Incorrect Handling of ".include" in Identifiers in Souffle Parser

[Mingtao521]

I noticed a issue in the Souffle parser when handling ".include" within identifiers. Here is an example illustrating the problem:

.decl test(b: symbol)
test(b) :- A.includeB(b).
.output test(IO=stdout)

.comp A_comp {
    .decl includeB(b: symbol)
    includeB("B").
}
.init A = A_comp

Compiling this Datalog program with Souffle results in the following errors:

Error: syntax error, unexpected invalid token in file test.dl at line 2
test(b) :- A.includeB(b).
--------------------^-----
Error: unexpected B in file test.dl at line 2
test(b) :- A.includeB(b).
--------------------^-----
2 errors generated, evaluation aborted

I suspect the issue lies within Souffle's lexer, which incorrectly recognizes ".include" as a keyword or directive, rather than part of an identifier. Relevant code can be found in src/parser/scanner.ll on lines 156-159:

".include"                            {
                                        yyinfo.LastIncludeDirectiveLoc = yylloc;
                                        BEGIN(INCLUDE);
                                      }

I think the correct handling of ".include" should be similar to other directives like ".decl" or ".functor," where it should only be treated as a directive declaration when followed by one or more spaces.

To test my hypothesis, I modified line 156 in scanner.ll to:

".include"/{WS}                       {

Compiling the above Datalog program with modified Souffle produces the following output:

---------------
test
===============
B
===============

If this issue is really a bug, I am happy to submit a pull request to fix it.

[quentin]

Hi, thanks for reporting this. Would you like to submit a fix ?

[Mingtao521]

Sure! I have submitted a PR.