Wiki page: http://wiki.smc.org.in/Automated_rendering_testing
A framework to test the correctness of output by rendering engines.
Files required to test rendering using this framework:
(Assuming the user has harfbuzz installed. If not build it from: https://github.com/behdad/harfbuzz)
Files required to test rendering using this framework:
Test cases file Reference file for a specific font File with rendering outputs by engines Font file in ttf format Create a test cases file that consists of all the words that you wish to test the rendering for. Here is a sample test cases file created for Malayalam lamguage: https://github.com/nandajavarma/Automated-Rendering-Testing/blob/master/ml-test-data/ml-test-cases.txt
Along with this create the reference file that contains the correct glyph names of the words in the test cases file in a particular font. The framework assumes that the glyph names are in the following format: [glyph_name1,glyph_name2,glyph_name3,..,glyph_nameN]
Now if the word has more than one correct rendering, provide the next correct one along with this seperated by a semi colon. For eg: [glyph_name1,glyph_name2,glyph_name3,..,glyph_nameN];[glyph_name1,glyph_name2...,glyph_nameN];.. Here is the reference file for the above mentioned test cases file in the font Rachana: https://github.com/nandajavarma/Automated-Rendering-Testing/blob/master/ml-test-data/lohit-ml-test-data/lohit-glyph.txt
Now the file with rendering outputs. If the engine you are testing for is Harfbuzz, you can create this file using the following command:
cat ml-test-data.txt | hb-shape /path/to/Font.ttf > output.txt
If that is not the case, you will have to create it for the font you wish and the rendering of each word must be in the form: [glyph_name1|glyph_name2|..] Here is the harfbuzz rendering of the above mentioned test cases file in font Rachana: https://github.com/nandajavarma/Automated-Rendering-Testing/blob/master/ml-test-data/lohit-ml-test-data/hb_lohit_rendering.txt
Now that you have all the necessary files, write these data to an .ini file for the main script to read. Here is the structure: [main] Reference-file: Rendered-output: Font-file: Test-cases-file: Output-file: Shaping-engine: Out of these, Reference-file and Rendered-output are mandatory. Comment out all the other lines, if a result file is not necessary. If the shaping-engine is harfbuzz, then the output file will also have harfbuzz rendered image of every word. These images will be stored inside a directory names 'hb_images'. Here is a sample: https://github.com/nandajavarma/Automated-Rendering-Testing/blob/master/ml-test-data/lohit-ml-test-data/Lohit-ml.ini
Now to test, run the script rendering_test.py passing the name of the .ini file as a parameter. For example:
./rendering_test.py Lohit-ml.ini
(In the repo one can find samples in four Malayalam fonts and one Devanagari font. Test cases file for Malayalam being https://github.com/nandajavarma/Automated-Rendering-Testing/blob/master/ml-test-data/ml-test-cases.txt and that for Devanagari being https://github.com/nandajavarma/Automated-Rendering-Testing/blob/master/devanagari-test-data/devanagari_test_cases.txt.)