import sys import os import string #//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////# #OPCODE TABLE FOR THE GIVEN 12 BIT ACCUMULATOR MACHINE OPTAB = {} OPTAB["CLA"] = ["0000",0] OPTAB["LAC"] = ["0001",1] OPTAB["SAC"] = ["0010",1] OPTAB["ADD"] = ["0011",1] OPTAB["SUB"] = ["0100",1] OPTAB["BRZ"] = ["0101",1] OPTAB["BRN"] = ["0110",1] OPTAB["BRP"] = ["0111",1] OPTAB["INP"] = ["1000",1] OPTAB["DSP"] = ["1001",1] OPTAB["MUL"] = ["1010",1] OPTAB["DIV"] = ["1011",1] OPTAB["STP"] = ["1100",0] #//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////# #Helper Functions for cleaning and processing data def isComment(line): if line.find("//")!=-1: return True else: return False def hasVariable(line): l = str(line) label = False tokens = l.split() if "DW" in tokens: return True else: return False def getVariable(line): l = str(line) label = False tokens = l.split() if hasVariable(line): return [tokens[0],tokens[2]] else: return False def hasLabel(line): if line.find(":")!=-1: return True else: return False def getLabel(line): l = str(line) label = False tokens = l.split() if hasLabel(line): return hasSymbol(line) else: return False def RepresentsInt(s): try: int(s) return True except ValueError: return False def hasSymbol(line): l = str(line) symbol = False tokens = l.split() for i in tokens: if i not in list(OPTAB.keys()) and not(RepresentsInt(i)): symbol = i break return symbol def getOpcode(line): l = str(line) tokens = l.split() opcode = None noofOpcodes = 0 if ":" not in tokens: if tokens[0] not in OPTAB and "DW" not in tokens: print("[Error] Invalid OPCODE Used" , end =' ') return -2 else: if tokens[2] not in OPTAB: print("[Error] Invalid OPCODE Used" , end =' ') return -2 for i in tokens: if i in OPTAB and noofOpcodes<1: opcode = i noofOpcodes+=1 if noofOpcodes>1: raise Exception("One of the Lines contains multiple OPCODES...") else: return opcode def getPsuedoOP(line): l = str(line) tokens = l.split() if "STP" in tokens: return "STP" elif "START" in tokens: return "START" else: return False def isEnd(line): l = str(line) tokens = l.split() if "END" in tokens: return True else: return False def isStart(line): l = str(line) tokens = l.split() if "START" in tokens: return True else: return False def getStart(line): l = str(line) tokens = l.split() if "START" in tokens and len(tokens)==2: if int(tokens[1])>=256: return -1 else: return tokens[1] elif "START" in tokens and len(tokens)!=2: return 0 else: return False def getOperand(line): l = str(line) tokens = l.split() opc = None error_flag = False if ":" not in tokens: opc = tokens[0] if len(tokens)>4: error_flag = True else: opc = tokens[2] if len(tokens)>4: error_flag = True if "DW" in tokens: opc = tokens[1] if error_flag: print("[Error] OPCODE "+str(tokens[0]) +" Supplied with too many arguments thab required at Line" , end =' ') return -2 if OPTAB[opc][1] == 1 and tokens[-1] in OPTAB: print("[Error] OPCODE "+str(tokens[-1]) +" Supplied with fewer arguments thab required at Line" , end =' ') return -2 if tokens[-1] not in OPTAB: return tokens[-1] else: return False OPTAB["START"] = ["",-1] OPTAB["END"] = ["",-1] #Assuming opcode for DW OPTAB["DW"] = ["1101",2] def delAllFiles(): os.remove("tempfile.txt") os.remove("LABTAB.txt") os.remove("SYMTAB.txt") os.remove("output.out") def deleteTempFiles(): os.remove("tempfile.txt") os.remove("LABTAB.txt") os.remove("SYMTAB.txt") def binary(n): n = int(n) return '{0:08b}'.format(n) #//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////# def pass_one(linebyline): """ Working of the first Pass: -------------------------- 1. The location counter is initialised 2. Label table and Symbol Tables are initialised 3. The input program is processed line by line, wherein only the non-commented lines are processed, and the comments are simply ignored 4. If the given line: a. Is a Start Statement, then the address given/specified is used starting value of location counter b. Has a label then it is put in the label table c. Has a Variable then it is put in the symbol table d. Has opcode then the opcode is verified whether it is valid, if not an Error is thrown, and assembly process is terminated. e. If the given opcode is correct then its corresponding operand is read (depending on whether it takes in any operand or not), else an error is thrown. f. If it is a variable declarative then its symbol is checked whether it is present in the symbol table, g. If it is a label then it is checked whether it is present in the symbol table, 5. If the code has variables/ symbols that have not been defined but have been used, then the assembler throws an error 6. If the code doesn’t encounter any END statement, then the code throws an error, and if it does have any then everything below the END statement is ignored. 7. The code prepares several temporary files, a. File that contains all the lines except the comments b. Label/Variable Address Table c. Symbol Table 8. All errors are handled in the first pass. The symbol table and label table are in the form of a python dictionary which are dumped in temperory files. """ LOC_CTR = 0 length = 12 value = 0 Type = None EOS = False f=open("tempfile.txt", "a+") f1 = open("LABTAB.txt", "a+") f2 = open("SYMTAB.txt", "a+") LABTAB ={} SYMTAB = {} err_end_flag = True err_start_flag = True tok = 0 lineno = 0 for line in linebyline: line.strip() k = line.split() if not(isComment(line)): lineno+=1 label = 0 symbol = 0 tok = getStart(line) if isEnd(line): err_end_flag = False break if tok != False: LOC_CTR = int(tok) if tok == -1: print("[Error] Start statement specifies an address that is beyond the memory limit of the system line "+str(lineno)) sys.exit(-1) delAllFiles() if isStart(line): err_start_flag = False continue if getLabel(line) != False: label = getLabel(line) if getLabel(line) in LABTAB: print("[Error] Multiple Declaration of Symbol "+str(getLabel(line))+" at line "+str(lineno)) sys.exit(-1) delAllFiles() if getLabel(line) not in LABTAB: LABTAB[getLabel(line)] = LOC_CTR f1.writelines(getLabel(line)+" "+str(LOC_CTR)+"\n") if hasSymbol(line) != False: symbol = hasSymbol(line) if hasSymbol(line) not in SYMTAB: SYMTAB[hasSymbol(line)] = LOC_CTR f2.writelines(hasSymbol(line)+" "+str(LOC_CTR)+"\n") if getVariable(line) != False: var = getVariable(line) if var[0] in LABTAB: print("[Error] Multiple Declaration of Variable "+var[0]+" at line "+str(lineno)) sys.exit(-1) delAllFiles() if var[0] not in LABTAB: try: LABTAB[var[0]] = LOC_CTR f1.writelines(var[0]+" "+str(LOC_CTR)+"\n") except: print("[Error] No inital value provided at Declaration of Variable "+str(getLabel(var[0]))+" at line "+str(lineno)) sys.exit(-1) delAllFiles() opcode = getOpcode(line) if opcode == -2: print("at line "+str(lineno)) sys.exit(-1) delAllFiles() if getOperand(line) == -2: print(str(lineno)) sys.exit(-1) delAllFiles() if "DW" in line: f.writelines(str(LOC_CTR)+ " " +line[0]+" " +opcode+ " " + str(getOperand(line)) +"\n") else: if getOperand(line) != False: f.writelines(str(LOC_CTR)+ " " +opcode+ " " + str(getOperand(line)) +"\n") else: f.writelines(str(LOC_CTR)+ " " +opcode+ " " + "None" +"\n") if not(isComment(line)): if OPTAB[opcode][1] == 0: LOC_CTR += 4 else: LOC_CTR += 12 if err_end_flag: print("[Error] Missing END statement...") sys.exit(-1) delAllFiles() if err_start_flag: print("[Error] Missing START statement...") sys.exit(-1) delAllFiles() error_flag = False var_not_defined = [] for i in SYMTAB: if i not in LABTAB: error_flag = True var_not_defined.append(i) for i in var_not_defined: print("[Error] Symbol/Variable "+str(i)+" has been used but has not been defined...") if error_flag: sys.exit(-1) delAllFiles() f.close() f1.close() f2.close() pass_two(SYMTAB,LABTAB) def pass_two(SYMTAB,LABTAB): """ Working of Pass Two: 1. Since the assembler runs this pass only if all the symbols, labels and opcodes have been declared the code can directly processed. 2. The Temporary file created during the first pass, is now processed line by line 3. The labels and variables are replaced by their corresponding addresses 4. The opcodes and the addresses are converted to their binary equivalents and all changes are written off to output.out 5. Clean-Up is done, removing all temporary files and symbol and label tables. """ try: symtable = open(source_filename) labtable = open(source_filename) temp = open("tempfile.txt") except: print("Problem in Assembly...") sys.exit(0) byteCode = None LOC = 0 linebyline = temp.readlines() final = open("output.out", "a+") for i in linebyline: line = i.split() addr = line[0] operand = None try: opc = OPTAB[line[1]][0] except: opc = OPTAB[line[2]][0] if opc=="1101": opc = binary(line[3]) else: if line[2] in LABTAB: operand = LABTAB[line[2]] operand = binary(operand) elif OPTAB[line[1]][1]==1 and line[2] not in LABTAB: operand = binary(int(line[2])) l=binary(int(addr)) if operand != None: final.writelines(l[-8:]+" : "+str(opc)+" "+str(operand)+"\n") else: final.writelines(l[-8:]+" : "+str(opc)+"\n") final.close() #//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////# #opening the input files... try: source_filename = str(sys.argv[1]) source = open(source_filename) except: print("Please Enter Valid filename as python3 assembler.py <FILENAME>") sys.exit(0) #//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////# #Run the main code... linebyline = source.readlines() pass_one(linebyline) deleteTempFiles() source.close()
Write, Run & Share Python code online using OneCompiler's Python online compiler for free. It's one of the robust, feature-rich online compilers for python language, supporting both the versions which are Python 3 and Python 2.7. Getting started with the OneCompiler's Python editor is easy and fast. The editor shows sample boilerplate code when you choose language as Python or Python2 and start coding.
OneCompiler's python online editor supports stdin and users can give inputs to programs using the STDIN textbox under the I/O tab. Following is a sample python program which takes name as input and print your name with hello.
import sys
name = sys.stdin.readline()
print("Hello "+ name)
Python is a very popular general-purpose programming language which was created by Guido van Rossum, and released in 1991. It is very popular for web development and you can build almost anything like mobile apps, web apps, tools, data analytics, machine learning etc. It is designed to be simple and easy like english language. It's is highly productive and efficient making it a very popular language.
When ever you want to perform a set of operations based on a condition IF-ELSE is used.
if conditional-expression
#code
elif conditional-expression
#code
else:
#code
Indentation is very important in Python, make sure the indentation is followed correctly
For loop is used to iterate over arrays(list, tuple, set, dictionary) or strings.
mylist=("Iphone","Pixel","Samsung")
for i in mylist:
print(i)
While is also used to iterate a set of statements based on a condition. Usually while is preferred when number of iterations are not known in advance.
while condition
#code
There are four types of collections in Python.
List is a collection which is ordered and can be changed. Lists are specified in square brackets.
mylist=["iPhone","Pixel","Samsung"]
print(mylist)
Tuple is a collection which is ordered and can not be changed. Tuples are specified in round brackets.
myTuple=("iPhone","Pixel","Samsung")
print(myTuple)
Below throws an error if you assign another value to tuple again.
myTuple=("iPhone","Pixel","Samsung")
print(myTuple)
myTuple[1]="onePlus"
print(myTuple)
Set is a collection which is unordered and unindexed. Sets are specified in curly brackets.
myset = {"iPhone","Pixel","Samsung"}
print(myset)
Dictionary is a collection of key value pairs which is unordered, can be changed, and indexed. They are written in curly brackets with key - value pairs.
mydict = {
"brand" :"iPhone",
"model": "iPhone 11"
}
print(mydict)
Following are the libraries supported by OneCompiler's Python compiler
Name | Description |
---|---|
NumPy | NumPy python library helps users to work on arrays with ease |
SciPy | SciPy is a scientific computation library which depends on NumPy for convenient and fast N-dimensional array manipulation |
SKLearn/Scikit-learn | Scikit-learn or Scikit-learn is the most useful library for machine learning in Python |
Pandas | Pandas is the most efficient Python library for data manipulation and analysis |
DOcplex | DOcplex is IBM Decision Optimization CPLEX Modeling for Python, is a library composed of Mathematical Programming Modeling and Constraint Programming Modeling |