Merge lp:~therp-nl/ocb-addons/6.1-lp1342360-fetchmail_file_storage_duplicate_attachment_ids into lp:ocb-addons/6.1

[Stefan Rijnhart (Opener)]

See bug description. The duplication is no longer performed in OpenERP 7.0, so it only affects 6.1

[Holger Brunn (Therp) (hbrunn)]

What is the deduplication based on? If it's the file's name, we'd rather have to turn that off. For a mail, it's perfectly valid to have two times 'attachment.pdf' containing different documents.

[Stefan Rijnhart (Opener) (stefan-opener)]

Ouch, you are right. While the file size is passed to _check_duplication, it is ignored further down and any file on the same resource with the same name is considered a dupe. The same thing happens on write(), in which case the data is being overwritten. This could be a 'feature' that other parts of the code depend on, and changing this behaviour could lead to an unexpected explosion of documents with the same name. I need to think about a possible solution to this.

Unmerged revisions

6848. By Stefan Rijnhart (Opener) on 2014-07-15

[FIX] Don't write duplicate attachment ids on incoming mail

=== modified file 'mail/mail_thread.py'
--- mail/mail_thread.py	2013-09-10 15:12:35 +0000
+++ mail/mail_thread.py	2014-07-15 21:27:25 +0000
@@ -221,7 +221,12 @@
                     'res_model': thread._name,
                     'res_id': thread.id,
                 }
-                to_attach.append(ir_attachment.create(cr, uid, data_attach, context=context))
+
+                attachment_id = ir_attachment.create(cr, uid, data_attach, context=context)
+                # Document module can return the same ID multiple times if the
+                # same document is attached twice due to deduplication
+                if attachment_id not in to_attach:
+                    to_attach.append(attachment_id)
 
             partner_id = hasattr(thread, 'partner_id') and (thread.partner_id and thread.partner_id.id or False) or False
             if not partner_id and thread._name == 'res.partner':