Rewrite (from bzed), making this script actually work correctly if the input data...

diff --git a/scripts/debian/insert_missing_changedby.py b/scripts/debian/insert_missing_changedby.py
index 7b817b5e7413843f10aeecc9e2edc6d27f500e65..e3148301e4516599b6234fac9e7d9232a9806cf6 100755 (executable)
--- a/scripts/debian/insert_missing_changedby.py
+++ b/scripts/debian/insert_missing_changedby.py
@@ -1,4 +1,5 @@
 #!/usr/bin/env python
+# -*- coding: utf-8 -*-
 
 # Adds yet unknown changedby fields when this column is added to an existing
 # database. If everything goes well, it needs to be run only once. Data is
@@ -6,6 +7,8 @@
 # merkel:/home/filippo/upload-history/*.db.
 
 # Copyright (C) 2008  Christoph Berg <myon@debian.org>
+# Copyright (C) 2008  Bernd Zeimetz <bzed@debian.org>
+
 
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
@@ -35,51 +38,104 @@ import daklib.database
 import daklib.queue
 import daklib.utils
 from pysqlite2 import dbapi2 as sqlite
+import pysqlite2.dbapi2
+import psycopg2
 
 projectB = None
-DBNAME = "uploads-ddc.db"
+projectBdb = None
+DBNAME = "uploads-queue.db"
 sqliteConn = None
+maintainer_id_cache={}
 
 ###############################################################################
 
+def get_or_set_maintainer_id (maintainer):
+    global maintainer_id_cache
+
+    if maintainer_id_cache.has_key(maintainer):
+        return maintainer_id_cache[maintainer]
+
+    if isinstance(maintainer, basestring):
+        if not isinstance(maintainer, unicode):
+           try:
+                maintainer = unicode(maintainer, 'utf-8')
+            except:
+                maintainer = unicode(maintainer, 'iso8859-15')
+    maintainer = maintainer.encode('utf-8')
+
+    print "%s" % maintainer
+    cursor = projectBdb.cursor()
+    cursor.execute("SELECT id FROM maintainer WHERE name=%s", (maintainer, ))
+    row = cursor.fetchone()
+    if not row:
+        cursor.execute("INSERT INTO maintainer (name) VALUES (%s)" , (maintainer, ))
+        cursor.execute("SELECT id FROM maintainer WHERE name=%s", (maintainer, ))
+        row = cursor.fetchone()
+    maintainer_id = row[0]
+    maintainer_id_cache[maintainer] = maintainer_id
+    cursor.close()
+
+    return maintainer_id
+
+
+def __get_changedby__(package, version):
+    cur = sqliteConn.cursor()
+    cur.execute("SELECT changedby FROM uploads WHERE package=? AND version=? LIMIT 1", (package, version))
+    res = cur.fetchone()
+    cur.close()
+    return res
+
 def insert ():
     print "Adding missing changedby fields."
 
-    projectB.query("BEGIN WORK")
-
-    q = projectB.query("SELECT id, source, version FROM source WHERE changedby IS NULL")
-
-    for i in q.getresult():
-        print i[1] + "/" + i[2] + ":",
-
-        cur = sqliteConn.cursor()
-        cur.execute("SELECT changedby FROM uploads WHERE package = '%s' AND version = '%s' LIMIT 1" % (i[1], i[2]))
-        res = cur.fetchall()
-        if len(res) != 1:
+    listcursor = projectBdb.cursor()
+    listcursor.execute("SELECT id, source, version FROM source WHERE changedby IS NULL")
+    row = listcursor.fetchone()
+
+    while row:
+        print repr(row)
+        try:
+            res = __get_changedby__(row[1], row[2])
+        except:
+            sqliteConn.text_factory = str
+            try:
+                res = __get_changedby__(row[1], row[2])
+            except:
+                print 'FAILED SQLITE'
+                res=None
+            sqliteConn.text_factory = unicode
+        if res:
+            changedby_id = get_or_set_maintainer_id(res[0])
+        
+            cur = projectBdb.cursor()
+            cur.execute("UPDATE source SET changedby=%s WHERE id=%s" % (changedby_id, row[0]))
+            cur.close()
+            print changedby_id, "(%d)" % row[0]
+
+        else:
             print "nothing found"
-            continue
 
-        changedby = res[0][0].replace("'", "\\'")
-        changedby_id = daklib.database.get_or_set_maintainer_id(changedby)
-
-        projectB.query("UPDATE source SET changedby = %d WHERE id = %d" % (changedby_id, i[0]))
-        print changedby, "(%d)" % changedby_id
-
-    projectB.query("COMMIT WORK")
+        row = listcursor.fetchone()
+    listcursor.close()
 
 ###############################################################################
 
+
 def main():
-    global projectB, sqliteConn
+    global projectB, sqliteConn, projectBdb
 
     Cnf = daklib.utils.get_conf()
     Upload = daklib.queue.Upload(Cnf)
     projectB = Upload.projectB
+    projectBdb = psycopg2.connect("dbname=%s" % Cnf["DB::Name"])
 
     sqliteConn = sqlite.connect(DBNAME)
 
     insert()
 
+    projectBdb.commit()
+    projectBdb.close()
+
 ###############################################################################
 
 if __name__ == '__main__':