kodexa 7.4.412833907659__tar.gz → 7.4.413039380629__tar.gz

{kodexa-7.4.412833907659 → kodexa-7.4.413039380629}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kodexa
-Version: 7.4.412833907659
+Version: 7.4.413039380629
 Summary: Python SDK for the Kodexa Platform
 Author: Austin Redenbaugh
 Author-email: austin@kodexa.com

{kodexa-7.4.412833907659 → kodexa-7.4.413039380629}/kodexa/dataclasses/__init__.py

@@ -70,6 +70,14 @@ class LLMDataAttribute(BaseModel):
         )
         self.exceptions.append(content_exception)
 
+    def to_dict(self, taxonomy: Taxonomy) -> dict:
+        """Convert attribute to JSON with normalized value"""
+
+        taxon_external_name = taxonomy.get_taxon_by_path(self.taxon_path).external_name
+        return {
+            taxon_external_name: self.normalized_text if self.normalized_text else self.value
+        }
+
 
 class LLMDataObject(BaseModel):
     """
@@ -154,6 +162,19 @@ 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-7.4.412833907659 → kodexa-7.4.413039380629}/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 → kodexa-7.4.413039380629}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "kodexa"
-version = "7.4.412833907659"
+version = "7.4.413039380629"
 description = "Python SDK for the Kodexa Platform"
 authors = ["Austin Redenbaugh <austin@kodexa.com>", "Philip Dodds <philip@kodexa.com>", "Romar Cablao <rcablao@kodexa.com>", "Amadea Paula Dodds <amadeapaula@kodexa.com>"]
 readme = "README.md"