Hyperstream is a lightweight, flexible and robust software package for processing streaming data.
Hyperstream overcomes the limitations of other computational engines and provides high-level interfaces to execute complex nesting, fusion, and prediction both in online and offline forms in streaming environments. Although developed specifically for SPHERE, Hyperstream is a general purpose tool that is well-suited for the design, development, and deployment of algorithms and predictive models in a wide space of sequential predictive problems.
This software has been designed from the outset to be domain-independent, in order to provide maximum value to the wider community. Key aspects of the software include the capability to create complex interlinked workflows, and a computational engine that is designed to be "compute-on-request", meaning that no unnecessary resources are used.
If you do not want to install all the packages separately you can use our Docker bundle available here.
Install via pip
pip install hyperstream python -c 'from hyperstream import HyperStream'
To get the latest version
pip install -U git+git://github.com/IRC-SPHERE/HyperStream.git#egg=hyperstream pip install -r requirements.txt
Or clone the repository
git clone firstname.lastname@example.org:IRC-SPHERE/HyperStream.git cd HyperStream virtualenv venv . venv/bin/activate pip install -r requirements.txt python -c 'from hyperstream import HyperStream'
Additionally, one of the requirements to run Hyperstream is a MongoDB server. By default, Hyperstream tries to connect to the port 27017 on the localhost.
To access via a cloud provider, you might try mLab, MongoDB Atlas, or search for a different one. If you are using mLab, here are a few notes:
- Set up an account via the mLab web site instructions. When asked to pick a server type (e.g. Amazon, Google, etc) you can just choose free option of 500MB. This is more than enough to get started.
- mLab will ask you to create a database; any name is fine, but make sure you write down what it is.
- After creating a database, note that you’ll need to create at least one database user in order to access the database.
- You can test your database connection using MongoDB’s built-in command line tools
To install MongoDB locally go to the official documentation. E.g. in a Debian OS it is possible to install with the following command
sudo apt-get install mongodb
Once the MongoDB server is installed, it can be started with the following command
service mongod start
If installing MongoDB on OSX
brew install mongodb
ln -sfv /usr/local/opt/mongodb/*.plist ~/Library/LaunchAgents
run MongoDB service with
launchctl load ~/Library/LaunchAgents/homebrew.mxcl.mongodb.plist
Run the following command
Note that for the MQTT logging test to succeed, you will need to have an MQTT broker running (e.g. Mosquitto). For example:
docker run -ti -p 1883:1883 -p 9001:9001 toke/mosquitto
or on OSX you will need pidof and mosquitto:
brew install pidof brew install mosquitto brew services start mosquitto
The following tutorials show how to use HyperStream in a step-by-step guide.
- Tutorial 1: Introduction
- Tutorial 2: Creating tools
- Tutorial 3: Stream composition
- Tutorial 4: Real-time streams
- Tutorial 5: Workflows
Running the tutorials in a docker container
It is possible to run all the tutorials in your own machine ussing Docker containers defined in IRC-SPHERE/Hyperstream-Dockerfiles. You can do that by running the following commands:
git clone https://github.com/IRC-SPHERE/Hyperstream-Dockerfiles.git cd Hyperstream-Dockerfiles docker-compose -f docker-compose-tutorials.yml -p hyperstream-tutorials up
And then open the url http://0.0.0.0:8888/tree in a web-browser
Running the tutorials in the cloned folder
To run the tutorials in the cloned repository you will need to install additional dependencies. First you should activate the virtual environment and installed the general requirements to run HyperStream following the instructions above. After that, install the dependencies for the tutorial with
pip install -r requirements_tutorial.txt
and go to the experiments folder
And run a Jupyter notebook
Now you can follow the instructions from the first tutorial.
from hyperstream import HyperStream, StreamId, TimeInterval from hyperstream.utils import utcnow, UTC from datetime import timedelta hs = HyperStream(loglevel=20) M = hs.channel_manager.memory T = hs.channel_manager.tools clock = StreamId(name="clock") clock_tool = T[clock].window().last().value() ticker = M.get_or_create_stream(stream_id=StreamId(name="ticker")) now = utcnow() before = (now - timedelta(seconds=30)).replace(tzinfo=UTC) ti = TimeInterval(before, now) clock_tool.execute(sources=, sink=ticker, interval=ti, alignment_stream=None) list(ticker.window().tail(5))
The last list contains
[StreamInstance(timestamp=datetime.datetime(2017, 7, 27, 10, 33, 45, tzinfo=<UTC>), value=datetime.datetime(2017, 7, 27, 10, 33, 45, tzinfo=<UTC>)), StreamInstance(timestamp=datetime.datetime(2017, 7, 27, 10, 33, 46, tzinfo=<UTC>), value=datetime.datetime(2017, 7, 27, 10, 33, 46, tzinfo=<UTC>)), StreamInstance(timestamp=datetime.datetime(2017, 7, 27, 10, 33, 47, tzinfo=<UTC>), value=datetime.datetime(2017, 7, 27, 10, 33, 47, tzinfo=<UTC>)), StreamInstance(timestamp=datetime.datetime(2017, 7, 27, 10, 33, 48, tzinfo=<UTC>), value=datetime.datetime(2017, 7, 27, 10, 33, 48, tzinfo=<UTC>)), StreamInstance(timestamp=datetime.datetime(2017, 7, 27, 10, 33, 49, tzinfo=<UTC>), value=datetime.datetime(2017, 7, 27, 10, 33, 49, tzinfo=<UTC>))]
The HyperStream Viewer is a python/Flask web-app for interacting with HyperStream. In order to keep HyperStream to a minimum, this web-app is released as a separate repository that takes the core as a dependency.
This code is released under the MIT license.
This work has been funded by the UK Engineering and Physical Sciences Research Council (EPSRC) under Grant EP/K031910/1 - "SPHERE Interdisciplinary Research Collaboration".