[Debburn-changes] r320 - nonameyet/trunk/3rd-party/dirsplit

[Eduard Bloch]

Author: blade
Date: 2006-09-17 21:19:58 +0000 (Sun, 17 Sep 2006)
New Revision: 320

Modified:
   nonameyet/trunk/3rd-party/dirsplit/dirsplit
Log:
Corrected dirsplit, splitting into multiple coalesced volumes in e4 mode

Modified: nonameyet/trunk/3rd-party/dirsplit/dirsplit
===================================================================
--- nonameyet/trunk/3rd-party/dirsplit/dirsplit	2006-09-17 19:17:09 UTC (rev 319)
+++ nonameyet/trunk/3rd-party/dirsplit/dirsplit	2006-09-17 21:19:58 UTC (rev 320)
@@ -75,6 +75,7 @@
 
 $opt_ver = 1 if $opt_sim;
 $opt_move=1 if ($opt_sln || $opt_ln);
+die "Invalid filesystem exploration mode!\n" if ($emode<1 || $emode>4);
 
 # big list @sizes containing the "items" (object sizes)
 # %names hash mapping "items" (size as key) to arrays with filenames/subarrays for coalesced files
@@ -238,7 +239,7 @@
 # parameter: directory
 # mode 1: descend as far as possible and index all non-directories
 # mode 2++:
-# put all files of a dir into coaleseced-object, then descend into each dir
+# put all files of a dir into coalesced-object, then descend into each dir
 sub explore {
    (my $dir) = @_;
    my @stuff;
@@ -248,9 +249,7 @@
    opendir(DIR, $dir) || die "Could not open $dir\n";
    @stuff=readdir(DIR);
    
-   if($opt_simple) {
-      @stuff=sort { lc($a) cmp lc($b) } @stuff;
-   }
+   @stuff=sort { lc($a) cmp lc($b) } @stuff;
       
    foreach my $f (@stuff) {
       next if ($f eq "." || $f eq "..");
@@ -299,46 +298,40 @@
          }
          $filesum += $tmp;
       };
+      #print "D: filesum: $filesum\n";
+      if($filesum<=$max) {
+         # ok, building a coalesced object, the simplest case
+         &insitem($filesum, \@files);
+      }
+      elsif($emode==2) {
+         die "Coalesced data object became too large for one medium, expriment with -e option.\n";
+      }
+      # oversized, deal with oversized coal. objects
+      elsif($emode==3) {
+         # don't coalesc in this mode, do like mode 1 above, leave them alone
+         &insitem(getsize($_), $_) for(@files);
+         return;
+      }
+      # a bit more complicated, split file set
+      elsif($emode==4) {
 
-      # handle coal. objects becoming too large
-      if($filesum>$max) {
-         # too large coal. object...
-         if($emode==3) {
-            # don't coalesc in this mode, do like mode 1 above, leave them alone
-            &insitem(getsize($_), $_) for(@files);
-            return;
-         }
-         # a bit complicated, split file set while creating coal.objects
-         if($emode==4) {
-            my $partsum=0;
-            my @sorted=sort(@files);
-            my @tmpvol;
-            for(my $i=0;$i<=$#sorted;$i++) {
-#            print "D: i: $i, partsum: $partsum, file: $sorted[$i]\n";
-               my $tmp=getsize($sorted[$i]);
-               $partsum+=$tmp;
-               if($partsum>$max) {
-                  # undo the last step then build the coal.object
-                  $partsum-=$tmp;
-                  $i--;
-
-                  &insitem($partsum, \@tmpvol);
-                  # reset temporaries
-                  undef @tmpvol;
-                  undef $partsum;
-               }
-               else {
-                  push(@tmpvol, $sorted[$i]);
-               }
+         my @tmpvol;
+         my $tmpsize=0;
+         for my $f (@files) {
+            my $newsize=getsize($f);
+            if($tmpsize+$newsize > $max) { # wouldn't fit
+               &insitem($tmpsize, [@tmpvol]);
+               $tmpsize=$newsize;
+               @tmpvol=($f);
             }
-            return;
+            else {
+               push(@tmpvol, $f);
+               $tmpsize+=$newsize;
+            }
          }
+         # the last incomplete volume, not copying the [array], it's local here already
+         &insitem($tmpsize, \@tmpvol)
       }
-
-      # ok, building a coalesced object for simple cases
-      if($filesum) {
-         &insitem($filesum, \@files);
-      }
    }
 }
 
@@ -348,6 +341,7 @@
 # args: size, object (filename or list reference)
 sub insitem {
    my ($size, $object) = @_;
+   #print "D: $size, $object\n";
    # normaly, put the items into the pool for calculation. In simple mode, calculate here
    
    push(@sizes, $size);