Automated Testing (Unit Tests) - pgRouting/pgrouting GitHub Wiki

CAUTION

This page has not been implemented. Currently the only test infrastructure that we have implemented and are using is the testing documented in 2.0-Development-Guidelines-and-Standards. All new functionality and regressions need to have test cases created to validate the fixes and or new code.

Goals

To provide unit tests for the core functions(c++) and also to the wrapper functions(sql) to make it easier to check for bugs before committing changes.

Testing frameworks/libraries used

CUnit Test framework

CUnit provides a framework for running tests against c/c++ code
It is used by PostGIS to test its geometry library(liblwgeom)
It is used to test core functions implementing routing algorithms and has the advantage that future integration with PostGIS will be a bit easier
[http://cunit.sourceforge.net/ Link to project page]

Python unittest

unittest is an easy to use python testing framework
It is used to test pl/pgsql functions using psycopg2 driver for PostgreSQL
[http://docs.python.org/library/unittest.html Link to unit test library]

CTest(CMake Testing System

CTest is provided by CMake, the build system used by pgrouting
Ties down the testing code into CMake configuration
CTest is used to invoke the above two frameworks to test the c/c++ code as well as the core and wrapper SQL functions
[http://www.itk.org/Wiki/CTest:FAQ Link to CTest wikipage]

Code

The code is under active development in a separate branch: [https://github.com/justjkk/pgrouting/commits/unittests]

Progress

C/C++ Functions testing

Tests are written for validating boost wrapper of dijkstra implementation
Have to write tests for other routing algorithm implementations like Astar, Shooting star, etc...

SQL functions testing

The basic framework for creating test database, loading just compiled SQL functions, creating and populating test tables are done
Very primitive testing for Dijkstra(proof of concept) is implemented
Have to write tests for all core and wrapper SQL functions with diverse test data

Writing New tests

SQL Testing

Add necessary sql files inside SetUpDB function in tests/init.py
Create ''yourdata''.sql and ''yourdata''.csv files under tests/loaders ( use simplegraph.csv and simplegraph.sql for the format )
Create ''yourtest''.py in tests/ using tests/test_dijkstra.py as reference
Use loadTable() inside SetUp function of ''yourtest'' class and pass ''yourdata'' as argument to load ''yourdata''.sql and ''yourdata''.csv during testing
Add ''yourtest''.py inside tests/CMakeLists.txt
Make sure you have a tests/settings.py copied from tests/settings.py.sample and filled with actual values
cmake; make; make test to test the application
Or run ''yourtest''.py to run your test alone