scientific-programming-exercises/ex_28.py

from collections import deque
import string
from io import StringIO

TOKEN_NUMERIC_LITERAL = "0123456789"
TOKEN_DOT = "."
TOKEN_PARENTHESIS_OPEN = "("
TOKEN_PARENTHESIS_CLOSE = ")"
TOKEN_BINARY_OPERATOR = "+-*/"
TOKEN_SIGN = "-"
TOKEN_WHITESPACE = string.whitespace

STATE_BEGIN_EXPRESSION = 0
STATE_INSIDE_TERM = 1
STATE_EXPECT_NEXT_ARGUMENT = 2
STATE_PARENTHESIS_CLOSED = 4

class CharacterDevice(object):
	def __init__(self, file_):
		self._file = file_
		self._offset = 0
		self.character = deque([0])
		self.line = 0

	def getc(self):
		self._offset += 1
		character = self._file.read(1)
		if(character == "\n"):
			self.line += 1
			self.character.append(0)
		else:
			self.character[-1] += 1
		return character

	def ungetc(self, c):
		if(c == "\n"):
			self.line -= 1
			self.character.pop()
		else:
			self.character[-1] -= 1
		self._offset -= 1
		self._file.seek(self._offset)

class ParsingError(Exception): 
	pass

class InternalError(Exception): 
	pass

class ArithmeticParser(object):
	def __init__(self, file_):
		self._file = CharacterDevice(file_)
		self._stack = deque()
		self._current_expr = deque()

	def parse_numeric_literal(self, carry = ""):
		literal = carry
		dot = False
		character = None
		while True:
			character = self._file.getc()
			if(not character):
				break
			if(character in TOKEN_NUMERIC_LITERAL):
				literal += character
				continue
			elif(character in TOKEN_DOT):
				if(dot):
					raise self.build_parsing_error("Unexpected dot while parsing literal")
				dot = True
				literal += character
				continue
			break
		self._file.ungetc(character)

		self._current_expr.append(literal)

	def parse_parenthesis(self):
		parenthesis = self._file.getc()
		if(parenthesis not in TOKEN_PARENTHESIS_OPEN):
			raise InternalError("parse_parenthesis did not recieve an opening parenthesis")
		
		self._stack.append(self._current_expr)
		self._current_expr = deque()
		state = self.parse_expression()
		parenthesis = self._file.getc()
		if(parenthesis not in TOKEN_PARENTHESIS_CLOSE):
			raise self.build_parsing_error("Expected closing parenthesis")
		if(not parenthesis):
			raise self.build_parsing_error("Unexpected EOF while parsing")
		nested = self._current_expr
		self._current_expr = self._stack.pop()
		self._current_expr.append(list(nested))

	def parse_expression(self):
		state = STATE_BEGIN_EXPRESSION
		while True:
			peek = self._file.getc()
			if(not peek):
				break
			if(peek in TOKEN_WHITESPACE):
				continue
			if(state == STATE_BEGIN_EXPRESSION):
				if(peek in TOKEN_NUMERIC_LITERAL):
					self._file.ungetc(peek)
					self.parse_numeric_literal()
					state = STATE_INSIDE_TERM
					continue
				if(peek in TOKEN_SIGN):
					self.parse_numeric_literal(carry=peek)
					state = STATE_INSIDE_TERM
					continue
				if(peek in TOKEN_PARENTHESIS_OPEN):
					self._file.ungetc(peek)
					self.parse_parenthesis()
					state = STATE_INSIDE_TERM
					continue

				raise self.build_parsing_error("Unexpected token: '{}'".format(peek))
			if(state == STATE_INSIDE_TERM):
				if(peek in TOKEN_BINARY_OPERATOR):
					self._current_expr.append(peek)
					state = STATE_EXPECT_NEXT_ARGUMENT
					continue
				if(peek in TOKEN_PARENTHESIS_CLOSE):
					state = STATE_PARENTHESIS_CLOSED
					self._file.ungetc(peek)
					break
				raise self.build_parsing_error("Unexpected token: '{}', expected binary operator".format(peek))
			if(state == STATE_EXPECT_NEXT_ARGUMENT):
				if(peek in TOKEN_NUMERIC_LITERAL):
					self._file.ungetc(peek)
					self.parse_numeric_literal()
					state = STATE_INSIDE_TERM
					continue
				if(peek in TOKEN_SIGN):
					self.parse_numeric_literal(carry=peek)
					state = STATE_INSIDE_TERM
					continue
				if(peek in TOKEN_PARENTHESIS_OPEN):
					self._file.ungetc(peek)
					self.parse_parenthesis()
					state = STATE_INSIDE_TERM
					continue

				raise self.build_parsing_error("Unexpected token: '{}'".format(peek))
		if(state not in (STATE_BEGIN_EXPRESSION
				, STATE_INSIDE_TERM
				, STATE_PARENTHESIS_CLOSED)):
			raise self.build_parsing_error("Unexpected EOF while parsing")
		return state


	def parse(self):
		state = self.parse_expression()
		if(state not in (STATE_BEGIN_EXPRESSION
				, STATE_INSIDE_TERM)):
			raise self.build_parsing_error("Unexpected EOF while parsing")

		return list(self._current_expr)

			



	def build_parsing_error(self, msg):
		return ParsingError("Error: Line {}, Characer {}: {}".format(self._file.line, self._file.character, msg))
		

def check_expression(string):
	parser = ArithmeticParser(StringIO(string))
	try:
		parser.parse()
		return True
	except ParsingError:
		return False


if( __name__ == "__main__"):

	print(ArithmeticParser(StringIO("((3.2+4)*(5)+7)")).parse())
	print(check_expression("((3+4)*(5)+7)"))
	print(check_expression("(-3+4)*(5/7)"))
	print(check_expression("(-3+4)*(5/(7+5))"))
	print(check_expression("(-3.3+4)*(5/(7+5)"))
	print(ArithmeticParser(StringIO("((3+4)*(5)+7")).parse())