Reference fields (#3267)

* Adds a configurable 'reference pattern' to the IndexingReferenceMixin class

* Expand tests for reference_pattern validator:

- Prevent inclusion of illegal characters
- Prevent multiple groups of hash (#) characters
- Add unit tests

* Validator now checks for valid strftime formatter

* Adds build order reference pattern

* Adds function for creating a valid regex from the supplied pattern

- More unit tests
- Use it to validate BuildOrder reference field

* Refactoring the whole thing again - try using python string.format

* remove datetime-matcher from requirements.txt

* Add some more formatting helper functions

- Construct a regular expression from a format string
- Extract named values from a string, based on a format string

* Fix validator for build order reference field

* Adding unit tests for the new format string functionality

* Adds validation for reference fields

* Require the 'ref' format key as part of a valid reference pattern

* Extend format extraction to allow specification of integer groups

* Remove unused import

* Fix requirements

* Add method for generating the 'next' reference field for a model

* Fix function for generating next BuildOrder reference value

- A function is required as class methods cannot be used
- Simply wraps the existing class method

* Remove BUILDORDER_REFERENCE_REGEX setting

* Add unit test for build order reference field validation

* Adds unit testing for extracting integer values from a reference field

* Fix bugs from previous commit

* Add unit test for generation of default build order reference

* Add data migration for BuildOrder model

- Update reference field with old prefix
- Construct new pattern based on old prefix

* Adds unit test for data migration

- Check that the BuildOrder reference field is updated as expected

* Remove 'BUILDORDER_REFERENCE_PREFIX' setting

* Adds new setting for SalesOrder reference pattern

* Update method by which next reference value is generated

* Improved error handling in api_tester code

* Improve automated generation of order reference fields

- Handle potential errors
- Return previous reference if something goes wrong

* SalesOrder reference has now been updated also

- New reference pattern setting
- Updated default and validator for reference field
- Updated serializer and API
- Added unit tests

* Migrate the "PurchaseOrder" reference field to the new system

* Data migration for SalesOrder and PurchaseOrder reference fields

* Remove PURCHASEORDER_REFERENCE_PREFIX

* Remove references to SALESORDER_REFERENCE_PREFIX

* Re-add maximum value validation

* Bug fixes

* Improve algorithm for generating new reference

- Handle case where most recent reference does not conform to the reference pattern

* Fixes for 'order' unit tests

* Unit test fixes for order app

* More unit test fixes

* More unit test fixing

* Revert behaviour for "extract_int" clipping function

* Unit test value fix

* Prevent build order notification if we are importing records

InvenTree/stock/models.py

@@ -30,8 +30,7 @@ import report.models
 from company import models as CompanyModels
 from InvenTree.fields import (InvenTreeModelMoneyField, InvenTreeNotesField,
                               InvenTreeURLField)
-from InvenTree.models import InvenTreeAttachment, InvenTreeTree
-from InvenTree.serializers import extract_int
+from InvenTree.models import InvenTreeAttachment, InvenTreeTree, extract_int
 from InvenTree.status_codes import StockHistoryCode, StockStatus
 from part import models as PartModels
 from plugin.events import trigger_event
@@ -1708,8 +1707,7 @@ class StockItem(MetadataMixin, MPTTModel):
             s += ' @ {loc}'.format(loc=self.location.name)
 
         if self.purchase_order:
-            s += " ({pre}{po})".format(
-                pre=InvenTree.helpers.getSetting("PURCHASEORDER_REFERENCE_PREFIX"),
+            s += " ({po})".format(
                 po=self.purchase_order,
             )
 

InvenTree/stock/serializers.py

@@ -20,7 +20,8 @@ import InvenTree.serializers
 import part.models as part_models
 from common.settings import currency_code_default, currency_code_mappings
 from company.serializers import SupplierPartSerializer
-from InvenTree.serializers import InvenTreeDecimalField, extract_int
+from InvenTree.models import extract_int
+from InvenTree.serializers import InvenTreeDecimalField
 from part.serializers import PartBriefSerializer
 
 from .models import (StockItem, StockItemAttachment, StockItemTestResult,
@@ -67,8 +68,8 @@ class StockItemSerializerBrief(InvenTree.serializers.InvenTreeModelSerializer):
 
     def validate_serial(self, value):
         """Make sure serial is not to big."""
-        if extract_int(value) > 2147483647:
-            raise serializers.ValidationError('serial is to to big')
+        if abs(extract_int(value)) > 0x7fffffff:
+            raise serializers.ValidationError(_("Serial number is too large"))
         return value
 
 

InvenTree/stock/tests.py

@@ -87,7 +87,7 @@ class StockTest(InvenTreeTestCase):
         # And there should be *no* items being build
         self.assertEqual(part.quantity_being_built, 0)
 
-        build = Build.objects.create(reference='12345', part=part, title='A test build', quantity=1)
+        build = Build.objects.create(reference='BO-4444', part=part, title='A test build', quantity=1)
 
         # Add some stock items which are "building"
         for _ in range(10):
@@ -395,13 +395,14 @@ class StockTest(InvenTreeTestCase):
         item.serial = "-123"
         item.save()
 
-        # Negative number should map to zero
-        self.assertEqual(item.serial_int, 0)
+        # Negative number should map to positive value
+        self.assertEqual(item.serial_int, 123)
 
         # Test a very very large value
         item.serial = '99999999999999999999999999999999999999999999999999999'
         item.save()
 
+        # The 'integer' portion has been clipped to a maximum value
         self.assertEqual(item.serial_int, 0x7fffffff)
 
         # Non-numeric values should encode to zero