Upgrade test engine about typecobol

[osmedile]

In order to ease tests writing, it should be useful to have only one file to describe tests cases.
In this file you could write the statements to test, if this statement is expected to be parsed correctly and then the expected processed statement.
Ex:
Ok: add a to b -> add a to b *> the test statement and processed statement are identical
Ok: add a to b *> same test as above. In this case the expected statement will be equals to the statement to test
Ko: add to b *> Parsing error expected
OK: add toto titi tata to X -> add toto titi tata to X *> in the processed statement all redundant spaces are trimmed

This kind of test could be written by Cobol developper without knowledge of the TypeCobol parser.

[wiztigers]

I think the form given to a specific result file could be given by its name.
For example, corresponding to source cobol file named ADD.cbl (put in Samples\Statements folder), could be (in ResultFiles\Statements folder) :

• ADDCodeElements.txt, containing the CodeElement.ToString() calls as we know it
• ADDRPN.txt, containing the conversion to Reverse Polish Notation strings
• ADDNY.txt, containing just "Y" or "N" depending of the correct (or incorrect) parsing of each line
• whatever else we think of

For a given cobol file, the test engine could find all files beginning with ADD, deciding wich comparison happens by looking at the suffix, and defaulting to CodeElement comparison if it only finds ADD.txt as a result file.
If no result file is to be found, an error would be thrown.

This solution is more flexible that what we have (even considering today's improvements), but would prevent having for example the files ADD.cbl and ADDErrors.cbl in the same folder. We should have ADDSuccesses.cbl and ADDErrors.cbl instead.

[wiztigers]

Since a few months, test engine automatically test files in subfolders of TypeCobol.Test/Compiler/Parser/Samples and compare results with expected result files in subfolders of TypeCobol.Test/Compiler/Parser.
Different comparators can be used, depending of the name of a given result file.
You can create more than one expected result file for a given sample file if you want to run different comparators for the same sample.

Let's say it's enough to close this issue.
Create other issue if additional needs arise.