kraken-trading-data / kraken-data-collection-script
GotThatData's picture
Update
0d94041 verified
raw
history blame
9.2 kB
```python
import krakenex
import pandas as pd
from datetime import datetime
import time
import os
from typing import Dict, List, Optional
import logging
from huggingface_hub import HfApi
# Set up logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.FileHandler(f'kraken_collection_{datetime.now().strftime("%Y%m%d")}.log'),
logging.StreamHandler()
]
)
logger = logging.getLogger(__name__)
class KrakenHuggingFaceCollector:
def __init__(self, kraken_key_path: str, repo_id: str):
self.kraken_api = krakenex.API()
try:
self.kraken_api.load_key(kraken_key_path)
logger.info("Successfully loaded Kraken API key")
except Exception as e:
logger.error(f"Failed to load Kraken API key: {e}")
raise
try:
self.hf_api = HfApi()
self.repo_id = repo_id
logger.info("Successfully connected to Hugging Face")
except Exception as e:
logger.error(f"Failed to initialize Hugging Face API: {e}")
raise
self.pairs = [
"XXBTZUSD", # Bitcoin/USD
"XETHZUSD", # Ethereum/USD
"XXRPZUSD", # Ripple/USD
"ADAUSD", # Cardano/USD
"XDGUSD", # Dogecoin/USD
"SOLUSD", # Solana/USD
"DOTUSD", # Polkadot/USD
"MATICUSD", # Polygon/USD
"LTCUSD" # Litecoin/USD
]
self.running = True
self.data_points_collected = 0
self.collection_start_time = None
self.api_calls = 0
self.last_api_reset = datetime.now()
def check_api_rate(self) -> bool:
"""Monitor API call rate"""
current_time = datetime.now()
if (current_time - self.last_api_reset).total_seconds() >= 30:
self.api_calls = 0
self.last_api_reset = current_time
return self.api_calls < 15
def fetch_ticker_data(self, pair: str) -> Optional[Dict]:
"""Fetch ticker data with rate limiting"""
if not self.check_api_rate():
logger.warning("API rate limit approaching, waiting...")
time.sleep(2)
try:
self.api_calls += 1
response = self.kraken_api.query_public('Ticker', {'pair': pair})
if 'error' in response and response['error']:
logger.error(f"Kraken API error for {pair}: {response['error']}")
return None
data = response['result']
pair_data = list(data.values())[0]
return {
'timestamp': datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S.%f'),
'pair': pair,
'price': float(pair_data['c'][0]),
'volume': float(pair_data['v'][0]),
'bid': float(pair_data['b'][0]),
'ask': float(pair_data['a'][0]),
'low': float(pair_data['l'][0]),
'high': float(pair_data['h'][0]),
'vwap': float(pair_data['p'][0]),
'trades': int(pair_data['t'][0])
}
except Exception as e:
logger.error(f"Error fetching data for {pair}: {e}")
return None
def upload_to_huggingface(self, df: pd.DataFrame, timestamp: str) -> None:
"""Upload DataFrame to Hugging Face as CSV"""
try:
# Create local directories
os.makedirs('data/continuous', exist_ok=True)
# Save locally first
local_path = f'data/continuous/kraken_trades_{timestamp}.csv'
df.to_csv(local_path, index=False)
# Upload to Hugging Face
self.hf_api.upload_file(
path_or_fileobj=local_path,
path_in_repo=f"data/continuous/kraken_trades_{timestamp}.csv",
repo_id=self.repo_id,
repo_type="dataset"
)
logger.info(f"Successfully uploaded batch to Hugging Face")
except Exception as e:
logger.error(f"Error uploading to Hugging Face: {e}")
logger.info(f"Data saved locally at: {local_path}")
def collect_continuous(self, interval_minutes: int = 3, batch_size: int = 30):
"""
Enhanced continuous data collection
Args:
interval_minutes: Minutes between each collection (default: 3)
batch_size: Number of snapshots per batch (default: 30)
"""
self.collection_start_time = datetime.now()
logger.info(f"Starting enhanced continuous collection at {self.collection_start_time}")
logger.info(f"Collecting {batch_size} snapshots every {interval_minutes} minutes")
logger.info(f"Total API calls per batch: ~{batch_size * len(self.pairs)}")
logger.info(f"Estimated daily data points: {(24 * 60 // interval_minutes) * batch_size * len(self.pairs)}")
logger.info("Press CTRL+C to stop collection")
while self.running:
try:
batch_start_time = datetime.now()
records = []
for i in range(batch_size):
if not self.running:
break
snapshot_start = datetime.now()
logger.info(f"Collecting snapshot {i+1}/{batch_size}")
for pair in self.pairs:
if self.check_api_rate():
record = self.fetch_ticker_data(pair)
if record:
records.append(record)
else:
time.sleep(1)
# Dynamic sleep calculation
elapsed = (datetime.now() - snapshot_start).total_seconds()
sleep_time = max(0.5, 1.5 - elapsed)
if i < batch_size - 1 and self.running:
time.sleep(sleep_time)
if records:
df = pd.DataFrame(records)
current_timestamp = datetime.now().strftime('%Y%m%d_%H%M')
self.upload_to_huggingface(df, current_timestamp)
self.data_points_collected += len(records)
collection_duration = (datetime.now() - self.collection_start_time)
logger.info("\nBatch Summary:")
logger.info(f"Records in batch: {len(records)}")
logger.info(f"Pairs collected: {len(df['pair'].unique())}")
logger.info(f"Total data points: {self.data_points_collected}")
logger.info(f"Collection duration: {collection_duration}")
logger.info(f"Data points per hour: {self.data_points_collected / collection_duration.total_seconds() * 3600:.2f}")
# Adaptive interval timing
batch_duration = (datetime.now() - batch_start_time).total_seconds()
sleep_time = max(0, interval_minutes * 60 - batch_duration)
if self.running and sleep_time > 0:
logger.info(f"Waiting {sleep_time:.2f} seconds until next batch...")
time.sleep(sleep_time)
except Exception as e:
logger.error(f"Error in continuous collection: {e}")
logger.info("Waiting 30 seconds before retry...")
time.sleep(30)
logger.info("Data collection stopped")
logger.info(f"Total data points collected: {self.data_points_collected}")
logger.info(f"Total collection time: {datetime.now() - self.collection_start_time}")
def main():
try:
collector = KrakenHuggingFaceCollector(
kraken_key_path="kraken.key",
repo_id="GotThatData/kraken-trading-data"
)
# Start collection with enhanced parameters
collector.collect_continuous(
interval_minutes=3, # Collect every 3 minutes
batch_size=30 # 30 snapshots per batch
)
except KeyboardInterrupt:
logger.info("Stopping collection (CTRL+C pressed)")
collector.running = False
except Exception as e:
logger.error(f"Fatal error: {e}")
raise
if __name__ == "__main__":
main()
```
To use this script:
1. Save it as `kraken_data_collector.py`
2. Make sure you have your `kraken.key` file with your API credentials
3. Install required packages if you haven't:
```bash
pip install krakenex pandas huggingface_hub
```
4. Run the script:
```bash
python kraken_data_collector.py
```
This will:
- Collect 30 snapshots every 3 minutes
- Save data locally and to Hugging Face
- Provide detailed logging
- Handle errors gracefully
- Respect API rate limits