Be sure to run "make proto" before "make".
--- orig/generator.c 2005-01-17 23:11:45
-+++ generator.c 2005-01-18 10:55:29
++++ generator.c 2005-01-18 19:25:55
@@ -44,6 +44,7 @@ extern int size_only;
extern OFF_T max_size;
extern int io_timeout;
extern int always_checksum;
extern char *partial_dir;
extern char *basis_dir[];
-@@ -242,6 +243,81 @@ static void generate_and_send_sums(int f
+@@ -242,6 +243,76 @@ static void generate_and_send_sums(int f
}
+ struct dirent *di;
+ char *basename, *dirname, *slash;
+ char bestname[MAXPATHLEN];
-+ int extlen, basename_len;
++ int suf_len, basename_len;
+ uint32 lowest_dist = 0x7FFFFFFF;
-+ const char *ext;
++ const char *suf;
+
+ strlcpy(buf, fname, MAXPATHLEN);
+ if ((slash = strrchr(buf, '/')) != NULL) {
+ if (slash)
+ *slash = '/';
+
-+ /* Get final extension, eg. .gz; never full basename though. */
-+ for (ext = basename; *ext == '.'; ext++) {}
-+ if (!(ext = strrchr(ext, '.')))
-+ ext = basename + basename_len; /* ext = "" */
-+ extlen = strlen(ext);
++ suf_len = basename_len;
++ suf = find_filename_suffix(basename, &suf_len);
+
+ bestname[0] = '\0';
+ while ((di = readdir(d)) != NULL) {
-+ const char *dname = d_name(di);
++ const char *dname_suf, *dname = d_name(di);
+ uint32 dist;
-+ int dname_len;
++ int dname_len, dname_suf_len;
+
+ if (dname[0] == '.' && (dname[1] == '\0'
+ || (dname[1] == '.' && dname[2] == '\0')))
+ continue;
+
-+ dname_len = strlen(dname);
-+
-+ /* Extensions must match */
-+ if (dname_len <= extlen
-+ || strcmp(dname + dname_len - extlen, ext) != 0)
-+ continue;
++ dname_len = dname_suf_len = strlen(dname);
++ dname_suf = find_filename_suffix(dname, &dname_suf_len);
+
+ dist = fuzzy_distance(dname, dname_len, basename, basename_len);
-+ if (verbose > 1) {
-+ rprintf(FINFO, "fuzzy distance for %s = %lx\n",
-+ dname, (unsigned long)dist);
++ /* Add some extra weight to how well the suffixes matched. */
++ dist += fuzzy_distance(dname_suf, dname_suf_len, suf, suf_len) * 10;
++ if (verbose > 4) {
++ rprintf(FINFO, "fuzzy distance for %s = %d (%d)\n",
++ dname, (int)(dist>>16), (int)(dist&0xFFFF));
+ }
+ if (dist < lowest_dist) {
+ strlcpy(bestname, dname, sizeof bestname);
/*
* Acts on file number @p i from @p flist, whose name is @p fname.
-@@ -496,6 +572,15 @@ static void recv_generator(char *fname,
+@@ -496,6 +567,15 @@ static void recv_generator(char *fname,
} else
partialptr = NULL;
if (statret == -1) {
if (preserve_hard_links && hard_link_check(file, HL_SKIP))
return;
-@@ -524,6 +609,8 @@ static void recv_generator(char *fname,
+@@ -524,6 +604,8 @@ static void recv_generator(char *fname,
if (!compare_dest && fnamecmp_type <= FNAMECMP_BASIS_DIR_HIGH)
;
else if (unchanged_file(fnamecmp, file, &st)) {
if (fnamecmp_type == FNAMECMP_FNAME)
set_perms(fname, file, &st, PERMS_REPORT);
-@@ -598,8 +685,24 @@ notify_others:
+@@ -598,8 +680,24 @@ notify_others:
write_int(f_out, i);
if (protocol_version >= 29 && inplace && !read_batch)
write_byte(f_out, fnamecmp_type);
the files that it sends to the destination machine. This
option is useful on slow connections. The compression method used is the
--- orig/util.c 2004-09-07 21:45:30
-+++ util.c 2005-01-18 11:18:46
-@@ -1217,3 +1217,50 @@ void *_realloc_array(void *ptr, unsigned
++++ util.c 2005-01-18 19:25:47
+@@ -1217,3 +1217,108 @@ void *_realloc_array(void *ptr, unsigned
return malloc(size * num);
return realloc(ptr, size * num);
}
+
++/* Take a filename and filename length and return the most significant
++ * filename suffix we can find. This ignores suffixes such as "~",
++ * ".bak", ".orig", ".~1~", etc. */
++const char *find_filename_suffix(const char *fn, int *len_ptr)
++{
++ const char *suf, *s;
++ int s_len, fn_len = *len_ptr;
++ BOOL had_tilde;
++
++ /* One or more dots at the start aren't a suffix. */
++ while (fn_len && *fn == '.') fn++, fn_len--;
++
++ /* Ignore the ~ in a "foo~" filename. */
++ if (fn_len > 1 && fn[fn_len-1] == '~')
++ fn_len--, had_tilde = True;
++ else
++ had_tilde = False;
++
++ /* Assume we don't find an suffix. */
++ suf = "";
++ *len_ptr = 0;
++
++ /* Find the last significant suffix. */
++ for (s = fn + fn_len - 1; fn_len > 1; ) {
++ while (*s != '.' && s != fn) s--;
++ if (s == fn)
++ break;
++ s_len = fn_len - (s - fn);
++ fn_len = s - fn;
++ if (s_len == 3) {
++ if (strcmp(s+1, "bak") == 0
++ || strcmp(s+1, "old") == 0)
++ continue;
++ } else if (s_len == 4) {
++ if (strcmp(s+1, "orig") == 0)
++ continue;
++ } else if (s_len > 2 && had_tilde
++ && s[1] == '~' && isdigit(s[2]))
++ continue;
++ *len_ptr = s_len;
++ suf = s;
++ /* Determine if the suffix is all digits. */
++ for (s++, s_len--; s_len > 0; s++, s_len--) {
++ if (!isdigit(*s))
++ return suf;
++ }
++ /* An all-digit suffix may not be that signficant. */
++ continue;
++ }
++
++ return suf;
++}
++
+/* This is an implementation of the Levenshtein distance algorithm. It
+ * was implemented to avoid needing a two-dimensional matrix (to save
+ * memory). It was also tweaked to try to factor in the ASCII distance
+ int32 cost;
+ int i1, i2;
+
-+ if (!len1)
-+ return (int32)len2 * UNIT;
-+ if (!len2)
-+ return (int32)len1 * UNIT;
++ if (!len1 || !len2) {
++ if (!len1) {
++ s1 = s2;
++ len1 = len2;
++ }
++ for (i1 = 0, cost = 0; i1 < len1; i1++)
++ cost += s1[i1];
++ return (int32)len1 * UNIT + cost;
++ }
+
+ for (i2 = 0; i2 < len2; i2++)
+ a[i2] = (i2+1) * UNIT;
Matt McCutchen's Web Site / rsync / rsync-patches.git / commitdiff