Assignment 4 (PostScript Interpreter - Part 2 Solution

An Interpreter for a Simple Postscript-like Language




Weight: The entire interpreter project (Part 1 and Part 2 together) will count for 12% of your course grade. Part 2 is worth 9%.




This assignment is to be your own work. Refer to the course academic integrity statement in the syllabus.




Turning in your assignment




Rename your Part -1 submission file as HW4_part2.py and continue developing your code in the HW4_part2.py file. I strongly encourage you to save a copy of periodically so you can go back in time if you really mess something up. To submit your assignment, turn in your file by uploading on the dropbox on Blackboard (under AssignmentSubmisions menu).




The file that you upload must be named HW4_part2.py . Be sure to include your name as a comment at the top of the file. You may turn in your assignment up to 4 times. Only the last one submitted will be graded.




Implement your code for Python 3. The TA will run all assignments using Python3 interpreter. You will lose points if your code is incompatible with Python 3.




The work you turn in is to be your own personal work. You may not copy another student's code or work together on writing code. You may not copy code from the web, or anything else that lets you avoid solving the problems for yourself.




Grading




The assignment will be marked for good programming style (appropriate algorithms, good indentation and appropriate comments -- refer to the Python style guide ) -- as well as thoroughness of testing and clean and correct execution. You will lose points if you don't (1) provide test functions / additional test cases, (2) explain your code with appropriate comments, and (3) follow a good programming style.




The Problem




In this assignment you will write an interpreter in Python for a simplified PostScript-like language, concentrating on key computational features of the abstract machine, omitting all PS features related to graphics, and using a somewhat-simplified syntax. The simplified language, SPS, has the following features of PS:




integer constants, e.g. 123: in Python3 there is no practical limit on the size of integers
string constants, e.g. (CptS355): string delimited in parenthesis (Make sure to keep the parenthesis delimiters when you store the string constants in the opstack and the dictstack.)



name constants, e.g. /fact: start with a / and letter followed by an arbitrary sequence of letters and numbers
names to be looked up in the dictionary stack, e.g. fact: as for name constants, without the /
code constants: code between matched curly braces { ... }













built-in operators on numbers: add, sub, mul, div, mod, eq, lt, gt

built-in operators on string values: length, get, getinterval, put (you will revise

your implementation for put operator in Part2).



built-in conditional operators: if, ifelse (you will implement if/ifelse operators in Part2)



built-in loop operator: for (you will implement for operator in Part2).



 stack operators: dup, copy, pop, clear, exch, roll

dictionary creation operator: dict; takes one operand from the operand stack, ignores it, and creates a new, empty dictionary on the operand stack (we will call this psDict)
dictionary stack manipulation operators: begin, end. begin requires one dictionary operand on the operand stack; end has no operands.
name definition operator: def.
defining (using def we will call this psDef) and calling functions
stack printing operator (prints contents of stack without changing it): stack



Part 2 - Requirements




In Part 2 you will continue building the interpreter, making use of everything you built in Part 1. The pieces needed to complete the interpreter are:




Revising the string put operator.



Parsing “Simple Postscript” code
Handling of code-arrays
Handling the if and ifelse operators (write the Python methods psIf and psIfelse)
Handling the for operator (write the Python method psFor)
Function calling
Interpreting input strings (code) in the simple Postscript language.



Revise the string put operator



Remember that the put operator gets a string, an index (integer), and an ASCII character (from the stack), and replaces the character at index with the new character in the string. Revise your string put operator implementation from part-1 as follows:




When a string is updated by the “put” operator, all copies of the same string (i.e., the strings that have the same object-id) in the opstack and the dictstack should be updated. Since Python strings are immutable, rather than changing the value of the string itself, you should update each stack entry in the opstack and the dictstack that refer to the same string with the updated string value.

Note: In Python, each object has an associated id which can be retrieved using the id() method (for example when s=’355’, id(s) will give the unique id for the specified string object.)



The above approach is not exactly replicating Postscript put. However, we are simplifying the language to make the implementation easier.




You can unit test your put implementation using the following function:




def testPut():

opPush("(This is a test _)")

dup()

opPush("/s")




exch()

psDef()

dup()

opPush(15)

opPush(48)

put()

if lookup("s") != "(This is a test 0)" or opPop()!= "(This is a test 0)":

return False

return True







2. Parsing




Parsing is the process by which a program is converted to a data structure that can be further processed by an interpreter or compiler. To parse the SPS programs, we will convert the continuous input text to a list of tokens and convert each token to our chosen representation for it. In SPS the tokens are: numbers with optional negative sign, multi-character names (with and without a preceding /), string constants enclosed in parenthesis (i.e., ( ) ) and the curly brace characters (i.e., “}” and “{“). We've already decided about how some of these will be represented: numbers as Python numbers, names as Python strings, booleans as Python booleans, string constants as Python strings, etc. For code-arrays, we will represent things falling between the braces using Python lists.




3-6. Handling of code-arrays: if/ifelse, for operators, and function calling




Recall that a code-array is pushed on the stack as a single unit when it is read from the input. Once a code-array is on the stack several things can happen:




if it is the top item on the stack when a def is executed, it is stored as the value of the name defined by the def.



if it is the body part of an if/ifelse operator, it is recursively interpreted as part of the evaluation of the if/ifelse. For the if operator, the code-array is interpreted only if the “condition” argument for if operator is true. For the ifelse operator, if the “condition” argument is true, first code-array is interpreted, otherwise the second code-array is evaluated.



if it is the body part of a for operator, it is recursively interpreted as part of the evaluation of the for loop. At each iteration of the for loop the loop index is pushed onto the stack.
finally, if when a name is looked up you find that its value is a code-array, the code-array is recursively interpreted.
(We will get to interpreting momentarily).




7. Interpreter




A key insight is that a complete SPS program is essentially a code-array. It doesn't have curly braces around it but it is a chunk of code that needs to be interpreted. This suggests how to proceed:




Convert the SPS program (a string of text) into a list of tokens and code-arrays.
Define a Python function interpret that takes one of these lists as input and processes it.
- Interpret the body of the if/ifelse, and for operators recursively.




When a name lookup produces a code-array as its result, recursively interpret it, thus implementing Postscript function calls.



Implementing Your Postscript Interpreter




I. Parsing




Parsing converts an SPS program in the form a string to a program in the form of a code-array. It will work in two stages:




1. Convert all the string to a list of tokens.




Given:

"/square {dup mul} def 0 1 1 5 {square add} for 55 eq stack"




will be converted to




['/square', '{', 'dup', 'mul', '}', 'def', '0', '1', '1', '5', '{', 'square', 'add', '}', 'for', '55', 'eq', 'stack']




Use the following code to tokenize your SPS program.




import re

def tokenize(s):

return re.findall("/?[a-zA-Z()][a-zA-Z0-9_()]*|[-]?[0-9]+|[}{]+|%.*|[^ \t\n]", s)
Important note: To simplify parsing, we will assume that SPS string space characters. (The regular expression in the above tokenize strings that include spaces.)

constant values don’t include any function won’t work with constant



Another tokenize example:

print (tokenize("""




/pow2 {/n exch def

(pow2_of_n_is) dup 8 n 48 add put

1 n -1 1 {pop 2 mul} for

} def




(Calculating_pow2_of_9) dup 20 get 48 sub pow2 stack

"""




))

returns




['/pow2', '{' , '/n', 'exch', 'def', '(Pow2_of_n _is)' , 'dup', '8', 'n', '48', 'add', 'put', '1', 'n', '-1', '1' , '{', 'pop', '2', 'mul', '}', 'for', '}' , 'def', '(Calculating_pow2_of_9)', 'dup', '20', 'get', '48', 'sub', 'pow2', 'stack']




2. Convert the token list to a code-array



The output of tokenize isn't fully suitable because things between matching curly braces are not themselves grouped into a code-array. We need to convert the output for the above example to:




['/pow2', ['/n', 'exch', 'def', '(Pow2_of_n_is)', 'dup', 8, 'n', 48, 'add',

'put', 1, 'n', -1, 1, ['pop', 2, 'mul'], 'for'], 'def',

'(Calculating_pow2_ of_ 9)', 'dup', 20, 'get', 48, 'sub', 'pow2', 'stack']




Notice how in addition to grouping tokens between curly braces into lists, we've also converted the strings that represent numbers to Python numbers, and the strings that represent booleans to Python boolean values. We kept the parenthesis delimiters for SPS string constants.




The main issue in how to convert to a code-array is how to group things that fall in between matching curly braces. There are several ways to do this. One possible way is find the matching opening and closing parenthesis (“{“ and “}”) recursively, and including all tokens between them in a Python list.




Here is some starting code to find the matching parenthesis using an iterator. Here we iterate over the characters of a string (rather than a list of tokens) using a Python iter and we try to find the matching curly braces. This code assumes that the input string includes opening and closing curly braces only (e.g.,




“{{}{{}}}”)




# The it argument is an iterator. The sequence of return characters should # represent a string of properly nested {} parentheses pairs, from which # the leasing '{' has been removed. If the parentheses are not properly # nested, returns False.




def groupMatching1(it):

res = []




for c in it:

if c == '}':

return res

else:




Note how we use a recursive call to group the inner matching
parenthesis string and append it as a whole to the list we are
constructing. Also note how we have already seen the leading
'{' of this inner group and consumed it from the iterator.
res.append(groupMatching1(it))




return False




# Function to parse a string of { and } braces. Properly nested parentheses

# are arranged into a list of properly nested lists.

def group(s):




res = []

it = iter(s)

for c in it:




if c=='}': #non matching closing parenthesis; return false return False




else:

res.append(groupMatching1(it))

return res