kodexa 7.4.416992731072__py3-none-any.whl → 7.4.417022235742__py3-none-any.whl

kodexa/model/persistence.py

@@ -1172,6 +1172,36 @@ class SqliteDocumentPersistence(object):
         if result[0] == 0:
             self.cursor.execute("INSERT INTO validations (obj) VALUES (?)", [sqlite3.Binary(msgpack.packb([]))])
 
+    def __deduplicate_document_taxon_validations(self, taxon_validations: List[DocumentTaxonValidation]) -> List[DocumentTaxonValidation]:
+        """
+        Deduplicate DocumentTaxonValidation objects based on taxon_path and exception_id.
+        
+        Args:
+            taxon_validations: List of DocumentTaxonValidation objects to deduplicate.
+            
+        Returns:
+            List of deduplicated DocumentTaxonValidation objects.
+        """
+        if taxon_validations is None or len(taxon_validations) == 0:
+            return []
+        
+        seen_keys = set()
+        deduplicated = []
+        
+        for tv in taxon_validations:
+            if tv is None:
+                continue
+            
+            taxon_path = tv.taxon_path if tv.taxon_path is not None else ""
+            exception_id = tv.validation.exception_id if tv.validation and tv.validation.exception_id else ""
+            key = f"{taxon_path}|{exception_id}"
+            
+            if key not in seen_keys:
+                seen_keys.add(key)
+                deduplicated.append(tv)
+        
+        return deduplicated
+
     def set_validations(self, validations: List[DocumentTaxonValidation]):
         """
         Sets the validations for the document.
@@ -1180,7 +1210,12 @@ class SqliteDocumentPersistence(object):
             validations (List[DocumentTaxonValidation]): The validations to store.
         """
         self.__ensure_validations_table_exists()
-        serialized_data = sqlite3.Binary(msgpack.packb([v.model_dump(by_alias=True) for v in validations]))
+        # Handle None input
+        if validations is None:
+            validations = []
+        # Deduplicate validations before storing
+        deduplicated_validations = self.__deduplicate_document_taxon_validations(validations)
+        serialized_data = sqlite3.Binary(msgpack.packb([v.model_dump(by_alias=True) for v in deduplicated_validations]))
         self.cursor.execute("UPDATE validations SET obj = ? WHERE rowid = 1", [serialized_data])
         self.connection.commit()
 
@@ -1194,7 +1229,9 @@ class SqliteDocumentPersistence(object):
         self.__ensure_validations_table_exists()
         result = self.cursor.execute("SELECT obj FROM validations WHERE rowid = 1").fetchone()
         if result and result[0]:
-            return [DocumentTaxonValidation.model_validate(v) for v in msgpack.unpackb(result[0])]
+            validations = [DocumentTaxonValidation.model_validate(v) for v in msgpack.unpackb(result[0])]
+            # Deduplicate validations before returning
+            return self.__deduplicate_document_taxon_validations(validations)
         return []
 
     def set_external_data(self, external_data: dict, key: str = "default"):

{kodexa-7.4.416992731072.dist-info → kodexa-7.4.417022235742.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: kodexa
-Version: 7.4.416992731072
+Version: 7.4.417022235742
 Summary: Python SDK for the Kodexa Platform
 Author: Austin Redenbaugh
 Author-email: austin@kodexa.com

{kodexa-7.4.416992731072.dist-info → kodexa-7.4.417022235742.dist-info}/RECORD

@@ -14,7 +14,7 @@ kodexa/model/entities/product_group.py,sha256=540fRGyUf34h1BzAN1DiWu6rGgvaj3xDFh
 kodexa/model/entities/product_subscription.py,sha256=UcmWR-qgLfdV7VCtJNwzgkanoS8nBSL6ngVuxQUK1M8,3810
 kodexa/model/model.py,sha256=o93SDgw8mj28LWbwJB78PF_Oim8qeCTH1a9x_GDc7yg,120895
 kodexa/model/objects.py,sha256=mPLc1QwVzNuyXLhaHArUMLRf7KTIwgTyY3cu5zb7yyY,204986
-kodexa/model/persistence.py,sha256=jUgQ8xwsAFIoZ_bEynxCDEWhUII42eN0e0Mum0dkQPg,72043
+kodexa/model/persistence.py,sha256=n5KfdyujKmBWc-x_LN0pmRpFCo50VSrTr4_1M98q6tc,73548
 kodexa/model/utils.py,sha256=YEG3f8YzBil06D0P3_dfx2S9GnHREzyrjnlWwQ7oPlU,3038
 kodexa/pipeline/__init__.py,sha256=sA7f5D6qkdMrpp2xTIeefnrUBI6xxEEWostvxfX_1Cs,236
 kodexa/pipeline/pipeline.py,sha256=zyNEpA7KlGhPs_l-vgV6m-OCb16dbxQhl8QezeylugA,25540
@@ -45,7 +45,7 @@ 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-7.4.416992731072.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
-kodexa-7.4.416992731072.dist-info/METADATA,sha256=d91Rzx4ahdirQhJ--DUkkUAImzahFMu5Vh3WcdOlAek,2916
-kodexa-7.4.416992731072.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
-kodexa-7.4.416992731072.dist-info/RECORD,,
+kodexa-7.4.417022235742.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
+kodexa-7.4.417022235742.dist-info/METADATA,sha256=Du0Gon9vD5l1DVOD2awarhL2j3uUpEpEZw_smgzEWEA,2916
+kodexa-7.4.417022235742.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
+kodexa-7.4.417022235742.dist-info/RECORD,,