Add OANDA RestV20 Historical Instrument Reader #257

davidandreoletti · 2016-11-08T05:36:34Z

No description provided.

davidandreoletti · 2016-11-08T05:47:41Z

@femtotrader I create a new Pull Request for several reasons:

The OANDA API used to fetch the data has changed from OANDA REST API: (oandapy dependency) to OANDA REST V20 API (oandapyV20 dependency) because there is no intention from OANDA to release oandapy on PyPy :( See Publish OANDAPY on https://pypi.python.org/pypi oanda/oandapy#39 (comment)

I will have a look at implementing some of your suggestions below.

Thanks @davidandreoletti

a) But you could simply give this PR a title starting with WIP (work in progress) and still push there.
b) In order to provide a better user API experience granularity could be a Pandas offset shortcut string (like '5Min', '5H' ... in resample function ).
c) And it could be convert to a string compatible with Oanda API.
d) We could have a parameter named freq for DataReader function (it's role is quite similar to pandas.date_range parameter)

See #199

I'd convert freq (string) to Pandas DateOffset
http://stackoverflow.com/questions/27368265/convert-freq-string-to-dateoffset-in-pandas

and have a dict to convert Pandas DateOffset to Oanda granularity string.

Ask me if you don't understand what I mean (because I'm not english native speaker) so I may miss something.

femtotrader · 2016-11-08T08:41:07Z

pandas_datareader/oandarest.py

+            "1T":"M1",
+            "1min":"M1",
+            "2T":"M2",
+            "2min":"M2",


When you define SUPPORTED_OFFSET_ALIASES

You should have Pandas Offset as key not string shortcut as key.
Because it urge you to repeat several times some values.

In [1]: from pandas.tseries.frequencies import to_offset In [2]: to_offset("1Min") Out[2]: <Minute> In [3]: offset = to_offset("1Min") In [4]: offset Out[4]: <Minute> In [5]: type(offset) Out[5]: pandas.tseries.offsets.Minute

@femtotrader OANDA REST v20 only accept predefined freqencies. Eg: 1T is supported since M1 is an accepted value for OANDA REST v20's endpoint however 7T -> M7 is not supported since M7 accepted by the endpoint.

Does it still make sense to have pandas.tseries.offsets.XXXX object instead of hardcoded Panda Offset aliases ?

I will use this to_offset in a try/catch

so if freq can't be convert to Pandas Offset using to_offset it will be passed to Oanda directly.

On the other side I don't think it's a problem if user pass freq="7T" which will be convert to <7 * Minutes> but it won't be in SUPPORTED_OFFSET_ALIASES dict keys so we can decide here to raise an Exception to inform user that it's no an accepted timeframe.

Should we define keys for "1min", "1MIN", "1Min"...

In [9]: to_offset("1min") Out[9]: <Minute> In [10]: to_offset("1Min") Out[10]: <Minute> In [11]: to_offset("1MIN") Out[11]: <Minute>

I don't think so!

I don't think so

@femtotrader Ok. So i am thinking of accepting a string or DateOffset for freq param

How can a DateOffset object be converted back to string ? Eg:

In[1]d = to_offset("1min") In[2] ds = to_offset_string(d) Out[2] print(ds) # Displays "1min"

I'm not sure a DateOffset can be convert back to a "normalized" offset string

In [20]: offset = to_offset("7T") In [21]: offset. offset.apply offset.delta offset.kwds offset.nanos offset.rollback offset.apply_index offset.freqstr offset.n offset.normalize offset.rollforward offset.copy offset.isAnchored offset.name offset.onOffset offset.rule_code

but it could be an interesting feature to have in Pandas

In [18]: offset.rule_code Out[18]: 'T' In [20]: offset = to_offset("7T") In [21]: offset.n Out[21]: 7 In [22]: offset.rule_code Out[22]: 'T' In [23]: offset = to_offset("7Min") In [24]: offset.n Out[24]: 7 In [25]: offset.rule_code Out[25]: 'T'

a quick and dirty approach could be

In [47]: def normalized_offset_string(offset): ...: if offset.n == 1: ...: return offset.rule_code ...: else: ...: return str(offset.n) + offset.rule_code

femtotrader · 2016-11-08T09:47:16Z

There is in fact an attribute to get normalized string from DateOffset it's named: freqstr

In [63]: from pandas.tseries.frequencies import to_offset

In [64]: to_offset("7Min").freqstr
Out[64]: '7T'

In [65]: to_offset("7MIN").freqstr
Out[65]: '7T'

In [66]: to_offset("7min").freqstr
Out[66]: '7T'

femtotrader · 2016-11-08T09:48:10Z

I'm sorry I don't know "all" Pandas ;-)

davidandreoletti · 2016-11-08T09:49:15Z

@femtotrader That's fine :) I only expect you to know the answers to the questions I ask :P

davidandreoletti · 2016-11-08T09:49:47Z

@femtotrader My PR is done.

femtotrader · 2016-11-08T09:57:09Z

pandas_datareader/oandarest.py

+            "5T":"M5",
+            "5min":"M5",
+            "15T":"M15",
+            "15min":"M15",


freqstr DateOffset attribute should avoid to repeat these keys

In [67]: to_offset("15min").freqstr Out[67]: '15T' In [68]: to_offset("15T").freqstr Out[68]: '15T' In [69]: to_offset("15Min").freqstr Out[69]: '15T'

@femtotrader Fixed.

femtotrader · 2016-11-08T10:00:08Z

pandas_datareader/oandarest.py

+            current_end_rfc3339 = current_end.strftime(rfc3339)
+
+            if isinstance(freq, DateOffset):
+                offsetString = self._normalized_offset_string(freq)


offsetString = freq.freqstr

Sorry;-)

@femtotrader Fixed.

PS: No worries at all :)

femtotrader · 2016-11-08T10:15:44Z

pandas_datareader/data.py

@@ -56,7 +57,8 @@ def get_quote_google(*args, **kwargs):


 def DataReader(name, data_source=None, start=None, end=None,
-               retry_count=3, pause=0.001, session=None, access_key=None):
+               retry_count=3, pause=0.001, session=None, 
+               custom=None):


access_key is required for Enigma datareader.

@femtotrader Fixed.

femtotrader · 2016-11-08T10:17:57Z

pandas_datareader/data.py

+    oa = DataReader("EUR_USD", data_source="oanda_rest_historical_currency", 
+                    custom={    "accountType"="practice",
+                                    "accountVersion="0"
+                                    "apiToken":"Your private API token" })


oa = DataReader("EUR_USD", data_source="oanda_rest_historical_currency", custom = { "accountType": "practice", "accountVersion": "0", "apiToken": "Your private API token" })

@femtotrader Fixed

femtotrader · 2016-11-08T10:21:16Z

You need to run locally

flake8 --ignore E501 pandas_datareader

to fix some Flake8 errors

davidandreoletti · 2016-11-08T14:39:31Z

@femtotrader PR done :)

femtotrader · 2016-11-08T14:45:53Z

pandas_datareader/oandarest.py

+        "15T": "M15",
+        "30T": "M30",
+        "H": "H1",
+        "1H": "H1",


from pandas.tseries.frequencies import to_offset to_offset("1H").freqstr Out[20]: 'H' to_offset("H").freqstr Out[21]: 'H'

so we need just "H", not "1H"

femtotrader · 2016-11-08T14:49:07Z

pandas_datareader/oandarest.py

+        "15S": "S15",
+        "30S": "S30",
+        "T": "M1",
+        "1T": "M1",


1T key should be removed (same as 1H)

femtotrader · 2016-11-08T14:49:33Z

pandas_datareader/oandarest.py

+        "6H": "H6",
+        "8H": "H8",
+        "12H": "H12",
+        "1D": "D",


I think it might be

"D": "D",

because

In: to_offset("1D").freqstr Out: 'D'

femtotrader · 2016-11-08T14:50:49Z

pandas_datareader/oandarest.py

+        "8H": "H8",
+        "12H": "H12",
+        "1D": "D",
+        "1W": "W",


"W": "W",

femtotrader · 2016-11-08T14:51:07Z

pandas_datareader/oandarest.py

+        "12H": "H12",
+        "1D": "D",
+        "1W": "W",
+        "1M": "M"


femtotrader · 2016-11-08T14:53:37Z

pandas_datareader/oandarest.py

+        df.rename(inplace=True,
+                  index=str,
+                  columns={
+                      "time": "date",


Most DataReader index name is Date

"time": "Date",

femtotrader · 2016-11-08T14:54:50Z

pandas_datareader/oandarest.py

+                      "bid.c": "bid.close",
+                  })
+
+        df = df.set_index('date')


df = df.set_index('Date')

A DataFrame with a MultiIndex as columns will be a great feature to have
http://pandas.pydata.org/pandas-docs/stable/advanced.html

Most other DataReader with candlestick data are using; Open, High, Low, Close column name (not open, high, low, close)

Here is an example to show how a DataFrame with MultiIndexed columns works

In: periods = 10 In: idx = pd.date_range("2010-01-01", freq="D", periods=10) In: columns = [np.array(['Mid', 'Mid', 'Mid', 'Mid', 'Ask', 'Ask', 'Ask', 'Ask', 'Bid', 'Bid', 'Bid', 'Bid']), np.array(['Open', 'High', 'Low', 'Close', 'Open', 'High', 'Low', 'Close', 'Open', 'High', 'Low', 'Close'])] In: df = pd.DataFrame(np.random.randn(periods, 12), columns=columns, index=idx) In: df Out[59]: Mid Ask \ Open High Low Close Open High 2010-01-01 -0.357908 0.638564 -0.808755 0.029079 0.150620 0.282146 2010-01-02 1.015372 -0.624254 0.353841 1.491935 -0.322171 0.353117 2010-01-03 0.051334 0.516878 -1.769581 0.812352 -0.528826 0.677380 2010-01-04 -1.681174 0.411583 -1.032651 1.924580 0.162397 -0.422782 2010-01-05 0.512275 -0.112695 -0.705641 0.715283 0.446572 -1.428909 2010-01-06 -1.197475 -0.025064 -0.997523 1.252269 0.217094 -1.086155 2010-01-07 -0.935145 -0.446475 -1.220699 -1.135570 0.369933 1.695358 2010-01-08 -0.259295 0.164924 0.371293 -0.136309 2.280143 -1.256588 2010-01-09 -0.392548 1.762788 -1.656464 1.548067 -0.333061 0.226948 2010-01-10 -0.431441 0.674132 -1.381039 0.892813 0.793012 -0.755067 Bid Low Close Open High Low Close 2010-01-01 -1.313252 -1.179413 1.873345 -0.719676 -2.595192 1.182884 2010-01-02 -0.670126 0.421320 0.318873 -0.756884 1.189023 -0.785939 2010-01-03 -1.027978 0.609855 0.877085 -0.767293 -1.859567 0.662125 2010-01-04 1.049496 -1.062435 0.577158 -0.068621 -0.080246 -0.786441 2010-01-05 -3.193001 -0.540472 2.350964 -1.983940 2.278458 -0.783962 2010-01-06 -0.101567 -0.145842 -1.600600 0.098544 0.133307 0.262338 2010-01-07 -0.263570 -1.012718 -0.579255 1.360744 1.661800 -2.185659 2010-01-08 -2.260959 -1.143676 -0.900669 0.721042 -0.421265 -0.716886 2010-01-09 -1.229502 -0.442861 -0.767872 0.734038 -0.403405 0.046703 2010-01-10 1.522349 0.463781 0.522694 -0.700728 -0.726378 -1.144720 df["Ask"] Out[60]: Open High Low Close 2010-01-01 0.150620 0.282146 -1.313252 -1.179413 2010-01-02 -0.322171 0.353117 -0.670126 0.421320 2010-01-03 -0.528826 0.677380 -1.027978 0.609855 2010-01-04 0.162397 -0.422782 1.049496 -1.062435 2010-01-05 0.446572 -1.428909 -3.193001 -0.540472 2010-01-06 0.217094 -1.086155 -0.101567 -0.145842 2010-01-07 0.369933 1.695358 -0.263570 -1.012718 2010-01-08 2.280143 -1.256588 -2.260959 -1.143676 2010-01-09 -0.333061 0.226948 -1.229502 -0.442861 2010-01-10 0.793012 -0.755067 1.522349 0.463781 df["Ask"]["Open"] Out[61]: 2010-01-01 0.150620 2010-01-02 -0.322171 2010-01-03 -0.528826 2010-01-04 0.162397 2010-01-05 0.446572 2010-01-06 0.217094 2010-01-07 0.369933 2010-01-08 2.280143 2010-01-09 -0.333061 2010-01-10 0.793012 Freq: D, Name: Open, dtype: float64

Fixed: Use Open instead open, etc ...

I will look at A DataFrame with a MultiIndex later this week.

Fixed: Added MultiIndex support. Thanks for the suggestion @femtotrader !

femtotrader · 2016-11-08T14:56:08Z

pandas_datareader/tests/test_oandarest.py

+
+    def get_credential(self):
+        return {'accountType': "practice",
+                'apiToken': "8106b9109c7c6b17d45ff215f6da9d42-1446c148ee0f4ac9274eb6c2be0784b7"


We can "hide" this in an environment variable if you want (like Enigma)

Fixed: Token removed from tests. Now users can specify the token via an env variable

femtotrader · 2016-11-08T15:26:30Z

pandas_datareader/oandarest.py

+            self.access_credential['apiToken'] = ""
+
+    def read(self):
+        dfs = {}


using an OrderedDict will keep order of symbols in this Panel

from pandas.compat import OrderedDict

femtotrader · 2016-11-08T15:28:50Z

pandas_datareader/tests/test_oandarest.py

+    def test_oanda_historical_currencypair(self):
+        start = "2014-03-19T09:00:00Z"
+        end = "2014-03-21T9:00:00Z"
+        symbols = ["EUR_USD"]


Does it works with simply

symbols = "EUR_USD"

?

Fixed: string is now also accepted to represent a single currency pair.

Nice.
Great job on this PR!

femtotrader · 2016-11-08T15:29:17Z

pandas_datareader/tests/test_oandarest.py

+            pn = OANDARestHistoricalInstrumentReader(
+                symbols=symbols,
+                start=start, end=end,
+                freq="S5",


freq="5T",

which will be convert to Pandas DateOffset
which will be convert to "5T" (using freqstr)
which will be find in dictionnary and return "S5"

femtotrader · 2016-11-08T15:31:28Z

pandas_datareader/tests/test_oandarest.py

+    def test_oanda_historical_currencypair2(self):
+        start = "2014-03-19T09:00:00Z"
+        end = "2014-03-21T09:00:00Z"
+        symbols = ["EUR_USD"]


A test with several symbols to ensure that we get a Panel will be a good idea.

femtotrader

OANDA_API_TOKEN is now set on Travis.

I'm not sure but we probably need to export this environment variable in .travis.yml

export OANDA_API_TOKEN=$OANDA_API_TOKEN

femtotrader · 2016-11-10T17:35:16Z

pandas_datareader/oandarest.py

+                    8H  ->  8 hour candlesticks, day alignment
+                    12H -> 12 hour candlesticks, day alignment
+                    1D  ->  1 day candlesticks, day alignment
+                    1W  ->  1 week candlesticks, aligned to start of week


femtotrader · 2016-11-10T17:35:26Z

pandas_datareader/oandarest.py

+                    6H  ->  6 hour candlesticks, day alignment
+                    8H  ->  8 hour candlesticks, day alignment
+                    12H -> 12 hour candlesticks, day alignment
+                    1D  ->  1 day candlesticks, day alignment


femtotrader · 2016-11-10T17:35:39Z

pandas_datareader/oandarest.py

+                    12H -> 12 hour candlesticks, day alignment
+                    1D  ->  1 day candlesticks, day alignment
+                    1W  ->  1 week candlesticks, aligned to start of week
+                    1M  ->  1 month candlesticks, aligned to first day of the month


femtotrader · 2016-11-10T17:37:38Z

pandas_datareader/tests/test_oandarest.py

+    def test_oanda_historical_currencypair(self):
+        start = "2014-03-19T09:00:00Z"
+        end = "2014-03-21T9:00:00Z"
+        symbols = ["EUR_USD"]


Nice.
Great job on this PR!

femtotrader · 2016-11-10T17:44:57Z

pandas_datareader/oandarest.py

@@ -0,0 +1,313 @@
+import pandas as pd


import os

is required

davidandreoletti · 2016-11-11T14:08:53Z

OANDA_API_TOKEN is now set on Travis.

I'm not sure but we probably need to export this environment variable in .travis.yml

export OANDA_API_TOKEN=$OANDA_API_TOKEN

The initial API TOKEN I committed was a mistake and it has now been revoked. So I will not provide an API TOKEN. Alternatively, someone (else) can provide an OANDA API TOKEN, encrypt it and then store it as encrypted env variable (supported by Travis).

davidandreoletti · 2016-11-13T04:36:52Z

@femtotrader PR done.

femtotrader · 2016-11-14T08:51:48Z

I added my API key in Travis. But I think you need to add

export OANDA_API_TOKEN=$OANDA_API_TOKEN

in .travis.yml

…ader API) waits indefinitly due threads not gracefully exiting in case of unhandled exception

davidandreoletti · 2016-12-11T07:22:55Z

@femtotrader Is OANDA_API_TOKEN now unset in Travis (or revoked) ? CI fails with 401 errors when running the tests for this reader.

femtotrader · 2016-12-11T08:05:55Z

docs/source/remote_data.rst

+    In [1]: import pandas_datareader.data as web
+    In [2]: start, end = "2016-01-01", "2016-02-01"
+    In [3]: symbols = ["EUR_USD"]
+    In [4]: credentials["accountType"]="practise"


femtotrader · 2016-12-11T08:06:15Z

docs/source/remote_data.rst

+    In [1]: import pandas_datareader.data as web
+    In [2]: start, end = "2016-01-01", "2016-02-01"
+    In [3]: symbols = ["EUR_USD"]
+    In [4]: credentials["accountType"]="practise"


empty credentials dict is required

femtotrader · 2016-12-11T08:48:03Z

docs/source/remote_data.rst

+    In [5]: credentials["apiToken"]="Your OANDA API token"
+    In [6]: pn = web.DataReader(
+                symbols, data_source="oanda_historical_currency",
+                start=start, end=end,


We might be able to pass the freq parameter

ischurov · 2016-12-11T10:15:30Z

Hi, could you please comment on the proposed solution to pandas-dev/pandas#14826 ? (See this comment and below.)

davidandreoletti · 2016-12-12T15:24:37Z

@femtotrader PR ready.

femtotrader · 2016-12-12T17:39:43Z

I'd be very pleased to merge it... but CI is still failing.

I don't understand why.

Could you try to add simply

    print(self.access_credential)

after

    if 'apiToken' not in self.access_credential:
        self.access_credential['apiToken'] = os.getenv('OANDA_API_TOKEN')
        if self.access_credential['apiToken'] is None:
            raise ValueError(
                """Please provide an OANDA API token or set the OANDA_API_TOKEN environment variable\n
                If you do not have an API key, you can get one here: http://developer.oanda.com/rest-live/authentication/""")

You should show my API key (so this approach is not very secure... but enough here because I don't mind if you display it in a Travis build.)

See https://github.com/femtotrader/test

davidandreoletti · 2016-12-13T00:07:30Z

@femtotrader The API token is missing. See output. Once added back, I will revert the last commit.

femtotrader · 2016-12-13T06:15:14Z

That's very odd because I set in Travis

OANDA_API_TOKEN=9b7baf...-...903cd6

Put this token in code because I'm fed up with Travis

femtotrader · 2016-12-13T06:20:21Z

About request and multithreading you might have a look at
http://stackoverflow.com/questions/38280094/python-requests-with-multithreading

davidandreoletti · 2016-12-13T09:57:03Z

@femtotrader

Put this token in code because I'm fed up with Travis

I advise you not to. OANDA may revoke this token at best or close your account for some breach (after all those are considered personal info)

I personally set the OANDA_API_TOKEN env variable in my bash scripts so my environment is quite close to Travis I suppose.

Are you using encrypted environment value or not ?

It seems like encrypted env value are not provided to untrusted builds such as the ones initiated by pull requests.

About request and multithreading you might have a look at http://stackoverflow.com/questions/38280094/python-requests-with-multithreading

Python is not well suited for multithreaded code due to GIL. I have limited GIL knownlegde and initially thought this was a concern for CPU intensive apps. Turns out it also has implications with IO as per the doc provided.

This PR uses oanda-api-v20 to access OANDA's endpoints. So I lodged a feature request to that project.

hootnot · 2016-12-16T09:21:14Z

@davidandreoletti
Use requests_mock to fake the API requests. That way you can bypass the token sharing problem.
Take a look at the tests in the oandapyV20 wrapper repo: https://github.com/hootnot/oanda-api-v20

…ata for a particular account

davidandreoletti · 2016-12-18T05:26:01Z

@hootnot

a)Do you know why the environment variable OANDA_API_TOKEN's value is not visible in unit test run on Travis while in my personal environment it is visible ? If you do, would you mind explaining it to me, please ? :)

@femtotrader

b) if @hootnot's solution cannot be used, I don't have access to Travis to investigate the environment variable issue, can you investigate it or grant me temporary access to Travis ?

c) I added another reader OANDARestAccountInstrumentReader. It is returning hard coded data and is missing in the documentation because I could not get it to fetch real data with my personal account. What do you want to do with it ("keep it undocumented in the documentation + warning in the source code", "something else") ?

d) I think we should move the oandrest.py classes (+ perhaps oanda.py) into a proper python module. What do you think ?

femtotrader · 2016-12-18T07:52:47Z

Pinging @sinhrks @davidastephens @jreback
Could you help on this ?

hootnot · 2016-12-18T10:18:04Z

@davidandreoletti @femtotrader

regarding a):

This is not how it works. Travis can never know the value of the OANDA_API_TOKEN this way. If you use it from your shell this works. But the shell Travis uses has the value not set, so it it empty. Travis gets its environment variables from .travis.yml. Structure described in: https://docs.travis-ci.com/user/environment-variables/

If you want to use the/your token in the tests (and I would advise you not to) you need to set it up the way Travis points out in their docs, using encryption. The owner of the repo must do this since it involves a procedure with keys.
You end up with a section in .travis.yml looking like:

env:
  global:
  - secure: xxxxxxxxxxxxxxxxxxxxxxx

But when you do it this way you share a token as a contributor to the repo-owner. You do not want that for both parties:

as a contributor you can revoke the token and tests would break, so the owner relies on a contributor here
as owner you have access to a token which was offered on certain conditions by the issueing party to the contributor.

If you do want to use the token in the tests this way I suggest the best solution is that the repo-owner obtains a token and does the Travis setup with that token. This way the owner has control and responsibility for the token and contributors can still run the tests.

So if you want to use encrypted environment variables, it is up to to repo-owner to set this up.

I use this encryption setup in https://github.com/hootnot/oanda-api-v20/ to deploy to pypi automatically when tagging a release.

I still think that you should use mocking to setup your tests. With mocking you can bypass the issue of the token. In the next example you find a Panda class that represents the same kind of flow as you use in OANDARestHistoricalInstrumentReader :

import sys
import json
import unittest

import oandapyV20
import oandapyV20.endpoints.instruments as instruments
import requests_mock

data = range(10)      # just some testdata, 
environment = "practice"
access_token = None
api = None

class Panda(object):
    # some test class supposing to read data from OANDA using oandapyV20
    # ( and it will if you pass it a valid token )
    # the access_token is not relevant since the request is mocked
    def __init__(self, access_token="xxx"):  
        self.client = oandapyV20.API(access_token=access_token)

    def get_data(self, instrument, params):
        r = instruments.InstrumentsCandles(instrument=instrument, params=params)
        # this will call the mocked URI and return the data you provided setting up the mock request
        rv = self.client.request(r)
        return rv


class OA(unittest.TestCase):

    def setUp(self):
            global access_token
            global api

            # override the environments values with the ones relevant for the mock requests
            setattr(sys.modules["oandapyV20.oandapyV20"],
                    "TRADING_ENVIRONMENTS",
                    {"practice": {
                     "stream": "https://test.com",
                     "api": "https://test.com",
                     }})
            # setup the client the regular way
            api = oandapyV20.API(environment=environment,
                      access_token=access_token)
            api.api_url = "https://test.com"


    @requests_mock.Mocker()
    def test_candles(self, mock_req):
        params = { "granularity": "M5", "count": 5}
        r = instruments.InstrumentsCandles(instrument="EUR_USD", params=params)
        mock_req.register_uri('GET', "{}/{}".format(api.api_url, r),
                              text=json.dumps(data))
        # the statements before registered the uri, when this one gets called with this parameters, it returns
        # the associated data

        # try that with this test class
        panda = Panda()
        pandadata = panda.get_data("EUR_USD", params=params)
        self.assertEqual(pandadata, data)



if __name__ == "__main__":

     unittest.main()

create some testdata representing the data you get in real from a request. Store this in a file or files along with your tests. Write some method to read this data to setup the mock request and you can simulate the IO with OANDA's REST V20.

Regarding d)

I think it is wise to split some things up. The oandarest.py contains a lot of constants:

the offset-aliases
the self.OANDA_* and self.DATAFRAME_*

These could be separated and that makes them importable in other code also, which is currently not the case. It also reduces the size of oandarest.py about 10% making it more readable.

gliptak · 2017-07-11T01:13:05Z

@davidandreoletti Support for Oanda was removed with #366 Consider closing this issue

femtotrader · 2017-07-11T07:18:24Z

@davidandreoletti you did a great work on this PR but
I'm closing because of OANDA restrictions.

davidandreoletti force-pushed the fx-oanda-historical branch from ae458ad to d8f8d32 Compare November 8, 2016 06:03

femtotrader mentioned this pull request Nov 8, 2016

Add OANDA Rest Historical Instrument Reader #255

Closed

femtotrader reviewed Nov 8, 2016

View reviewed changes

femtotrader mentioned this pull request Nov 8, 2016

DateOffset should have a property for a "normalized" shortcut string pandas-dev/pandas#14614

Closed

davidandreoletti changed the title ~~[WIP] Add OANDA RestV20 Historical Instrument Reader~~ Add OANDA RestV20 Historical Instrument Reader Nov 8, 2016

femtotrader reviewed Nov 8, 2016

View reviewed changes

davidandreoletti mentioned this pull request Nov 8, 2016

Publish OANDAPY on https://pypi.python.org/pypi oanda/oandapy#39

Closed

femtotrader reviewed Nov 8, 2016

View reviewed changes

davidandreoletti changed the title ~~Add OANDA RestV20 Historical Instrument Reader~~ [WIP] Add OANDA RestV20 Historical Instrument Reader Nov 10, 2016

femtotrader reviewed Nov 10, 2016

View reviewed changes

davidandreoletti changed the title ~~[WIP] Add OANDA RestV20 Historical Instrument Reader~~ Add OANDA RestV20 Historical Instrument Reader Nov 13, 2016

davidandreoletti added 4 commits December 11, 2016 13:51

Added support for importing Python 2.x Queue module

210d0c9

Log OANDA specific errors

57957ec

Prevented condition where the calling thread (calling the PandaDataRe…

257d108

…ader API) waits indefinitly due threads not gracefully exiting in case of unhandled exception

Return None for each symbol where irrecoverable exception occured

0dfb503

femtotrader reviewed Dec 11, 2016

View reviewed changes

davidandreoletti added 2 commits December 12, 2016 20:47

Fixed documentation

e6f01d8

Fixed F841 error

2b53d24

Show credentials data (for debugging purpose)

d063ab6

davidandreoletti mentioned this pull request Dec 13, 2016

Forex support quantopian/zipline#1285

Closed

davidandreoletti mentioned this pull request Dec 13, 2016

Asynchronous API support hootnot/oanda-api-v20#61

Closed

davidandreoletti added 2 commits December 18, 2016 13:12

Use logging.logger facility for logging statements instead of print

56c28cf

Added OANDARestAccountInstrumentReader to get available instruments d…

0830029

…ata for a particular account

davidandreoletti mentioned this pull request Jul 11, 2017

OANDA is broken #296

Closed

femtotrader closed this Jul 11, 2017

Add OANDA RestV20 Historical Instrument Reader #257

Add OANDA RestV20 Historical Instrument Reader #257

Uh oh!

Conversation

davidandreoletti commented Nov 8, 2016

Uh oh!

davidandreoletti commented Nov 8, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

davidandreoletti Nov 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

femtotrader Nov 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

davidandreoletti Nov 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

femtotrader Nov 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

femtotrader commented Nov 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

femtotrader commented Nov 8, 2016

Uh oh!

davidandreoletti commented Nov 8, 2016

Uh oh!

davidandreoletti commented Nov 8, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

femtotrader Nov 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

femtotrader commented Nov 8, 2016

Uh oh!

davidandreoletti commented Nov 8, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

davidandreoletti Nov 8, 2016 •

edited

Loading

femtotrader Nov 8, 2016 •

edited

Loading

davidandreoletti Nov 8, 2016 •

edited

Loading

femtotrader Nov 8, 2016 •

edited

Loading

femtotrader commented Nov 8, 2016 •

edited

Loading

femtotrader Nov 8, 2016 •

edited

Loading

femtotrader Nov 8, 2016 •

edited

Loading