From 52870eb2d820ab58e83669f6f22e36af5f2611d9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jonas=20Borgstr=C3=B6m?= <jonas@borgstrom.se>
Date: Tue, 19 Oct 2010 00:00:52 +0200
Subject: [PATCH] bandstore: More robust band cleaning

---
 dedupestore/archiver.py  |  3 ++-
 dedupestore/bandstore.py | 26 +++++++++++++++++++-------
 2 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/dedupestore/archiver.py b/dedupestore/archiver.py
index 31c55f666..4450f4879 100644
--- a/dedupestore/archiver.py
+++ b/dedupestore/archiver.py
@@ -98,7 +98,7 @@ def extract(self, dest=None):
         dest = dest or os.getcwdu()
         for item in self.items:
             assert item['path'][0] not in ('/', '\\', ':')
-            path = os.path.join(dest, item['path'])
+            path = os.path.join(dest, item['path'].decode('utf-8'))
             if item['type'] == 'DIRECTORY':
                 logging.info(path)
                 if not os.path.exists(path):
@@ -127,6 +127,7 @@ def extract(self, dest=None):
     def verify(self):
         for item in self.items:
             if item['type'] == 'FILE':
+                item['path'] = item['path'].decode('utf-8')
                 for chunk in item['chunks']:
                     id = self.chunk_idx[chunk]
                     data = self.store.get(NS_CHUNKS, id)
diff --git a/dedupestore/bandstore.py b/dedupestore/bandstore.py
index 6c143e5cd..d99c57ebd 100644
--- a/dedupestore/bandstore.py
+++ b/dedupestore/bandstore.py
@@ -62,6 +62,7 @@ def _begin(self):
         while os.path.exists(self.band_filename(band)):
             os.unlink(self.band_filename(band))
             band += 1
+        self.delete_bands()
 
     def create(self, path):
         os.mkdir(path)
@@ -69,11 +70,14 @@ def create(self, path):
         cnx = sqlite3.connect(os.path.join(path, 'dedupestore.db'))
         cnx.execute('CREATE TABLE objects(ns TEXT NOT NULL, id NOT NULL, '
                     'band NOT NULL, offset NOT NULL, size NOT NULL)')
+        cnx.execute('CREATE UNIQUE INDEX objects_pk ON objects(ns, id)')
+        cnx.execute('CREATE TABLE to_delete(band NOT NULL)')
+        cnx.execute('CREATE UNIQUE INDEX to_delete_pk ON to_delete(band)')
         cnx.execute('CREATE TABLE system(uuid NOT NULL, tid NOT NULL, '
                     'nextband NOT NULL, version NOT NULL, bandlimit NOT NULL)')
         cnx.execute('INSERT INTO system VALUES(?,?,?,?,?)',
                     (uuid.uuid1().hex, 0, 0, 1, self.BAND_LIMIT))
-        cnx.execute('CREATE UNIQUE INDEX objects_pk ON objects(ns, id)')
+        cnx.commit()
 
     def close(self):
         self.rollback()
@@ -85,7 +89,18 @@ def commit(self):
         """
         """
         self.band = None
-        for b in self.to_delete:
+        self.cursor.executemany('INSERT INTO to_delete(band) VALUES(?)',
+                                [[d] for d in self.to_delete])
+        self.cursor.execute('UPDATE system SET tid=tid+1, nextband=?',
+                            (self.nextband,))
+        self.cnx.commit()
+        self.tid += 1
+        self._begin()
+
+    def delete_bands(self):
+        self.cursor.execute('SELECT band FROM to_delete')
+        to_delete = [r[0] for r in self.cursor.fetchall()]
+        for b in to_delete:
             objects = self.cursor.execute('SELECT ns, id, offset, size '
                                           'FROM objects WHERE band=? ORDER BY offset',
                                           (b,)).fetchall()
@@ -93,13 +108,10 @@ def commit(self):
                 band, offset, size = self.store_data(self.retrieve_data(b, *o[2:]))
                 self.cursor.execute('UPDATE objects SET band=?, offset=?, size=? '
                                     'WHERE ns=? AND id=?', (band, offset, size, o[0], o[1]))
+            self.cursor.execute('DELETE FROM to_delete WHERE band=?', (b,))
+            self.cursor.execute('UPDATE system SET nextband=?', (self.nextband,))
             self.cnx.commit()
             os.unlink(self.band_filename(b))
-        self.cursor.execute('UPDATE system SET tid=tid+1, nextband=?',
-                            (self.nextband,))
-        self.cnx.commit()
-        self.tid += 1
-        self._begin()
 
     def rollback(self):
         """