Refactor pk preservation; Add special transforms

legacy/scripts/migrate.py

@@ -3,46 +3,63 @@ from django.db import connection, models
 from field_mappings import field_mappings
 from migration_base import appconfs, legacy_app
 
+from parlamentares.models import Parlamentar
+from django.core.exceptions import ObjectDoesNotExist
 
-def migrate_all():
+
+def migrate_all(count_limit=None):
     for app in appconfs:
         for model in app.models.values():
-            migrate(model)
+            migrate(model, count_limit)
+
+
+special_transforms = {
+    # Field is defined as not null in legacy db, but data includes null values
+    #  => transform None to False
+    Parlamentar._meta.get_field('unid_deliberativa'): bool,
+}
+
 
+def migrate(model, count_limit=None):
 
-def migrate(model):
+    print 'Migrating %s...' % model.__name__
 
-        print 'Migrating %s...' % model.__name__
+    # clear all model entries
+    model.objects.all().delete()
 
-        # clear all model entries
+    def reset_seq(id):
-        model.objects.all().delete()
+            # resets id sequence
-        # resets id sequence
+        sql_reset_seq = 'ALTER SEQUENCE %s_id_seq RESTART WITH %s;' % (model._meta.db_table, id)
-        sql_reset_seq = 'ALTER SEQUENCE %s_id_seq RESTART WITH 1;' % model._meta.db_table
         cursor = connection.cursor()
         cursor.execute(sql_reset_seq)
 
-        legacy_model = legacy_app.get_model(model.__name__)
+    legacy_model = legacy_app.get_model(model.__name__)
-        old_pk_name = legacy_model._meta.pk.name
+    old_pk_name = legacy_model._meta.pk.name
-        if old_pk_name == 'id':
+    if old_pk_name == 'id':
-            # There is no pk in the legacy table
+        # There is no pk in the legacy table
-            pass
+        reset_seq(1)
-        else:
+        # ...
-            for old in legacy_model.objects.all().order_by(old_pk_name):
+    else:
-                old_pk = getattr(old, old_pk_name)
+        for old in legacy_model.objects.all().order_by(old_pk_name)[:count_limit]:
-                new = model()
+            old_pk = getattr(old, old_pk_name)
-                while not new.id:
+            reset_seq(old_pk)
-                    for new_field, old_field in field_mappings[model].items():
+            new = model()
-                        value = getattr(old, old_field)
+            for new_field, old_field in field_mappings[model].items():
-                        # check for a relation
+                value = getattr(old, old_field)
-                        model_field = model._meta.get_field(new_field)
+                model_field = model._meta.get_field(new_field)
-                        # TODO ... test for special transformations first (e.g. Parlamentar.localidade_residencia)
+                transform = special_transforms.get(model_field)
-                        # elfi ...
+                if transform:
-                        if isinstance(model_field, models.ForeignKey):
+                    value = transform(value)
+                else:
+                    # check for a relation
+                    if isinstance(model_field, models.ForeignKey) and value is not None:
+                        try:
                             value = model_field.related_model.objects.get(id=value)
+                        except ObjectDoesNotExist:
+                            print 'WARNING! FK [%s : %s] not found for value %s (in pk = %s)' % (model.__name__, model_field.name, value, old_pk)
+                            value = None
+                        else:
                             assert value
-                        setattr(new, new_field, value)
+                setattr(new, new_field, value)
-                    new.save()
+            new.save()
-                    assert new.id <= old_pk, 'New id exceeds old pk!'
+            assert new.id == old_pk, 'New id is different from old pk!'
-                    if new.id < old_pk:
-                        new.delete()
-                        new = model()