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 |
054f3f90 | 8 | +++ generator.c 2005-01-18 10:55:29 |
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[]; |
054f3f90 | 17 | @@ -242,6 +243,81 @@ 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]; |
054f3f90 WD |
27 | + int extlen, basename_len; |
28 | + uint32 lowest_dist = 0x7FFFFFFF; | |
47dd7a31 WD |
29 | + const char *ext; |
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 WD |
48 | + |
49 | + /* Get final extension, eg. .gz; never full basename though. */ | |
0f626034 WD |
50 | + for (ext = basename; *ext == '.'; ext++) {} |
51 | + if (!(ext = strrchr(ext, '.'))) | |
054f3f90 WD |
52 | + ext = basename + basename_len; /* ext = "" */ |
53 | + extlen = strlen(ext); | |
47dd7a31 | 54 | + |
054f3f90 | 55 | + bestname[0] = '\0'; |
47dd7a31 WD |
56 | + while ((di = readdir(d)) != NULL) { |
57 | + const char *dname = d_name(di); | |
054f3f90 WD |
58 | + uint32 dist; |
59 | + int dname_len; | |
47dd7a31 WD |
60 | + |
61 | + if (dname[0] == '.' && (dname[1] == '\0' | |
62 | + || (dname[1] == '.' && dname[2] == '\0'))) | |
63 | + continue; | |
64 | + | |
054f3f90 WD |
65 | + dname_len = strlen(dname); |
66 | + | |
67 | + /* Extensions must match */ | |
68 | + if (dname_len <= extlen | |
69 | + || strcmp(dname + dname_len - extlen, ext) != 0) | |
70 | + continue; | |
71 | + | |
72 | + dist = fuzzy_distance(dname, dname_len, basename, basename_len); | |
73 | + if (verbose > 1) { | |
74 | + rprintf(FINFO, "fuzzy distance for %s = %lx\n", | |
75 | + dname, (unsigned long)dist); | |
47dd7a31 | 76 | + } |
054f3f90 | 77 | + if (dist < lowest_dist) { |
8c5b8235 | 78 | + strlcpy(bestname, dname, sizeof bestname); |
054f3f90 | 79 | + lowest_dist = dist; |
47dd7a31 WD |
80 | + } |
81 | + } | |
82 | + closedir(d); | |
83 | + | |
84 | + /* Found a candidate. */ | |
054f3f90 | 85 | + if (bestname[0] != '\0') { |
e55625fb | 86 | + strlcpy(basename, bestname, MAXPATHLEN - (basename - buf)); |
054f3f90 WD |
87 | + if (verbose > 2) { |
88 | + rprintf(FINFO, "fuzzy match %s->%s\n", | |
89 | + safe_fname(fname), buf); | |
90 | + } | |
58118c25 | 91 | + return link_stat(buf, st_ptr, 0); |
47dd7a31 WD |
92 | + } |
93 | + return -1; | |
94 | +} | |
58118c25 WD |
95 | + |
96 | ||
97 | /* | |
98 | * Acts on file number @p i from @p flist, whose name is @p fname. | |
054f3f90 | 99 | @@ -496,6 +572,15 @@ static void recv_generator(char *fname, |
f48a237e WD |
100 | } else |
101 | partialptr = NULL; | |
824abc86 | 102 | |
09fb2223 WD |
103 | + if (statret == -1 && fuzzy_basis) { |
104 | + if (find_fuzzy(fname, fnamecmpbuf, &st) == 0 | |
105 | + && S_ISREG(st.st_mode)) { | |
106 | + statret = 0; | |
107 | + fnamecmp = fnamecmpbuf; | |
0edc7d7f | 108 | + fnamecmp_type = FNAMECMP_FUZZY; |
09fb2223 | 109 | + } |
824abc86 WD |
110 | + } |
111 | + | |
09fb2223 WD |
112 | if (statret == -1) { |
113 | if (preserve_hard_links && hard_link_check(file, HL_SKIP)) | |
7628f156 | 114 | return; |
054f3f90 | 115 | @@ -524,6 +609,8 @@ static void recv_generator(char *fname, |
241013b4 | 116 | |
9cf86680 | 117 | if (!compare_dest && fnamecmp_type <= FNAMECMP_BASIS_DIR_HIGH) |
0808daa5 WD |
118 | ; |
119 | + else if (fnamecmp_type == FNAMECMP_FUZZY) | |
120 | + ; | |
121 | else if (unchanged_file(fnamecmp, file, &st)) { | |
0edc7d7f | 122 | if (fnamecmp_type == FNAMECMP_FNAME) |
8c5b8235 | 123 | set_perms(fname, file, &st, PERMS_REPORT); |
054f3f90 | 124 | @@ -598,8 +685,24 @@ notify_others: |
0edc7d7f | 125 | write_int(f_out, i); |
0f626034 WD |
126 | if (protocol_version >= 29 && inplace && !read_batch) |
127 | write_byte(f_out, fnamecmp_type); | |
0edc7d7f WD |
128 | - if (f_out_name >= 0) |
129 | + if (f_out_name >= 0) { | |
09fb2223 | 130 | write_byte(f_out_name, fnamecmp_type); |
0edc7d7f | 131 | + if (fnamecmp_type == FNAMECMP_FUZZY) { |
09fb2223 WD |
132 | + uchar lenbuf[3], *lb = lenbuf; |
133 | + int len = strlen(fnamecmpbuf); | |
134 | + if (len > 0x7F) { | |
135 | +#if MAXPATHLEN > 0x7FFF | |
136 | + *lb++ = len / 0x10000 + 0x80; | |
137 | + *lb++ = len / 0x100; | |
138 | +#else | |
139 | + *lb++ = len / 0x100 + 0x80; | |
140 | +#endif | |
141 | + } | |
142 | + *lb = len; | |
143 | + write_buf(f_out_name, lenbuf, lb - lenbuf + 1); | |
144 | + write_buf(f_out_name, fnamecmpbuf, len); | |
145 | + } | |
0edc7d7f | 146 | + } |
09fb2223 | 147 | |
0edc7d7f WD |
148 | if (dry_run || read_batch) |
149 | return; | |
0f626034 | 150 | --- orig/main.c 2005-01-17 23:11:45 |
3eabe3a3 | 151 | +++ main.c 2005-01-14 18:33:15 |
d5753a22 | 152 | @@ -48,6 +48,7 @@ extern int keep_dirlinks; |
495f1899 WD |
153 | extern int preserve_hard_links; |
154 | extern int protocol_version; | |
155 | extern int recurse; | |
09fb2223 | 156 | +extern int fuzzy_basis; |
495f1899 WD |
157 | extern int relative_paths; |
158 | extern int rsync_port; | |
159 | extern int whole_file; | |
3eabe3a3 | 160 | @@ -464,7 +465,8 @@ static int do_recv(int f_in,int f_out,st |
495f1899 WD |
161 | int pid; |
162 | int status = 0; | |
163 | int error_pipe[2], name_pipe[2]; | |
3eabe3a3 WD |
164 | - BOOL need_name_pipe = (basis_dir[0] || partial_dir) && !dry_run; |
165 | + BOOL need_name_pipe = (basis_dir[0] || partial_dir || fuzzy_basis) | |
166 | + && !dry_run; | |
495f1899 | 167 | |
d5753a22 WD |
168 | /* The receiving side mustn't obey this, or an existing symlink that |
169 | * points to an identical file won't be replaced by the referent. */ | |
0f626034 | 170 | --- orig/options.c 2005-01-17 23:11:45 |
9cf86680 | 171 | +++ options.c 2005-01-15 21:08:13 |
0808daa5 | 172 | @@ -86,6 +86,7 @@ int copy_unsafe_links = 0; |
f6c3b300 | 173 | int size_only = 0; |
0808daa5 | 174 | int daemon_bwlimit = 0; |
f6c3b300 | 175 | int bwlimit = 0; |
09fb2223 | 176 | +int fuzzy_basis = 0; |
f6c3b300 WD |
177 | size_t bwlimit_writemax = 0; |
178 | int delete_after = 0; | |
179 | int only_existing = 0; | |
0808daa5 | 180 | @@ -288,6 +289,7 @@ void usage(enum logcode F) |
f0533c4c | 181 | rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n"); |
0808daa5 WD |
182 | rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n"); |
183 | rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n"); | |
dc800efa | 184 | + rprintf(F," --fuzzy find similar file for basis when no dest file\n"); |
f0533c4c WD |
185 | rprintf(F," -P equivalent to --partial --progress\n"); |
186 | rprintf(F," -z, --compress compress file data\n"); | |
187 | rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n"); | |
37da98ae | 188 | @@ -384,6 +386,7 @@ static struct poptOption long_options[] |
0808daa5 WD |
189 | {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 }, |
190 | {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 }, | |
191 | {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 }, | |
09fb2223 | 192 | + {"fuzzy", 0, POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 }, |
f0533c4c WD |
193 | /* TODO: Should this take an optional int giving the compression level? */ |
194 | {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 }, | |
5388f859 | 195 | {"stats", 0, POPT_ARG_NONE, &do_stats, 0, 0, 0 }, |
0f626034 | 196 | @@ -1234,6 +1237,9 @@ void server_options(char **args,int *arg |
37da98ae | 197 | args[ac++] = "--no-relative"; |
241013b4 | 198 | } |
7b675ff5 | 199 | |
09fb2223 | 200 | + if (fuzzy_basis && am_sender) |
241013b4 | 201 | + args[ac++] = "--fuzzy"; |
7b675ff5 | 202 | + |
241013b4 | 203 | *argc = ac; |
f74d2272 | 204 | return; |
7b675ff5 | 205 | |
0f626034 | 206 | --- orig/receiver.c 2005-01-17 23:11:45 |
9cf86680 WD |
207 | +++ receiver.c 2005-01-15 21:21:02 |
208 | @@ -324,6 +324,27 @@ static int receive_data(int f_in, char * | |
09fb2223 WD |
209 | } |
210 | ||
211 | ||
212 | +static void read_gen_name(int fd, char *buf) | |
213 | +{ | |
214 | + int len = read_byte(fd); | |
215 | + if (len & 0x80) { | |
216 | +#if MAXPATHLEN > 32767 | |
217 | + uchar lenbuf[2]; | |
218 | + read_buf(fd, (char *)lenbuf, 2); | |
219 | + len = (len & ~0x80) * 0x10000 + lenbuf[0] * 0x100 + lenbuf[1]; | |
220 | +#else | |
221 | + len = (len & ~0x80) * 0x100 + read_byte(fd); | |
222 | +#endif | |
223 | + } | |
224 | + if (len >= MAXPATHLEN) { | |
225 | + rprintf(FERROR, "bogus data on generator name pipe\n"); | |
226 | + exit_cleanup(RERR_PROTOCOL); | |
227 | + } | |
228 | + | |
229 | + read_sbuf(fd, buf, len); | |
230 | +} | |
231 | + | |
232 | + | |
233 | static void discard_receive_data(int f_in, OFF_T length) | |
234 | { | |
235 | receive_data(f_in, NULL, -1, 0, NULL, -1, length); | |
9cf86680 | 236 | @@ -454,6 +475,10 @@ int recv_files(int f_in, struct file_lis |
0edc7d7f | 237 | case FNAMECMP_BACKUP: |
f48a237e | 238 | fnamecmp = get_backup_name(fname); |
09fb2223 | 239 | break; |
0edc7d7f | 240 | + case FNAMECMP_FUZZY: |
09fb2223 WD |
241 | + read_gen_name(f_in_name, fnamecmpbuf); |
242 | + fnamecmp = fnamecmpbuf; | |
243 | + break; | |
244 | default: | |
9cf86680 WD |
245 | if (j >= basis_dir_cnt) { |
246 | rprintf(FERROR, | |
0f626034 | 247 | --- orig/rsync.h 2005-01-17 23:11:45 |
9cf86680 | 248 | +++ rsync.h 2005-01-15 21:24:09 |
0f626034 | 249 | @@ -128,6 +128,7 @@ |
0edc7d7f WD |
250 | #define FNAMECMP_FNAME 0x80 |
251 | #define FNAMECMP_PARTIAL_DIR 0x81 | |
252 | #define FNAMECMP_BACKUP 0x82 | |
253 | +#define FNAMECMP_FUZZY 0x83 | |
09fb2223 WD |
254 | |
255 | ||
256 | /* Log-message categories. FLOG is only used on the daemon side to | |
0f626034 | 257 | --- orig/rsync.yo 2005-01-17 23:11:46 |
637c560e | 258 | +++ rsync.yo 2005-01-15 21:48:52 |
0808daa5 | 259 | @@ -358,6 +358,7 @@ verb( |
f0533c4c | 260 | --compare-dest=DIR also compare received files relative to DIR |
0808daa5 WD |
261 | --copy-dest=DIR ... and include copies of unchanged files |
262 | --link-dest=DIR hardlink to files in DIR when unchanged | |
dc800efa | 263 | + --fuzzy find similar file for basis when no dest |
f0533c4c WD |
264 | -P equivalent to --partial --progress |
265 | -z, --compress compress file data | |
266 | -C, --cvs-exclude auto ignore files in the same way CVS does | |
0f626034 WD |
267 | @@ -878,6 +879,11 @@ Note that rsync versions prior to 2.6.1 |
268 | (or implied by -a). You can work-around this bug by avoiding the -o option | |
269 | when sending to an old rsync. | |
637c560e WD |
270 | |
271 | +dit(bf(--fuzzy)) This option tells rsync that it should look around for a | |
272 | +basis file for any destination file that is missing. The current algorithm | |
273 | +looks for a similarly-named file in the same directory as the destination | |
274 | +file, and, if found, uses that to try to speed up the transfer. | |
275 | + | |
276 | dit(bf(-z, --compress)) With this option, rsync compresses any data from | |
277 | the files that it sends to the destination machine. This | |
278 | option is useful on slow connections. The compression method used is the | |
054f3f90 WD |
279 | --- orig/util.c 2004-09-07 21:45:30 |
280 | +++ util.c 2005-01-18 11:18:46 | |
281 | @@ -1217,3 +1217,50 @@ void *_realloc_array(void *ptr, unsigned | |
282 | return malloc(size * num); | |
283 | return realloc(ptr, size * num); | |
284 | } | |
285 | + | |
286 | +/* This is an implementation of the Levenshtein distance algorithm. It | |
287 | + * was implemented to avoid needing a two-dimensional matrix (to save | |
288 | + * memory). It was also tweaked to try to factor in the ASCII distance | |
289 | + * between changed characters as a minor distance quantity. The normal | |
290 | + * Levenshtein units of distance (each signifying a single change between | |
291 | + * the two strings) are defined as a "UNIT". */ | |
292 | + | |
293 | +#define UNIT (1 << 16) | |
294 | + | |
295 | +uint32 fuzzy_distance(const char *s1, int len1, const char *s2, int len2) | |
296 | +{ | |
297 | + uint32 a[MAXPATHLEN], diag, above, left, diag_inc, above_inc, left_inc; | |
298 | + int32 cost; | |
299 | + int i1, i2; | |
300 | + | |
301 | + if (!len1) | |
302 | + return (int32)len2 * UNIT; | |
303 | + if (!len2) | |
304 | + return (int32)len1 * UNIT; | |
305 | + | |
306 | + for (i2 = 0; i2 < len2; i2++) | |
307 | + a[i2] = (i2+1) * UNIT; | |
308 | + | |
309 | + for (i1 = 0; i1 < len1; i1++) { | |
310 | + diag = i1 * UNIT; | |
311 | + above = (i1+1) * UNIT; | |
312 | + for (i2 = 0; i2 < len2; i2++) { | |
313 | + left = a[i2]; | |
314 | + if ((cost = *((uchar*)s1+i1) - *((uchar*)s2+i2)) != 0) { | |
315 | + if (cost < 0) | |
316 | + cost = UNIT - cost; | |
317 | + else | |
318 | + cost = UNIT + cost; | |
319 | + } | |
320 | + diag_inc = diag + cost; | |
321 | + left_inc = left + UNIT + *((uchar*)s1+i1); | |
322 | + above_inc = above + UNIT + *((uchar*)s2+i2); | |
323 | + a[i2] = above = left < above | |
324 | + ? (left_inc < diag_inc ? left_inc : diag_inc) | |
325 | + : (above_inc < diag_inc ? above_inc : diag_inc); | |
326 | + diag = left; | |
327 | + } | |
328 | + } | |
329 | + | |
330 | + return a[len2-1]; | |
331 | +} |