More cleanup work

dedupestore/archive.py

@@ -4,6 +4,7 @@ import logging
 import msgpack
 import os
 import stat
+import sys
 import zlib
 
 from .cache import NS_ARCHIVES, NS_CHUNKS
@@ -30,6 +31,9 @@ class Archive(object):
         if hashlib.sha256(data).digest() != id:
             raise Exception('Archive hash did not match')
         archive = msgpack.unpackb(zlib.decompress(data))
+        version = archive.get('version')
+        if version != 1:
+            raise Exception('Archive version %r not supported' % version)
         self.items = archive['items']
         self.name = archive['name']
         self.chunks = archive['chunks']
@@ -38,7 +42,9 @@ class Archive(object):
 
     def save(self, name):
         archive = {
+            'version': 1,
             'name': name,
+            'cmdline': ' '.join(sys.argv),
             'ts': datetime.utcnow().isoformat(),
             'items': self.items,
             'chunks': self.chunks
@@ -58,17 +64,15 @@ class Archive(object):
             return idx
 
     def stats(self, cache):
-        total_osize = 0
-        total_csize = 0
-        total_usize = 0
+        osize = csize = usize = 0
         for item in self.items:
             if item['type'] == 'FILE':
-                total_osize += item['size']
+                osize += item['size']
         for id, size in self.chunks:
-            total_csize += size
+            csize += size
             if self.cache.seen_chunk(id) == 1:
-                total_usize += size
-        return dict(osize=total_osize, csize=total_csize, usize=total_usize)
+                usize += size
+        return osize, csize, usize
 
     def list(self):
         for item in self.items:

dedupestore/archiver.py

@@ -55,10 +55,10 @@ class Archiver(object):
     def do_info(self, args):
         store, cache = self.open_store(args.archive)
         archive = Archive(store, cache, args.archive.archive)
-        stats = archive.stats(cache)
-        print 'Original size:', pretty_size(stats['osize'])
-        print 'Compressed size:', pretty_size(stats['csize'])
-        print 'Unique data:', pretty_size(stats['usize'])
+        osize, csize, usize = archive.stats(cache)
+        print 'Original size:', pretty_size(osize)
+        print 'Compressed size:', pretty_size(csize)
+        print 'Unique data:', pretty_size(usize)
         return self.exit_code_from_logger()
 
     def run(self, args=None):

dedupestore/cache.py

@@ -1,6 +1,7 @@
 import hashlib
-import os
+import logging
 import msgpack
+import os
 import zlib
 
 NS_ARCHIVES = 'A'
@@ -29,7 +30,11 @@ class Cache(object):
         if hashlib.sha256(data).digest() != id:
             raise Exception('Cache hash did not match')
         data = msgpack.unpackb(zlib.decompress(data))
-        if data['uuid'] != self.store.uuid:
+        version = data.get('version')
+        if version != 1:
+            logging.error('Unsupported cache version %r' % version)
+            return
+        if data['store'] != self.store.uuid:
             raise Exception('Cache UUID mismatch')
         self.chunkmap = data['chunkmap']
         self.archives = data['archives']
@@ -38,6 +43,7 @@ class Cache(object):
     def init(self):
         """Initializes cache by fetching and reading all archive indicies
         """
+        logging.info('Initialzing cache...')
         self.chunkmap = {}
         self.archives = {}
         self.tid = self.store.tid
@@ -59,7 +65,8 @@ class Cache(object):
 
     def save(self):
         assert self.store.state == self.store.OPEN
-        data = {'uuid': self.store.uuid,
+        data = {'version': 1,
+                'store': self.store.uuid,
                 'chunkmap': self.chunkmap,
                 'tid': self.store.tid, 'archives': self.archives}
         cachedir = os.path.dirname(self.path)
@@ -77,11 +84,8 @@ class Cache(object):
         data = hashlib.sha256(data).digest() + data
         csize = len(data)
         self.store.put(NS_CHUNKS, id, data)
-        return self.init_chunk(id, csize)[1]
-
-    def init_chunk(self, id, size):
-        self.chunkmap[id] = (1, size)
-        return id, size
+        self.chunkmap[id] = (1, csize)
+        return csize
 
     def seen_chunk(self, id):
         count, size = self.chunkmap.get(id, (0, 0))