CS 3723 Programming Languages

Assignments:

• Quadruple Interpreter/Translator
• FSM Homework
• Initial Scanner

Handouts:

• A scanner in c for c-style comments, based on a FSM.
  A similar scanner in Java.
• Bare parser for simple arithmetic expressions, with extra code to print a trace of the parse.
• A parser with extra code to translate to Reverse Polish Notation (RPN).

Topics:

• Contrasts:
  • Syntax versus semantics.
  • Run time versus compile time.
  • Translation versus interpretation.
• Overview of a compiler (text, pages 10 and 13):
  • Lexical analysis.
  • Syntax analysis.
  • Semantic analysis.
  • Intermediate code generation.
  • Code optimization.
  • Code generation.
• Quadruples:
  • Basic form: operator, argument 1, argument 2, result.
  • Coded as four integers.
  • The types of instructions:
    • Arithmetic: ADD, SUB, MUL, DIV, MOD, ABS, CHS.
    • Jumps: JMP, JEQ, JNE, JGE, JGT, JLE, JLT.
    • Assignment: ASG.
    • I/O: WRT, RDM.
    • Halt execution: HLT.
    • Literal (for constants): LIT.
  • The form of a quadruple interpreter and of a quadruple translator.
• Finite-state machines (FTMs):
  • Diagrams with states (circles), transitions (arrows), and labels on arrows (character absorbed in following the transition). There is also exactly one start state and one or more final (or terminal states).
  • Non-deterministic Finite Automata (NFAs). This is a FSM with at least one case of several arrows labeled with the same symbol.
  • Deterministic Finite Automata (DFAs). Here there is a unique arrow for each symbol.
  • Algorithm for converting an NFA to a DFA. The new states in the constructed DFA are sets of states of the old NFA.
  • The language described by a DFA (= all possible sequences of characters absorbed by the machine when it ends up in a final state). We also say the language recognized by a DFA.
• Use of FSMs in constructing a scanner:
  • Simulate the FSM for a scanner. Usually based on a DFA, though it is possible to simulate an NFA. (Just keep track of a set of states at each step.)
  • Construct a simulator for a DFA by letting each state be a label's location, and each transition a goto, conditional on the character (or type of symbol) above the transition.
  • Simulate the same DFA using a state variable and a while and switch to pick off the correct code at each step, corresponding to the state number.
  • Examples, especially for c-style comments and for the language of our term project.
• Formal grammars and formal languages:
  • Terminology: Terminal (symbol), non-terminal (symbol), meta-symbol, start symbol, grammar rule, left side (= single non-terminal) and right side (= sequence of terminals and non-terminals), a formal grammar (= finite set of rules).
  • Formal grammars are also called BNF, Backus-Naur, and Context-free grammars. (``Context-free'' because no part depends on the context in which it occurs.)
  • Derivation sequences, starting with the start symbol and ending with a sequence of terminals (called a sentence). At each step of the derivation, a single non-terminal is replaced using a grammar rule with that non-terminal on its left side. The non-terminal is replaced by the right side.
  • Language associated with a grammar (= all strings of terminals that are generated by some derivation sequence). We also say language recognized by the grammar.
  • Leftmost derivations (leftmost non-terminal replaced at each stage). Rightmost derivations.
  • Parse tree for a sentence, one tree for each derivation. There can be several derivations for a given tree, but only one leftmost and one rightmost derivation.
  • Ambiguity of a grammar. (There is a sentence with two distinct parse trees, or equivalently, with two distinct leftmost derivations.)
  • Unambiguous grammars. (Each sentence has a unique parse tree, or equivalently a unique leftmost derivation.)
  • Disambiguating a grammar. Two ways: use special rules, or change the grammar.
  • Example: simple sentences in English, with noun, verb, etc.
  • Example: various grammars for simple forms of arithmetic expressions.
  • Example: simple grammar for if-then with optional else. In its simple form it is ambiguous, but a special rule disambiguates it.
• Parsers:
  • A program that constructs a derivation sequence for an input sentence, or that constructs a parse tree (or that goes through the motions of doing these things).
  • Top-down parser: constructs leftmost derivation from the start symbol, aiming toward the given sentence.
  • Recursive descent parser: can be coded by hand and is useful for compilers. One writes a recursive function for each grammar rule. The code of the function mirrors the right side of the rule. Not all grammars can be handled by recursive descent, so one might have to rewrite the grammar.
  • Bottom-up parser: constructs a rightmost derivation in reverse, from the input sentence, aiming toward the start symbol. These parsers are usually driven by a table that is constructed separately.
• Use of parsers to construct translators:
  • Syntax-directed translation: Add code to a parser so as to carry out a translation from one language to another.
  • Example that translates from arithmetic expressions to rpn.

Sample Questions:

• About FSMs:
  • Given a diagram for a specific DFA or NFA, describe the language recognized by the machine.
  • Given a diagram for a specific NFA, construct
• Quadruples:
  • Explain how execution of a specific sequence of quadruples would work.
• Scanners:
  • Examples of the kind of DFA that would be used to implement a scanner.
  • Given a simple DFA, write a simulator for it.
  • Some specific question about your particular scanner code written for the course assignment.
• Formal grammars:
  • Given a specific grammar, say whether or not it is ambiguous.
  • Given a specific grammar, describe the language recognized by it.
  • Given a specific grammar and a specific sentence, construct:
    • A leftmost derivation for the sentence.
    • A parse tree for the sentence.
  • Given a description of a language, construct a grammar for it.