Skip to main content

Integration test harness useful for distributed deployments

Project description

nile

About

Build Status codecov PyPI version PyPI - Python Version PyPI - License PyPI - Downloads

nile is a library that aims to simplify building applications that leverage real time data collected from one or more sensors or media devices. nile simplifies this process by implementing the boiler plate code that orchestrates configurations composed of these key abstractions:

  • Device abstraction that yields data
  • Functional layer to transform data
  • Processing layer that applies logic to transformation output

nile will handle all device orchestration including wiring up the correct device, transformer and processing layers, managing device lifecycle and cleaning up unused resources.

Installation

nile currently supports Python versions 3.4, 3.5, 3.6 and 3.7. It can be installed by running pip install nile.

Samples

A dockerized sample is included with this repository. In this sample nile orchestrates streaming from multiple webcam devices pipes the feeds to a camera-specific set of consumers which print output in a table format to stdout. More specifically, the following configurations are used:

  • Intersection Webcam
    • Frames converted to grayscale
    • Output piped to ML model that counts vehicals
    • Output piped to brightness detection
  • College Campus Webcam
    • Frames converted to grayscale
    • Output piped to brightness detection

The sample can be run locally by running the following commands:

# clone the repo
git clone https://github.com/microsoft/nile.git
# enter the nile directory
cd nile/
# ensure Docker is running
docker --version
# build the project
docker build . -t nile
# run the samples
docker run nile

Basic Usage

In order to use nile you must implement three abstractions (documentation here):

  • Device: Represents a data stream adapter to a physical media device. For example you may build separate devices to stream audio from microphone and HTTP endpoints.
  • Transformer: Transform source data. For example, you may transform an MPEG audio stream into PCM/WAV format.
  • Consumer: Represents a consumer of the transformed device data. There may be many of these per device. For example, in an audio-streaming application you may build separate consumers to (1) analyze audio wave frequencies, (2) analyze audio wave amplitudes and (3) archive audio onto a cloud-provided storage endpoint.

The samples directory contains real-world implementations that target live webcam video analysis. Examples can be found here.

Once you have these you can orchestrate them using nile:

from time import sleep
from nile.controllers import DeviceController, DeviceFleetManager
# import your implementations here...

# the controller manages the configuration for a single device
device_controller = DeviceController(
    YourCustomDevice(),
    YourCustomTransformer(),
    (YourCustomDeviceConsumer(),),
    throttle_time_ms=-1     # a positive value specifies the polling delay on the device
)

# the fleet manager handles configurations for multiple devices
fleet_manager = DeviceFleetManager()
fleet_manager.set_controllers({device_controller})

# process for 20 seconds
sleep(20)

# this closes all devices under management and cleans any resources for those devices
fleet_controller.shutdown()

Advanced usage

nile supports non-trivial configurations as well and can update in real-time to configuration updates. Here is a more complex orchestration that leverages these features. The source code for the providers can be found here:

# example runs brigtness detection on video frames for 20 seconds, and then adds a car count consumer
# to one of the video feeds.
from time import sleep
from nile.controllers import DeviceController, DeviceFleetManager
from providers import BrightnessMeasureConsumer, CarCounterConsumer, FrameToGrayscaleTransformer, HttpVideoCaptureDevice


# ML model to count cars
car_count_consumer = CarCounterConsumer()

# Measure avg. brightness
brightness_measure_consumer = BrightnessMeasureConsumer()

# Convert img to grayscale
image_converter = FrameToGrayscaleTransformer()

road_cam = HttpVideoCaptureDevice('Tennasee Road', 'http://traf3.murfreesborotn.gov/axis-cgi/mjpg/video.cgi')
campus_cam = HttpVideoCaptureDevice('Bowdoin Campus', 'http://webcam1.bowdoin.edu:4200/axis-cgi/mjpg/video.cgi')
fleet_controller = DeviceFleetManager()

# start with both feeds running through brightness measuring
fleet_controller.set_controllers({
    DeviceController(
        device=road_cam,
        transformer=image_converter,
        consumers=(brightness_measure_consumer,),
        throttle_time_ms=-1),
    DeviceController(
        device=campus_cam,
        transformer=image_converter,
        consumers=(brightness_measure_consumer,),
        throttle_time_ms=-1)
})

# process for 20 seconds
sleep(20)

# add car count detection on road cam
fleet_controller.set_controllers({
    DeviceController(
        device=road_cam,
        transformer=image_converter,
        consumers=(brightness_measure_consumer, car_count_consumer),
        throttle_time_ms=-1),
    DeviceController(
        device=campus_cam,
        transformer=image_converter,
        consumers=(brightness_measure_consumer,),
        throttle_time_ms=-1)
})

# process for 20 seconds
sleep(20)

# close devices and clean up all resources
fleet_controller.shutdown()

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.

When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

For additional guidance, refer to the project-specific contributing guidelines as well as the pull request template

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

nile-0.0.3.tar.gz (14.1 kB view details)

Uploaded Source

Built Distribution

nile-0.0.3-py3-none-any.whl (15.0 kB view details)

Uploaded Python 3

File details

Details for the file nile-0.0.3.tar.gz.

File metadata

  • Download URL: nile-0.0.3.tar.gz
  • Upload date:
  • Size: 14.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/1.13.0 pkginfo/1.5.0.1 requests/2.21.0 setuptools/41.0.1 requests-toolbelt/0.9.1 tqdm/4.31.1 CPython/3.6.3

File hashes

Hashes for nile-0.0.3.tar.gz
Algorithm Hash digest
SHA256 6f4e1975aebe17c8fb8a47163d40b3849a29745e2e6f9fd51ee04f98ffa06626
MD5 4e069e91f0d3ebf5cbebe7a81a06b52a
BLAKE2b-256 dc927723c3bd0aaa7dfe0f86c4613ed949946934935fc86e01f08fe752ca6503

See more details on using hashes here.

File details

Details for the file nile-0.0.3-py3-none-any.whl.

File metadata

  • Download URL: nile-0.0.3-py3-none-any.whl
  • Upload date:
  • Size: 15.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/1.13.0 pkginfo/1.5.0.1 requests/2.21.0 setuptools/41.0.1 requests-toolbelt/0.9.1 tqdm/4.31.1 CPython/3.6.3

File hashes

Hashes for nile-0.0.3-py3-none-any.whl
Algorithm Hash digest
SHA256 21b776235797a25cf2c4638327ece4a3e7e1ad3abab2cae7d17ceb2f73af186a
MD5 9d335e9b10c28a13bdc6f984450f286d
BLAKE2b-256 e29f6f984df3b50a018ac1814611be9fb6cd84d22ce0756e495c238844edb8ac

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page