Purpose: this code snippet illustrates one of our test cases for determining the correctness of the Tools in libphoto. This snippet tests the Blur Tool and compares it to a ‘correct’ result from iteration two.
//make a new blur tool and buffer to modify
Tool * blur_tool = new BlurTool();
PixelBuffer * test_buffer = new PixelBuffer(800, 800, ColorData(1,1,0.95));
//load the picture to be blurred into the buffer
ImageFileHandler::loadImageIntoBuffer("blur_tool_before_test.png", test_buffer);
//blur specific parts of the buffer
blur_tool->apply(100,200,test_buffer);
blur_tool->apply(300,300,test_buffer);
blur_tool->apply(100,299,test_buffer);
blur_tool->apply(180,270,test_buffer);
blur_tool->apply(130,260,test_buffer);
blur_tool->apply(345,400,test_buffer);
blur_tool->apply(400,400,test_buffer);
for(int i = 100; i < 400; i++)
{
blur_tool->apply(i,i,test_buffer);
}
for(int i = 50; i < 300; i++)
{
blur_tool->apply(i,100,test_buffer);
blur_tool->apply(100,i,test_buffer);
}
//import the 'correct' buffer and image from iteration 2
PixelBuffer * ground_buffer = new PixelBuffer(800, 800, ColorData(1,1,0.95));
ImageFileHandler::loadImageIntoBuffer("blur_tool_test.png", ground_buffer);
//see if they're the same
TSM_ASSERT("Failed blur tool test", test_buffer->isEqualTo(ground_buffer));
How it works: this code is relatively straightforward, however it illustrates several characteristics of of our program design and style. These can be seen as we walk through the code.
First, I create a pointer to a new Blur Tool, called blur_tool. This is in-line with our naming conventions of using all lowercase letters and separating words with underscores. Next, I create a pointer to a new PixelBuffer called test_buffer, which I load an image into.
This illustrates the use of our image handling class, ImageFileHandler. It is a static class to use for loading and saving png and jpg images. In this case I am loading an image called “blur_tool_before_test.png” into the test_buffer with a call to ImageFileHandler::loadImageIntoBuffer. This image is of a cartoon bowl of fruit to be blurred.
I then call blur_tool’s apply function to many different parts of test_buffer. First, I do it on certain select locations, and then iterate over many parts in a line, which will test many repeated calls to apply the tool. Note that not every line is commented - just the ones which help to describe what I am trying to do with the code. This is because we want to let most of the code speak for itself, as to avoid over-commenting.
Lastly, it is time to compare the changes made to test_buffer to a buffer which has had the same modifications made to it in our code for iteration two, which was visually inspected and deemed as correct. In order to do this, I simply create a new buffer called ground_buffer (for ‘ground truth’) and load the saved image from iteration two into it. Then I use the TSM_ASSERT function from the cxxtest library to assert that if I compared the two buffers I am left with, using a function from PixelBuffer to compare them, it would return a value of true, confirming the (at least partial) correctness of the Blur Tool.
As is evident from the code, this snippet provides a good example of our variable naming strategies, code documentation practices, brevity and use of not only our own library functions and classes but also the third party tool, cxxtest.