Use the Interactive Brokers API (IBAPI) asynchonouslywith asyncio or PyQt5
This project has evolved into IB-insync and all development is happening there now.
To port code:
- TWSClient has moved to ib_insync.client.Client (with much improvements);
- TWSClientQt: The new Client can be used with PyQt and quamash.
This offers a simpler, safer and more performant approach to concurrency than multithreading.
Install using pip:
pip3 install -U tws_async
Note that on some systems the pip3 command is just pip.
This package offers two clients that can be used as a drop-in replacement for the standard EClient as provided by IBAPI:
- TWSClient, for use with the asyncio event loop;
- TWSClientQt, for use with the PyQt5 event loop.
These clients also inherit from ibapi.wrapper.EWrapper and can be used exactly as one would use the standard IBAPI version. The asynchronous clients use their own event-driven networking code that replaces the networking code of the standard EClient, and they also replace the infinite loop of EClient.run() with an event loop.
To simplify working with contracts, this package provides Contract, Stock, Option, Future, Forex, Index, CFD and Commodity classes that can be used anywhere where a ibapi.contract.Contract is expected. Examples of some simple cases are Stock('AMD'), Forex('EURUSD'), CFD('IBUS30') or Future('ES', '201612', 'GLOBEX'). To specify more complex contracts, any property can be given as a keyword.
To learn more, consult the official IBAPI documentation or have a look at these sample use cases:
Historical data downloader
Realtime streaming ticks
The tick streamer subscribes to realtime tick data.
To use the Interactive Brokers API fully interactively in a Jupyter notebook, have a look at the example notebook.
Jupyter can be started with the command jupyter notebook.
This notebook uses the Qt version of the client, where the Qt event loop is started with the %gui qt5 directive at the very top. It is not necessary to call the run() method of the client.
Notes on using asycio in a notebook
Currently there does not seem to be a single-threaded way to directly run the asyncio event loop in Jupyter. What can be done is to use the Qt event loop (which does have good integration with the Jupyter kernel) with the quamash adaptor. With quamash the Qt event loop is used to drive the asyncio event loop. This can be done by placing the following code at the top of the notebook:
%gui qt5 import asyncio, quamash loop = quamash.QEventLoop() asyncio.set_event_loop(loop)
One thing that does not work in the combination of quamash and Jupyter is the loop.run_until_finished method. It can be patched like this:
def run_until_complete(self, future): future = asyncio.ensure_future(future) qApp = qt.QApplication.instance() while not future.done(): qApp.processEvents(qt.QEventLoop.WaitForMoreEvents) return future.result() quamash.QEventLoop.run_until_complete = run_until_complete
The asyncio version of the client relies on loop.run_until_finished to connect synchonously. So in order to run the asyncio client in the notebook, apply the patch or just connect asynchonously (i.e. give asyncConnect=True to the connect call).
- HistRequester fix for endDateTime formatting
- HistRequester updated to version 9.73.04 of the API
- small simplifications
- connect() call of the clients will now by default block until client is ready to serve requests.
- getReqId() method added to both clients.
- dataHandlingPre() and dataHandlingPost() event hooks added to clients.
- logging added.
- util module aded.
- file tws_async.py renamed to twsclient.py, tws_asyncqt.py to twsclientqt.py.
- Added optional asyncConnect argument to client.connect() method. The default is now to connect synchronously (block until connected).
- Fixed bug in HistRequester when downloading daily data.
- Initial pip package release.
Good luck and enjoy,
|author:||Ewald de Wit <firstname.lastname@example.org>|
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size tws_async-0.5.7-py3-none-any.whl (25.8 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
Hashes for tws_async-0.5.7-py3-none-any.whl