CSCI-E113
Finite State Machines
_________________________________________________________________
0. Introduction: Processing Text
A lot of Unix programming involves working with text.
Unix text processing programs read data in, do something
with the data, and write data out. Text files have struc-
ture, the program has to extract meaning and contentfrom
that structure. Forexample, the train schedule file has
the following structure:
a. Each line represents a train stopping at a station.
b. Each record contains six pieces of information.
c. The pieces of information are separated by semicolons.
d. The pieces of information are identified by a tag and an equals sign.
A program that wants to get meaningful content fromthis
file has to be able to process lines, fields, tags, and
values. Spotting these components requires identifying the
different sections of each line. A programming idea that is
extremely helpful for spotting sections of a pattern is the
idea of modes or states.
We begin with an example.
1. Capitalizing Schedule Data
Consider typical lines of schedule data:
TR=002;dir=i;day=m-f;TI=5:25;stn=midd./lakeville;Line=middle
TR=002;dir=i;day=m-f;TI=5:35;stn=bridgewater;Line=middle
The names of the stops and the names of the lines are all
storedas lower case strings. When we present data to
users, it would look better if the names had upper case ini-
tial letters.
The program capitalize2.c reads text input, capitalizes any
letterthat follows a non-letter, and writes the resulting
text to standard output. Here is the code:
CSCI-E113 Handout September 21, 2008 page 1
Finite State Machines
#include <stdio.h>
#include <ctype.h>
/*
* capitalize
* purpose: convert leading chars to upper case
* usage: capitalize < input > output
*/
#define NOCHANGE_MODE0
#define UPPERCASE_MODE1
main()
{
intc;
intmode = NOCHANGE_MODE;
while( ( c=getchar() ) != EOF ){
if ( mode == UPPERCASE_MODE )
c = toupper(c);
if ( c == '=' || c == ' ' || c == '/' )
mode = UPPERCASE_MODE;
if ( isalpha(c) )
mode = NOCHANGE_MODE;
putchar(c);
}
}
This program processes the text one characterat a time.
When the program sees an equals sign, a space, or a slash,
it sets the variable called mode to UPPERCASE_MODE to remind
itself to capitalize the next character it processes. Then,
after processing a character, it sets that variable back to
NOCHANGE_MODE.
A useful way to think about this behavior is to say the pro-
gram operatesin two modes: nochange and uppercase. In
nochange mode, the program reads in characters and writes
them out. In uppercase mode, the program reads in charac-
ters, converts them to upper case, then writes themout.
The program switches from one mode to another based on data
it sees. When the program sees an equals, space, or slash,
the program changes into uppercase mode. When it sees an
alphabetic character, it changes back to nochange mode.
Some programs operate in more than two modes.
2. Example 2 - Word Count
The Unix tool called wc counts words, characters, and lines
in a file. Counting characters and lines is easy, counting
words is easily done with a program that uses a mode vari-
able. Consider input that looks like:
CSCI-E113 Handout September 21, 2008 page 2
Finite State Machines
this some
words is.a.test/of to-be-counted.
The following code can handle the wide spaces and punctua-
tion.
#include <stdio.h>
#include <ctype.h>
/* wordcount.c - count words in input */
#define IN_A_WORD1
#define BETWEEN_WORDS2
main()
{
intc;
intmode = BETWEEN_WORDS;
intnumwords = 0;
while( (c=getchar()) != EOF )
{
if ( isalpha(c) && mode == BETWEEN_WORDS ){
numwords++;
mode = IN_A_WORD;
}
else if ( !isalpha(c) && mode == IN_A_WORD ){
mode = BETWEEN_WORDS;
}
}
printf("%d\n", numwords);
}
Trace through this code processing the sample data. The
program spotsthe beginning of a word by keeping track of
whether it is processing a word or if it is between words.
If the program is between words, then the appearance of an
alphabetic character indicates the beginning of a new word.
Once the program has seen that initial letter and has incre-
mented the word counter, it sets the mode so it knows it is
already in a word. In that mode, alphabetic characters are
considered just part of the same word. Whenit finds a
non-alphabetic character, it changes the mode to reflect the
fact it is no longer processing a word.
3. Example 3 - Passing Mode
On some roads one sees signs that say "No Passing". Later
on one sees a sign saying "Passing Allowed".As you drive
along that road, you change from one mode to another.When
you are in Passing Mode, you respond to slow drivers by
passing them. When you are in NoPassing Mode,you respond
to slow drivers by following along behind them. The signs
CSCI-E113 Handout September 21, 2008 page 3
Finite State Machines
cause you to change from one mode to another.
The world is full of real-life situations where you, a
human,operate in differentmodes. In many cases, the
change from one mode to another is triggeredby specific
inputs. In the driving example, the inputs are visual input
in the form of signs.
4. Programming Finite State Machines
The capit...