Remove Histdata; too many nonces, not worth it

master
Bradlee Speice 2016-08-16 21:44:03 -04:00
parent 5cf75ebc47
commit c42f5a6980
2 changed files with 0 additions and 130 deletions

View File

@ -1,98 +0,0 @@
from zipfile import ZipFile
from io import BytesIO
import requests
import pandas as pd
from datetime import datetime
from metrik.tasks.base import MongoCreateTask
from metrik.utils import masked_get
from metrik.conf import USER_AGENT
class HistDataWrapper(MongoCreateTask):
@staticmethod
def fetch_zip(currency, year, month, side):
"""
:param currency: Currency pair specified as EURUSD
:type currency: str
:param year: Year for retrieval as an integer
:param month: Month for retrieval as an integer
:param side: BID or ASK for which side of the trade you want
:return: zip file content with CSV of tick data
"""
# HistData has some interesting tricks they go through to make it
# hard to bulk download data. Specifically: you post to one page with
# your `Referer` as a different page to get the data, plus some extra
# POST body things in order to get the data to come.
request_url = 'http://www.histdata.com/get.php'
month_leading = '{:02d}'.format(month)
# No idea what this parameter does, because I'm not convinced it's a nonce:
if side == 'BID':
tk = '6b95868ebd96e92f3f28196354e1901f'
else:
tk = 'fa921f9b65d175d05c5976b5f78e5333'
zip_fname = 'HISTDATA_COM_NT_{pair}_T_{side}_{year}{month_leading}.zip'.format(
year=year,
month_leading=month_leading,
side=side,
pair=currency
)
zip_url = 'http://www.histdata.com/' \
'download-free-forex-historical-data/?/ninjatrader/' \
'tick-{side}-quotes/{pair}/{year}/{month_nonleading}/' \
'{zip_fname}' \
.format(
year=year,
side=side.lower(),
month_nonleading=str(month),
pair=currency.lower(),
zip_fname=zip_fname
)
request_content = requests.post(request_url, headers={
'Host': 'www.histdata.com',
'Referer': zip_url,
'User-Agent': USER_AGENT,
'Cookie': '__cfduid=d42d8f540c786afd0568ab63eb20c50eb1471099760; complianceCookie=off',
}, data={
'tk': tk,
'date': year,
'datemonth': '{year}{month_leading}'.format(
year=year, month_leading=month_leading),
'platform': 'NT',
'timeframe': 'T_{}'.format(side),
'fxpair': currency
}).content
return BytesIO(request_content)
@staticmethod
def fetch_csv(zip_content, currency, year, month, side):
csv_fname = 'DAT_NT_{pair}_T_{side}_{year}{month_leading}.csv'.format(
pair=currency,
side=side,
year=year,
month_leading='{:02d}'.format(month)
)
z = ZipFile(zip_content)
bytes = z.read(csv_fname)
return BytesIO(bytes)
@staticmethod
def parse_csv(csv_content):
def parse_timestamp(string):
return datetime.strptime(string, '%Y%m%d %H%M%S')
table = pd.read_csv(
filepath_or_buffer=csv_content,
names=['Timestamp', 'Price', 'Unknown'],
parse_dates=['Timestamp'],
date_parser=parse_timestamp,
sep=';'
).drop('Unknown', axis=1)
return table.to_dict()

View File

@ -1,32 +0,0 @@
from unittest import TestCase
from io import BytesIO
import pandas as pd
from metrik.tasks.histdata import HistDataWrapper
class HistDataWrapperTest(TestCase):
def test_eur_usd_july2016_bid(self):
zip_content = HistDataWrapper.fetch_zip('EURUSD', 2016, 7, 'BID')
assert len(zip_content.getvalue()) > 0
csv_content = HistDataWrapper.fetch_csv(zip_content, 'EURUSD', 2016, 7, 'BID')
assert len(csv_content.getvalue()) > 0
content_dict = HistDataWrapper.parse_csv(csv_content)
assert len(content_dict) > 0
def test_usd_cad_july2016_bid(self):
zip_content = HistDataWrapper.fetch_zip('USDCAD', 2016, 7, 'BID')
assert len(zip_content.getvalue()) > 0
csv_content = HistDataWrapper.fetch_csv(zip_content, 'USDCAD', 2016, 7, 'BID')
assert len(csv_content.getvalue()) > 0
content_dict = HistDataWrapper.parse_csv(csv_content)
assert len(content_dict) > 0
def test_eur_usd_july2016_ask(self):
zip_content = HistDataWrapper.fetch_zip('EURUSD', 2016, 7, 'ASK')
assert len(zip_content.getvalue()) > 0
csv_content = HistDataWrapper.fetch_csv(zip_content, 'EURUSD', 2016, 7, 'ASK')
assert len(csv_content.getvalue()) > 0
content_dict = HistDataWrapper.parse_csv(csv_content)
assert len(content_dict) > 0