From 5c3deb341db21c9b3d22ffcbdd452e20b235f082 Mon Sep 17 00:00:00 2001
From: Thomas Rast <trast@student.ethz.ch>
Date: Mon, 9 Feb 2009 23:08:30 +0100
Subject: [PATCH] Refactor blob cache code into separate module

---
 blobtracker.py | 20 +++++---------------
 dbcache.py     | 27 +++++++++++++++++++++++++++
 2 files changed, 32 insertions(+), 15 deletions(-)
 create mode 100644 dbcache.py

diff --git a/blobtracker.py b/blobtracker.py
index 0761090..08dffbc 100644
--- a/blobtracker.py
+++ b/blobtracker.py
@@ -1,31 +1,21 @@
 import sys
 
 import db
+import dbcache
 from git import git
 
-_blob_cache = {}
-
-def get_blob(session, sha1):
-    if sha1 in _blob_cache:
-        return _blob_cache[sha1]
-    b = session.query(db.Blob).filter(db.Blob.sha1==sha1).first()
-    _blob_cache[sha1] = b
-    return b
-
-def scan_commit(session, commit):
+def scan_commit(commit):
     for line in git('ls-tree', '-r', commit.sha1, ret_pipe=True):
         assert line.endswith('\n')
         rest, name = line[:-1].split('\t', 1)
         mode, type, sha1 = rest.split(' ')
-        blob = get_blob(session, sha1)
-        if not blob:
-            blob = db.Blob(sha1)
-            session.add(blob)
-            _blob_cache[sha1] = blob
+        blob = blob_cache.get(sha1)
         blob.update_contained_in(commit)
 
 def scan_history(excludes=[]):
+    global blob_cache
     session = db.Session()
+    blob_cache = dbcache.Cache(db.Blob, db.Blob.sha1, session)
     refdata = git('for-each-ref', '--format=%(objectname)')[0]
     refs = refdata.split()
     boundaries = session.query(db.Boundary).all()
diff --git a/dbcache.py b/dbcache.py
new file mode 100644
index 0000000..0012d4c
--- /dev/null
+++ b/dbcache.py
@@ -0,0 +1,27 @@
+class Cache(object):
+
+    def __init__(self, cls, attr, session):
+        self.cls = cls
+        self.attr = attr
+        self._cache = {}
+        self.session = session
+
+    def __getitem__(self, key):
+        try:
+            return self._cache[key]
+        except KeyError:
+            pass
+        v = session.query(self.cls).filter(self.attr==key).first()
+        if not v:
+            raise KeyError
+        self._cache[key] = v
+        return v
+
+    def get(self, key):
+        try:
+            v = self[key]
+        except KeyError:
+            v = cls(key)
+            self._cache[key] = v
+            session.add(v)
+        return v
-- 
2.11.4.GIT