CSCI-E113:
Class 4
Strings, CGI, Structs
0. Introduction
Last week we continued to extend our HTML and shell skills
as we improved our train schedule web site. On the C end of
the course, we looked at arrays in general and strings in
particular. We saw how to write programs that worked with
arrays of strings.
This week we continue to improve our website and extend our
C skills.
The outline for tonight's class is:
1. Quick review of the big picture
2. Strings can be dangerous
3. HTML and CGI scripts: select, case, generating HTML
4. A Bigger Project: trip planner
5. structs -- the basics
6. structs -- arrays to model train runs
1. Quick Review
The big picture for C/Unix/CGI programs is:
user --> html form --> CGI connector --> shell script --> tools
2. Strings Can Be Dangerous
Most of the data flowing through this sequence of connec-
tions is plain text. C programs store text in strings --
arrays of characters terminated by a nul character. Working
with strings in C is not as easy as it is in other
languages. There are two main types of problems.
String Operations Require Functions
The first problem new C programmers face with strings is
that simple assignment, comparison, and concatenation opera-
tors do not work with strings.For example:
char s[10], t[10], u[10];
s = "abc";
t = s;
u = "xyz";
if ( s == t )...
t = u + s;
do not work. Some do not compile, and some compile but do
not do what you want.
Last Update: 11/14/0-9November 14, 2005 page 1
class 4: strings, cgi, structs
Strings are arrays, and loading, comparing, or append-
ing arrays require looping through elements. These func-
tions are performed with the functions:
strcpy(dest, src)
strcmp(str1, str2)
strcat(dest, newstr)
Read the man page onthese functions for details. We
explore the program called string-ops.c to see how these
operations work and how the functions work.
String Overruns
The most common C programming error is probably unchecked
bufferoverflows. These bugs cause programs to crash and
allow computervandals to break into computer systems.
Understanding why they occur and how to prevent them is an
essential skill. We explore two example programs:
memdemo1.c
#include <stdio.h>
/*
* memdemo - show how array over-runs are easy
* variables are allocated on the stack in order of request
*/
main()
{
char c;
int i;
char a[6];
char b[6];
while(1){
printf("c=%c i=%d a=%s b=%s a[-6]=%c b[10]=%c\n",
c, i, a, b, a[-6], b[10]);
printf("enter a: ");
scanf("%s", a);
printf("c=%c i=%d a=%s b=%s a[-6]=%c b[10]=%c\n",
c, i, a, b, a[-6], b[10]);
printf("enter b: ");
scanf("%s", b);
printf("c=%c i=%d a=%s b=%s a[-6]=%c b[10]=%c\n",
c, i, a, b, a[-6], b[10]);
printf("enter i: ");
scanf("%d", &i);
}
}
Notice that this function contains four variables, two of
them char arrays. What happens to these variables if we
enter more characters than the length of the array?Data
from one string appears in other variables! What is going
on?
Last Update: 11/14/0-9November 14, 2005 page 2
class 4: strings, cgi, structs
Local variables are stored in memory on the stack.
When a function begins, space needed for the variables is
allocated on the stack by moving the stack pointer down the
correct amount. When the function ends, the stack pointer
moves back up, effectively deallocating the local variables,
although leaving their contents unchanged.
Most C compilers allocate variables in the order requested.
Therefore the six chars for a appear above the six chars for
b. If we put too many characters in b, they run overinto
a. If we put enough extra characters into b, they run all
the way into c and i and maybe farther.
memdemo2.c:
#include <stdio.h>
/*
* memdemo2.c -- show how variables in functions can affect
* variables in calling functions
*/
main()
{
char a[10] = "abc";
printf("before function call: a is %s\n", a);
f();
printf("after function call: a is %s\n", a);
}
f()
{
char b[3];
int i;
int c;
printf("in f, please enter string for b: ");
for(i=0; (c=getchar())!='\n'; i++)
b[i] = c;
b[i] = ' ';
printf("in f, b is now %s\n",b);
}
What about functions and variables? Can overruns in a func-
tion affectmemory in the calling function?Yes.
memdemo2.c shows this.
Preventing Overruns
Only you can prevent buffer overruns. Always check indexes
as you load arrays.Never assume the size will be 'large
enough' and never assume the source of data will know how
much space is available.
Last Update: 11/14/0-9November 14, 2005 page 3
class 4: strings, cgi, structs
3. HTML and CGI scripts: select, checkboxes, case,
generating HTML
We now return to our train schedule website. As we did last
time, we list some improvements to make, and then we learn
new techniques that allow us to make the improvements.
We now have three CGI programs, one to find times of trains
through a station, one to print a schedule, and one to print
the names of stations.We continue toimprove this site.
This week we add the following:
1. Nicer Reports -- use HTML on the response page
2. Easier Input -- User selects a station from a list
3. Better Integration -- include hyperlinks on response page
3.1. Nicer Reports -- use HTML on response
We start with trainsched2.cgi. This pagereturns the
schedule for a specified train number.The output is simple
and clear: a list of times and stations in the right order.
But the plain text format is boring. We can change this to
send back html text with tags to set the color, drawhor-
izontal lines, and make sure each stop is on its own line.
#!/bin/sh
#
# trainsched3.cgi
# connector for getting train schedule
# v3:added html format
# v2:this one uses and if then else to capitalize or not
#
eval `./qryparse`
TITLE="Schedule for Train $trainnumber"
# announce countent type
echo "Content-type: text/html"
echo ""
# start of html, include title and body start
echo "<html>
<head><title>$TITLE</title></head>
<body bgcolor='darkblue' text='white'>
<B>$TITLE</B>
<hr>
"
# content here
if test "$caps" = "y"
then
./trainsched2 $trainnumber | ./capitalize | sed 's/$/<br>/'
Last Update: 11/14/0-9November 14, 2005 page 4
class 4: strings, cgi, structs
else
./trainsched2 $trainnumber | sed 's/$/<br>/'
fi
# footer and end
echo "<hr>"
echo "<small>Generated by <i>$0</i> on `date`</small>"
echo "</body></html>"
The new techniques here include:
a. Setting a variable TITLE to be used later in the script
b. The sed command to put a new tag at the end of each line
The only complicated item is the sed command.
Sed is a "stream editor". It is designed tobe used in
pipelines to modify text as it flows through the pipeline.
A common use of sed is for text replacement. The syntax:
s/original/replacement/
is the substitute command. The symbol $ stands for end
of line.
Running the Program from the Browser Command Line
Before we move on, look carefully at the location line in
the browser.This line contains the name of the computer,
the name of the program, and the variables as name=value
pairs. We do not have to use a browser form to fill in
those values. Instead, we can type the values directly into
the little window and run the program. The browser simply
prepares that string and sendsthe string to the remote
machine. We could send that string without using an html
form. Remember this fact. It is very important.
Nicer Output 2 -- Improve train-times
We now look at the train-times page. The user enters a sta-
tion name, and the program lists all the stops at that sta-
tion. There are two problems -- input and output. For
input, the user has to know the correct spelling of the sta-
tion. It would be nicer to allow the user to pick a station
from a list.
On the output side, it would be nicer to list just theuse-
ful information: train number, time, station. We know how
to do that from our work with trainsched. We spruce up the
shell script so it looks like:
#!/bin/sh
#
# train-times3
# purpose: list train times for a station
# usage: train-times3 stationname direction [day]
Last Update: 11/14/0-9November 14, 2005 page 5
class 4: strings, cgi, structs
#where: direction is "i" or "o", day is "m-f" or "sa" or "su"
#
# 1st - check there are two arguments or more
if test $# -lt 2
then
echo "usage: train-times3 station direction [day]"
exit 1
fi
STATION=$1
DIR=$2
grep "stn=$STATION" sched | grep "dir=$DIR" | grep "day=$3" |
cut -d";" -f1,4,5 | ./semi2tab3 -t | sort -k2n
and then we spruce up the cgi so it wraps the data inhtml
tags:
#!/bin/sh
# train-times3.cgi
# processing script for train-times3.html
# this one produces html
eval `./qryparse`
# announce type
echo "Content-type: text/html"
echo ""
# build up some strings for response
case $dir in
i) DIR="Inbound" ;;
o) DIR="Outbound" ;;
esac
TITLE="$DIR Trains through $station"
# generate html
echo "<html><head><title>$TITLE</title></head>
<body bgcolor='darkblue' text='white'>
<b>$TITLE</b>
<hr/>" | ./capitalize
./train-times3 "$station" "$dir" "$when" | ./capitalize |
sed 's/$/<br>/'
./footer "$0"
echo "</body></html>"
This example uses the following tools:
a) sed -- to add the <br> tag to end of each line
b) case -- part of the sh language
c) continuation -- if | is the last char, the pipeline continues
d) footer - a short script to append a footer to the page
Last Update: 11/14/0-9November 14, 2005 page 6
class 4: strings, cgi, structs
The case control structure is like switch in C. It
compares a string to a set of patterns. If the string
matches a pattern, the code associated with that pattern is
executed. Thecode for a pattern can contain any number of
comamnds. The end of the code is marked with ;; .
The case control structure works well with radio buttons on
a webpage. The user picks one of several items, and the
script can take different action based on the choice.
3.2. Easier Input -- Select a station from a list
We now modify the formfor train-times so the user can
selecta station from a list rather than have to type the
name. To do this, we replace the textinput area with a
<select> input control. A select has the format:
<select name='NAME'>
<option value='V1'>Str1
<option value='V2'>Str2
...
</select>
The 'value' attribute the of option items is not required.
If omitted, the value of the selection is the string. We
replace the text input with a select, and to demonstrate the
use of the value attribute, we change the when choice to use
a select. The result is:
<html>
<!-- version 3.2 of the train-times html form
version 3.2 allows user to select station from a list -->
<body>
Find train times for a station
<p>
<form action="train-times3.cgi" method="get">
Station:
<select name='station'>
<option>abington
<option>anderson/woburn
<option>andover
...
<option>wyoming hill
<option>yawkey
</select>
<br>
Direction: <input type='radio' name='dir' value='i'>Inbound
<input type='radio' name='dir' value='o'>Outbound
<br>
Day: <select name='when'>
<option value='m-f'>Weekday
<option value='sa'>Saturday
<option value='su'>Sunday
</select>
<br>
<input type="submit">
</form>
Last Update: 11/14/0-9November 14, 2005 page 7
class 4: strings, cgi, structs
<body></html>
3.3. Better Integration -- include hyperlinkson response
page
This page is certainly looking better. But, the page is
cumbersome touse. We pick a station and get a list of
trains. To see the schedule for that page, we then need to
go tothe trainsched.html page to enter the train number.
It would be nicer if we could enter the trainnumberfrom
the result page, oreven better, just click on the train
number.
For our first attempt, we put the form to request a schedule
on the result from the train time searcher. This is train-
times3.2.cgi:
#!/bin/sh
# train-times3.2.cgi
# processing script for train-times3.html
#
# This one produces html AND includes the form for trainsched.html
# as part of the form.
eval `./qryparse`
# announce type
echo "Content-type: text/html"
echo ""
# build up some strings for response
case $dir in
i) DIR="Inbound" ;;
o) DIR="Outbound" ;;
esac
TITLE="$DIR Trains through $station"
# generate html
echo "<html><head><title>$TITLE</title></head>
<body bgcolor='darkblue' text='white'>
<b>$TITLE</b>
<hr/>" | ./capitalize
./train-times3 "$station" "$dir" "$when" | ./capitalize |
sed 's/$/<br>/'
# add form to request train schedule
echo "
<form action='trainsched3.cgi'>
Train number please?
<input type='text' name='trainnumber' size='10'>
<p>
Last Update: 11/14/0-9November 14, 2005 page 8
class 4: strings, cgi, structs
Capitalize Names: <input type='radio' name='caps' value='y'>Yes
<input type='radio' name='caps' value='n'>No
<p>
<input type='submit'>
</form>"
./footer "$0"
echo "</body></html>"
This works better than the old method, but it is still not
the best solution. Think about how to make the train
numbers clickable links that call up the schedule directly.
4. A Bigger Project: A trip Planner
Our site is now looking better, and allows users to search
for trains to places they want to visit. A more powerful
feature would be the trip planner. The user would select a
starting station andan ending station, and the program
would figure out which trains to take to get from one place
to another.
We shall solve this problem over several weeks. We start
with an htmlpage that allows the user to select two sta-
tions and a cgi program and script that tells the user if
the two stations are on the same line.
Here is the form:
<html>
<head><title>Trip Planner Version 1</title></head>
<body>
<center>Trip Planner Version 1</center>
<form action='trip_plan1.cgi'>
Starting station:
<select name='startstn'>
<option>abington
<option>anderson/woburn
...
<option>wyoming hill
<option>yawkey
</select>
Ending station:
<select name='endstn'>
<option>abington
<option>anderson/woburn
...
<option>wyoming hill
<option>yawkey
</select>
<input type='submit' value='find trains'>
</form>
Last Update: 11/14/0-9November 14, 2005 page 9
class 4: strings, cgi, structs
</center>
</body>
</html>
and here is the processing script:
#!/bin/sh
#
# trip_plan1 -- first version of trip planner
# Allow user to specify starting station and ending station
# if both are on same line, then list trains that go from Start to End
# if both are not on same line, then say we cannot do that now
#
eval `./qryparse`
echo "Content-type: text/plain"
echo ""
if test "$startstn" = "$endstn"
then
echo "$startstn is the same as $endstn"
exit 0
fi
./lines_through "$startstn" | sort > lines1.tmp
./lines_through "$endstn" | sort > lines2.tmp
COMMON=`comm -12 lines1.tmp lines2.tmp`
if test -z "$COMMON"
then
echo "$startstn and $endstn are on different lines"
else
echo "$startstn and $endstn are on $COMMON"
fi
STARTLINES=`./lines_through "$START"`
ENDLINES=`./lines_through "$END"`
rm lines1.tmp lines2.tmp
There is not a simple solution to a general trip planner.
To figure out how to get from one place to another, we need
some way to store and process a train line and figure out
when trains get to which stops. There are no standard unix
tools to manage this, so we shall have to write some new C
code.Our new C code will need to record train trips. A
train trip is a sequence of stops. A stop is a station, a
time, a train number. Therefore we need a way to create a
way to store a sequence of stops.
C has a data type perfect for representing a stop: the
struct.
5. Structs - the basics
Say we wanted to record a train trip in a C program. A trip
is a sequence of stops. Our datafile records for each stop
a a train number, a direction, a time, a station, a day, and
Last Update: 11/14/0-9November 14, 2005 page 10
class 4: strings, cgi, structs
a line. Say we wanted to record for each stop the sta-
tion, the direction, the train number, and the time. C pro-
vides the struct asa way to store several values in one
container. We can define a new data type called a struct
tstop by writing this code:
struct tstop {
char station[SLEN]; /* name of station */
char dir; /* direction 'i' or 'o' */
char tnum[NLEN]; /* the train number */
int hr, min; /* stop time */
};
This defines what a struct tstop looks like. This is a
blueprint, not a house. To instantiate actual variables, we
write:
struct tstop s1,s2; /* create trainstops */
And we can work with the members of these structs using:
strcpy(s1.station, "salem");
s1.dir = 'o';
s1.hr = 9;
s1.min = 40;
Operations on Structs
The three basic operations one may perform on structs are:
access Use the dotoperator to select members in a
struct. Inthe examples above, we see how to
refer to a string, a char, and ints.The name of
the struct is on the left, and the name of the
member is on the right.
assign Structs, unlike arrays, may be copied with the =
operator. Thus, s2 = s1, will copy the entire
struct. No special function is needed.
functions Structs are passed to functions by value and may
be returned by value.
See structdemo1.c for details:
#include <stdio.h>
/* structdemo1.c -- show basic struct operations */
#define SLEN 100
#define NLEN 10
/* type declaration -- a blueprint, not a house */
struct tstop {
char station[SLEN];
char dir;
char tnum[NLEN];
int hr, min;
};
Last Update: 11/14/0-9November 14, 2005 page 11
class 4: strings, cgi, structs
main()
{
struct tstop s1, s2; /* two houses */
strcpy(s1.station, "salem");
s1.dir = 'o';
s1.hr = 9;
s1.min = 45;
strcpy(s1.tnum, "114");
show_stop("struct s1:\n", s1);
s2 = s1;
show_stop("struct s2:\n", s2);
}
/*
* print stop info
*/
show_stop(char msg[], struct tstop s)
{
printf("%s", msg);
printf("TR=%s;dir=%c;TI=%02d:%02d;stn=%s\n",
s.tnum, s.dir, s.hr, s.min, s.station);
}
6. structs -- arrays to model train runs
A train trip is a sequence of stops. We model that with an
array of stops:
struct tstop trip[NUM_STOPS];
Each element in this array is of the form trip[i] and is a
structtstop. The name ofthe station, for example, is
trip[i].station . We now write a program to read datafrom
the user, build an array, then print the array.
#include <stdio.h>
/* readtrip.c
* shows: how to use an array of structs
* use: reads in one train trip worth of stops
*/
#define MAXSTOPS 100
#define LINELEN 200
#define SLEN 40
#define NLEN 10
#define YES 1
#define NO 0
struct tstop {
char station[SLEN];
char dir;
char tnum[NLEN];
int hr, min;
};
Last Update: 11/14/0-9November 14, 2005 page 12
class 4: strings, cgi, structs
main()
{
struct tstop trip[MAXSTOPS];
int n = 0;
while( n<MAXSTOPS && read_in_stop(trip, n) == YES )
n++;
show_stops(trip, n);
}
/*
* prompt user for items, load the struct at pos
* Return NO for no more data, YES for more
*/
int read_in_stop(struct tstop a[], int pos)
{
char line[LINELEN];
int hh, mm;
if ( getline("station:", line, LINELEN) == NO )
return NO;
strcpy(a[pos].station, line);
if ( getline("direction: ", line, LINELEN) == NO )
return NO;
a[pos].dir = line[0];
if ( getline("number: ", line, LINELEN) == NO )
return NO;
strcpy(a[pos].tnum, line);
if ( getline("time (hh:mm): ", line, LINELEN) == NO)
return NO;
if ( 2 == sscanf(line,"%d:%d", &hh, &mm) ){
a[pos].hr = hh;
a[pos].min = mm;
}
return YES;
}
int getline(char prompt[], char buf[], int len)
{
printf("%s", prompt);
if ( fgets(buf, len, stdin) == NULL )
return NO;
buf[strlen(buf)-1] = ' ';
return YES;
}
/*
* list all the stops recorded in the array of structs
*/
show_stops(struct tstop trip[], int len)
Last Update: 11/14/0-9November 14, 2005 page 13
class 4: strings, cgi, structs
{
int i;
struct tstop s;
for(i=0; i<len; i++){
printf("Stop %d: ", i);
s = trip[i];
printf("TR=%s;dir=%c;TI=%02d:%02d;stn=%s\n",
s.tnum, s.dir, s.hr, s.min, s.station);
}
}
Last Update: 11/14/0-9November 14, 2005 page 14