RbGCCXML allows one to easily parse out and query C++ code. This library uses GCC-XML to parse out the C++ code into XML, and then nokogiri for parsing and querying.
GCC-XML (www.gccxml.org) is an application that takes takes the parse tree of g++ and constructs a very parsable and queryable XML file with all related information. GCC-XML currently only works with code declarations; there currently are no plans to support code bodies.
Note: For those familiar with pygccxml, the similarities are minimal. Outside of the purpose of both libraries, rbgccxml was built from scratch to provide a Ruby-esque query API instead of being a port. However, many thanks to Roman for his work, without which this library would also not exist.
nokogiri gccxml_gem (binary gem, choose proper platform)
gem install rbgccxml
RbGCCXML will work on all platforms that GCC-XML supports, which includes Windows, Mac, and *nix. If you're on a platform for which there isn't a current gccxml_gem build, please open an Issue.
RbGCCXML’s source is in a git repository hosted on github:
git clone git://github.com/jasonroelofs/rbgccxml.git
All rbgccxml parses start with the RbGCCXML.parse method:
# Parse a single header file RbGCCXML.parse("/path/to/header/file.h") # Parse out all files that match a given glob RbGCCXML.parse("/my/headers /*.h") # Parse out a specified set of files" RbGCCXML.parse(["/path/to/file1.h", "/path/to/file2.h", ...]) # Parse out multiple globs RbGCCXML.parse(["/my/headers/ *.h", "/other/headers/*.hpp"])
As GCC-XML runs on top of GCC, it will need to know about locations of other header files that may be included by the header files being parsed out. Adding these paths is simple:
RbGCCXML.parse(..., :includes => *directories)
directories can be a single directory string or an array
of directories, just like RbGCCXML.parse.
Also, if there are other CXXFLAGS that need to be added to the command line for GCC-XML to properly parse the source headers (say, -D defines), add those via the :cxxflags option.
RbGCCXML.parse(..., :cxxflags => *flags)
Once the header files have been parsed, RbGCCXML.parse returns a Namespace node that references the global namespace “::”. From here, all the function, class, etc declarations are easily queryable.
source = RbGCCXML.parse("header.h") #=> <Namespace ...>
Each major C++ node (class, struct, function, method, argument (of functions, methods, and constructors)) have a related query method that can be called in various ways:
# Get all classes in the current scope source.classes # Explicitly call #find on these classes source.classes.find(...) # Or use a short cut, if you're just looking for a class by a given name source.classes("ClassName") # Find supports regular expressions as well source.classes(%rManager$/)
These queries are also nestable as long as there’s more code available to query. To find the class “Math” inside the namespace “core::utils”, you can do:
or in a short-hand / C++ qualified form:
source.namespaces("core::utils").classes("Math") # or source.classes("core::utils::Math")
Of course, querying for names is only the tip of the powerful querying that RbGCCXML supports.
What if you want to find all methods on a class that return an int and have three arguments of any type? This is easy with RbGCCXML:
source.classes("TheClass").methods.find(:returns => :int, :arguments => [nil, nil, nil])
arguments can be used on
their own or together as seen above. The
must be an array, and if the type of the argument doesn’t matter, placing
‘nil’ in it’s place acts as a wildcard. QueryResult#find is also chainable,
provided that there are always more than one result. Otherwise, if there is
just one result, only that Node will be returned and any further chained
QueryResult.find methods will fail with NoMethodError.
See RbGCCXML::Node for all methods available on all C++ nodes.
See RbGCCXML::QueryResult for the full run-down on find
Querying for unsigned types is currently not implemented.