kodexa 7.4.412833907659__py3-none-any.whl → 7.4.413041559723__py3-none-any.whl

kodexa/dataclasses/__init__.py

@@ -70,6 +70,18 @@ class LLMDataAttribute(BaseModel):
         )
         self.exceptions.append(content_exception)
 
+    def to_dict(self, taxonomy: Taxonomy) -> dict:
+        """Convert attribute to JSON with normalized value"""
+
+        target_taxon = taxonomy.get_taxon_by_path(self.taxon_path)
+        if target_taxon is None:
+            return {}
+
+        taxon_external_name = target_taxon.external_name
+        return {
+            taxon_external_name: self.normalized_text if self.normalized_text else self.value
+        }
+
 
 class LLMDataObject(BaseModel):
     """
@@ -154,6 +166,19 @@ class LLMDataObject(BaseModel):
         if 'normalized_text' in field_data:
             attr.normalized_text = field_data['normalized_text']
 
+    def to_dict(self, taxonomy: Taxonomy) -> dict:
+        """Convert data object to JSON using normalized values and taxon paths"""
+        result = {}
+        for field in self.__fields__:
+            value = getattr(self, field)
+            if isinstance(value, list):
+                result[field] = [item.to_dict(taxonomy) for item in value if isinstance(item, (LLMDataObject, LLMDataAttribute))]
+            elif isinstance(value, LLMDataAttribute):
+                result.update(value.to_dict(taxonomy))
+            elif isinstance(value, LLMDataObject):
+                result[field] = value.to_dict(taxonomy)
+        return result
+
     def to_review(self, page_number=None):
         """
         Build a representation of the data object and its data attributes that is a dict that includes the

kodexa/model/objects.py

@@ -2731,6 +2731,16 @@ class Taxon(BaseModel):
             structure["taxonType"] = self.taxon_type
         return structure
 
+    def get_taxon_by_path(self, path):
+        if self.path == path:
+            return self
+
+        if self.children:
+            for child in self.children:
+                result = child.get_taxon_by_path(path)
+                if result:
+                    return result
+
 
 class ContentObject(BaseModel):
     """
@@ -5667,6 +5677,16 @@ class Taxonomy(ExtensionPackProvided):
         description="A list of references to an external data taxonomy",
     )
 
+    def get_taxon_by_path(self, path):
+        for taxon in self.taxons:
+            if taxon.path == path:
+                return taxon
+            if taxon.children is not None:
+                child_taxon = taxon.get_taxon_by_path(path)
+                if child_taxon is not None:
+                    return child_taxon
+        return None
+
     def update_paths(self):
         for taxon in self.taxons:
             taxon.update_path()

{kodexa-7.4.412833907659.dist-info → kodexa-7.4.413041559723.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kodexa
-Version: 7.4.412833907659
+Version: 7.4.413041559723
 Summary: Python SDK for the Kodexa Platform
 Author: Austin Redenbaugh
 Author-email: austin@kodexa.com

{kodexa-7.4.412833907659.dist-info → kodexa-7.4.413041559723.dist-info}/RECORD

@@ -3,7 +3,7 @@ kodexa/assistant/__init__.py,sha256=nlXm_YnV_50hgn0TIT2Fkc2fQ-86OjmctY_j8My9nc4,
 kodexa/assistant/assistant.py,sha256=5KFdbqFSLIZJyDRyZdpcfr448fT-CW4JhYu9A6B9DGY,14663
 kodexa/connectors/__init__.py,sha256=WF6G_MUeU32TlKSUKkpNoNX7dq8iBPliFMep4E8BmZc,328
 kodexa/connectors/connectors.py,sha256=FpUZDkSyHld2b9eYRuVOWzaFtuGoaRuPXXicJB7THbc,10413
-kodexa/dataclasses/__init__.py,sha256=DKLWMALnUWYFDMraEc9We85GI_rvpp-z7Q1fKX-3jQI,18561
+kodexa/dataclasses/__init__.py,sha256=eLVjCFB3Sp_36FEf_wbw2OdQyu2B9FaXVvtoe1wn1xM,19618
 kodexa/dataclasses/templates/llm_data_class.j2,sha256=YWjStW136chV_59JM3AYis3i-0jdrqDvLXsISUW9zDU,660
 kodexa/model/__init__.py,sha256=rtLXYJBxB-rnukhslN9rlqoB3--1H3253HyHGbD_Gc8,796
 kodexa/model/base.py,sha256=CaZK8nMhT1LdCpt4aLhebJGcorjq9qRID1FjnXnP14M,521
@@ -13,7 +13,7 @@ kodexa/model/entities/product.py,sha256=StUhTEeLXmc05cj6XnZppQfeJsqCPbX1jdhsysHH
 kodexa/model/entities/product_group.py,sha256=540fRGyUf34h1BzAN1DiWu6rGgvaj3xDFhZ2k-RvSFY,3617
 kodexa/model/entities/product_subscription.py,sha256=UcmWR-qgLfdV7VCtJNwzgkanoS8nBSL6ngVuxQUK1M8,3810
 kodexa/model/model.py,sha256=qh1YUew3UgtjU0t4fAwSXYYuzQjXTOZWZkafyFp_w8M,118801
-kodexa/model/objects.py,sha256=D90kTi8RZSnWOATchu0DmGLd3CD6Eg8At8c_HjqpCX0,185074
+kodexa/model/objects.py,sha256=CK8T-JqSBU9BK0Q8Raul6uw6PZqZlgu8u38pudZN9Jg,185695
 kodexa/model/persistence.py,sha256=HX_uIkGs8bqHwqyE5wB2qMlGIG5ZnjuTu7xMdvKhEzA,72033
 kodexa/model/utils.py,sha256=6R-3rFiW9irBwj0Mq5yhp7EDXkNUFaeFhr3bWmnlW4g,2961
 kodexa/pipeline/__init__.py,sha256=sA7f5D6qkdMrpp2xTIeefnrUBI6xxEEWostvxfX_1Cs,236
@@ -44,7 +44,7 @@ kodexa/testing/test_utils.py,sha256=v44p__gE7ia67W7WeHN2HBFCWSCUrCZt7G4xBNCmwf8,
 kodexa/training/__init__.py,sha256=xs2L62YpRkIRfslQwtQZ5Yxjhm7sLzX2TrVX6EuBnZQ,52
 kodexa/training/train_utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 kodexa/utils/__init__.py,sha256=Pnim1o9_db5YEnNvDTxpM7HG-qTlL6n8JwFwOafU9wo,5928
-kodexa-7.4.412833907659.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
-kodexa-7.4.412833907659.dist-info/METADATA,sha256=eMueirzULKg5-Co0jpJxcl9UC9t9oINr3fyEDo51n3Y,3528
-kodexa-7.4.412833907659.dist-info/WHEEL,sha256=Zb28QaM1gQi8f4VCBhsUklF61CTlNYfs9YAZn-TOGFk,88
-kodexa-7.4.412833907659.dist-info/RECORD,,
+kodexa-7.4.413041559723.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
+kodexa-7.4.413041559723.dist-info/METADATA,sha256=-ktNcWfsonqxZUIAMF4lV6H4sEpHo5u03j_YDPX56wQ,3528
+kodexa-7.4.413041559723.dist-info/WHEEL,sha256=Zb28QaM1gQi8f4VCBhsUklF61CTlNYfs9YAZn-TOGFk,88
+kodexa-7.4.413041559723.dist-info/RECORD,,