Update sync_data.sh
Browse files- sync_data.sh +20 -20
sync_data.sh
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
#!/bin/sh
|
| 2 |
|
| 3 |
-
#
|
| 4 |
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
|
| 5 |
-
echo "
|
| 6 |
exit 1
|
| 7 |
fi
|
| 8 |
|
| 9 |
-
#
|
| 10 |
cat > /tmp/hf_sync.py << 'EOL'
|
| 11 |
from huggingface_hub import HfApi
|
| 12 |
import sys
|
|
@@ -22,9 +22,9 @@ def manage_backups(api, repo_id, max_files=50):
|
|
| 22 |
for file_to_delete in files_to_delete:
|
| 23 |
try:
|
| 24 |
api.delete_file(path_in_repo=file_to_delete, repo_id=repo_id, repo_type="dataset")
|
| 25 |
-
print(f'
|
| 26 |
except Exception as e:
|
| 27 |
-
print(f'
|
| 28 |
|
| 29 |
def upload_backup(file_path, file_name, token, repo_id):
|
| 30 |
api = HfApi(token=token)
|
|
@@ -35,13 +35,13 @@ def upload_backup(file_path, file_name, token, repo_id):
|
|
| 35 |
repo_id=repo_id,
|
| 36 |
repo_type="dataset"
|
| 37 |
)
|
| 38 |
-
print(f"
|
| 39 |
|
| 40 |
manage_backups(api, repo_id)
|
| 41 |
except Exception as e:
|
| 42 |
-
print(f"
|
| 43 |
|
| 44 |
-
#
|
| 45 |
def download_latest_backup(token, repo_id):
|
| 46 |
try:
|
| 47 |
api = HfApi(token=token)
|
|
@@ -49,7 +49,7 @@ def download_latest_backup(token, repo_id):
|
|
| 49 |
backup_files = [f for f in files if f.startswith('webui_backup_') and f.endswith('.db')]
|
| 50 |
|
| 51 |
if not backup_files:
|
| 52 |
-
print("
|
| 53 |
return
|
| 54 |
|
| 55 |
latest_backup = sorted(backup_files)[-1]
|
|
@@ -63,10 +63,10 @@ def download_latest_backup(token, repo_id):
|
|
| 63 |
if filepath and os.path.exists(filepath):
|
| 64 |
os.makedirs('./data', exist_ok=True)
|
| 65 |
os.system(f'cp "{filepath}" ./data/webui.db')
|
| 66 |
-
print(f"
|
| 67 |
|
| 68 |
except Exception as e:
|
| 69 |
-
print(f"
|
| 70 |
|
| 71 |
if __name__ == "__main__":
|
| 72 |
action = sys.argv[1]
|
|
@@ -81,35 +81,35 @@ if __name__ == "__main__":
|
|
| 81 |
download_latest_backup(token, repo_id)
|
| 82 |
EOL
|
| 83 |
|
| 84 |
-
#
|
| 85 |
-
echo "
|
| 86 |
python3 /tmp/hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}"
|
| 87 |
|
| 88 |
-
#
|
| 89 |
sync_data() {
|
| 90 |
while true; do
|
| 91 |
-
echo "
|
| 92 |
|
| 93 |
if [ -f "./data/webui.db" ]; then
|
| 94 |
timestamp=$(date +%Y%m%d_%H%M%S)
|
| 95 |
backup_file="webui_backup_${timestamp}.db"
|
| 96 |
|
| 97 |
-
#
|
| 98 |
cp ./data/webui.db "/tmp/${backup_file}"
|
| 99 |
|
| 100 |
-
echo "
|
| 101 |
python3 /tmp/hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "/tmp/${backup_file}" "${backup_file}"
|
| 102 |
|
| 103 |
rm -f "/tmp/${backup_file}"
|
| 104 |
else
|
| 105 |
-
echo "
|
| 106 |
fi
|
| 107 |
|
| 108 |
SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
|
| 109 |
-
echo "
|
| 110 |
sleep $SYNC_INTERVAL
|
| 111 |
done
|
| 112 |
}
|
| 113 |
|
| 114 |
-
#
|
| 115 |
sync_data &
|
|
|
|
| 1 |
#!/bin/sh
|
| 2 |
|
| 3 |
+
# Check environment variables
|
| 4 |
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
|
| 5 |
+
echo "Missing required environment variables HF_TOKEN or DATASET_ID"
|
| 6 |
exit 1
|
| 7 |
fi
|
| 8 |
|
| 9 |
+
# Upload backup
|
| 10 |
cat > /tmp/hf_sync.py << 'EOL'
|
| 11 |
from huggingface_hub import HfApi
|
| 12 |
import sys
|
|
|
|
| 22 |
for file_to_delete in files_to_delete:
|
| 23 |
try:
|
| 24 |
api.delete_file(path_in_repo=file_to_delete, repo_id=repo_id, repo_type="dataset")
|
| 25 |
+
print(f'Deleted old backup: {file_to_delete}')
|
| 26 |
except Exception as e:
|
| 27 |
+
print(f'Error deleting {file_to_delete}: {str(e)}')
|
| 28 |
|
| 29 |
def upload_backup(file_path, file_name, token, repo_id):
|
| 30 |
api = HfApi(token=token)
|
|
|
|
| 35 |
repo_id=repo_id,
|
| 36 |
repo_type="dataset"
|
| 37 |
)
|
| 38 |
+
print(f"Successfully uploaded {file_name}")
|
| 39 |
|
| 40 |
manage_backups(api, repo_id)
|
| 41 |
except Exception as e:
|
| 42 |
+
print(f"File upload error: {str(e)}")
|
| 43 |
|
| 44 |
+
# Download latest backup
|
| 45 |
def download_latest_backup(token, repo_id):
|
| 46 |
try:
|
| 47 |
api = HfApi(token=token)
|
|
|
|
| 49 |
backup_files = [f for f in files if f.startswith('webui_backup_') and f.endswith('.db')]
|
| 50 |
|
| 51 |
if not backup_files:
|
| 52 |
+
print("Backup file not found")
|
| 53 |
return
|
| 54 |
|
| 55 |
latest_backup = sorted(backup_files)[-1]
|
|
|
|
| 63 |
if filepath and os.path.exists(filepath):
|
| 64 |
os.makedirs('./data', exist_ok=True)
|
| 65 |
os.system(f'cp "{filepath}" ./data/webui.db')
|
| 66 |
+
print(f"Successfully restored backup from {latest_backup}")
|
| 67 |
|
| 68 |
except Exception as e:
|
| 69 |
+
print(f"Error downloading backup: {str(e)}")
|
| 70 |
|
| 71 |
if __name__ == "__main__":
|
| 72 |
action = sys.argv[1]
|
|
|
|
| 81 |
download_latest_backup(token, repo_id)
|
| 82 |
EOL
|
| 83 |
|
| 84 |
+
# Download latest backup on first startup
|
| 85 |
+
echo "Downloading latest backup from HuggingFace..."
|
| 86 |
python3 /tmp/hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}"
|
| 87 |
|
| 88 |
+
# Sync function
|
| 89 |
sync_data() {
|
| 90 |
while true; do
|
| 91 |
+
echo "Starting sync process $(date)"
|
| 92 |
|
| 93 |
if [ -f "./data/webui.db" ]; then
|
| 94 |
timestamp=$(date +%Y%m%d_%H%M%S)
|
| 95 |
backup_file="webui_backup_${timestamp}.db"
|
| 96 |
|
| 97 |
+
# Copy database file
|
| 98 |
cp ./data/webui.db "/tmp/${backup_file}"
|
| 99 |
|
| 100 |
+
echo "Uploading backup to HuggingFace..."
|
| 101 |
python3 /tmp/hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "/tmp/${backup_file}" "${backup_file}"
|
| 102 |
|
| 103 |
rm -f "/tmp/${backup_file}"
|
| 104 |
else
|
| 105 |
+
echo "Database file does not exist, waiting for next sync..."
|
| 106 |
fi
|
| 107 |
|
| 108 |
SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
|
| 109 |
+
echo "Next sync will be in ${SYNC_INTERVAL} seconds..."
|
| 110 |
sleep $SYNC_INTERVAL
|
| 111 |
done
|
| 112 |
}
|
| 113 |
|
| 114 |
+
# Start sync process in background
|
| 115 |
sync_data &
|