STM2IR.py

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"STM2IR.py","provenance":[],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"}},"cells":[{"cell_type":"code","metadata":{"id":"RYFB_pMZbR8Q","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1613495124205,"user_tz":-180,"elapsed":1567,"user":{"displayName":"Deniz DİKBIYIK","photoUrl":"","userId":"07706369359170137388"}},"outputId":"22335145-6c2c-4034-82c1-e70718aefb09"},"source":["import sys\r\n","\r\n","# this class is created to use stack and use its methods\r\n","class Stack:\r\n","    def __init__(self):\r\n","            self.item = []\r\n","    def isEmpty(self):\r\n","            return self.item == []\r\n","    def push(self,value):\r\n","            self.item.insert(0,value)\r\n","    def pop(self):\r\n","            return self.item.pop(0)\r\n","    def peek(self):\r\n","            return self.item[0]\r\n","\r\n","# inputFileOpen function takes the file as parameter and opens it\r\n","import io\r\n","def inputFileOpen(fileName):\r\n","  try:\r\n","    f = io.open(fileName, 'r', encoding='utf8')\r\n","    return f\r\n","  except OSError as osError:\r\n","    print(osError)\r\n","  except UnicodeDecodeError as decodeError:\r\n","    print(decodeError)    \r\n","\r\n","input_file = inputFileOpen(\"file.stm\")\r\n","# here the output file is opened with writing type\r\n","output_file = open(\"file.ll\",\"w\")\r\n","lines = input_file.readlines()\r\n","output_writer = inputFileOpen(\"file.ll\")\r\n","\r\n","# operation counter keeps the number of % for lines\r\n","operationCounter = 0\r\n","# varList stores variables\r\n","varList = []\r\n","# var is the part of the line before equal sign\r\n","var = \"\"\r\n","# expression is the part of the line after equal sign\r\n","expression = \"\"\r\n","# result is defined for switcher function which returns add, sub, mul or sdiv\r\n","result = \"\" \r\n","# error is taken from syntaxanalyzer function and if the input file has errors,\r\n","# it does not continue to the other functions\r\n","error = \"no\"\r\n","# postfix is taken from infix to postfix function to which expression is given\r\n","postfix = []\r\n","# token is the operation sign to give to switcher function and get the text of it\r\n","token = \"\"\r\n","tempList = []\r\n","priority = {'+': 1, '-': 1, '*': 2, '/': 2}\r\n","# temporary lists are taken for going over postfix partly\r\n","tempPostfixVar = []\r\n","tempPostfixOp = []  \r\n","tempList = []\r\n","\r\n","# isOperator checks whether operator or not\r\n","def isOperator(x):\r\n","  ops = ['+', '-', '/', '*']\r\n","  return x in ops\r\n","\r\n","# isOperand checks whether operand or not\r\n","def isOperand(ch):\r\n","  return ch.isalnum()\r\n","\r\n","# the three functions below check the characters\r\n","def isOpenParenthesis(ch):\r\n","  return ch == '('\r\n","\r\n","def isCloseParenthesis(ch):\r\n","  return ch == ')'\r\n","\r\n","def isEqualSign(ch):\r\n","  return ch == '='\r\n","\r\n","def switcher(token):\r\n","  # This function returns operators' IR text to write proper expression in ll file\r\n","  global result\r\n","  if (token == \"+\"):\r\n","      result = \"add\"\r\n","  elif (token == \"-\"):\r\n","      result = \"sub\"\r\n","  elif (token == \"/\"):\r\n","      result = \"sdiv\"\r\n","  elif (token == \"*\"):\r\n","      result = \"mul\"\r\n","\r\n","# main function has all functions' call in it\r\n","def main():\r\n","  try:\r\n","    writeHeader()\r\n","    toIRConverter()\r\n","    global output_writer\r\n","    output_file.writelines([\"\\n  ret i32 0\"])\r\n","    output_file.writelines([\"\\n}\"])\r\n","    print(output_writer.read())\r\n","  except OSError:\r\n","    print(\"\\nAn error occured.\")\r\n","\r\n","# writeHeader adds header to the output file\r\n","def writeHeader():\r\n","  try:\r\n","    output_file.writelines([\"; ModuleID = 'stm2ir'\", \"\\ndeclare i32 @printf(i8*, ...)\", \r\n","                            \"\\n@print.str = constant [4 x i8] c\\\"%d\\\\0A\\\\00\\\"\", \"\\n\\ndefine i32 @main() {\"])\r\n","  except IOError:\r\n","    print(\"\\nAn error occured.\")\r\n","\r\n","# toIRConverter line converter function for each line in input text\r\n","def toIRConverter():\r\n","  global operationCounter\r\n","  operationCounter = 1\r\n","  try:\r\n","    for i in lines:\r\n","      convertLine(i)\r\n","  except OSError:\r\n","    print(\"\\nAn error occured.\")\r\n","\r\n","# syntaxAnalyzer analyzes each lines from the input and checks if there are any syntax errors\r\n","def syntaxAnalyzer(i):\r\n","  try:\r\n","    numLeftPar = 0\r\n","    numRightPar = 0\r\n","    c0 = ''\r\n","    c = ''\r\n","    c1 = ''\r\n","    global error\r\n","    inputLine = i\r\n","    lineLength = len(inputLine)\r\n","    for j in range(0, lineLength):\r\n","      c = inputLine[j]\r\n","      if (isCloseParenthesis(c)):\r\n","          numLeftPar = numLeftPar + 1\r\n","      elif (isOpenParenthesis(c)):\r\n","          numRightPar = numRightPar + 1\r\n","      # this checks if the inspected character is an operator\r\n","      elif (isOperator(c) or isEqualSign(c)):\r\n","        # if the next character is also an operator, there is an error\r\n","        if (j == 0):\r\n","          # this checks if inspected character is at the first index\r\n","          c1 = inputLine[j + 1]\r\n","          if (isOperator(c1) or isEqualSign(c)):\r\n","            print(\"\\nExtra operator error.\")\r\n","            error = \"yes\"\r\n","        elif (j == lineLength - 1):\r\n","          # this checks if inspected character is at the last index\r\n","          c0 = inputLine[j - 1]\r\n","          if (isOperator(c0) or isEqualSign(c)):\r\n","            print(\"\\nExtra operator error.\")\r\n","            error = \"yes\"\r\n","        else:\r\n","          # for other cases of inspected character\r\n","          c0 = inputLine[j - 1]\r\n","          c1 = inputLine[j + 1]\r\n","          if (isOperator(c0) or isEqualSign(c0) or isOperator(c1) or isEqualSign(c1)):\r\n","            print(\"\\nExtra operator error.\")\r\n","            error = \"yes\"\r\n","    # this is resulted with if number of right and left parentheses are not the same, then there is an error\r\n","    if (numRightPar != numLeftPar):\r\n","      print(\"\\nMissing parenthesis error.\")\r\n","      error = \"yes\"\r\n","  except OSError:\r\n","    print(\"\\nAn error occured.\")\r\n","\r\n","# convertLine converts the line to the intermediate representation, prints the representation to the output file via calling other functions\r\n","# it only writes alloca and store text in it, other texts are created by other called functions\r\n","import re\r\n","def convertLine(i):\r\n","  try:\r\n","    syntaxAnalyzer(i)\r\n","    global error\r\n","    global var\r\n","    global expression\r\n","    if error == \"yes\":\r\n","      output_file.writelines([\"\\nThere are some errors in your expression file.\"])\r\n","    else:\r\n","      global varList\r\n","      line = i.replace(\"\\\\s\", \"\")\r\n","      line = line.replace(\"\\s\", \"\")\r\n","      line = line.replace('\\n', \"\")\r\n","      equalSign = \"=\"\r\n","      if equalSign in line:\r\n","        var = line.split(\"=\")[0]\r\n","        expression = line.split(\"=\")[1]\r\n","        # variable is added into the variable list\r\n","        if var not in varList:\r\n","          varList.append(var.strip())\r\n","          output_file.writelines([\"\\n  %\" + var.strip() + \" = alloca i32\"])\r\n","      else:\r\n","        # if there is no equals sign, then whole line is an expression\r\n","        expression = line        \r\n","        # If expression is numeric, that is, it doesn't have any arithmetic operation,\r\n","      # then intermediate representation is printed\r\n","      expression = expression.replace(\" \", \"\")\r\n","      if var != \"\" and expression.isnumeric():\r\n","        output_file.writelines([\"\\n  store i32 \" + expression + \", i32* %\" + var])\r\n","      else:\r\n","        # else, expression is first converted into postfix notation\r\n","        # then the LLVM representation is printed to output file\r\n","        infixToPostfix(expression, line)\r\n","        equalsign = \"=\"\r\n","        if equalsign in line:\r\n","          output_file.writelines([\"\\n  store i32 %\" + str (operationCounter-1) + \", i32* %\" + var])\r\n","  except OSError:\r\n","    print(\"\\nAn error occured.\")\r\n","\r\n","# infixToPostfix function converts expression which is the right side of equal sign \r\n","# or the expression without equal sign in the line to postfix\r\n","def infixToPostfix(infixStr, line):\r\n","  global postfix\r\n","  # to be able to divide expression into list of variables and numbers, \r\n","  # expression is divided according to operation signs\r\n","  infixStr = infixStr.replace(\"+\", \"\\n+\\n\")\r\n","  infixStr = infixStr.replace(\"-\", \"\\n-\\n\")\r\n","  infixStr = infixStr.replace(\"*\", \"\\n*\\n\")\r\n","  infixStr = infixStr.replace(\"/\", \"\\n/\\n\")\r\n","  infixStr = infixStr.replace(\"(\", \"\\n(\\n\")\r\n","  infixStr = infixStr.replace(\")\", \"\\n)\\n\")\r\n","  infixStr = infixStr.splitlines()\r\n","  x = \"\"\r\n","  for i in infixStr:\r\n","    x += \" \" + i + \" \"\r\n","  infixStr = x\r\n","  infixStr = infixStr.replace(\"  \", \" \")\r\n","  prec = {}\r\n","  prec[\"^\"] = 4\r\n","  prec[\"*\"] = 3\r\n","  prec[\"/\"] = 3\r\n","  prec[\"+\"] = 2\r\n","  prec[\"-\"] = 2\r\n","  prec[\"(\"] = 1\r\n","  opStack = Stack()\r\n","  postfixList = []\r\n","  infixList = infixStr.split()\r\n","  correctData(infixList)\r\n","  for token in infixList:\r\n","      if len(token) > 1:\r\n","          postfixList.append(token)\r\n","          continue\r\n","      if token in \"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwyz\":\r\n","          postfixList.append(token)\r\n","      elif token == \"(\":\r\n","          opStack.push(token)\r\n","      elif token == \")\":\r\n","          topToken = opStack.pop()\r\n","          while topToken != \"(\":\r\n","              postfixList.append(topToken)\r\n","              topToken = opStack.pop()\r\n","      else:\r\n","          while (not opStack.isEmpty()) and  \\\r\n","          prec[opStack.peek()] >= prec[token]:\r\n","              postfixList.append(opStack.pop())\r\n","          opStack.push(token)\r\n","  while (not opStack.isEmpty()):\r\n","      postfixList.append(opStack.pop())\r\n","  postfixstr = \" \".join(postfixList)\r\n","  list1 = []\r\n","  txt = \"\"\r\n","  nextTxt = \"\"\r\n","  for i in range (0, len(postfixstr)):\r\n","    txt = txt + postfixstr[i]\r\n","    if i != len(postfixstr)-1:\r\n","      nextTxt = postfixstr [i + 1]\r\n","    elif i == len(postfixstr)-1:\r\n","      nextTxt == \" \"\r\n","    if nextTxt == \" \":\r\n","      list1.append(txt.strip())\r\n","      txt = \"\"\r\n","  list1.append(postfixstr[len(postfixstr)-1].strip())\r\n","  postfix = list1\r\n","  postfixInterpreter(line)\r\n","\r\n","def correctData(infixList):\r\n","  # correctData cleans the data if there are spaces between the\r\n","  # operands and parentheses, or multi-digit numbers or both\r\n","  for index, token in enumerate(infixList):\r\n","      if len(token) > 1 and (not token.isalnum()):\r\n","          if len(token) >= 3:\r\n","              rear_index = index + token.index(token[-1])\r\n","              newindex, i = (rear_index, -1) if token[:-1].isalnum() else (index, +1)\r\n","              parenth, rest = (token[0], token[1:]) if token[1:].isalnum() \\\r\n","                         else (token[-1], token[:-1])\r\n","              infixList.insert(newindex, parenth)\r\n","              infixList.insert(newindex + i, rest)\r\n","          else:\r\n","              for subindex, subtoken in enumerate(token):\r\n","                  infixList.insert(index + subindex, subtoken)\r\n","          infixList.remove(token)\r\n","  return infixList\r\n","\r\n","# postconverter is the function which is called from postfixInterpreter. \r\n","# this can be counted as the hearth of the project because it creates load and operational lines\r\n","def postconverter(varList):\r\n","  global postfix  \r\n","  global result\r\n","  global output_writer\r\n","  global operationCounter\r\n","  myOpList = ['+', '-', '*', '/']\r\n","  previousi = \"\"\r\n","  global tempPostfixVar\r\n","  global tempPostfixOp  \r\n","  global tempList\r\n","  for i in postfix:\r\n","    if i in myOpList:\r\n","      tempPostfixOp.append(i)\r\n","      postfix = postfix[1:]\r\n","    else:\r\n","      if previousi in myOpList:\r\n","        break\r\n","      else:\r\n","        tempPostfixVar.append(i)\r\n","        postfix = postfix[1:]\r\n","    previousi = i\r\n","  tempPostfixOp.reverse()\r\n","  while len(tempPostfixVar) > 1 and len(tempPostfixOp) > 0:\r\n","    num2 = tempPostfixVar.pop()\r\n","    num2 = num2.replace(\" \", \"\")    \r\n","    if num2 in varList or num2 == \"%\":\r\n","      if num2 in varList:\r\n","        output_file.writelines([\"\\n  %\" + str (operationCounter) + \" = load i32* %\" + num2])\r\n","        tempList.append(operationCounter)\r\n","        operationCounter = operationCounter + 1       \r\n","      tempWriter2 = tempList.pop()\r\n","      tempWriterToPrint2 = \"%\" + str (tempWriter2)\r\n","    else:\r\n","      tempWriter2 = num2\r\n","      tempWriterToPrint2 = str (num2)\r\n","    num1 = tempPostfixVar.pop()\r\n","    num1 = num1.replace(\" \", \"\")\r\n","    if num1 in varList or num1 == \"%\":\r\n","      if num1 in varList:\r\n","        output_file.writelines([\"\\n  %\" + str (operationCounter) + \" = load i32* %\" + num1])\r\n","        tempList.append(operationCounter)\r\n","        operationCounter = operationCounter + 1        \r\n","      tempWriter1 = tempList.pop()\r\n","      tempWriterToPrint1 = \"%\" + str (tempWriter1)\r\n","    else:\r\n","      tempWriter1 = num1\r\n","      tempWriterToPrint1 = str (num1)\r\n","    opTemper = tempPostfixOp.pop()\r\n","    switcher(opTemper)\r\n","    output_file.writelines([\"\\n  %\" + str (operationCounter) + \" = \" + result + \" i32 \" + tempWriterToPrint1 + \" ,\" + tempWriterToPrint2]) \r\n","    tempPostfixVar.append(\"%\")  \r\n","    tempList.append(operationCounter)\r\n","    operationCounter = operationCounter + 1\r\n","  if len(tempPostfixOp) > 0 and len(tempPostfixVar) > 1:\r\n","    output_file.writelines([\"\\nThere are some errors in your input file.\"])\r\n","\r\n","# postfixInterpreter goes over the postfix and writes IR with the parts of postfix\r\n","# it goes until seeing next operation sign to be able to check operational priority,\r\n","# after analyzing and writing to output file with that part, it continues\r\n","# till the end of postfix until it is being empty\r\n","def postfixInterpreter(line): \r\n","  try:\r\n","    global varList\r\n","    global postfix\r\n","    global tempPostfixVar\r\n","    global tempPostfixOp  \r\n","    global tempList\r\n","    global operationCounter\r\n","    tempPostfixVar = []\r\n","    tempPostfixOp = []\r\n","    while len(postfix) > 0:\r\n","      postconverter(varList) \r\n","      equalSign = \"=\"\r\n","      if equalSign not in line:\r\n","        tempWriterNow = tempList.pop()\r\n","        operationCounter = operationCounter + 1\r\n","        output_file.writelines([\"\\n  call i32 (i8*, ...)* @printf(i8* getelementptr ([4 x i8]* @print.str, i32 0, i32 0), i32 %\" + str (tempWriterNow)+')'])\r\n","  except OSError:\r\n","    print(\"\\nAn error occured.\")\r\n","\r\n","main()"],"execution_count":2,"outputs":[{"output_type":"stream","text":["; ModuleID = 'stm2ir'\n","declare i32 @printf(i8*, ...)\n","@print.str = constant [4 x i8] c\"%d\\0A\\00\"\n","\n","define i32 @main() {\n","  %x1 = alloca i32\n","  store i32 3, i32* %x1 \n","  %y = alloca i32\n","  store i32 5, i32* %y\n","  %zvalue = alloca i32\n","  %1 = load i32* %y\n","  %2 = add i32 1 ,%1\n","  %3 = load i32* %x1\n","  %4 = mul i32 %3 ,%2\n","  %5 = add i32 23 ,%4\n","  store i32 %5, i32* %zvalue\n","  call i32 (i8*, ...)* @printf(i8* getelementptr ([4 x i8]* @print.str, i32 0, i32 0), i32 %5)\n","  %k = alloca i32\n","  %7 = load i32* %y\n","  %8 = load i32* %x1\n","  %9 = sub i32 %8 ,%7\n","  %10 = load i32* %zvalue\n","  %11 = sub i32 %9 ,%10\n","  store i32 %11, i32* %k\n","  %12 = load i32* %y\n","  %13 = mul i32 3 ,%12\n","  %14 = add i32 2 ,5\n","  %15 = mul i32 1 ,%14\n","  %16 = mul i32 %13 ,%15\n","  %17 = load i32* %x1\n","  %18 = add i32 %17 ,%16\n","  store i32 %18, i32* %k\n","  %19 = load i32* %k\n","  %20 = add i32 %19 ,1\n","  call i32 (i8*, ...)* @printf(i8* getelementptr ([4 x i8]* @print.str, i32 0, i32 0), i32 %20)\n","  ret i32 0\n","}\n"],"name":"stdout"}]}]}