[cfe-dev] check whether a `case` statement has a `break` statement?

[kevinlynx]

I only want to know if there's a break between two cases (or between a case and default). I dump a CastStmt AST, but not found a BreakStmt. That's why i asked for help here. 




------------------ Original ------------------
From:  "Alexander Kornienko"<alexfh at google.com>;
Date:  Wed, Jan 2, 2013 10:17 AM
To:  "kevinlynx"<kevinlynx at gmail.com>; 
Cc:  "cfe-dev"<cfe-dev at cs.uiuc.edu>; 
Subject:  Re: [cfe-dev] check whether a `case` statement has a `break` statement?



I guess you're trying to do something similar to -Wimplicit-fallthrough diagnostics. If this is the case, take a look at http://llvm.org/svn/llvm-project/cfe/trunk/lib/Sema/AnalysisBasedWarnings.cpp, DiagnoseSwitchLabelsFallthrough function. In short: you can use CFG (control-flow graph) to reason about execution paths, and, for example, detect if there's an execution path between two case labels.
 

If you need something simpler, and just detect if there's any 'break' somewhere between two case labels (I'm not sure I understand why you would need this, though), you can just use RecursiveASTVisitor and handle SwitchStmt, SwitchCase (which can be either CaseStmt or DefaultStmt) and BreakStmt.
 
On Tue, Jan 1, 2013 at 3:30 PM, kevinlynx <kevinlynx at gmail.com> wrote:
 In my parser, I get a `SwitchStmt`, and I can traverse all `case` statements of it. I want to check whether a `case` statement has a `break` statement.. But i can not figure it out. Here is my codes:

    for (SwitchCase *c = stmt->getSwitchCaseList(); c != NULL; c = c->getNextSwitchCase()) {
             if (isa<CaseStmt>(c)) {
                CaseStmt *caseStmt = cast<CaseStmt>(c)
                // I do not know how to check `CaseStmt` has a `BreakStmt`
            }
    }

Thanks.
 
BTW, I'm writing a static c code analyzer recently, this tool will check a number of c coding style rules for c codes. And because I'm a newbie to clang, so there will be more newbie questions coming, i'm sorry if i bother you guys. 
 




-- 
Regards,
Alexander Kornienko
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20130102/196a09db/attachment.html>