This actually stores file chunks now

1 files changed, 90 insertions, 3 deletions

diff --git a/fusearchive.py b/fusearchive.py
index b427a5f..f3b5093 100755
--- a/fusearchive.py
+++ b/fusearchive.py
@@ -7,7 +7,7 @@
 #    See the file COPYING.
 #
 
-import os, sys, shutil, fcntl, fuse, re, tempfile
+import os, sys, shutil, fcntl, fuse, re, tempfile, sha, pickle
 from errno import *
 from stat import *
 from fuse import Fuse
@@ -22,6 +22,8 @@ fuse.fuse_python_api = (0, 2)
 
 fuse.feature_assert('stateful_files', 'has_init')
 
+magic_blocksize = 1024 * 32
+magic_depth = 5
 
 def flag2mode(flags):
     md = {os.O_RDONLY: 'r', os.O_WRONLY: 'w', os.O_RDWR: 'w+'}
@@ -32,6 +34,91 @@ def flag2mode(flags):
 
     return m
 
+# Inflate a file, src is a packed file, dest is where the unpacked file
+# should go
+# we assume our chunks are in storage/
+def inflate( src, dest ):
+    print "inflate!"
+
+# Deflate a file, src is the unpacked file, dest is where we want to pack
+# to, and we assume storage/ is where chunks are stored
+def deflate( src, dest ):
+    print "deflate!"
+    inp = open( src, "r" )
+
+    hashs = [];
+    # This is retarded:
+    # http://groups.google.com/group/comp.lang.python/browse_thread/thread/ed25388487b3ac7b
+    #
+    # Why can't I just do:
+    # while( chunk = inp.read( magic_blocksize ) ):
+    # I though python was supposed to be easier! :(
+
+    while True:
+        chunk = inp.read( magic_blocksize )
+        if len( chunk ) == 0:
+            break
+
+        # Save this hash string, similar to the backuppc algo
+        digest = sha.new( str(len(chunk)) + chunk ).digest()
+
+        # Write out our chunk
+        chars = list( digest )
+        print chars
+        subparts = [];
+        for l in range( 0, magic_depth ):
+            subparts.append( "%02x" % ord( chars.pop( 0 ) ) )
+
+        # We make the hexdigest here, yeah we could just call hexdigest()
+        # but we need to essentially do this same thing to reassemble the
+        # file anyway
+        hexdigest = ''.join( [ "%02x" % ord( x ) for x in chars ] );
+
+        print subparts
+        subpath = '/'.join( subparts );
+        print "Subpath: " + subpath
+
+        # Make sure this sub path exists
+        nextpart = "./storage"
+        for part in subparts:
+            nextpart += "/" + part
+            if not os.path.exists( nextpart ):
+                print "Creating subdir: " + nextpart
+                os.mkdir( nextpart )
+
+        # Find a chunk slot
+        sub = 0
+        while True:
+            checkpath = "./storage/" + subpath + "/" + hexdigest + "_" + str( sub )
+            print "Checking: " + checkpath
+            if os.path.exists( checkpath ):
+                # Check if this is our data
+                verify = open( checkpath, "r" )
+                verify_contents = verify.read()
+                verify.close()
+
+                if verify_contents == chunk:
+                    print "Found existing block"
+                    break
+                else:
+                    print "Block exists but is not the same"
+                    sub += 1
+            else:
+                # We found a spot, dump our data here
+                print "No block here, creating new block"
+                savechunk = open( checkpath, "w" )
+                savechunk.write( chunk )
+                savechunk.close
+                break
+
+        print "Got chunk slot: " + str( sub )
+        hashs.append( [ digest, sub ] )
+
+    inp.close()
+
+    out = open( dest, "w" )
+    pickle.dump( { 'data': hashs }, out )
+    out.close()
 
 class FuseArchive(Fuse):
 
@@ -150,7 +237,7 @@ class FuseArchive(Fuse):
             #os.close( fdnum );
 
             if os.path.exists( "./tree" + self.orig_path ):
-                shutil.copy( "./tree" + path, self.tmp_name )
+                inflate( "./tree" + path, self.tmp_name )
 
             print "Shadow file: " + self.tmp_name + " for " + self.orig_path
             print "Going to open shadow file with flags: " + str(flags) + " mode " + str(mode)
@@ -191,7 +278,7 @@ class FuseArchive(Fuse):
                 self.tmp_name + " -> " + self.orig_path
 
             #pdb.set_trace()
-            shutil.copy( self.tmp_name, "./tree" + self.orig_path );
+            deflate( self.tmp_name, "./tree" + self.orig_path );
 
             print "Deleting old file: " + self.tmp_name
             os.unlink( self.tmp_name );