Fix duplicated property problems

Added duplicated_property_types function in CLI

bam_masterdata/cli/cli.py

@@ -8,6 +8,7 @@
 from openpyxl import Workbook
 from rdflib import Graph
 
+from bam_masterdata.cli.duplicated_property_types import duplicated_property_types
 from bam_masterdata.cli.entities_to_excel import entities_to_excel
 from bam_masterdata.cli.entities_to_json import entities_to_json
 from bam_masterdata.cli.entities_to_rdf import entities_to_rdf
@@ -166,6 +167,12 @@ def export_to_json(force_delete, python_path):
 
     # Process each module using the `model_to_json` method of each entity
     for module_path in py_modules:
+        if module_path.endswith("property_types.py"):
+            if duplicated_property_types(module_path=module_path, logger=logger):
+                click.echo(
+                    "Please fix the duplicated property types before exporting to RDF/XML."
+                )
+                return
         entities_to_json(module_path=module_path, export_dir=export_dir, logger=logger)
 
     click.echo(f"All entity artifacts have been generated and saved to {export_dir}")
@@ -216,6 +223,12 @@ def export_to_excel(force_delete, python_path):
     masterdata_file = os.path.join(export_dir, "masterdata.xlsx")
     wb = Workbook()
     for i, module_path in enumerate(py_modules):
+        if module_path.endswith("property_types.py"):
+            if duplicated_property_types(module_path=module_path, logger=logger):
+                click.echo(
+                    "Please fix the duplicated property types before exporting to RDF/XML."
+                )
+                return
         if i == 0:
             ws = wb.active
         else:
@@ -280,6 +293,12 @@ def export_to_rdf(force_delete, python_path):
     # Process each module using the `model_to_rdf` method of each entity
     graph = Graph()
     for module_path in py_modules:
+        if module_path.endswith("property_types.py"):
+            if duplicated_property_types(module_path=module_path, logger=logger):
+                click.echo(
+                    "Please fix the duplicated property types before exporting to RDF/XML."
+                )
+                return
         entities_to_rdf(graph=graph, module_path=module_path, logger=logger)
 
     # Saving RDF/XML to file

bam_masterdata/cli/duplicated_property_types.py

@@ -0,0 +1,34 @@
+import inspect
+import re
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from structlog._config import BoundLoggerLazyProxy
+
+from bam_masterdata.utils import import_module
+
+
+def duplicated_property_types(module_path: str, logger: "BoundLoggerLazyProxy") -> dict:
+    duplicated_props: dict = {}
+    module = import_module(module_path=module_path)
+    source_code = inspect.getsource(module)
+    for name, _ in inspect.getmembers(module):
+        if name.startswith("_") or name == "PropertyTypeDef":
+            continue
+
+        pattern = rf"^\s*{name} *= *PropertyTypeDef"
+
+        # Find all matching line numbers
+        matches = [
+            i + 1  # Convert to 1-based index
+            for i, line in enumerate(source_code.splitlines())
+            if re.match(pattern, line)
+        ]
+        if len(matches) > 1:
+            duplicated_props[name] = matches
+    if duplicated_props:
+        logger.critical(
+            f"Found {len(duplicated_props)} duplicated property types. These are stored in a dictionary "
+            f"where the keys are the names of the variables in property_types.py and the values are the lines in the module: {duplicated_props}"
+        )
+    return duplicated_props

bam_masterdata/datamodel/property_types.py

@@ -3259,7 +3259,8 @@
 )
 
 
-ProductCategory = PropertyTypeDef(
+# ! Duplicated variable name for the property type definition (manually fixed)
+ProductCategory1 = PropertyTypeDef(
     code="PRODUCT_CATEGORY",
     description="""Product Category (corresponds to field `Product Category` in the Hazardous Materials Inventory (GSM) of BAM)//Produktkategorie (entspricht Feld `Verwendungstypen/Produktkategorie` aus dem Gefahrstoffmanagement (GSM) der BAM))""",
     data_type="CONTROLLEDVOCABULARY",
@@ -5897,7 +5898,8 @@
 )
 
 
-ProductCategory = PropertyTypeDef(
+# ! Duplicated variable name for the property type definition (manually fixed)
+ProductCategory2 = PropertyTypeDef(
     code="PRODUCT.CATEGORY",
     description="""Category""",
     data_type="VARCHAR",

bam_masterdata/metadata/definitions.py

@@ -143,7 +143,10 @@ def model_id(cls, data: Any) -> Any:
         Returns:
             Any: The data with the validated fields.
         """
-        data.id = code_to_class_name(data.code)
+        if "PropertyType" in data.name:
+            data.id = code_to_class_name(code=data.code, entity_type="property")
+        else:
+            data.id = code_to_class_name(code=data.code, entity_type="object")
         return data