(fname, file, &st, PERMS_REPORT); |
4cd1daea | 119 | @@ -598,8 +680,24 @@ notify_others: |
0edc7d7f | 120 | write_int(f_out, i); |
0f626034 WD |
121 | if (protocol_version >= 29 && inplace && !read_batch) |
122 | write_byte(f_out, fnamecmp_type); | |
0edc7d7f WD |
123 | - if (f_out_name >= 0) |
124 | + if (f_out_name >= 0) { | |
09fb2223 | 125 | write_byte(f_out_name, fnamecmp_type); |
0edc7d7f | 126 | + if (fnamecmp_type == FNAMECMP_FUZZY) { |
09fb2223 WD |
127 | + uchar lenbuf[3], *lb = lenbuf; |
128 | + int len = strlen(fnamecmpbuf); | |
129 | + if (len > 0x7F) { | |
130 | +#if MAXPATHLEN > 0x7FFF | |
131 | + *lb++ = len / 0x10000 + 0x80; | |
132 | + *lb++ = len / 0x100; | |
133 | +#else | |
134 | + *lb++ = len / 0x100 + 0x80; | |
135 | +#endif | |
136 | + } | |
137 | + *lb = len; | |
138 | + write_buf(f_out_name, lenbuf, lb - lenbuf + 1); | |
139 | + write_buf(f_out_name, fnamecmpbuf, len); | |
140 | + } | |
0edc7d7f | 141 | + } |
09fb2223 | 142 | |
0edc7d7f WD |
143 | if (dry_run || read_batch) |
144 | return; | |
0f626034 | 145 | --- orig/main.c 2005-01-17 23:11:45 |
3eabe3a3 | 146 | +++ main.c 2005-01-14 18:33:15 |
d5753a22 | 147 | @@ -48,6 +48,7 @@ extern int keep_dirlinks; |
495f1899 WD |
148 | extern int preserve_hard_links; |
149 | extern int protocol_version; | |
150 | extern int recurse; | |
09fb2223 | 151 | +extern int fuzzy_basis; |
495f1899 WD |
152 | extern int relative_paths; |
153 | extern int rsync_port; | |
154 | extern int whole_file; | |
3eabe3a3 | 155 | @@ -464,7 +465,8 @@ static int do_recv(int f_in,int f_out,st |
495f1899 WD |
156 | int pid; |
157 | int status = 0; | |
158 | int error_pipe[2], name_pipe[2]; | |
3eabe3a3 WD |
159 | - BOOL need_name_pipe = (basis_dir[0] || partial_dir) && !dry_run; |
160 | + BOOL need_name_pipe = (basis_dir[0] || partial_dir || fuzzy_basis) | |
161 | + && !dry_run; | |
495f1899 | 162 | |
d5753a22 WD |
163 | /* The receiving side mustn't obey this, or an existing symlink that |
164 | * points to an identical file won't be replaced by the referent. */ | |
0f626034 | 165 | --- orig/options.c 2005-01-17 23:11:45 |
9cf86680 | 166 | +++ options.c 2005-01-15 21:08:13 |
0808daa5 | 167 | @@ -86,6 +86,7 @@ int copy_unsafe_links = 0; |
f6c3b300 | 168 | int size_only = 0; |
0808daa5 | 169 | int daemon_bwlimit = 0; |
f6c3b300 | 170 | int bwlimit = 0; |
09fb2223 | 171 | +int fuzzy_basis = 0; |
f6c3b300 WD |
172 | size_t bwlimit_writemax = 0; |
173 | int delete_after = 0; | |
174 | int only_existing = 0; | |
0808daa5 | 175 | @@ -288,6 +289,7 @@ void usage(enum logcode F) |
f0533c4c | 176 | rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n"); |
0808daa5 WD |
177 | rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n"); |
178 | rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n"); | |
dc800efa | 179 | + rprintf(F," --fuzzy find similar file for basis when no dest file\n"); |
f0533c4c WD |
180 | rprintf(F," -P equivalent to --partial --progress\n"); |
181 | rprintf(F," -z, --compress compress file data\n"); | |
182 | rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n"); | |
37da98ae | 183 | @@ -384,6 +386,7 @@ static struct poptOption long_options[] |
0808daa5 WD |
184 | {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 }, |
185 | {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 }, | |
186 | {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 }, | |
09fb2223 | 187 | + {"fuzzy", 0, POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 }, |
f0533c4c WD |
188 | /* TODO: Should this take an optional int giving the compression level? */ |
189 | {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 }, | |
5388f859 | 190 | {"stats", 0, POPT_ARG_NONE, &do_stats, 0, 0, 0 }, |
0f626034 | 191 | @@ -1234,6 +1237,9 @@ void server_options(char **args,int *arg |
37da98ae | 192 | args[ac++] = "--no-relative"; |
241013b4 | 193 | } |
7b675ff5 | 194 | |
09fb2223 | 195 | + if (fuzzy_basis && am_sender) |
241013b4 | 196 | + args[ac++] = "--fuzzy"; |
7b675ff5 | 197 | + |
241013b4 | 198 | *argc = ac; |
f74d2272 | 199 | return; |
7b675ff5 | 200 | |
0f626034 | 201 | --- orig/receiver.c 2005-01-17 23:11:45 |
9cf86680 WD |
202 | +++ receiver.c 2005-01-15 21:21:02 |
203 | @@ -324,6 +324,27 @@ static int receive_data(int f_in, char * | |
09fb2223 WD |
204 | } |
205 | ||
206 | ||
207 | +static void read_gen_name(int fd, char *buf) | |
208 | +{ | |
209 | + int len = read_byte(fd); | |
210 | + if (len & 0x80) { | |
211 | +#if MAXPATHLEN > 32767 | |
212 | + uchar lenbuf[2]; | |
213 | + read_buf(fd, (char *)lenbuf, 2); | |
214 | + len = (len & ~0x80) * 0x10000 + lenbuf[0] * 0x100 + lenbuf[1]; | |
215 | +#else | |
216 | + len = (len & ~0x80) * 0x100 + read_byte(fd); | |
217 | +#endif | |
218 | + } | |
219 | + if (len >= MAXPATHLEN) { | |
220 | + rprintf(FERROR, "bogus data on generator name pipe\n"); | |
221 | + exit_cleanup(RERR_PROTOCOL); | |
222 | + } | |
223 | + | |
224 | + read_sbuf(fd, buf, len); | |
225 | +} | |
226 | + | |
227 | + | |
228 | static void discard_receive_data(int f_in, OFF_T length) | |
229 | { | |
230 | receive_data(f_in, NULL, -1, 0, NULL, -1, length); | |
9cf86680 | 231 | @@ -454,6 +475,10 @@ int recv_files(int f_in, struct file_lis |
0edc7d7f | 232 | case FNAMECMP_BACKUP: |
f48a237e | 233 | fnamecmp = get_backup_name(fname); |
09fb2223 | 234 | break; |
0edc7d7f | 235 | + case FNAMECMP_FUZZY: |
09fb2223 WD |
236 | + read_gen_name(f_in_name, fnamecmpbuf); |
237 | + fnamecmp = fnamecmpbuf; | |
238 | + break; | |
239 | default: | |
9cf86680 WD |
240 | if (j >= basis_dir_cnt) { |
241 | rprintf(FERROR, | |
0f626034 | 242 | --- orig/rsync.h 2005-01-17 23:11:45 |
9cf86680 | 243 | +++ rsync.h 2005-01-15 21:24:09 |
0f626034 | 244 | @@ -128,6 +128,7 @@ |
0edc7d7f WD |
245 | #define FNAMECMP_FNAME 0x80 |
246 | #define FNAMECMP_PARTIAL_DIR 0x81 | |
247 | #define FNAMECMP_BACKUP 0x82 | |
248 | +#define FNAMECMP_FUZZY 0x83 | |
09fb2223 WD |
249 | |
250 | ||
251 | /* Log-message categories. FLOG is only used on the daemon side to | |
0f626034 | 252 | --- orig/rsync.yo 2005-01-17 23:11:46 |
637c560e | 253 | +++ rsync.yo 2005-01-15 21:48:52 |
0808daa5 | 254 | @@ -358,6 +358,7 @@ verb( |
f0533c4c | 255 | --compare-dest=DIR also compare received files relative to DIR |
0808daa5 WD |
256 | --copy-dest=DIR ... and include copies of unchanged files |
257 | --link-dest=DIR hardlink to files in DIR when unchanged | |
dc800efa | 258 | + --fuzzy find similar file for basis when no dest |
f0533c4c WD |
259 | -P equivalent to --partial --progress |
260 | -z, --compress compress file data | |
261 | -C, --cvs-exclude auto ignore files in the same way CVS does | |
0f626034 WD |
262 | @@ -878,6 +879,11 @@ Note that rsync versions prior to 2.6.1 |
263 | (or implied by -a). You can work-around this bug by avoiding the -o option | |
264 | when sending to an old rsync. | |
637c560e WD |
265 | |
266 | +dit(bf(--fuzzy)) This option tells rsync that it should look around for a | |
267 | +basis file for any destination file that is missing. The current algorithm | |
268 | +looks for a similarly-named file in the same directory as the destination | |
269 | +file, and, if found, uses that to try to speed up the transfer. | |
270 | + | |
271 | dit(bf(-z, --compress)) With this option, rsync compresses any data from | |
272 | the files that it sends to the destination machine. This | |
273 | option is useful on slow connections. The compression method used is the | |
054f3f90 | 274 | --- orig/util.c 2004-09-07 21:45:30 |
4cd1daea WD |
275 | +++ util.c 2005-01-18 19:25:47 |
276 | @@ -1217,3 +1217,108 @@ void *_realloc_array(void *ptr, unsigned | |
054f3f90 WD |
277 | return malloc(size * num); |
278 | return realloc(ptr, size * num); | |
279 | } | |
280 | + | |
4cd1daea WD |
281 | +/* Take a filename and filename length and return the most significant |
282 | + * filename suffix we can find. This ignores suffixes such as "~", | |
283 | + * ".bak", ".orig", ".~1~", etc. */ | |
284 | +const char *find_filename_suffix(const char *fn, int *len_ptr) | |
285 | +{ | |
286 | + const char *suf, *s; | |
287 | + int s_len, fn_len = *len_ptr; | |
288 | + BOOL had_tilde; | |
289 | + | |
290 | + /* One or more dots at the start aren't a suffix. */ | |
291 | + while (fn_len && *fn == '.') fn++, fn_len--; | |
292 | + | |
293 | + /* Ignore the ~ in a "foo~" filename. */ | |
294 | + if (fn_len > 1 && fn[fn_len-1] == '~') | |
295 | + fn_len--, had_tilde = True; | |
296 | + else | |
297 | + had_tilde = False; | |
298 | + | |
299 | + /* Assume we don't find an suffix. */ | |
300 | + suf = ""; | |
301 | + *len_ptr = 0; | |
302 | + | |
303 | + /* Find the last significant suffix. */ | |
304 | + for (s = fn + fn_len - 1; fn_len > 1; ) { | |
305 | + while (*s != '.' && s != fn) s--; | |
306 | + if (s == fn) | |
307 | + break; | |
308 | + s_len = fn_len - (s - fn); | |
309 | + fn_len = s - fn; | |
310 | + if (s_len == 3) { | |
311 | + if (strcmp(s+1, "bak") == 0 | |
312 | + || strcmp(s+1, "old") == 0) | |
313 | + continue; | |
314 | + } else if (s_len == 4) { | |
315 | + if (strcmp(s+1, "orig") == 0) | |
316 | + continue; | |
317 | + } else if (s_len > 2 && had_tilde | |
318 | + && s[1] == '~' && isdigit(s[2])) | |
319 | + continue; | |
320 | + *len_ptr = s_len; | |
321 | + suf = s; | |
322 | + /* Determine if the suffix is all digits. */ | |
323 | + for (s++, s_len--; s_len > 0; s++, s_len--) { | |
324 | + if (!isdigit(*s)) | |
325 | + return suf; | |
326 | + } | |
327 | + /* An all-digit suffix may not be that signficant. */ | |
328 | + continue; | |
329 | + } | |
330 | + | |
331 | + return suf; | |
332 | +} | |
333 | + | |
054f3f90 WD |
334 | +/* This is an implementation of the Levenshtein distance algorithm. It |
335 | + * was implemented to avoid needing a two-dimensional matrix (to save | |
336 | + * memory). It was also tweaked to try to factor in the ASCII distance | |
337 | + * between changed characters as a minor distance quantity. The normal | |
338 | + * Levenshtein units of distance (each signifying a single change between | |
339 | + * the two strings) are defined as a "UNIT". */ | |
340 | + | |
341 | +#define UNIT (1 << 16) | |
342 | + | |
343 | +uint32 fuzzy_distance(const char *s1, int len1, const char *s2, int len2) | |
344 | +{ | |
345 | + uint32 a[MAXPATHLEN], diag, above, left, diag_inc, above_inc, left_inc; | |
346 | + int32 cost; | |
347 | + int i1, i2; | |
348 | + | |
4cd1daea WD |
349 | + if (!len1 || !len2) { |
350 | + if (!len1) { | |
351 | + s1 = s2; | |
352 | + len1 = len2; | |
353 | + } | |
354 | + for (i1 = 0, cost = 0; i1 < len1; i1++) | |
355 | + cost += s1[i1]; | |
356 | + return (int32)len1 * UNIT + cost; | |
357 | + } | |
054f3f90 WD |
358 | + |
359 | + for (i2 = 0; i2 < len2; i2++) | |
360 | + a[i2] = (i2+1) * UNIT; | |
361 | + | |
362 | + for (i1 = 0; i1 < len1; i1++) { | |
363 | + diag = i1 * UNIT; | |
364 | + above = (i1+1) * UNIT; | |
365 | + for (i2 = 0; i2 < len2; i2++) { | |
366 | + left = a[i2]; | |
367 | + if ((cost = *((uchar*)s1+i1) - *((uchar*)s2+i2)) != 0) { | |
368 | + if (cost < 0) | |
369 | + cost = UNIT - cost; | |
370 | + else | |
371 | + cost = UNIT + cost; | |
372 | + } | |
373 | + diag_inc = diag + cost; | |
374 | + left_inc = left + UNIT + *((uchar*)s1+i1); | |
375 | + above_inc = above + UNIT + *((uchar*)s2+i2); | |
376 | + a[i2] = above = left < above | |
377 | + ? (left_inc < diag_inc ? left_inc : diag_inc) | |
378 | + : (above_inc < diag_inc ? above_inc : diag_inc); | |
379 | + diag = left; | |
380 | + } | |
381 | + } | |
382 | + | |
383 | + return a[len2-1]; | |
384 | +} |