module testing.dummy_applications

Short summary

module lightmlrestapi.testing.dummy_applications

Machine Learning Post request

source on GitHub

Functions

function

truncated documentation

_distance_img

Computes the distance between two images. The function uses Pillow.

_distance_img_b64

Calls _distance_img() on an image encoded with base64.

dummy_application

Defines a dummy application using this API. It returns a score produced by a model trained on Iris datasets

dummy_application_auth

Defines a dummy application using this API including authentification. It returns a score produced by a model trained …

dummy_application_fct

Defines an application as defined in the tutorial Create your first REST API.

dummy_application_image

Defines a dummy application using this API and processing one image. The API ingests an image, resizes it to 224x224 …

dummy_application_neighbors

Defines a dummy application using this API. It returns a list of neighbors with a score on Iris datasets. …

dummy_application_neighbors_image

Defines a dummy application using this API. It returns a list of one neighbor for an image and metadata (random). …

dummy_mlstorage

Defines a dummy application using this API. It stores a model and it returns a score produced by a model trained …

Documentation

Machine Learning Post request

source on GitHub

lightmlrestapi.testing.dummy_applications._distance_img(img1, img2, arr1=None)

Computes the distance between two images. The function uses Pillow.

Parameters
  • img1 – reference PIL.Image.Image

  • img2 – new image PIL.Image.Image

  • arr1 – img1 as an array if available (to avoid converting the same image multiple times)

Returns

distance (in [0, 1]) or list of distances

source on GitHub

lightmlrestapi.testing.dummy_applications._distance_img_b64(img1, img2, arr1=None)

Calls _distance_img on an image encoded with base64.

Parameters
  • img1 – reference PIL.Image.Image

  • img2 – new image or list of images encoded with base64

  • arr1 – img1 as an array if available (to avoid converting the same image multiple times)

Returns

distance (in [0, 1]) or list of distances

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_application(app=None, **params)

Defines a dummy application using this API. It returns a score produced by a model trained on Iris datasets and two features.

Parameters
Returns

app

Query a REST API with features

This example shows how to query a REST API by sending a vector of features. You can start it by running:

start_mlrestapi --name=dummy

And then query it with:

import requests
import ujson
features = ujson.dumps({'X': [0.1, 0.2]})
r = requests.post('http://127.0.0.1:8081', data=features)
print(r)
print(r.json())

It should return:

{'Y': [[0.4994216179, 0.4514893599, 0.0490890222]]}

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_application_auth(app=None, algo='sha224', **params)

Defines a dummy application using this API including authentification. It returns a score produced by a model trained on Iris datasets and two features.

Parameters
  • app – application, if None, creates one

  • algo – algorithm used to encrypt the passwords

  • params – parameters sent to MachineLearningPost

Returns

app

It adds one users with the login ‘me’ and the password ‘dummy’.

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_application_fct(restapi_load, restapi_predict, users=None, algo='sha224', app=None, **params)

Defines an application as defined in the tutorial Create your first REST API.

Parameters
  • restapi_load – function to load a model

  • restapi_predict – predict function

  • params – parameters sent to MachineLearningPost

  • users – restrict to authenticated users, load_passwords

  • algo – algorithm used to encrypt password

  • app – application, if None, creates one

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_application_image(app=None, options=None, **params)

Defines a dummy application using this API and processing one image. The API ingests an image, resizes it to 224x224 and returns a distance to an original image from subfolder data.

Parameters
  • app – application, if None, creates one

  • options – if not empty, path to an image

  • params – parameters sent to MachineLearningPost

Returns

app

Query a REST API with an image

This example shows how to query a REST API by sending an image. You can start it by running:

start_mlrestapi --name=dummyimg

And then query it with:

import requests
import ujson
from lightmlrestapi.args import image2base64
img = "path_to_image"
b64 = image2base64(img)[1]
features = ujson.dumps({'X': b64}, reject_bytes=False)
r = requests.post('http://127.0.0.1:8081', data=features)
print(r)
print(r.json())

It should return something like:

{'distance': [0.21]}

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_application_neighbors(app=None, **params)

Defines a dummy application using this API. It returns a list of neighbors with a score on Iris datasets.

Parameters
Returns

app

Query a REST API with an image and get neighbors

A previous example shows how to send an image, this one illustrates a result which is a classifier result neither a regressor one but neighbors. You can start it by running:

start_mlrestapi --name=dummyknn

And then query it with:

import requests
import ujson
features = ujson.dumps({'X': [0.1, 0.2]})
r = requests.post('http://127.0.0.1:8081', data=features)
print(r)
print(r.json())

It should return:

{'Y': [[[41, 4.8754486973], [13, 5.0477717856], [8, 5.0774009099], [38, 5.1312766443], [60, 5.2201532545]]]}

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_application_neighbors_image(app=None, options=None, **params)

Defines a dummy application using this API. It returns a list of one neighbor for an image and metadata (random).

Parameters
  • app – application, if None, creates one

  • options – if not empty, path to an image

  • params – parameters sent to MachineLearningPost

Returns

app

You can start it by running:

start_mlrestapi --name=dummyknnimg

And then query it with:

import requests
import ujson
from lightmlrestapi.args import image2base64
img = "path_to_image"
b64 = image2base64(img)[1]
features = ujson.dumps({'X': b64}, reject_bytes=False)
r = requests.post('http://127.0.0.1:8081', data=features)
print(r)
print(r.json())

It should return:

{'Y': [[[41, 4.8754486973, {'name': 'wiki.png', description='something'}]]]}

source on GitHub

lightmlrestapi.testing.dummy_applications.dummy_mlstorage(app=None, **params)

Defines a dummy application using this API. It stores a model and it returns a score produced by a model trained on Iris datasets and two features.

Parameters
  • app – application, if None, creates one

  • params – parameters sent to MLStoragePost

Returns

app

source on GitHub