Add support for external-id properties on lexemes

LexData/claim.py

@@ -110,6 +110,8 @@ def pure_value(self) -> Union[str, int, float, Tuple[float, float]]:
             return value["id"]
         if vtype == "string":
             return value
+        if vtype == "external-id":
+            return value
         if vtype == "monolingualtext":
             return value["text"]
         if vtype == "quantity":

LexData/entity.py

@@ -3,6 +3,7 @@
 from typing import Dict, List, Union
 
 from .claim import Claim
+from .utils import getPropertyType
 from .wikidatasession import WikidataSession
 
 
@@ -87,19 +88,51 @@ def __setEntityClaim__(self, idProp: str, idStr: str):
         :param idProp: id of the property (example: "P31")
         :param idItem: id of the entity (example: "Q1")
         """
-        entityId = int(idStr[1:])
-        claim_value = json.dumps({"entity-type": "item", "numeric-id": entityId})
-        self.__setClaim__(idProp, claim_value)
+        # Check if this is an external-id property
+        datatype = None
+        try:
+            datatype = getPropertyType(idProp)
+        except Exception:
+            # If we can't get the property type, assume it's an entity
+            pass
+        if datatype == "external-id":
+            # For external-id properties, create a Claim object and use __setClaims__
+            claim = Claim(propertyId=idProp, value=idStr)
+            self.__setClaims__([claim])
+            return
+        # Handle entity-type properties as before
+        if idStr.startswith(('Q', 'P', 'L')):
+            entityId = int(idStr[1:])
+            claim_value = json.dumps({"entity-type": "item", "numeric-id": entityId})
+            self.__setClaim__(idProp, claim_value)
+        else:
+            raise ValueError(f"Invalid entity ID format: {idStr}. Expected Q, P, or L prefix.")
 
     def __setClaim__(self, idProp: str, claim_value):
+        from .utils import getPropertyType
+        import LexData.lexeme
+        is_lexeme = isinstance(self, LexData.lexeme.Lexeme)
+        if isinstance(claim_value, Claim):
+            datatype = getPropertyType(idProp)
+            if datatype == "external-id":
+                claim_value = claim_value.pure_value
+            else:
+                snak_data = claim_value["mainsnak"]
+                claim_value = json.dumps(snak_data["datavalue"])
+        datatype = getPropertyType(idProp)
+        if datatype == "external-id":
+            claim_value_json = json.dumps(claim_value)
+        else:
+            claim_value_json = claim_value
+        # Use wbcreateclaim for both lexemes and other entities
         PARAMS = {
             "action": "wbcreateclaim",
             "format": "json",
             "entity": self.id,
             "snaktype": "value",
             "bot": "1",
             "property": idProp,
-            "value": claim_value,
+            "value": claim_value_json,
             "token": "__AUTO__",
         }
 

LexData/utils.py

@@ -39,6 +39,13 @@ def buildDataValue(datatype: str, value):
             raise TypeError(
                 f"Can not convert type {type(value)} to datatype {datatype}"
             )
+    elif datatype == "external-id":
+        if type(value) == str:
+            return {"value": value, "type": "external-id"}
+        else:
+            raise TypeError(
+                f"Can not convert type {type(value)} to datatype {datatype}"
+            )
     elif datatype in [
         "string",
         "tabular-data",

README.md

@@ -8,6 +8,14 @@ power of the access to the internals.
 LexData is still in beta phase and there fore some features are missing and
 functions might be renamed in future.
 
+## Features
+
+- Create and manage Wikidata Lexemes
+- Add forms and senses to lexemes
+- Add claims to lexemes, forms, and senses (including external-id properties)
+- Search and find existing lexemes
+- Support for various Wikidata data types (entities, strings, external-ids, etc.)
+
 The code of AitalvivemBot was used as a starting point, but probably theres not
 a single line of code that wasn't rewritten.
 

example.py

@@ -29,7 +29,7 @@
 if len(L2.forms) == 0:
     L2.createForm("firsts", ["Q146786"])
 
-# …or senses, with or without additional claims
+# …or senses, with or without additional claims…
 if len(L2.senses) == 0:
     L2.createSense(
         {
@@ -38,3 +38,8 @@
         },
         claims={"P5137": ["Q19269277"]},
     )
+
+# …and add external-id claim to lexeme
+if len(L2.claims.get("P12682", [])) == 0:
+    external_id_claim = LexData.Claim("P12682", "example_50bcf7bc0a0ae2bab9011b09139f6f8a")
+    L2.addClaims([external_id_claim])