splitting doc number in parts

themis/libellioimport/configure.zcml

@@ -42,4 +42,12 @@
       permission="cmf.ManagePortal"
       />
 
+  <browser:view
+      name="splitdocnumbers"
+      for="*"
+      class=".migration.UpdateToBlobs"
+      permission="cmf.ManagePortal"
+      />
+
+
 </configure>

themis/libellioimport/migration.py

@@ -6,6 +6,7 @@ import datetime
 import cStringIO
 import xml.etree.ElementTree as ET
 
+import transaction
 from zope.app.intid.interfaces import IIntIds
 from z3c.relationfield import RelationValue
 from Products.Five.browser import BrowserView
@@ -22,6 +23,9 @@ except ImportError:
     from plone.namedfile.file import NamedImage
 from Products.CMFCore.WorkflowCore import WorkflowException
 
+from zope.event import notify
+from zope.lifecycleevent import ObjectAddedEvent, ObjectModifiedEvent
+
 from plone.registry.interfaces import IRegistry
 from zope import component
 
@@ -513,3 +517,43 @@ class UpdateToBlobs(BrowserView):
                 print 'transforming file in', object.id
                 newv = NamedFile(v.data, filename=v.filename)
                 setattr(object, attr, newv)
+
+
+class SplitDocNumbers(BrowserView):
+    def __call__(self):
+        catalog = getToolByName(self.context, 'portal_catalog')
+        for count, brain in enumerate(catalog.search({'portal_type': ['bqrD', 'biqD',
+                'cr_seanceD', 'ProjetD', 'proposition_decret_reglementD',
+                'proposition_resolution_modificationD', 'PropositionD',
+                'rapports_commissionsD']})):
+            object = brain.getObject()
+            if object.numero_suite:
+                continue # already filled
+            attr_name = None
+            attr_value = None
+            for attr in ('numero_biq', 'numero_document'):
+                if hasattr(object, attr):
+                    attr_name = attr
+                    attr_value = getattr(object, attr)
+                    break
+            if not attr_name:
+                continue
+            try:
+                new_number = re.findall(r'(\d+) \(', attr_value)[0]
+            except IndexError:
+                print 'failed to get doc number for', attr_value
+                continue
+            no_suite = None
+            if '°' in attr_value:
+                try:
+                    no_suite = re.findall(r'°\s*(\d+)', attr_value)[0]
+                except IndexError:
+                    print 'failed to get suite number for', attr_value
+                    continue
+            setattr(object, attr, new_number)
+            if no_suite:
+                object.numero_suite = no_suite
+
+            notify(ObjectModifiedEvent(object))
+            if count%100 == 0:
+                transaction.commit()