如何寫一個Python Interpreter

最近一直在搗鼓Python，老想學別人在自己的SolidMCP之類搞一個Python Interpreter。

起初覺得很高深，就直接把將某開源軟體基於Qt的Python Console實現剝離開來，成功移植到SolidMCP內。

在這個過程中發現，其實寫一個蹩腳的Python Interpreter並不是太難，Piaoger決定沉下心來，研究一把，並把研究成果記錄下來如下：

>> 如何寫一個Python Interpreter

可以先用Python來寫一個Prototype，再用C++翻譯過來就是啦。

import sysimport osimport coderedirectedOutput = r'C:\RedirectedPythonConsole.txt'# Redirected Channelclass MyConsole:    def write(self, input):                myFile = file(redirectedOutput, 'a')        output = '\nPython>' + input        myFile.write(output)        myFile.close()# Redirect stdout to MyConsoledefaultStdOut = sys.stdoutmyConsole = MyConsole()sys.stdout = myConsoleif os.path.exists(redirectedOutput):    os.remove(redirectedOutput)        # Compile input code to code object# Run Code ObjectinputCode = '2+3'interpreter = code.InteractiveInterpreter()codeObject = interpreter.compile(inputCode)interpreter.runcode(codeObject)sys.stdout = myConsole

 

>>C++版本原型

    // Demonstrates how to use Interactive Interpreter and Interactive Console    void testInterpreter()    {        Py_Initialize();        // #load code module to use InteractiveInterpreter and InteractiveConsole        // # Utilities used to emulate Python's interactive interpreter        // # InteractiveConsole closely emulates the behavior of the interactive Python interpreter        // # InteractiveConsole builds on InteractiveInterpreter        // import code module        PyObject* module = PyImport_ImportModule("code");        // Construct InteractiveInterpreter from code module        PyObject* func = PyObject_GetAttrString(module, "InteractiveInterpreter");        PyObject* args = Py_BuildValue("()");        PyObject* interpreter = PyEval_CallObject(func,args);        // Provide input as code        const char* source = "2+3";        PyObject* sourceArgs = Py_BuildValue("(s)", source);        PyObject* compilefunc = PyObject_GetAttrString(interpreter, "compile");        PyObject* codeObject = PyEval_CallObject(compilefunc, sourceArgs);        // run compiled bytecode        PyObject* mainModule = PyImport_AddModule("__main__");        PyObject* dict = PyModule_GetDict(mainModule);        PyObject* presult = PyEval_EvalCode((PyCodeObject*)codeObject, dict, dict);        // Finally, release everything by decrementing their reference counts.        //        Py_DECREF(mainModule);        Py_DECREF(presult);        Py_DECREF(dict);        Py_DECREF(codeObject);        Py_DECREF(compilefunc);        Py_DECREF(sourceArgs);        Py_DECREF(interpreter);        Py_DECREF(args);        Py_DECREF(func);        Py_Finalize();    }

基本上原理就是這樣的，其餘的事情不外乎也搞一個sys.stdout的重新導向，然後就是處理TextEdit控制項的事件啦。

 

>> 利用InteractiveConsole直接用Python寫

Python提供了一個InteractiveConsole幫我們來幹這個事情，這玩意其實也是繼承自InteractiveInterpreter.

那些基於PyQt的Python Console大抵就是這樣乾的。

# -----------------------------------------------------------------------------------------------
# Copy from ActiveState Code Recipes
#  http://code.activestate.com/recipes/355319-using-codeinteractiveconsole-to-embed-a-python-she/
# -----------------------------------------------------------------------------------------------
import sysimport codefrom code import InteractiveConsoleclass FileCacher:    "Cache the stdout text so we can analyze it before returning it"    def __init__(self): self.reset()    def reset(self): self.out = []    def write(self,line): self.out.append(line)    def flush(self):        output = '\n'.join(self.out)        self.reset()        return outputclass Shell(InteractiveConsole):    "Wrapper around Python that can filter input/output to the shell"    def __init__(self):        self.stdout = sys.stdout        self.cache = FileCacher()        InteractiveConsole.__init__(self)        return    def get_output(self):
        sys.stdout = self.cache
def return_output(self):
        sys.stdout = self.stdout    def push(self,line):        self.get_output()        # you can filter input here by doing something like        # line = filter(line)        InteractiveConsole.push(self,line)        self.return_output()        output = self.cache.flush()        # you can filter the output here by doing something like        # output = filter(output)        print output # or do something else with it        return if __name__ == '__main__':     sh = Shell()     sh.interact()