All of the test input files, some with their own config file, are in the `input` directory. These tests can be run using the various cmd/bat and sh script files supplied.
In essence it is an attempt to automate some regression testing. The idea is that after making a code change to tidy, the alltest.cmd can be run using the new executable. This would produce an output in the `temp-5` folder.
Comparing the `testbase` folder with the new `temp-5` folder will show you what file output was changed by your code modification, if any. In WIN32 there should be none.
If unix the `$ diff -ua testbase tmp` will normally yield 3 changes, tests 431895, 500236 and 616606. 431895 is because it uses the `gnu-emacs: yes` option and we can thus expect the path separator in the file names to change.
The other two 500236 and 616606 just seem to have some spaces changes. Not sure exactly why. If `-w` or `-b` option is use there should be no difference. So these 3 tests must be especially checked.
Difficult, and tedious! Yes, but is a sure way to see if your changes adversely effected tidy. Unfortunately, only such a visual comparison would show the results. If the output changes are fully acceptable, like a warning message changed, then this should become the new base file for that test.
Of course some of the tests were to say avoid a segfault found. Other tests were to visually compare the original input test file in a browser, with how the new output displayed in a browser. This is a purely VISUAL compare, and can not be done in code.
And what about if there was NO current test existing to test what you were trying to fix. Well that means a NEW test should be added. Its output added to the 'base', then there would be a comparison.
The chain is alltest.cmd runs alltest1.cmd, which then uses onetest.cmd for each test. And a similar chain for unix stating with testall.sh which uses testone.sh.
Both tool chains use the testcases.txt file for the list of some 227 cases. There is now a `testinfo.txt` file which give some desciption of the test. Well actually the title of the original bug report filed at https://sourceforge.net/p/tidy/bugs/#number/
There is a similar unix tool chain of testaccess.sh, which uses testaccessone.sh for each test, from the same accesscases.txt list.
#### Files: alltestc.bat
This is essentially similar to the above, except it includes a COMPARE of the previous established output in the `testbase` folder with the NEW output in temp-5, hence the addition of a 'c'.
As indicated above, this is an attempt to create such a BASE set of output files when tidy is run on the input test cases.
So the difference between this and the above is, it further immediately compares the output of tidy, if there is one, with the equivalent file in the `testbase` directory using a windows port of diff.
It is a success if there is **NO** diff! A difference means this newer version of tidy has modified the output. That modification needs to be carefully inspected, and if it is thought exact and suitable, then that new output should be copied to the `testbase` folder for future compares.
