Spaces:
Running
on
T4
Running
on
T4
Update app.py
Browse files
app.py
CHANGED
@@ -738,11 +738,22 @@ if __name__ == '__main__':
|
|
738 |
|
739 |
embeddings_path = os.path.join(os.getenv('HF_HOME'), 'embeddings')
|
740 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
741 |
for filename in os.listdir(embeddings_path):
|
742 |
if '_ids' in filename:
|
743 |
-
|
|
|
|
|
|
|
|
|
744 |
|
745 |
-
ids = weakDict.all_ids()
|
746 |
payload_docs = [{ 'text': documents[i], 'metadata': metadata } for i, metadata in enumerate(weakDict.all_metadatas())]
|
747 |
|
748 |
client.upsert(
|
|
|
738 |
|
739 |
embeddings_path = os.path.join(os.getenv('HF_HOME'), 'embeddings')
|
740 |
|
741 |
+
def generate_unique_id(existing_ids):
|
742 |
+
while True:
|
743 |
+
new_id = uuid.uuid4().int
|
744 |
+
if new_id not in existing_ids:
|
745 |
+
return new_id
|
746 |
+
|
747 |
+
ids = weakDict.all_ids()
|
748 |
+
|
749 |
for filename in os.listdir(embeddings_path):
|
750 |
if '_ids' in filename:
|
751 |
+
list_ids = np.load(os.path.join(embeddings_path, filename), allow_pickle=True).tolist()
|
752 |
+
|
753 |
+
for i, ids_ in enumerate(ids):
|
754 |
+
if ids_ in list_ids:
|
755 |
+
ids[i] = generate_unique_id(list_ids)
|
756 |
|
|
|
757 |
payload_docs = [{ 'text': documents[i], 'metadata': metadata } for i, metadata in enumerate(weakDict.all_metadatas())]
|
758 |
|
759 |
client.upsert(
|