brozzler - distributed browser-based web crawler
Find a file
2014-05-30 23:07:39 -07:00
bin fancy --version that includes git branch and timestamp of last commit if available 2014-05-29 20:43:00 -07:00
umbra dump stack trace and don't return browser to pool on critical error where chrome process might still be running 2014-05-30 23:07:39 -07:00
.gitignore not sure why /bin/ et al were in .gitignore... replace with a couple of useful things 2014-05-20 17:06:26 -07:00
README.md improve readme, mentioning archive-it per kristine 2014-05-23 13:34:51 -07:00
setup.py for the version string, use abbreviated commit hash instead of attempting to use the branch name 2014-05-29 23:33:14 -07:00

umbra

Umbra is a browser automation tool, developed for the web archiving service https://archive-it.org/.

Umbra receives urls via AMQP. It opens them in the chrome or chromium browser, with which it communicates using the chrome remote debug protocol (see https://developer.chrome.com/devtools/docs/debugger-protocol). It runs javascript behaviors to simulate user interaction with the page. It publishes information about the the urls requested by the browser back to AMQP. The format of the incoming and outgoing AMQP messages is described in pydoc umbra.controller.

Umbra can be used with the Heritrix web crawler, using these heritrix modules:

Install

Install via pip from this repo, e.g.

pip install git+https://github.com/internetarchive/umbra.git

Umbra requires an AMQP messaging service like RabbitMQ. On Ubuntu, sudo apt-get install rabbitmq-server will install and start RabbitMQ at amqp://guest:guest@localhost:5672/%2f, which the default AMQP url for umbra.

Run

The command umbra will start umbra with default configuration. umbra --help describes all command line options.

Umbra also comes with these utilities:

  • browse-url - open urls in chrome/chromium and run behaviors (without involving AMQP)
  • queue-url - send url to umbra via AMQP
  • drain-queue - consume messages from AMQP queue