diff --git a/src/linkahead/apiutils.py b/src/linkahead/apiutils.py
index 53c9957e045055d4bc69fb240ef651fe9a5e71b0..35d33c8c6ab2d9fc0224f9755a4f06f376a14797 100644
--- a/src/linkahead/apiutils.py
+++ b/src/linkahead/apiutils.py
@@ -244,6 +244,8 @@ def compare_entities(entity0: Entity, entity1: Entity,
# - Do different inheritance levels for parents count as a difference?
# - Do we care about parents and properties of properties?
# - Should there be a more detailed comparison of parents without id?
+ # - Revisit filter - do we care about RecordType when matching?
+ # How to treat None?
# Suggestions for enhancements:
# - For the comparison of entities in value and properties, consider
# keeping a list of traversed entities, not only look at first layer
@@ -321,8 +323,12 @@ def compare_entities(entity0: Entity, entity1: Entity,
# Compare Entities
if (compare_referenced_records and
isinstance(val0, Entity) and isinstance(val1, Entity)):
- if empty_diff(val0, val1, False,
- entity_name_id_equivalency):
+ try:
+ same = empty_diff(val0, val1, False,
+ entity_name_id_equivalency)
+ except:
+ same = False
+ if same:
continue
# Compare Entity name and id
if entity_name_id_equivalency:
@@ -381,10 +387,10 @@ def compare_entities(entity0: Entity, entity1: Entity,
# As the importance of a property is an attribute of the record
# and not the property, it is not contained in the diff returned
# by compare_entities and needs to be added separately
- if (entity0.get_importance(prop.name) !=
- entity1.get_importance(prop.name)):
- propdiff[0]["importance"] = entity0.get_importance(prop.name)
- propdiff[1]["importance"] = entity1.get_importance(prop.name)
+ if (entity0.get_importance(prop) !=
+ entity1.get_importance(matching[0])):
+ propdiff[0]["importance"] = entity0.get_importance(prop)
+ propdiff[1]["importance"] = entity1.get_importance(matching[0])
# in case there is no difference, we remove the dict keys again
if len(propdiff[0]) == 0 and len(propdiff[1]) == 0:
diff --git a/src/linkahead/common/models.py b/src/linkahead/common/models.py
index ee42e06a8baac03f0e4ffbfdf9a30626f4b0d9ab..d3ef1e525a347833d341d399bb08a3dc511b32ae 100644
--- a/src/linkahead/common/models.py
+++ b/src/linkahead/common/models.py
@@ -114,7 +114,7 @@ if TYPE_CHECKING:
IMPORTANCE = Literal["OBLIGATORY", "RECOMMENDED", "SUGGESTED", "FIX", "NONE"]
ROLE = Literal["Entity", "Record", "RecordType", "Property", "File"]
-SPECIAL_ATTRIBUTES = ["name", "role", "datatype", "description",
+SPECIAL_ATTRIBUTES = ["name", "role", "datatype", "description", "file",
"id", "path", "checksum", "size", "value", "unit"]
diff --git a/unittests/test_apiutils.py b/unittests/test_apiutils.py
index fe2155e5881f600a213149a93d9e38ab5ef0467b..c11cecdf2175b12723fc9ddecf963266d967b6c9 100644
--- a/unittests/test_apiutils.py
+++ b/unittests/test_apiutils.py
@@ -25,6 +25,7 @@
# Test apiutils
# A. Schlemmer, 02/2018
+from io import StringIO
import linkahead as db
import linkahead.apiutils
@@ -166,6 +167,207 @@ def test_compare_entities():
assert diff_r2["properties"]["test"]["unit"] == "m"
+def test_compare_entities_battery():
+ par1, par2, par3 = db.Record(), db.Record(), db.RecordType()
+ r1, r2, r3 = db.Record(), db.Record(), db.Record()
+ prop1 = db.Property()
+ prop2 = db.Property(name="Property 2")
+ prop3 = db.Property()
+
+ ## Basic tests for Properties
+ prop_settings = {"datatype": db.REFERENCE, "description": "desc of prop",
+ "value": db.Record().add_parent(par3), "unit": '°'}
+ t1 = db.Record().add_parent(db.RecordType())
+ t2 = db.Record().add_parent(db.RecordType())
+ # Change datatype
+ t1.add_property(db.Property(name="datatype", **prop_settings))
+ prop_settings["datatype"] = par3
+ t2.add_property(db.Property(name="datatype", **prop_settings))
+ # Change description
+ t1.add_property(db.Property(name="description", **prop_settings))
+ prop_settings["description"] = "diff desc"
+ t2.add_property(db.Property(name="description", **prop_settings))
+ # Change value to copy
+ t1.add_property(db.Property(name="value copy", **prop_settings))
+ prop_settings["value"] = db.Record().add_parent(par3)
+ t2.add_property(db.Property(name="value copy", **prop_settings))
+ # Change value to something different
+ t1.add_property(db.Property(name="value", **prop_settings))
+ prop_settings["value"] = db.Record(name="n").add_parent(par3)
+ t2.add_property(db.Property(name="value", **prop_settings))
+ # Change unit
+ t1.add_property(db.Property(name="unit", **prop_settings))
+ prop_settings["unit"] = db.Property(unit='°')
+ t2.add_property(db.Property(name="unit", **prop_settings))
+ # Change unit again
+ t1.add_property(db.Property(name="unit 2", **prop_settings))
+ prop_settings["unit"] = db.Property()
+ t2.add_property(db.Property(name="unit 2", **prop_settings))
+ # Compare
+ diff_0 = compare_entities(t1, t2)
+ diff_1 = compare_entities(t1, t2, compare_referenced_records=True)
+ # Check correct detection of changes
+ assert diff_0[0]["properties"]["datatype"] == {"datatype": db.REFERENCE}
+ assert diff_0[1]["properties"]["datatype"] == {"datatype": par3}
+ assert diff_0[0]["properties"]["description"] == {"description": "desc of prop"}
+ assert diff_0[1]["properties"]["description"] == {"description": "diff desc"}
+ assert "value" in diff_0[0]["properties"]["value copy"]
+ assert "value" in diff_0[1]["properties"]["value copy"]
+ assert "value" in diff_0[0]["properties"]["value"]
+ assert "value" in diff_0[1]["properties"]["value"]
+ assert "unit" in diff_0[0]["properties"]["unit"]
+ assert "unit" in diff_0[1]["properties"]["unit"]
+ assert "unit" in diff_0[0]["properties"]["unit 2"]
+ assert "unit" in diff_0[1]["properties"]["unit 2"]
+ # Check correct result for compare_referenced_records=True
+ assert "value copy" not in diff_1[0]["properties"]
+ assert "value copy" not in diff_1[1]["properties"]
+ diff_0[0]["properties"].pop("value copy")
+ diff_0[1]["properties"].pop("value copy")
+ assert diff_0 == diff_1
+
+ ## Basic tests for Parents
+ t3 = db.Record().add_parent(db.RecordType("A")).add_parent(db.Record("B"))
+ t4 = db.Record().add_parent(db.RecordType("A"))
+ assert compare_entities(t3, t4)[0]['parents'] == ['B']
+ assert len(compare_entities(t3, t4)[1]['parents']) == 0
+ t4.add_parent(db.Record("B"))
+ assert empty_diff(t3, t4)
+ # The two following assertions document current behaviour but do not make a
+ # lot of sense
+ t4.add_parent(db.Record("B"))
+ assert empty_diff(t3, t4)
+ t3.add_parent(db.RecordType("A")).add_parent(db.Record("B"))
+ t4.add_parent(db.RecordType("B")).add_parent(db.Record("A"))
+ assert empty_diff(t3, t4)
+
+ ## Basic tests for special attributes
+ prop_settings = {"id": 42, "name": "Property",
+ "datatype": db.LIST(db.REFERENCE), "value": [db.Record()],
+ "unit": '€', "description": "desc of prop"}
+ alt_settings = {"id": 64, "name": "Property 2",
+ "datatype": db.LIST(db.TEXT), "value": [db.RecordType()],
+ "unit": '€€', "description": " ę Ě ப ཾ ཿ ∛ ∜ ㅿ ㆀ 값 "}
+ t5 = db.Property(**prop_settings)
+ t6 = db.Property(**prop_settings)
+ assert empty_diff(t5, t6)
+ # ID
+ t5.id = alt_settings['id']
+ diff = compare_entities(t5, t6)
+ assert diff[0] == {'properties': {}, 'parents': [], 'id': alt_settings['id']}
+ assert diff[1] == {'properties': {}, 'parents': [], 'id': prop_settings['id']}
+ t6.id = alt_settings['id']
+ assert empty_diff(t5, t6)
+ # Name
+ t5.name = alt_settings['name']
+ diff = compare_entities(t5, t6)
+ assert diff[0] == {'properties': {}, 'parents': [], 'name': alt_settings['name']}
+ assert diff[1] == {'properties': {}, 'parents': [], 'name': prop_settings['name']}
+ t6.name = alt_settings['name']
+ assert empty_diff(t5, t6)
+ # Description
+ t6.description = alt_settings['description']
+ diff = compare_entities(t5, t6)
+ assert diff[0] == {'properties': {}, 'parents': [], 'description': prop_settings['description']}
+ assert diff[1] == {'properties': {}, 'parents': [], 'description': alt_settings['description']}
+ t5.description = alt_settings['description']
+ assert empty_diff(t5, t6)
+ # Unit
+ t5.unit = alt_settings['unit']
+ diff = compare_entities(t5, t6)
+ assert diff[0] == {'properties': {}, 'parents': [], 'unit': alt_settings['unit']}
+ assert diff[1] == {'properties': {}, 'parents': [], 'unit': prop_settings['unit']}
+ t6.unit = alt_settings['unit']
+ assert empty_diff(t5, t6)
+ # Value
+ t6.value = alt_settings['value']
+ diff = compare_entities(t5, t6)
+ assert diff[0] == {'properties': {}, 'parents': [], 'value': prop_settings['value']}
+ assert diff[1] == {'properties': {}, 'parents': [], 'value': alt_settings['value']}
+ t5.value = alt_settings['value']
+ assert empty_diff(t5, t6)
+ # Datatype
+ t6.datatype = alt_settings['datatype']
+ diff = compare_entities(t5, t6)
+ assert diff[0] == {'properties': {}, 'parents': [], 'datatype': prop_settings['datatype']}
+ assert diff[1] == {'properties': {}, 'parents': [], 'datatype': alt_settings['datatype']}
+ t5.datatype = alt_settings['datatype']
+ assert empty_diff(t5, t6)
+ # All at once
+ diff = compare_entities(db.Property(**prop_settings), db.Property(**alt_settings))
+ assert diff[0] == {'properties': {}, 'parents': [], **prop_settings}
+ assert diff[1] == {'properties': {}, 'parents': [], **alt_settings}
+ # Entity Type
+ diff = compare_entities(db.Property(value=db.Property(id=101)),
+ db.Property(value=db.Record(id=101)))
+ assert "value" in diff[0]
+ assert "value" in diff[1]
+ diff = compare_entities(db.Property(value=db.Record(id=101)),
+ db.Property(value=db.Record(id=101)))
+ assert "value" in diff[0]
+ assert "value" in diff[1]
+ assert empty_diff(db.Property(value=db.Record(id=101)),
+ db.Property(value=db.Record(id=101)),
+ compare_referenced_records=True)
+
+ ## Special cases
+ # Files
+ assert not empty_diff(db.File(path='ABC', file=StringIO("ABC")),
+ db.File(path='ABC', file=StringIO("Other")))
+ # Importance
+ assert empty_diff(db.Property().add_property(prop1),
+ db.Property().add_property(prop1))
+ assert not empty_diff(db.Property().add_property(prop1, importance=db.SUGGESTED),
+ db.Property().add_property(prop1, importance=db.OBLIGATORY))
+ # Mixed Lists
+ assert empty_diff(db.Property(value=[1, 2, 'a', r1]),
+ db.Property(value=[1, 2, 'a', r1]))
+ # entity_name_id_equivalency
+ assert not empty_diff(db.Property(value=[1, db.Record(id=2), 3, db.Record(id=4)]),
+ db.Property(value=[db.Record(id=1), 2, db.Record(id=3), 4]))
+ assert empty_diff(db.Property(value=[1, db.Record(id=2), 3, db.Record(id=4)]),
+ db.Property(value=[db.Record(id=1), 2, db.Record(id=3), 4]),
+ entity_name_id_equivalency=True)
+ assert empty_diff(db.Property(value=1), db.Property(value=db.Record(id=1)),
+ entity_name_id_equivalency=True)
+ # entity_name_id_equivalency
+ prop4 = db.Property(**prop_settings).add_parent(par1).add_property(prop2)
+ prop4_c = db.Property(**prop_settings).add_parent(par1).add_property(prop2)
+ prop4.value = db.Record(id=12)
+ prop4_c.value = '12'
+ prop4.add_property(db.Property(name="diff", datatype=db.LIST(db.REFERENCE),
+ value=[12, db.Record(id=13), par1, "abc%"]))
+ prop4_c.add_property(db.Property(name="diff", datatype=db.LIST(db.REFERENCE),
+ value=[db.Record(id=12), "13", par1, "abc%"]))
+ assert not empty_diff(prop4, prop4_c, entity_name_id_equivalency=False)
+ assert empty_diff(prop4, prop4_c, entity_name_id_equivalency=True)
+ # Order invariance
+ t7 = db.Property(**prop_settings).add_parent(par1).add_property(prop1)
+ t8 = db.Property(**alt_settings).add_parent(par3).add_property(prop3)
+ try:
+ diffs_0 = compare_entities(t7, t8), compare_entities(t7, t8, True)
+ except:
+ diffs_0 = None
+ try:
+ diffs_1 = compare_entities(t8, t7)[::-1], compare_entities(t8, t7, True)[::-1]
+ except:
+ diffs_1 = None
+ assert diffs_0 == diffs_1
+ prop_settings = {"datatype": db.REFERENCE, "description": "desc of prop",
+ "value": db.Record().add_parent(par3), "unit": '°'}
+ t1.add_property(db.Property(name="description", **prop_settings))
+ t2.add_property(db.Property(name="description", **prop_settings))
+ try:
+ diffs_0 = compare_entities(t1, t2), compare_entities(t1, t2, True)
+ except Exception as e:
+ diffs_0 = type(e)
+ try:
+ diffs_1 = compare_entities(t2, t1)[::-1], compare_entities(t2, t1, True)[::-1]
+ except Exception as e:
+ diffs_1 = type(e)
+ assert diffs_0 == diffs_1
+
+
def test_compare_special_properties():
# Test for all known special properties:
INTS = ("size", "id")