import io import logging import os import requests import shutil import zipfile import time from pandas.tseries.offsets import BDay import pandas as pd logger = logging.getLogger(__name__) def convertToNone(v): return v if v else None def download_cds_data(payload): r = requests.get('https://www.markit.com/export.jsp', params=payload) f2 = open(os.path.join(os.environ['BASE_DIR'], "Tranche_data", "CDS", "cds eod {0}.csv".format(payload['date'])), "wb") with zipfile.ZipFile(io.BytesIO(r.content)) as z: for f in z.namelist(): if "csv" in f: f1 = z.open(f) next(f1) next(f1) shutil.copyfileobj(f1, f2) f1.close() f2.close() def download_composite_data(payload, historical=False): # if historical, we want to maintain the invariant mtime(f)== payload['date'] + BDay(1) if historical: ts = (pd.Timestamp(payload['date']) + BDay(1)).timestamp() for report in ['COMPOSITES', 'TRANCHE_COMPOSITES']: for family in ['CDX', 'ITRAXX-EUROPE']: payload.update({'family': family, 'report': report}) while True: r = requests.get('https://www.markit.com/export.jsp', params=payload) try: with zipfile.ZipFile(io.BytesIO(r.content)) as z: for f in z.namelist(): if "csv" in f: path = z.extract(f, path=os.path.join(os.environ['BASE_DIR'], "Tranche_data", "Composite_reports")) if historical: os.utime(path, (ts, ts)) except zipfile.BadZipfile: logger.error(r.content) time.sleep(5) continue else: break