From ea4c1389ba7fd3f77160ea68d0e56e510e42a15b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Georges=20Dup=C3=A9ron?= <jahvascriptmaniac+github@free.fr>
Date: Thu, 22 Sep 2011 12:16:50 +0200
Subject: [PATCH] Move database entries for deleted files to a separate history
 table.

---
 updatehash.py  | 11 ++++++++++-
 updatehash.sql |  8 ++++++++
 2 files changed, 18 insertions(+), 1 deletion(-)
 create mode 100644 updatehash.sql

diff --git a/updatehash.py b/updatehash.py
index d437c4b..c583b8f 100755
--- a/updatehash.py
+++ b/updatehash.py
@@ -28,6 +28,7 @@ def initdb(cursor):
 	cursor.execute("create table if not exists files(tag,timestamp,path primary key,md5,sha1,mtime,size)")
 	cursor.execute("create index if not exists i_files_tag on files(tag)")
 	cursor.execute("create index if not exists i_files_path_md5_sha1 on files(path,md5,sha1)")
+	cursor.execute("create table removedfiles(rmtimestamp,tag,timestamp,path,md5,sha1,mtime,size)")
 
 def cacheFileInfo(cursor, path):
 	cursor.execute('select mtime,size from files where path = ?', (path,))
@@ -35,6 +36,8 @@ def cacheFileInfo(cursor, path):
 	return data and {'mtime':data[0], 'size':data[1]}
 
 def update(connection,cursor,path):
+	cursor.execute("create temp table newfiles(path)")
+	cursor.execute("create index i_newfiles_path on newfiles(path)")
 	timestamp = time.time()
 	currentTime = time.clock()
 	lastTime = currentTime
@@ -48,6 +51,7 @@ def update(connection,cursor,path):
 					print "!skipping", fpath
 					continue
 				cfi = cacheFileInfo(cursor,fpath)
+				cursor.execute("insert into newfiles(path) values(?)", (fpath,))
 				if fi != cfi:
 					print " updating", fpath
 					sums = checksumFile(fpath)
@@ -59,6 +63,12 @@ def update(connection,cursor,path):
 						lastTime = currentTime
 						connection.commit()
 						print "commit!"
+	connection.commit()
+	print "commit!"
+	print "cleaning up..."
+	cursor.execute("insert into removedfiles(rmtimestamp,tag,timestamp,path,md5,sha1,mtime,size) select ?,tag,timestamp,path,md5,sha1,mtime,size from files where path not in newfiles", (timestamp,))
+	cursor.execute("delete from files where path not in (select path from newfiles)")
+	connection.commit()
 
 def walk(db,path):
 	connection = sqlite3.connect(db)
@@ -66,7 +76,6 @@ def walk(db,path):
 	cursor = connection.cursor()
 	initdb(cursor)
 	update(connection, cursor, path)
-	connection.commit()
 	cursor.close()
 
 def help():
diff --git a/updatehash.sql b/updatehash.sql
new file mode 100644
index 0000000..08ffee5
--- /dev/null
+++ b/updatehash.sql
@@ -0,0 +1,8 @@
+# Size of duplicates that can be removed (doesn't count the size of the copy we leave)
+select round((B.tot-A.tot)/(1024.*1024.*1024.),2)||' Gb' from (select sum(size) as tot from (select distinct md5,sha1,size from files)) as A, (select sum(size) as tot from (select md5,sha1,size from files)) as B;
+
+# List of duplicates (all copies)
+select size,path from files where md5||'#'||sha1||'#'||size in (select md5||'#'||sha1||'#'||size from files group by md5,sha1,size having count(path) > 1) order by size;
+
+# Total count of files and total weight in Gb
+select round(sum(size)/(1024.*1024.*1024.),2)||' Gb        '||count(size)||' files' from files;