broadfield-dev commited on
Commit
679e83a
·
verified ·
1 Parent(s): a66e49f

Update database.py

Browse files
Files changed (1) hide show
  1. database.py +5 -4
database.py CHANGED
@@ -260,9 +260,9 @@ def save_chromadb_to_hf(dataset_name=HF_DATASET_NAME, token=os.getenv("HF_KEY"))
260
  dataset = Dataset.from_dict(data)
261
  logger.info(f"Created Hugging Face Dataset with {len(data['code'])} entries")
262
 
263
- # Push to Hugging Face Hub
264
- dataset.push_to_hub(dataset_name, token=token)
265
- logger.info(f"Dataset pushed to Hugging Face Hub as {dataset_name}")
266
  # Verify push (optional, could check dataset on Hub)
267
  logger.info(f"Verified Hugging Face dataset push with {len(dataset)} entries")
268
  except Exception as e:
@@ -291,4 +291,5 @@ def load_chromadb_from_hf(dataset_name=HF_DATASET_NAME, token=os.getenv("HF_KEY"
291
 
292
  if __name__ == '__main__':
293
  client = load_chromadb_from_hf()
294
- logger.info("Database initialized or loaded from Hugging Face Hub, contains {client.get_collection(DB_NAME).count()} entries")
 
 
260
  dataset = Dataset.from_dict(data)
261
  logger.info(f"Created Hugging Face Dataset with {len(data['code'])} entries")
262
 
263
+ # Push to Hugging Face Hub, overwriting existing dataset
264
+ dataset.push_to_hub(dataset_name, token=token, exist_ok=True)
265
+ logger.info(f"Dataset pushed to Hugging Face Hub as {dataset_name}, overwriting existing dataset")
266
  # Verify push (optional, could check dataset on Hub)
267
  logger.info(f"Verified Hugging Face dataset push with {len(dataset)} entries")
268
  except Exception as e:
 
291
 
292
  if __name__ == '__main__':
293
  client = load_chromadb_from_hf()
294
+ collection = create_collection(client, DB_NAME)
295
+ logger.info(f"Database initialized or loaded from Hugging Face Hub, contains {collection.count()} entries")