diff options
Diffstat (limited to 'python')
| -rw-r--r-- | python/parse_emails.py | 15 |
1 files changed, 7 insertions, 8 deletions
diff --git a/python/parse_emails.py b/python/parse_emails.py index 65a8b091..f9ad9d9c 100644 --- a/python/parse_emails.py +++ b/python/parse_emails.py @@ -5,7 +5,7 @@ import pdb from download_emails import update_emails import datetime -def makedf(r, indextype, ref): +def makedf(r, indextype): if indextype=='IG': cols = ['Strike', 'RecBid', 'RecOffer', 'DeltaRec', 'PayBid', 'PayOffer', 'DeltaPay', 'Vol', 'Gamma'] @@ -13,7 +13,6 @@ def makedf(r, indextype, ref): cols = ['Strike', 'RecBid', 'RecOffer', 'DeltaRec', 'PayBid', 'PayOffer', 'DeltaPay', 'Vol', 'PxVol', 'Gamma'] df = pd.DataFrame.from_records(r, columns = cols) - df['ref'] = ref for col in ['DeltaRec', 'DeltaPay', 'Vol', 'PxVol', 'Gamma']: if col in df: df[col] = df[col].str.strip("%").astype('float')/100 @@ -60,7 +59,7 @@ def parse_email(email_path): try: d = m.groupdict() d['quotedate'] = quotedate - d['index'] = indextype + d['indextype'] = indextype d['series'] = series d['expiry'] = pd.to_datetime(d['expiry'], format='%d-%b-%y') except AttributeError: @@ -77,13 +76,13 @@ def parse_email(email_path): r.append(vals) continue else: - option_stack[d['expiry']] = makedf(r, indextype, d['ref']) + option_stack[d['expiry']] = makedf(r, indextype) fwd_index.append(d) flag = False r = [] continue if flag: - option_stack[d['expiry']] = makedf(r, indextype, d['ref']) + option_stack[d['expiry']] = makedf(r, indextype) fwd_index.append(d) if option_stack: fwd_index = pd.DataFrame.from_records(fwd_index, @@ -103,13 +102,13 @@ if __name__=="__main__": except RuntimeError as e: print(e) else: - swaption_stack[key] = pd.concat(option_stack, names=['expiry', 'strike']) + swaption_stack[key] = pd.concat(option_stack, names=['expiry', 'Strike']) index_data = index_data.append(fwd_index) for col in ['fwdbpv', 'fwdprice', 'fwdspread', 'ref']: index_data[col] = index_data[col].astype('float') - index_data['index'] = index_data['index'].astype('category') + index_data['indextype'] = index_data['indextype'].astype('category') swaption_stack = pd.concat(swaption_stack, names=['quotedate', 'indextype', 'series']) - with pd.HDFStore('swaptions.hdf', mode = 'w', complevel=4, + with pd.HDFStore('../../data/swaptions.hdf', mode = 'w', complevel=4, complib='blosc', fletcher32=True) as swaptions: swaptions.append('swaptions', swaption_stack) swaptions.append('index_data', index_data) |