fuzzy.diff

   1 From: Rusty Russell <rusty@rustcorp.com.au>
   2 Date: Wed, 03 Apr 2002 17:08:57 +1000
   3
   4 Found old patch on google, and updated it for 2.5.4 (I know, but that's what
   5 apt-get source gave me).
   6
   7 Compiles, otherwise untested.
   8 Rusty.
   9
  10 [Updated for current CVS version by Wayne Davison.  Passes *MOST* of the
  11 test suite, but otherwise UNTESTED.]
  12 --
  13   Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
  14
  15 --- Makefile.in 15 May 2004 00:48:11 -0000      1.101
  16 +++ Makefile.in 18 Jun 2004 17:32:53 -0000
  17 @@ -32,7 +32,7 @@ ZLIBOBJ=zlib/deflate.o zlib/infblock.o z
  18         zlib/inflate.o zlib/inftrees.o zlib/infutil.o zlib/trees.o \
  19         zlib/zutil.o zlib/adler32.o
  20  OBJS1=rsync.o generator.o receiver.o cleanup.o sender.o exclude.o util.o \
  21 -       main.o checksum.o match.o syscall.o log.o backup.o
  22 +       main.o checksum.o match.o syscall.o log.o backup.o alternate.o
  23  OBJS2=options.o flist.o io.o compat.o hlink.o token.o uidlist.o socket.o \
  24         fileio.o batch.o clientname.o
  25  OBJS3=progress.o pipe.o
  26 --- generator.c 18 Jun 2004 16:30:24 -0000      1.88
  27 +++ generator.c 18 Jun 2004 17:32:53 -0000
  28 @@ -41,6 +41,7 @@ extern int ignore_times;
  29  extern int size_only;
  30  extern int io_timeout;
  31  extern int protocol_version;
  32 +extern int fuzzy;
  33  extern int always_checksum;
  34  extern char *compare_dest;
  35  extern int link_dest;
  36 @@ -259,7 +260,61 @@ static void generate_and_send_sums(struc
  37         }
  38  }
  39
  40 +/* Returns -1 for can't open (null file), -2 for skip */
  41 +static int open_base_file(struct file_struct *file, char *fname, int statret,
  42 +                         STRUCT_STAT *st)
  43 +{
  44 +       int fd = -1;
  45
  46 +       if (statret == 0) {
  47 +               if (S_ISREG(st->st_mode)) {
  48 +                       if (update_only
  49 +                           && cmp_modtime(st->st_mtime, file->modtime) > 0) {
  50 +                               if (verbose > 1)
  51 +                                       rprintf(FINFO, "%s is newer\n", fname);
  52 +                               return -2;
  53 +                       }
  54 +                       if (skip_file(fname, file, st)) {
  55 +                               set_perms(fname, file, st, PERMS_REPORT);
  56 +                               return -2;
  57 +                       }
  58 +                       fd = do_open(fname, O_RDONLY, 0);
  59 +                       if (fd == -1) {
  60 +                               rsyserr(FERROR, errno, "failed to open %s, continuing",
  61 +                                   full_fname(fname));
  62 +                               return -1;
  63 +                       }
  64 +                       return fd;
  65 +               } else {
  66 +                       /* Try to use symlink contents */
  67 +                       if (S_ISLNK(st->st_mode)) {
  68 +                               fd = do_open_regular(fname);
  69 +                               /* Don't delete yet; receiver will need it */
  70 +                       } else {
  71 +                               if (delete_file(fname) != 0) {
  72 +                                       if (fd != -1)
  73 +                                               close(fd);
  74 +                                       return -2;
  75 +                               }
  76 +                       }
  77 +               }
  78 +       }
  79 +
  80 +       if (fd == -1 && compare_dest != NULL)
  81 +               fd = open_alternate_base_comparedir(fname);
  82 +
  83 +       if (fd == -1 && fuzzy)
  84 +               fd = open_alternate_base_fuzzy(fname);
  85 +
  86 +       /* Update stat to understand size */
  87 +       if (fd != -1) {
  88 +               if (do_fstat(fd, st) != 0) {
  89 +                       rsyserr(FERROR, errno, "fstat %s", full_fname(fname));
  90 +               }
  91 +       }
  92 +
  93 +       return fd;
  94 +}
  95
  96  /**
  97   * Acts on file number @p i from @p flist, whose name is @p fname.
  98 @@ -275,8 +330,6 @@ void recv_generator(char *fname, struct
  99         STRUCT_STAT st;
 100         struct map_struct *mapbuf;
 101         int statret;
 102 -       char *fnamecmp;
 103 -       char fnamecmpbuf[MAXPATHLEN];
 104
 105         if (list_only)
 106                 return;
 107 @@ -416,109 +469,39 @@ void recv_generator(char *fname, struct
 108         }
 109  #endif
 110
 111 -       if (preserve_hard_links && hard_link_check(file, HL_CHECK_MASTER))
 112 -               return;
 113 -
 114 -       if (!S_ISREG(file->mode)) {
 115 -               rprintf(FINFO, "skipping non-regular file \"%s\"\n",fname);
 116 -               return;
 117 -       }
 118 -
 119 -       fnamecmp = fname;
 120 -
 121 -       if (statret == -1 && compare_dest != NULL) {
 122 -               /* try the file at compare_dest instead */
 123 -               int saveerrno = errno;
 124 -               pathjoin(fnamecmpbuf, sizeof fnamecmpbuf, compare_dest, fname);
 125 -               statret = link_stat(fnamecmpbuf, &st, 0);
 126 -               if (!S_ISREG(st.st_mode))
 127 -                       statret = -1;
 128 -               if (statret == -1)
 129 -                       errno = saveerrno;
 130 -#if HAVE_LINK
 131 -               else if (link_dest && !dry_run) {
 132 -                       if (do_link(fnamecmpbuf, fname) != 0) {
 133 -                               if (verbose > 0) {
 134 -                                       rsyserr(FINFO, errno, "link %s => %s",
 135 -                                               fnamecmpbuf, fname);
 136 -                               }
 137 -                       }
 138 -                       fnamecmp = fnamecmpbuf;
 139 -               }
 140 -#endif
 141 -               else
 142 -                       fnamecmp = fnamecmpbuf;
 143 -       }
 144 -
 145 -       if (statret == -1) {
 146 -               if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 147 -                       return;
 148 -               if (errno == ENOENT) {
 149 -                       write_int(f_out,i);
 150 -                       if (!dry_run)
 151 -                               write_sum_head(f_out, NULL);
 152 -               } else if (verbose > 1) {
 153 +       /* Failed to stat for some reason besides "not found". */
 154 +       if (statret == -1 && errno != ENOENT) {
 155 +               if (verbose > 1) {
 156                         rsyserr(FERROR, errno,
 157 -                               "recv_generator: failed to open %s",
 158 +                               "recv_generator failed to stat %s",
 159                                 full_fname(fname));
 160                 }
 161                 return;
 162         }
 163
 164 -       if (!S_ISREG(st.st_mode)) {
 165 -               if (delete_file(fname) != 0) {
 166 -                       return;
 167 -               }
 168 +       if ((fd = open_base_file(file, fname, statret, &st)) == -2)
 169 +               return;
 170
 171 -               /* now pretend the file didn't exist */
 172 +       if ((disable_deltas_p() || dry_run) && fd != -1) {
 173 +               close(fd);
 174 +               fd = -1;
 175 +       }
 176 +
 177 +       if (fd == -1) {
 178 +               /* the file didn't exist, or we can pretend it doesn't */
 179                 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 180                         return;
 181 -               write_int(f_out,i);
 182 +               write_int(f_out, i);
 183                 if (!dry_run)
 184                         write_sum_head(f_out, NULL);
 185 -               return;
 186 -       }
 187 -
 188 -       if (opt_ignore_existing && fnamecmp == fname) {
 189 -               if (verbose > 1)
 190 -                       rprintf(FINFO,"%s exists\n",fname);
 191 -               return;
 192 -       }
 193 +               return;
 194 +       }
 195
 196 -       if (update_only && cmp_modtime(st.st_mtime,file->modtime)>0 && fnamecmp == fname) {
 197 -               if (verbose > 1)
 198 -                       rprintf(FINFO,"%s is newer\n",fname);
 199 -               return;
 200 -       }
 201 -
 202 -       if (skip_file(fname, file, &st)) {
 203 -               if (fnamecmp == fname)
 204 -                       set_perms(fname, file, &st, PERMS_REPORT);
 205 -               return;
 206 -       }
 207 -
 208 -       if (dry_run) {
 209 -               write_int(f_out,i);
 210 -               return;
 211 -       }
 212 -
 213 -       if (disable_deltas_p()) {
 214 -               write_int(f_out,i);
 215 -               write_sum_head(f_out, NULL);
 216 +       if (preserve_hard_links && hard_link_check(file, HL_CHECK_MASTER))
 217                 return;
 218 -       }
 219
 220 -       /* open the file */
 221 -       fd = do_open(fnamecmp, O_RDONLY, 0);
 222 -
 223 -       if (fd == -1) {
 224 -               rsyserr(FERROR, errno, "failed to open %s, continuing",
 225 -                       full_fname(fnamecmp));
 226 -               /* pretend the file didn't exist */
 227 -               if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 228 -                       return;
 229 -               write_int(f_out,i);
 230 -               write_sum_head(f_out, NULL);
 231 +       if (!S_ISREG(file->mode)) {
 232 +               rprintf(FINFO, "skipping non-regular file \"%s\"\n",fname);
 233                 return;
 234         }
 235
 236 @@ -528,7 +511,7 @@ void recv_generator(char *fname, struct
 237                 mapbuf = NULL;
 238
 239         if (verbose > 3) {
 240 -               rprintf(FINFO,"gen mapped %s of size %.0f\n", fnamecmp,
 241 +               rprintf(FINFO, "gen mapped %s of size %.0f\n", fname,
 242                         (double)st.st_size);
 243         }
 244
 245 --- options.c   7 Jun 2004 22:05:22 -0000       1.156
 246 +++ options.c   18 Jun 2004 17:32:54 -0000
 247 @@ -94,6 +94,7 @@ int ignore_errors = 0;
 248  int modify_window = 0;
 249  int blocking_io = -1;
 250  int checksum_seed = 0;
 251 +int fuzzy = 0;
 252  unsigned int block_size = 0;
 253
 254
 255 @@ -270,6 +271,7 @@ void usage(enum logcode F)
 256    rprintf(F," -T  --temp-dir=DIR          create temporary files in directory DIR\n");
 257    rprintf(F,"     --compare-dest=DIR      also compare destination files relative to DIR\n");
 258    rprintf(F,"     --link-dest=DIR         create hardlinks to DIR for unchanged files\n");
 259 +  rprintf(F,"     --fuzzy                 use similar file as basis if basis doesn't exist\n");
 260    rprintf(F," -P                          equivalent to --partial --progress\n");
 261    rprintf(F," -z, --compress              compress file data\n");
 262    rprintf(F," -C, --cvs-exclude           auto ignore files in the same way CVS does\n");
 263 @@ -368,6 +370,7 @@ static struct poptOption long_options[]
 264    {"temp-dir",        'T', POPT_ARG_STRING, &tmpdir, 0, 0, 0 },
 265    {"compare-dest",     0,  POPT_ARG_STRING, &compare_dest, 0, 0, 0 },
 266    {"link-dest",        0,  POPT_ARG_STRING, &compare_dest,  OPT_LINK_DEST, 0, 0 },
 267 +  {"fuzzy",            0,  POPT_ARG_NONE,   &fuzzy, 0, 0, 0 },
 268    /* TODO: Should this take an optional int giving the compression level? */
 269    {"compress",        'z', POPT_ARG_NONE,   &do_compression, 0, 0, 0 },
 270    {"daemon",           0,  POPT_ARG_NONE,   &daemon_opt, 0, 0, 0 },
 271 @@ -989,6 +992,9 @@ void server_options(char **args,int *arg
 272                 }
 273         }
 274
 275 +       if (fuzzy && am_sender)
 276 +               args[ac++] = "--fuzzy";
 277 +
 278         *argc = ac;
 279         return;
 280
 281 --- receiver.c  14 Jun 2004 15:09:36 -0000      1.82
 282 +++ receiver.c  18 Jun 2004 17:32:54 -0000
 283 @@ -48,6 +48,7 @@ extern int ignore_errors;
 284  extern int orig_umask;
 285  extern int keep_partial;
 286  extern int checksum_seed;
 287 +extern int fuzzy;
 288
 289  static void delete_one(char *fn, int is_dir)
 290  {
 291 @@ -300,8 +301,6 @@ int recv_files(int f_in,struct file_list
 292         char *fname, fbuf[MAXPATHLEN];
 293         char template[MAXPATHLEN];
 294         char fnametmp[MAXPATHLEN];
 295 -       char *fnamecmp;
 296 -       char fnamecmpbuf[MAXPATHLEN];
 297         struct map_struct *mapbuf;
 298         struct file_struct *file;
 299         struct stats initial_stats;
 300 @@ -364,35 +363,31 @@ int recv_files(int f_in,struct file_list
 301                 if (verbose > 2)
 302                         rprintf(FINFO,"recv_files(%s)\n",fname);
 303
 304 -               fnamecmp = fname;
 305 -
 306                 /* open the file */
 307 -               fd1 = do_open(fnamecmp, O_RDONLY, 0);
 308 +               fd1 = do_open(fname, O_RDONLY, 0);
 309
 310 -               if (fd1 == -1 && compare_dest != NULL) {
 311 -                       /* try the file at compare_dest instead */
 312 -                       pathjoin(fnamecmpbuf, sizeof fnamecmpbuf,
 313 -                                compare_dest, fname);
 314 -                       fnamecmp = fnamecmpbuf;
 315 -                       fd1 = do_open(fnamecmp, O_RDONLY, 0);
 316 -               }
 317 +               if (fd1 == -1 && compare_dest != NULL)
 318 +                       fd1 = open_alternate_base_comparedir(fname);
 319 +
 320 +               if (fd1 == -1 && fuzzy)
 321 +                       fd1 = open_alternate_base_fuzzy(fname);
 322
 323                 if (fd1 != -1 && do_fstat(fd1,&st) != 0) {
 324                         rsyserr(FERROR, errno, "fstat %s failed",
 325 -                               full_fname(fnamecmp));
 326 +                               full_fname(fname));
 327                         receive_data(f_in,NULL,-1,NULL,file->length);
 328                         close(fd1);
 329                         continue;
 330                 }
 331
 332 -               if (fd1 != -1 && S_ISDIR(st.st_mode) && fnamecmp == fname) {
 333 +               if (fd1 != -1 && S_ISDIR(st.st_mode)) {
 334                         /* this special handling for directories
 335                          * wouldn't be necessary if robust_rename()
 336                          * and the underlying robust_unlink could cope
 337                          * with directories
 338                          */
 339                         rprintf(FERROR,"recv_files: %s is a directory\n",
 340 -                               full_fname(fnamecmp));
 341 +                               full_fname(fname));
 342                         receive_data(f_in, NULL, -1, NULL, file->length);
 343                         close(fd1);
 344                         continue;
 345 @@ -413,8 +408,10 @@ int recv_files(int f_in,struct file_list
 346
 347                 if (fd1 != -1 && st.st_size > 0) {
 348                         mapbuf = map_file(fd1,st.st_size);
 349 -                       if (verbose > 2)
 350 -                               rprintf(FINFO,"recv mapped %s of size %.0f\n",fnamecmp,(double)st.st_size);
 351 +                       if (verbose > 2) {
 352 +                               rprintf(FINFO, "recv mapped %s of size %.0f\n",
 353 +                                   fname, (double)st.st_size);
 354 +                       }
 355                 } else
 356                         mapbuf = NULL;
 357
 358 --- rsync.yo    5 Jun 2004 16:16:30 -0000       1.171
 359 +++ rsync.yo    18 Jun 2004 17:32:54 -0000
 360 @@ -325,6 +325,7 @@ verb(
 361   -T  --temp-dir=DIR          create temporary files in directory DIR
 362       --compare-dest=DIR      also compare received files relative to DIR
 363       --link-dest=DIR         create hardlinks to DIR for unchanged files
 364 +     --fuzzy                 use similar file as basis if basis is gone
 365   -P                          equivalent to --partial --progress
 366   -z, --compress              compress file data
 367   -C, --cvs-exclude           auto ignore files in the same way CVS does