SigaaParser
Pet-project. Use at your own risk. Low test coverage ratio.
Gem to scrap and parse the academic web system (SIGAA) used at UFPB.
Powers the API: https://github.com/fernandobrito/ufpb_sigaa_api
Features
Authenticates with given username and password
To access data from the system, we first need to log in. If the student has more than one enrollment ID, the most recent one is selected.
Parses the curricular structure of a bachelor program
Given the ID of a bachelor program, it is possible to parse all mandatory course names that make this bachelor program, including its dependencies (prerequisites).
Parsers a Transcript of Record (PDF)
Given the Transcript of Records PDF file generated from SIGAA, returns the parsed data (courses completed and grades).
Caches HTML responses
To make development faster, course and bachelor program pages are cached under
the cache folder.
Installation
Add this line to your application's Gemfile:
gem 'sigaa_parser'And then execute:
$ bundle
Or install it yourself as:
$ gem install sigaa_parser
Usage
Check the examples folder.
Update .env file with your UFPB SIGAA username and password.
Development
To run all the tests: $ rspec
To exclude tests that makes real requests to SIGAA (and require authentication): $ rake specs:ci. Travis-CI runs this test suit.
To-do
- Parse program by name (instead of ID);
- Parse grades and completed courses by student (either by login and password, or transcript PDF file);
- Add validations after every request to see if we are on right page.
Contributing
History
At first we tried using Mechanize gem to automate the HTTP requests. However, it does not support
JavaScript. As SIGAA is developed using JSF, the requests to work, we had to keep track of javax.faces.ViewState ID,
which is an integer returned on every response. This number had to be sent back on every request, and sometimes
the server would simply ignore my number and reset my session.
Afterwards we tried using capybara with poltergeist, but as capybara is mainly designed to be
used on acceptance tests, it was hard to make use of the lib outside that context.
Finally, we moved to Watir with PhantomJS driver. Slow, but works fine and code is much more
readable.