dcn01 / link-extractor

Simple producer/consumer web link extractor

Simple producer/consumer web link extractor

Installation

Needed:

• Python 3.5
• Redis
• Supervisord

Preferably under virtualenv:

pip install pip-tools (once)

pip-sync requirements*.txt (keeping the PyPI dependencies up-to-date)

Configuration

Customized settings are expected in extractor/settings_local.py. (But they shouldn't be needed.)

Usage

supervisord starts Supervisord with background services (Redis, Celery workers – "consumers"). They can be controlled by supervisorctl then. Logs are stored in the log directory.

./app.py is the "producer", expecting list of URLs on the standard input. They are parsed, so you can use HTML as input: ./app.py < index.html.

Each URL from input is processed by consumers in a way that the referenced webpage is downloaded and parsed for absolute URLs, which are then saved in a JSON file in the out directory. The output file name is an MD5 hash of the input URL.

Example

$ supervisord  # if not already done before
$ ./app.py
http://example.com
Ctrl+D
$ jq < out/a9b9f04336ce0181a08e774e01113b31.json
{
  "url": "http://example.com",
  "links": [
    "http://www.iana.org/domains/example"
  ],
  "version": "0.1.0"
}

Testing

./test.sh (also generates a coverage)