[cfe-dev] accessing template arguments from libclang

Geoffrey Irving irving at naml.us
Wed Oct 30 15:58:05 PDT 2013

Another ping for this message: let me know if this is the wrong forum
for questions about libclang.


On Thu, Aug 15, 2013 at 10:36 AM, Geoffrey Irving <irving at naml.us> wrote:
> In case the previous email was too long, here's a quick summary:
> I am trying to parse the code
>     class Toaster : Base<int,7> { ... };
> I can get to the base specifier corresponding to Base<int,7> and from there
> to a RECORD type, but there appears to be no way to access the template
> arguments of a template reference from either the Python or C bindings to
> clang.  If I'm correct that this isn't exposed yet, what's the right way to
> do it?
> Thanks,
> Geoffrey
> On Tue, Aug 13, 2013 at 12:26 PM, Geoffrey Irving <irving at naml.us> wrote:
>> I am trying to parse the following code with the Python interface to
>> libclang (full listing attached):
>>     class Toaster : Base<int,7> { ... };
>> Unfortunately, traversing cursors via getChildren gives me the following
>> tree (see clang_dump() in autodescribe.py for the code that generated this
>> tree):
>> CLASS_DECL Toaster
>>   CXX_BASE_SPECIFIER None : Base<int,7>
>>     TEMPLATE_REF None : Base
>>     INTEGER_LITERAL None : 7
>>   ...
>> The base specifier only seems to mention "Base" and "7", not "int".
>> Should I be able to fully inspect Base<int,7> using cursors?
>> I also tried traversing the Type object corresponding to Base<int,7>, but
>> I don't see any methods for accessing template parameters either in
>> libclang's Python bindings or in the original C.  If I'm correct that the
>> functionality is missing from the C layer, what's the easiest way to add it?
>> The C++ structure of template references seems quite complicated (I imagine
>> by necessity :)).
>> Background: I'm working on a clang-based front end for the xdress Python
>> binding generator project.  I've already made a couple minor improvements to
>> the libclang Python bindings (exposing formatDiagnostic and the details of
>> access specifiers), and am happy to contribute these back once I'm done.
>> Thanks for any suggestions!
>> Geoffrey

More information about the cfe-dev mailing list