Chapter 3

Lexical Analysis

Compiler Design by Varun Arora 1

Outline
 Role of lexical analyzer
 Specification of tokens
 Recognition of tokens
 Lexical analyzer generator
 Finite automata
 Design of lexical analyzer generator

Compiler Design by Varun Arora 2

 The role of lexical analyzer
token
Source Lexical To semantic
program Parser analysis
Analyzer
getNextToken

Symbol
table

Compiler Design by Varun Arora 3

Why to separate Lexical analysis
and parsing
1. Simplicity of design
2. Improving compiler efficiency
3. Enhancing compiler portability

Compiler Design by Varun Arora 4

Tokens, Patterns and Lexemes
 A token is a pair a token name and an optional token
value
 A pattern is a description of the form that the lexemes
of a token may take
 A lexeme is a sequence of characters in the source
program that matches the pattern for a token

Compiler Design by Varun Arora 5

Example
Token Informal description Sample lexemes
if Characters i, f if
else Characters e, l, s, e else
comparison < or > or <= or >= or == or != <=, !=

id Letter followed by letter and digits pi, score, D2

number Any numeric constant 3.14159, 0, 6.02e23
literal Anything but “ sorrounded by “ “core dumped”

printf(“total = %d\n”, score);

Compiler Design by Varun Arora 6

Attributes for tokens
 E = M * C ** 2
 <id, pointer to symbol table entry for E>
 <assign-op>
 <id, pointer to symbol table entry for M>
 <mult-op>
 <id, pointer to symbol table entry for C>
 <exp-op>
 <number, integer value 2>

Compiler Design by Varun Arora 7

Lexical errors
 Some errors are out of power of lexical analyzer to
recognize:
 fi (a == f(x)) …
 However it may be able to recognize errors like:
 d = 2r
 Such errors are recognized when no pattern for tokens
matches a character sequence

Compiler Design by Varun Arora 8

Error recovery
 Panic mode: successive characters are ignored until we
reach to a well formed token
 Delete one character from the remaining input
 Insert a missing character into the remaining input
 Replace a character by another character
 Transpose two adjacent characters

Compiler Design by Varun Arora 9

Input Buffering
 How to speed the reading of source program ?
 to look one additional character ahead
 e.g.
 to see the end of an identifieryou must see a character
 which is not a letter or a digit
 not a part of the lexeme for id
 in C
 -,= , <
 ->, ==, <=
 two buffer scheme that handles large lookaheadssafely
 sentinels –improvement which saves time checking
buffer ends
Compiler Design by Varun Arora 10
Buffer pairs
 Buffer size N
 N = size of a disk block (4096)
 read N characters into a buffer
 one system call
 not one call per character
 read < N characters we encounter eof
 two pointers to the input are maintained
 lexemeBegin–marks the beginning of the current lexeme
 forward–scans ahead until a pattern match is found

Compiler Design by Varun Arora 11

Sentinels
 Forward pointer
 to test if it is at the end of the buffer
 to determine what character is read (multiwaybranch)
 sentinel
 added at each buffer end
 can not be part of the source program
 character eofis a natural choice
 retains the role of entire input end
 when appears other than at the end of a buffer it means
that the input is at an end

Compiler Design by Varun Arora 12

Sentinels
E = M eof * C * * 2 eof eof

Switch (*forward++) {
case eof:
if (forward is at end of first buffer) {
reload second buffer;
forward = beginning of second buffer;
}
else if {forward is at end of second buffer) {
reload first buffer;\
forward = beginning of first buffer;
}
else /* eof within a buffer marks the end of input */
terminate lexical analysis;
break;
cases for the other characters;
} Compiler Design by Varun Arora 13
Specification of tokens
 In theory of compilation regular expressions are used
to formalize the specification of tokens
 Regular expressions are means for specifying regular
languages
 Example:
 Letter_(letter_ | digit)*
 Each regular expression is a pattern specifying the
form of strings

Compiler Design by Varun Arora 14

Regular expressions
 Ɛ is a regular expression, L(Ɛ) = {Ɛ}
 If a is a symbol in ∑then a is a regular expression, L(a)
= {a}
 (r) | (s) is a regular expression denoting the language
L(r) ∪ L(s)
 (r)(s) is a regular expression denoting the language
L(r)L(s)
 (r)* is a regular expression denoting (L9r))*
 (r) is a regular expression denting L(r)

Compiler Design by Varun Arora 15

Regular definitions
d1 -> r1
d2 -> r2
…
dn -> rn

 Example:
letter_ -> A | B | … | Z | a | b | … | Z | _
digit -> 0 | 1 | … | 9
id -> letter_ (letter_ | digit)*

Compiler Design by Varun Arora 16

Extensions
 One or more instances: (r)+
 Zero of one instances: r?
 Character classes: [abc]

 Example:
 letter_ -> [A-Za-z_]
 digit -> [0-9]
 id -> letter_(letter|digit)*

Compiler Design by Varun Arora 17

Recognition of tokens
 Starting point is the language grammar to understand
the tokens:
stmt -> if expr then stmt
| if expr then stmt else stmt
|Ɛ
expr -> term relop term
| term
term -> id
| number

Compiler Design by Varun Arora 18

Recognition of tokens (cont.)
 The next step is to formalize the patterns:
digit -> [0-9]
Digits -> digit+
number -> digit(.digits)? (E[+-]? Digit)?
letter -> [A-Za-z_]
id -> letter (letter|digit)*
If -> if
Then -> then
Else -> else
Relop -> < | > | <= | >= | = | <>
 We also need to handle whitespaces:
ws -> (blank | tab | newline)+

Compiler Design by Varun Arora 19

Transition diagrams
 Transition diagram for relop

Compiler Design by Varun Arora 20

Transition diagrams (cont.)
 Transition diagram for reserved words and identifiers

Compiler Design by Varun Arora 21

Transition diagrams (cont.)
 Transition diagram for unsigned numbers

Compiler Design by Varun Arora 22

Transition diagrams (cont.)
 Transition diagram for whitespace

Compiler Design by Varun Arora 23

Architecture of a transition-
diagram-based lexical analyzer
TOKEN getRelop()
{
TOKEN retToken = new (RELOP)
while (1) { /* repeat character processing until a
return or failure occurs */
switch(state) {
case 0: c= nextchar();
if (c == ‘<‘) state = 1;
else if (c == ‘=‘) state = 5;
else if (c == ‘>’) state = 6;
else fail(); /* lexeme is not a relop */
break;
case 1: …
…
case 8: retract();
[Link] = GT;
return(retToken);
}
Compiler Design by Varun Arora 24
Lexical Analyzer Generator - Lex
Lex Source program Lexical [Link].c
lex.l Compiler

[Link].c
C [Link]
compiler

Sequence
Input stream [Link]
of tokens

Compiler Design by Varun Arora 25

Structure of Lex programs

declarations
%%
translation rules Pattern {Action}
%%
auxiliary functions

Compiler Design by Varun Arora 26

Example
%{
Int installID() {/* funtion to install the
/* definitions of manifest constants
lexeme, whose first character is
LT, LE, EQ, NE, GT, GE, pointed to by yytext, and whose
IF, THEN, ELSE, ID, NUMBER, RELOP */ length is yyleng, into the symbol
%} table and return a pointer thereto
*/
/* regular definitions }
delim [ \t\n]
ws {delim}+ Int installNum() { /* similar to
installID, but puts numerical
letter [A-Za-z]
constants into a separate table */
digit [0-9]
}
id {letter}({letter}|{digit})*
number {digit}+(\.{digit}+)?(E[+-]?{digit}+)?

%%
{ws} {/* no action and no return */}
if {return(IF);}
then {return(THEN);}
else {return(ELSE);}
{id} {yylval = (int) installID(); return(ID); }
{number} {yylval = (int) installNum(); return(NUMBER);}
…
Compiler Design by Varun Arora 27
Finite Automata
 Regular expressions = specification
 Finite automata = implementation

 A finite automaton consists of

 An input alphabet 
 A set of states S
 A start state n
 A set of accepting states F  S
 A set of transitions state input state

Compiler Design by Varun Arora 28

Finite Automata
 Transition
s1 a s2
 Is read
In state s1 on input “a” go to state s2

 If end of input
 If in accepting state => accept, othewise => reject
 If no transition possible => reject

Compiler Design by Varun Arora 29

Finite
 A state
Automata State Graphs

• The start state

• An accepting state

a
• A transition

Compiler Design by Varun Arora 30

A ASimple Example
finite automaton that accepts only “1”

 A finite automaton accepts a string if we can follow

transitions labeled with the characters in the string
from the start to some accepting state

Compiler Design by Varun Arora 31

Another Simple Example
 A finite automaton accepting any number of 1’s
followed by a single 0
 Alphabet: {0,1}

 Check that “1110” is accepted but “110…” is not

Compiler Design by Varun Arora 32

And Another
 Alphabet {0,1}
Example
 What language does this recognize?

1 0

0 0

1
1

Compiler Design by Varun Arora 33

And Another Example
 Alphabet still { 0, 1 }
1

 The operation of the automaton is not completely

defined by the input
 On input “11” the automaton could be in either state

Compiler Design by Varun Arora 34

Epsilon Moves
 Another kind of transition: -moves

A B

• Machine can move from state A to state B

without reading input

Compiler Design by Varun Arora 35

Deterministic and
Nondeterministic Automata
 Deterministic Finite Automata (DFA)
 One transition per input per state
 No -moves
 Nondeterministic Finite Automata (NFA)
 Can have multiple transitions for one input in a given
state
 Can have -moves
 Finite automata have finite memory
 Need only to encode the current state

Compiler Design by Varun Arora 36

Execution of Finite Automata
 A DFA can take only one path through the state graph
 Completely determined by input

 NFAs can choose

 Whether to make -moves
 Which of multiple transitions for a single input to take

Compiler Design by Varun Arora 37

Acceptance of NFAs
 An NFA can get into multiple states
1

0 1

• Input: 1 0 1

• Rule: NFA accepts if it can get in a final state

Compiler Design by Varun Arora 38

NFA vs. DFA (1)
 NFAs and DFAs recognize the same set of languages
(regular languages)

 DFAs are easier to implement

 There are no choices to consider

Compiler Design by Varun Arora 39

NFA vs. DFA (2)
 For a given language the NFA can be simpler than the
DFA
1
0 0
NFA
0

1 0
0 0
DFA
1
1

• DFA can be exponentially larger than NFA

Compiler Design by Varun Arora 40

Regular Expressions to Finite
Automata
 High-level sketch

NFA

Regular
expressions DFA

Lexical Table-driven
Specification Implementation of DFA

Compiler Design by Varun Arora 41

Regular Expressions to NFA (1)
 For each kind of rexp, define an NFA
 Notation: NFA for rexp A

• For 


• For input a
a

Compiler Design by Varun Arora 42

Regular Expressions to NFA (2)
 For AB
A 
B

• For A | B


B 

 A

Compiler Design by Varun Arora 43

Regular Expressions to NFA (3)
 For A*


A


Compiler Design by Varun Arora 44

Example of RegExp -> NFA
conversion
 Consider the regular expression
(1 | 0)*1
 The NFA is


 C
1 
E
1
 0
A B G H  I J
 D F 


Compiler Design by Varun Arora 45

Next
NFA

Regular
expressions DFA

Lexical Table-driven
Specification Implementation of DFA

Compiler Design by Varun Arora 46

NFA to DFA. The Trick
 Simulate the NFA
 Each state of resulting DFA
= a non-empty subset of states of the NFA
 Start state
= the set of NFA states reachable through -moves from
NFA start state
 Add a transition S a S’ to DFA iff
 S’ is the set of NFA states reachable from the states in S
after seeing the input a
 considering -moves as well
Compiler Design by Varun Arora 47
NFA -> DFA Example


 C 1 E 
1
 0
A B G H  I J
 D F 

0
0 FGABCDHI
0 1
ABCDHI
1
1 EJGABCDHI

Compiler Design by Varun Arora 48

NFA to DFA. Remark
 An NFA may be in many states at any time

 How many different states ?

 If there are N states, the NFA must be in some subset

of those N states

 How many non-empty subsets are there?

 2N - 1 = finitely many, but exponentially many

Compiler Design by Varun Arora 49

Implementation
 A DFA can be implemented by a 2D table T
 One dimension is “states”
 Other dimension is “input symbols”
 For every transition Si a Sk define T[i,a] = k
 DFA “execution”
 If in state Si and input a, read T[i,a] = k and skip to state
Sk
 Very efficient

Compiler Design by Varun Arora 50

Table Implementation of a DFA
0
0 T
0 1
S
1
1 U

0 1
S T U
T T U
U T U

Compiler Design by Varun Arora 51

Implementation (Cont.)
 NFA -> DFA conversion is at the heart of tools such as
flex or jflex

 But, DFAs can be huge

 In practice, flex-like tools trade off speed for space in

the choice of NFA and DFA representations

Compiler Design by Varun Arora 52

Readings
 Chapter 3 of the book

Compiler Design by Varun Arora 53