Fix issue with empty string embeddings indexing, closes #496

src/python/txtai/database/filedb.py

@@ -377,15 +377,12 @@ def insertdocument(self, uid, document, tags, entry):
         if document:
             self.cursor.execute(FileDB.INSERT_DOCUMENT, [uid, json.dumps(document, allow_nan=False), tags, entry])
 
-        # Get value of text field
-        text = document.get("text")
-
-        # If both text and object are set, insert object as it won't otherwise be used
-        if text and obj:
+        # If text and object are both available, insert object as it won't otherwise be used
+        if "text" in document and obj:
             self.insertobject(uid, obj, tags, entry)
 
         # Return value to use for section - use text if available otherwise use object
-        return text if text else obj
+        return document["text"] if "text" in document else obj
 
     def insertobject(self, uid, obj, tags, entry):
         """

test/python/testdatabase/testsqlite.py

@@ -146,6 +146,19 @@ def testEmpty(self):
         embeddings.upsert([])
         self.assertIsNotNone(embeddings.ann)
 
+    def testEmptyString(self):
+        """
+        Test empty string indexing
+        """
+
+        # Test empty string
+        self.embeddings.index([(0, "", None)])
+        self.assertTrue(self.embeddings.search("test"))
+
+        # Test empty string with dict
+        self.embeddings.index([(0, {"text": ""}, None)])
+        self.assertTrue(self.embeddings.search("test"))
+
     def testExplain(self):
         """
         Test query explain

test/python/testembeddings.py

@@ -92,6 +92,19 @@ def testEmpty(self):
         embeddings.upsert([])
         self.assertIsNotNone(embeddings.ann)
 
+    def testEmptyString(self):
+        """
+        Test empty string indexing
+        """
+
+        # Test empty string
+        self.embeddings.index([(0, "", None)])
+        self.assertTrue(self.embeddings.search("test"))
+
+        # Test empty string with dict
+        self.embeddings.index([(0, {"text": ""}, None)])
+        self.assertTrue(self.embeddings.search("test"))
+
     def testExternal(self):
         """
         Test embeddings backed by external vectors