From 05b5c7a569c0ca964bf032995334a32f716dce9c Mon Sep 17 00:00:00 2001
From: Stefan Rijnhart <stefan@therp.nl>
Date: Wed, 12 Feb 2014 11:05:38 +0100
Subject: [PATCH] [RFR] Group models per table when detecting columns to purge 
      to prevent problems with models sharing the same table

---
 database_cleanup/model/purge_columns.py | 32 ++++++++++++++++++-------
 1 file changed, 23 insertions(+), 9 deletions(-)

diff --git a/database_cleanup/model/purge_columns.py b/database_cleanup/model/purge_columns.py
index e46a8f0da..2d4f270ef 100644
--- a/database_cleanup/model/purge_columns.py
+++ b/database_cleanup/model/purge_columns.py
@@ -85,18 +85,21 @@ class CleanupPurgeWizardColumn(orm.TransientModel):
             res['name'] = _('Purge columns')
         return res
 
-    def get_orphaned_columns(self, cr, uid, model_pool, context=None):
+    def get_orphaned_columns(self, cr, uid, model_pools, context=None):
         """
         From openobject-server/openerp/osv/orm.py
         Iterate on the database columns to identify columns
         of fields which have been removed
         """
-        columns = [
-            c for c in model_pool._columns
-            if not (isinstance(model_pool._columns[c], fields.function)
-                    and not model_pool._columns[c].store)]
+        
+        columns = list(set([
+            column for model_pool in model_pools 
+            for column in model_pool._columns
+            if not (isinstance(model_pool._columns[column], fields.function)
+                    and not model_pool._columns[column].store)
+            ]))
         columns += orm.MAGIC_COLUMNS
-        columns += self.blacklist.get(model_pool._table, [])
+        columns += self.blacklist.get(model_pools[0]._table, [])
 
         cr.execute("SELECT a.attname"
                    "  FROM pg_class c, pg_attribute a"
@@ -106,26 +109,37 @@ class CleanupPurgeWizardColumn(orm.TransientModel):
                    "   AND pg_catalog.format_type(a.atttypid, a.atttypmod)"
                    "        NOT IN ('cid', 'tid', 'oid', 'xid')"
                    "   AND a.attname NOT IN %s",
-                   (model_pool._table, False, tuple(columns))),
+                   (model_pools[0]._table, False, tuple(columns))),
         return [column[0] for column in cr.fetchall()]
 
     def find(self, cr, uid, context=None):
         """
         Search for columns that are not in the corresponding model.
+
+        Group models by table to prevent false positives for columns
+        that are only in some of the models sharing the same table.
+        Example of this is 'sale_id' not being a field of stock.picking.in
         """
         res = []
         model_pool = self.pool['ir.model']
         model_ids = model_pool.search(cr, uid, [], context=context)
         line_pool = self.pool['cleanup.purge.line.column']
+
+        # mapping of tables to tuples (model id, [pool1, pool2, ...])
+        table2model = {}
+
         for model in model_pool.browse(cr, uid, model_ids, context=context):
             model_pool = self.pool.get(model.model)
             if not model_pool or not model_pool._auto:
                 continue
+            table2model.setdefault(model_pool._table, (model.id, []))[1].append(model_pool)
+        
+        for table, model_spec in table2model.iteritems():
             for column in self.get_orphaned_columns(
-                    cr, uid, model_pool, context=context):
+                    cr, uid, model_spec[1], context=context):
                 res.append((0, 0, {
                             'name': column,
-                            'model_id': model.id}))
+                            'model_id': model_spec[0]}))
         if not res:
             raise orm.except_orm(
                 _('Nothing to do'),