transform dalvik format (androguard) to soot format (flowdroid)

.py

from androguard import misc
"""
The info of methods given by androguard has a different format between that given by Soot. 
To facilitate the FlowDroid tool, I implemented the tool to transform the format.

The script works on Androguard 3.0-dev.

The main function is getFuncNames(), we use androguard API AnalyzeAPK to traverse all functions in all classes
and transform them into soot formats.
"""
def classname_handler(class_name):
    if class_name[0] == 'L':
        class_name = class_name[1:]
    elif class_name[0] == '[' and class_name[1] == 'L':
        class_name = class_name[2:]
    class_name = class_name.replace('/', '.')
    class_name = class_name[:-1]  # remove ';'
    return class_name

def basictype_handler(typ):
    res = None
    if typ == 'Z':
        res = 'boolean'
    elif typ == 'B':
        res = 'byte'
    elif typ == 'S':
        res = 'short'
    elif typ == 'C':
        res = 'char'
    elif typ == 'I':
        res = 'int'
    elif typ == 'J':
        res = 'long'
    elif typ == 'F':
        res = 'float'
    elif typ == 'D':
        res = 'double'
    elif typ == 'V':
        res = 'void'
    return res

def type_handler(typ):
    if len(typ) == 1:
        return basictype_handler(typ)
    elif len(typ) == 2:
        return basictype_handler(typ[1:])+'[]'
    else:
        if typ[0] == '[':
            return classname_handler(typ[1:])+'[]'
        else:
            return classname_handler(typ)

def descriptor_handler(descriptor):
    #(Ljava/lang/String; Ljava/lang/String; Ljavax/naming/directory/SearchControls;)Ljavax/naming/NamingEnumeration;
    soot_param = '('
    tmp = descriptor.split(')')
    # handle parameters
    param = tmp[0][1:] # remove '('
    if len(param)==0:
        soot_param += ')'
    else:
        params = param.split(' ')
        for p in params:
            soot_param += type_handler(p)+','
        soot_param = soot_param[:-1] # remove ','
        soot_param += ')'
    # handle return value
    ret = tmp[1]
    soot_ret = type_handler(ret)
    return soot_param, soot_ret

def combiner(classname, funcname, descriptor):
    """
    
    :param classname: result of clazz.get_vm_class().name
    :param funcname: result of func.get_method().name
    :param descriptor: result of func.get_method().get_descriptor()
    :return: soot format
    """
    """
    TODO: Note that we do not handle <init> and <clinit>
    """
    print('classname', classname)
    print('funcname', funcname)
    print('descriptor', descriptor)
    soot_param, soot_ret = descriptor_handler(descriptor)
    soot_val = '<' + classname_handler(classname) + ': ' + soot_ret + ' ' + funcname + soot_param + '>'
    return soot_val
  
def getFuncNames(sess, apkPath, patterns=None):
    """
    :param sess: a session object
    :param apkPath: The path to apk file
    :return:
    """
    clazzList = []
    funcList = []
    a, d, dx = misc.AnalyzeAPK(apkPath)
    for clazz in dx.get_classes():
        name = clazz.get_vm_class().name
        """
        Here add your filter 
        """
        clazzList.append(clazz)
        print(name)
    for clazz in clazzList:
        funcs = clazz.get_methods()
        class_name = clazz.get_vm_class().name
        for func in funcs:
            func_desp = func.get_method().get_descriptor()
            func_name = func.get_method().name
            soot_val = combiner(class_name, func_name, func_desp)
            funcList.append(soot_val)
    return funcList