code_analyzer/ccleaner.py

import re

VAR = r"(?:int|float|double|char|long|string)\s*\*?\s+([^;\(\)]+);$"
FUNC = re.compile(r"(?:int|float|double|char|long|string|void)\s*\*?\s+([a-zA-Z0-9_]+)\s*\([^)]*\)\s*(?:{|;)")
FUNC_PARAM = re.compile(r"\(\s*(?:int|float|double|char|long|string)\s*\*?\s+([^;\(\)]+)\)\s*(?:{|;)")
PARAM_DECL = re.compile(r"(?:int|float|double|char|long|string)\s*\*?\s+([a-zA-Z0-9_]+)")

class CodeCleaner:
    def __init__ (self):
        self.varCount = 0
        self.funcCount = 0
        self.symbolMap = {}

#    def cleanCode (self,text):
#        lines = text.splitlines()
#        for line in lines:
#            if FUNC.search(line) != None:
#                self.cleanFunc(line)
#                if FUNC_PARAM.search(line) != None:
#                    self.cleanFuncParam(line)
#            elif VAR.search(line) != None:
#                self.cleanVar(line)
#        regex = {}
#        for k in self.symbolMap:
#            regex[k] = re.compile(r"[^\"](\W|^)({})(\W)".format(k))
#        finalText = []
#        for line in lines:
#            for k in regex:
#                line = regex[k].sub(r"\1{}\3".format(self.symbolMap[k]), line)
#            finalText.append(line)
#
#        return "\n".join(finalText)
    def cleanCode (self,text):
        funcs = FUNC.finditer(text)
        for _,m in enumerate(funcs, start=1):
            #print("Func delc: {}".format(m.group()))
            if m == None:
                continue
            self.cleanFunc(m.group())
            params = FUNC_PARAM.finditer(m.group())
            #print("p",list(params))
            for _, p in enumerate(params, start=1):
                if p == None:
                    continue
                #print("Func param: {}".format(p.group()))
                self.cleanFuncParam(p.group())
        nvars = re.finditer(VAR,text,re.MULTILINE)
        for _, m in enumerate(nvars, start=1):
            if m == None:
                continue
            #print("vars: {}".format(m.group()))
            self.cleanVar(m.group())
        #print(self.symbolMap)
        stringMode = False
        lineComment = False
        blockComment = False
        lastChar = ''
        output = []
        alpha = ""
        for c in text:
            if stringMode:
                stringMode = not (c == '"')
                output.append(c)
            elif lineComment:
                lineComment = c != '\n'
                output.append(c)
            elif blockComment:
                blockComment = (lastChar+c) != '*/'
                output.append(c)
            elif c == '/':
                if lastChar == '/':
                    lineComment = True
                output.append(c)
            elif c == '*':
                if lastChar == '/':
                    blockComment = True
                output.append(c)
            elif c == '"':
                alpha = ""
                stringMode = True
                output.append(c)
            elif re.match("[a-zA-Z0-9_]",c) != None:
                alpha += c
            else:
                if len(alpha) > 0 and alpha in self.symbolMap:
                    #print("Replacing {} with {}.".format(alpha, self.symbolMap[alpha]))
                    output.append(self.symbolMap[alpha])
                    alpha = ""
                elif len(alpha) > 0:
                    #print("Reinserting {}".format(alpha))
                    output.append(alpha)
                    alpha = ""
                output.append(c)
            lastChar = c

        return "".join(output)

    def cleanFunc (self,line):
        match = FUNC.search(line)
        if match == None:
            print(line)
        varID = match.group(1)
        self.symbolMap[varID] = "f{}".format(self.funcCount)
        self.funcCount += 1

    def cleanFuncParam (self,line):
        match = FUNC_PARAM.findall(line)[0]
        ids = match.split(",")
        for i in range(len(ids)):
            varID = ids[i].strip()
            if varID.find(" ") == -1:
                varID = re.search("[a-zA-Z_][a-zA-Z0-9_]*",varID).group()
                if varID not in self.symbolMap:
                    self.symbolMap[varID] = "v{}".format(self.varCount)
                    self.varCount += 1
            else:
                varID = varID.split(" ")[1].strip()
                isMatch = re.search("[a-zA-Z_][a-zA-Z0-9_]*",varID)
                if isMatch == None:
                    print(match, ids[i])
                    continue
                if varID not in self.symbolMap:
                    self.symbolMap[varID] = "v{}".format(self.varCount)
                    self.varCount += 1

    def cleanVar (self,line):
        match = re.findall(VAR, line)[0]
        ids = match.split(",")
        for i in range(len(ids)):
            varID = ids[i].strip()
            if varID.find("=") != -1:
                varID = varID.split("=")[0].strip()
            isMatch = re.search("[a-zA-Z_][a-zA-Z0-9_]*",varID)
            if isMatch == None:
                continue
            varID = isMatch.group()
            if varID not in self.symbolMap:
                self.symbolMap[varID] = "v{}".format(self.varCount)
                self.varCount += 1