[cfe-dev] How to get the string from the CFGStmt

[Manuel Klimek]

On Fri, Feb 17, 2012 at 10:05 AM, redder_0210 <redder_0210 at yeah.net> wrote:
> Thanks for you replying.
> The new question is that I cannot get the SourceManager only by the Stmt. So
> the "SM.getSpellingLoc(Range.getBegin())" cannot work.
> Thus, I want to how to get SourceManager from the Stmt or from the
> SourceRange.

You can't. You need to get the SourceManager from the context. I could
help you more if you give a little more context on how you're hooking
into clang.

Cheers,
/Manuel

> Hope Best!
> /Redder
>
> At 2012-02-13 17:25:26,"Manuel Klimek" <klimek at google.com> wrote:
>>On Mon, Feb 13, 2012 at 5:48 AM, redder_0210 <redder_0210 at yeah.net> wrote:
>>> Dear all,
>>> I am interested in that how to translate from a statemnt to the
>>> corresponding string.
>>> For example,
>>> in the CFG, there is a CFGStmt "int i = 0;",
>>> now I want to get the string "int i = 0;",namely modify the type CFGStmt
>>> to
>>> the type string.
>>
>>CFGStmt has a getStmt() which gets you a Stmt, which has a function
>>getSourceRange which gives you the SourceRange.
>>When you have a SourceRange you can get a CharSourceRange by calling
>>CharSourceRange::getTokenRange(...).
>>Then you get the start location by calling
>>SM.getSpellingLoc(Range.getBegin()) (where Range is the
>>CharSourceRange).
>>You can calculate the length of the text by calling getDecomposedLoc
>>on the getBegin() and getEnd() of the spelling locations. You'll also
>>need to add Lexer::MeasureTokenLength(EndLocation, ...) to the size,
>>as the last element of a token range points to the begin of the last
>>token.
>>Finally you can call SM.getCharacterData with the begin and you'll
>>have your text.
>>
>>Unfortunately this is a little harder than it could be, mostly due to
>>the different types of locations. (Note that if there are for example
>>macros involved, it matters whether you use getSpellingLocation or
>>not).
>>
>>Cheers,
>>/Manuel
>
>