contentaiextractor 1.1.0

makes it easier to implement a ContentAI extractor

contentai-extractor-runtime-python

This is a python package used for implementing a custom extractor that runs on the ContentAI platform.

https://pypi.org/project/contentaiextractor/

1. Usage
2. API Documentation
3. Dependencies
4. Develop
5. Changes

Usage

pip install contentaiextractor

import contentaiextractor as contentai

# download content locally
content_path = contentai.download_content()

# access metadata that was supplied when running a job
# contentai run s3://bucket/video.mp4 -d '{ "input": "value" }'
inputData = contentai.metadata()["input"]

# get output from another extractor
csv = contentai.get("extractor", "data.csv")
json = contentai.get_json("extractor", "data.json")

# extract some data
outputData = []
outputData.append({"frameNumber": 1})

# output data from this extractor
contentai.set("output", outputData)

API Documentation

ContentAIError Objects

class ContentAIError(Exception)

represents a contentai error

Fields

• extractor_name - name of the extractor being run
• job_id - current job id
• content_url - URL of the content the extractor is run against
• content_path - local path where the extractor can access the content
• result_path - local path where the extractor should write the results
• running_in_contentai - boolean set to True; useful for testing code locally
• metadata_json - raw string (or None if not set) for active extractor run (also, see parsed metadata())

Functions

download_content

download_content()

download content to work with locally

returns local path where content is written

metadata

metadata()

returns a dict containing input metadata

example:

access metadata that was supplied when running a job

contentai run s3://bucket/video.mp4 -d '{ "input: "value" }'

input = contentai.metadata()["input"]

extractors

extractors()

get list of all extractors executed against this content url

returns a list of strings

[
    "extractor1",
    "extractor2"
],

example:

# get all data from all extractors
for extractor in contentai.extractors():
    for key in contentai.keys(extractor):
        data = contentai.get(extractor, key)

keys

keys(extractor_name)

get a list of keys for specified extractor

returns a dict containing a list of keys

[
  "data.json",
  "data.csv",
  "data.txt,"
]

example:

keys = contentai.keys("azure_videoindexer")
for key in keys:
    data = contentai.get("azure_videoindexer", key)

get

get(extractor_name, key)

get the contents of a particular key

example:

# get another extractor's output
data = contentai.get("some_extractor", "output.csv")

get_json

get_json(extractor_name, key)

get the json contents of a particular key

example:

# get another extractor's output
data = contentai.get_json("some_extractor", "data.json")

get_bytes

get_bytes(extractor_name, key)

get the contents of a particular key in raw bytes

example:

# get another extractor's output
data = contentai.get_bytes("some_extractor", "output.bin")

set

set(key, value)

set results data for this extractor

can be called multiple times with different keys

value is a string

example:

contentai.set("output", "hello world")

set_json

set_json(key, value)

set results data for this extractor

can be called multiple times with different keys

value can be anything

example:

data = {}
data["foo"] = bar
contentai.set_json("output", data)

set_bytes

set_bytes(key, value)

set results data for this extractor

can be called multiple times with different keys

value is bytes

example:

some_file = open("some-file", "rb")
contentai.set_bytes("output", some_file.read())

save_results

save_results()

save results immediately, instead of waiting until process exits

parse_content_url

parse_content_url()

extract details from content url

returns

• source_bucket_name - the s3 bucket name derived from content_url
• source_bucket_key - the s3 bucket key derived from content_url
• source_bucket_region - the s3 bucket region derived from content_url

the following content url formats are supported:

• Simple (CLI) Format - s3://{bucket}/{key}
• Virtual Hosted Format - https://{bucket}.s3.amazonaws.com/{key}
• Virtual Hosted Format with Region - https://{bucket}.s3.{region}.amazonaws.com/{key}

Dependencies

pip install -r requirements.txt

Develop

 Choose a make command to run

  build    build package
  deploy   upload package to pypi
  docs     generates api docs in markdown

Release

To publish a new release to pypi, increment the version number in setup.py, tag the commit and push it.

Changes

• 1.1.0

  • add extractors()

• 1.0.4

  • updated changelog

• 1.0.3

  • fixes issue where EXTRACTOR_METADATA envvar was indavertently required

• 1.0.2

  • add safety to setting retrieval on local runs
  • documentation updates

• 1.0.1

  • api docs for publish to pypi

• 1.0.0

  • initial release