Need asnwer for D and E only

4. When describing the structure of a programming language, we often use the formalisms of regular languages (to represent legal tokens) and context-free languages (to represent legal structures built up of these tokens). However, the formal machines that recognize these language families (finite-state automata and push-down automata) aren't immediately useful for scanning and parsing, respectively. Answer the following questions concerning how these machines are adapted for use in a compiler: (a) (2 points) To start off, the automata are recognizers: they simply say whether a string is in the language or not. What additional actions are needed for the tasks of scanning and parsing? (b) (2 points) A program usually consists of more than one token, and yet a finite-state automaton only recognizes a single string. How can a compiler use a finite state automaton? (c) (2 points) Tokens are often not self-delimiting. For example, the string a 2i could be a single identifier token or could be three tokens (the identifier a followed by the number 2 followed by the identifier i). How is this ambiguity resolved in a compiler? PhD (Day 2) Qualifying Exam (d) (2 points) The grammar used for parsing a language is usually not convenient for describing program structure to the rest of the compiler. Give an example grammar for arithmetic expressions with addition and multiplication that is acceptable to a top-down (that is, predictive or LL) parser, but which does not give the "true" structure of the language. Explain why the need for the grammar to be top-downparseable prevents it from intuitively describing the structure of expressions. (e) (2 points) Given a parser-generator such as yacc or bison, how do we handle the situation that the grammar used for parsing is not convenient for performing static semantic checking