get_function_names_in_a_directory.py

#!/usr/bin/python

import sys
import glob
import os
import re

# define base directory
baseDirectory = "" 

# define sub directory
formsDirectory = ""

tabuFunctions = ['rules', 'attributeLabels']

def read_directory(baseDirectory, directoryName):
	
	fullDirectory = baseDirectory + directoryName
	os.chdir(fullDirectory)	

	files = []

	for phpFile in glob.glob("*.php"):
		files.append(phpFile)

	return files


def get_function_names(baseDirectory, directory, files):

	regex = re.compile("public function (\w+)")
	functionList = {}

	for fileName in files:
		
		fileWithDirectory = baseDirectory + directory + fileName		
		f = open(fileWithDirectory, 'r')	
		f.readline()

		fileFunctionsList = []

		for line in f:

			r = regex.search(line)

			if r:
				matchedFunction = r.groups()[0]
				fileFunctionsList.append(matchedFunction)

		functionList[fileName] = fileFunctionsList

	return functionList

def remove_tabu_functions(filesAndFunctions, tabuFunctions):
	for fileName, fileFunctions in filesAndFunctions.iteritems():
		for tabuFunction in tabuFunctions:
			if tabuFunction in fileFunctions:
				filesAndFunctions[fileName].remove(tabuFunction)

	return filesAndFunctions


files = read_directory(baseDirectory, formsDirectory)

filesAndFunctions = get_function_names(baseDirectory, formsDirectory, files)

filesAndFunctions = remove_tabu_functions(filesAndFunctions, tabuFunctions)

print "starting...\n"

for fileName, fileFunctions in filesAndFunctions.iteritems():
	print "file name: " + fileName + " :"
	print "functions: "
	print fileFunctions
	print "******************"

• read all files in a given directory and return file list
• read lines and return function names list with regex
• remove tabu function names from given list