Skip to content

armgit / txtai

Public Imported from GitHub
0

💡 All-in-one AI framework for semantic search, LLM orchestration and language model workflows

Fix unique constraint error with DuckDB, closes #475

davidmezzettidc2829488157
diff --git a/src/python/txtai/database/duckdb.py b/src/python/txtai/database/duckdb.py
index 8b38b796..1f66ff1c 100644
--- a/src/python/txtai/database/duckdb.py
+++ b/src/python/txtai/database/duckdb.py
@@ -22,12 +22,32 @@ class DuckDB(FileDB):
Database instance backed by DuckDB.
"""
+ # Delete single document and object
+ DELETE_DOCUMENT = "DELETE FROM documents WHERE id = ?"
+ DELETE_OBJECT = "DELETE FROM objects WHERE id = ?"
+
def __init__(self, config):
super().__init__(config)
if not DUCKDB:
raise ImportError('DuckDB is not available - install "database" extra to enable')
+ def insertdocument(self, uid, document, tags, entry):
+ if document:
+ # Delete existing document
+ self.cursor.execute(DuckDB.DELETE_DOCUMENT, [uid])
+
+ # Call parent logic
+ return super().insertdocument(uid, document, tags, entry)
+
+ def insertobject(self, uid, obj, tags, entry):
+ if self.encoder:
+ # Delete existing object
+ self.cursor.execute(DuckDB.DELETE_OBJECT, [uid])
+
+ # Call parent logic
+ super().insertobject(uid, obj, tags, entry)
+
def connect(self, path=":memory:"):
# Create connection and start a transaction
# pylint: disable=I1101
diff --git a/test/python/testdatabase/testencoder.py b/test/python/testdatabase/testencoder.py
index c7eb268d..57b7c074 100644
--- a/test/python/testdatabase/testencoder.py
+++ b/test/python/testdatabase/testencoder.py
@@ -54,16 +54,22 @@ class TestEncoder(unittest.TestCase):
try:
# Set default encoder
self.embeddings.config["objects"] = True
- data = [(0, {"object": bytearray([1, 2, 3]), "text": "default test"}, None)]
- # Create an index
- self.embeddings.index(data)
+ # Test all database providers
+ for content in ["duckdb", "sqlite"]:
+ self.embeddings.config["content"] = content
- result = self.embeddings.search("select object from txtai limit 1")[0]
+ data = [(0, {"object": bytearray([1, 2, 3]), "text": "default test"}, None)]
+
+ # Create an index
+ self.embeddings.index(data)
+
+ result = self.embeddings.search("select object from txtai limit 1")[0]
- self.assertEqual(result["object"].getvalue(), bytearray([1, 2, 3]))
+ self.assertEqual(result["object"].getvalue(), bytearray([1, 2, 3]))
finally:
self.embeddings.config["objects"] = "image"
+ self.embeddings.config["content"] = True
def testImages(self):
"""