Extract Comments in C/C++ Source File

extract-comments.py

#!/usr/bin/python3

from clang.cindex import Index, Config, CursorKind
import json as JsonUtil
import os, io, sys



class MyTreeWalker:
    def __init__(self):
        self.namespace = []

    def walkThroughClass(self, cursor, visitor):
        if cursor.kind == CursorKind.CXX_METHOD:
            name = "::".join(self.namespace + [ cursor.spelling ])
            visitor.visitClassFunction(name, cursor)
        elif cursor.kind in ( CursorKind.STRUCT_DECL, CursorKind.CLASS_DECL, CursorKind.CLASS_TEMPLATE ):
            self.namespace += [ cursor.spelling ]

            name = "::".join(self.namespace + [ cursor.spelling ])
            visitor.visitClassDefinition(name, cursor)

            for child in cursor.get_children():
                self.walkThroughClass(child, visitor)

            self.namespace.pop()

    def walkThrough(self, cursor, visitor):
        if cursor.kind == CursorKind.NAMESPACE:
            self.namespace += [ cursor.spelling ]

            for child in cursor.get_children():
                self.walkThrough(child, visitor)
            
            self.namespace.pop()
        elif cursor.kind in ( CursorKind.STRUCT_DECL, CursorKind.CLASS_DECL, CursorKind.CLASS_TEMPLATE ):
            self.namespace += [ cursor.spelling ]

            name = "::".join(self.namespace + [ cursor.spelling ])
            visitor.visitClassDefinition(name, cursor)

            for child in cursor.get_children():
                self.walkThroughClass(child, visitor)

            self.namespace.pop()
        elif cursor.kind == CursorKind.FUNCTION_DECL:
            name = "::".join(self.namespace + [ cursor.spelling ])
            visitor.visitGlobalFunction(name, cursor)

    def startWalk(self, cursor, visitor):  
        if cursor.kind == CursorKind.TRANSLATION_UNIT:
            for child in cursor.get_children():
                self.walkThrough(child, visitor)



class JsonGeneratingVisitor:
    def __init__(self, stream: io.TextIOBase):
        self.stream = stream
        self.definitions = []

    def visitGlobalFunction(self, name: str, cursor):
        comment = cursor.raw_comment
        mangledName = cursor.mangled_name
        
        if mangledName and comment:
            self.definitions.append({ "name": mangledName, "comment": comment })

    def visitClassDefinition(self, name, cursor):
        pass
    
    def visitClassFunction(self, name, cursor):
        comment = cursor.raw_comment
        mangledName = cursor.mangled_name

        if mangledName and comment:
            self.definitions.append({ "name": mangledName, "comment": comment })

    def flush(self):
        JsonSerialized = JsonUtil.dumps(self.definitions, ensure_ascii=False, indent=2)
        self.stream.write(JsonSerialized)



class Executor:
    def __init__(self, walker, generator):
        self.walker = walker
        self.generator = generator

        self.libClangPath = os.environ.get("EMSDK") + "/upstream/lib/"
        self.path = None
        self.args = None

    def run(self):
        Config.set_library_path(self.libClangPath)

        index = Index.create()
        tree = index.parse(self.path, args=self.args)

        self.walker.startWalk(tree.cursor, self.generator)
        self.generator.flush()



def main(argv):
    path = argv[1]
    args = argv[2:]

    executor = Executor(MyTreeWalker(), JsonGeneratingVisitor(sys.stdout))
    executor.path = path
    executor.args = args
    executor.run()

if __name__ == "__main__":
    main(sys.argv)