This section contains a brief introduction to the C language. It is intended as a tutorial on the language, and aims at getting a reader new to C started as quickly as possible. It is certainly not intended as a substitute for any of the numerous textbooks on C.
The best way to learn a new ``human'' language is to speak it right from the outset, listening and repeating, leaving the intricacies of the grammar for later. The same applies to computer languages--to learn C, we must start writing C programs as quickly as possible.
An excellent textbook on C by two well-known and widely respected authors is:
The C Programming Language -- ANSI C
Brian W. C. Kernighan & Dennis M. Ritchie
Prentice Hall, 1988
Dennis Ritchie designed and implemented the first C compiler on a PDP-11 (a prehistoric machine by today's standards, yet one which had enormous influence on modern scientific computation). The C language was based on two (now defunct) languages: BCPL, written by Martin Richards, and B, written by Ken Thompson in 1970 for the first UNIX system on a PDP-7. The original ``official'' C language was the ``K & R'' C, the nickname coming from the names of the two authors of the original ``The C Programming Language''. In 1988, the American National Standards Institute (ANSI) adopted a ``new and improved'' version of C, known today as ``ANSI C''. This is the version described in the current edition of ``The C Programming Language -- ANSI C''. The ANSI version contains many revisions to the syntax and the internal workings of the language, the major ones being improved calling syntax for procedures and standarization of most (but, unfortunately, not quite all!) system libraries.
Let's
be polite and start by saluting the world! Type the following program into your
favorite editor:
#include < stdio.h>
void main()
{
printf("\nHello
World\n");
}
gcc hello.c
This creates an executable file a.out, which is then executed simply by typing its name. The result is
that the characters `` Hello World'' are printed out,
preceded by an empty line.
A C program contains functions and variables. The functions specify the tasks to be
performed by the program. The ``main'' function establishes the overall logic
of the code. It is normally kept short and calls different functions to perform
the necessary sub-tasks. All C codes must have a ``main'' function.
Our hello.c code calls printf, an output function from the I/O (input/output) library (defined
in the file stdio.h). The original C
language did not have any built-in I/O statements whatsoever. Nor did it have
much arithmetic functionality. The original language was really not intended
for ''scientific'' or ''technical'' computation.. These functions are now
performed by standard libraries, which are now part of ANSI C. The K & R
textbook lists the content of these and other standard libraries in an
appendix.
The printf line prints the message ``Hello World'' on ``stdout'' (the output stream corresponding to the X-terminal window in
which you run the code); ``\n'' prints a ``new line''
character, which brings the cursor onto the next line. By construction, printf never inserts this character on its own: the
following program would produce the same result:
#include < stdio.h>
void main()
{
printf("\n");
printf("Hello
World");
printf("\n");
}
Try
leaving out the ``\n'' lines and see what
happens.
The first statement ``#include <
stdio.h>'' includes a
specification of the C I/O library. All variables in C must be explicitly
defined before use: the ``.h'' files are by
convention ``header files'' which contain definitions of variables and
functions necessary for the functioning of a program, whether it be in a
user-written section of code, or as part of the standard C libaries. The
directive ``#include'' tells the C compiler to insert the contents
of the specified file at that point in the code. The ``< ...>'' notation instructs the compiler to look for
the file in certain ``standard'' system directories.
The void preceeding ``main'' indicates that main is of ``void'' type--that is, it has no type associated with it, meaning that it cannot
return a result on execution.
The ``;'' denotes the
end of a statement. Blocks of statements are put in braces {...}, as in the
definition of functions. All C statements are defined in free format, i.e.,
with no specified layout or column assignment. Whitespace (tabs or spaces) is
never significant, except inside quotes as part of a character string. The
following program would produce exactly the same result as our earlier example:
#include < stdio.h>
void main(){printf("\nHello World\n");}
The reasons for
arranging your programs in lines and indenting to show structure should be
obvious!
The
following program, sine.c, computes a table of the sine function for angles between 0 and
360 degrees.
/************************/
/* Table of */
/* Sine Function */
/************************/
/*
Michel Vallieres */
/*
Written: Winter 1995 */
#include < stdio.h>
#include < math.h>
void main()
{
int angle_degree;
double angle_radian, pi,
value;
/*
Print a header */
printf ("\nCompute
a table of the sine function\n\n");
/*
obtain pi once for all */
/*
or just use pi = M_PI, where
M_PI is defined in math.h */
pi = 4.0*atan(1.0);
printf ( " Value of
PI = %f \n\n", pi );
printf ( "
angle Sine \n" );
angle_degree=0; /* initial angle value */
/*
scan over angle */
while ( angle_degree <= 360 ) /* loop until angle_degree > 360 */
{
angle_radian = pi *
angle_degree/180.0 ;
value =
sin(angle_radian);
printf ( "
%3d %f \n ", angle_degree,
value );
angle_degree =
angle_degree + 10; /* increment the loop index
*/
}
}
The
code starts with a series of comments indicating its the purpose, as well as
its author. It is considered good programming style to identify and document
your work (although, sadly, most people only do this as an afterthought).
Comments can be written anywhere in the code: any characters between /* and */
are ignored by the compiler and can be used to make the code easier to
understand. The use of variable names that are meaningful within the context of
the problem is also a good idea.
The #include statements now also include the header file for
the standard mathematics library math.h. This statement is needed to define the calls to the
trigonometric functions atan and sin. Note also that thecompilation must include the mathematics library explicitly
by typing
gcc sine.c -lm
Variable names are
arbitrary (with some compiler-defined maximum length, typically 32 characters).
C uses the following standard variable types:
int -> integer
variable
short -> short integer
long -> long integer
float -> single
precision real (floating point) variable
double -> double precision real (floating point) variable
char -> character
variable (single byte)
The
compilers checks for consistency in the types of all variables used in any
code. This feature is intended to prevent mistakes, in particular in mistyping
variable names. Calculations done in the math library routines are usually done
in double precision arithmetic (64 bits on most workstations). The actual
number of bytes used in the internal storage of these data types depends on the
machine being used.
The printf function can be instructed to print integers,
floats and strings properly. The general syntax is
printf(
"format", variables );
where "format" specifies the
converstion specification and variables is a list of quantities to print. Some
useful formats are
%.nd integer (optional n =
number of columns; if 0, pad with zeroes)
%m.nf float or double
(optional m = number of columns,
n = number of decimal places)
%ns string (optional n =
number of columns)
%c character
\n \t to introduce new line
or tab
\g ring the bell
(``beep'') on the terminal
Most
real programs contain some construct that loops within the program, performing
repetitive actions on a stream of data or a region of memory. There are several
ways to loop in C. Two of the most common are thewhile loop:
while (expression)
{
...block of
statements to execute...
}
and
the for loop:
for (expression_1;
expression_2; expression_3)
{
...block of
statements to execute...
}
The while loop continues to
loop until the conditional expression becomes false. The condition is tested
upon entering the loop. Any logical construction (see below for a list) can be
used in this context.
The for loop is a special case, and is equivalent to the
following while loop:
expression_1;
while (expression_2)
{
...block of statements...
expression_3;
}
For
instance, the following structure is often encountered:
i = initial_i;
while (i <= i_max)
{
...block of statements...
i = i + i_increment;
}
This
structure may be rewritten in the easier syntax of the for loop as:
for (i = initial_i; i
<= i_max; i = i + i_increment)
{
...block of
statements...
}
Infinite
loops are possible (e.g. for(;;)), but not too good for your computer budget! C permits you to
write an infinite loop, and provides the break statement to ``breakout '' of the loop.
For example, consider the following (admittedly not-so-clean) re-write of the
previous loop:
angle_degree = 0;
for ( ; ; )
{
...block of
statements...
angle_degree =
angle_degree + 10;
if (angle_degree ==
360) break;
}
The
conditional if simply asks
whether angle_degree is equal to 360 or
not; if yes, the loop is stopped.
You
can define constants of any type by using the #define compiler
directive. Its syntax is simple--for instance
#define ANGLE_MIN 0
#define ANGLE_MAX 360
would
define ANGLE_MIN and ANGLE_MAX to the values 0
and 360, respectively. C distinguishes between lowercase and uppercase letters
in variable names. It is customary to use capital letters in defining global
constants.
Conditionals
are used within the if and while constructs:
if (conditional_1)
{
...block of
statements executed if conditional_1 is true...
}
else if (conditional_2)
{
...block of statements
executed if conditional_2 is true...
}
else
{
...block of
statements executed otherwise...
}
and
any variant that derives from it, either by omitting branches or by including
nested conditionals.
Conditionals are logical
operations involving comparison of quantities (of the same type) using the
conditional operators:
< smaller than
<= smaller
than or equal to
== equal to
!= not equal to
>= greater than or equal to
> greater than
and
the boolean operators
&& and
|| or
! not
Another conditional use
is in the switch construct:
switch (expression)
{
case
const_expression_1:
{
...block of statements...
break;
}
case
const_expression_2:
{
...block of statements...
break;
}
default:
{
...block of statements..
}
}
The
appropriate block of statements is executed according to the value of the
expression, compared with the constant expressions in the case statement.
The break statements insure
that the statements in the cases following the chosen one will not be executed.
If you would want to execute these statements, then you would leave out
the break statements. This
construct is particularly useful in handling input variables.
The
C language allows the programmer to ``peek and poke'' directly into memory
locations. This gives great flexibility and power to the language, but it also
one of the great hurdles that the beginner must overcome in using the language.
All variables in a
program reside in memory; the statements
float x;
x = 6.5;
request
that the compiler reserve 4 bytes of memory (on a 32-bit computer) for the
floating-point variable x, then put the ``value'' 6.5 in it.
Sometimes we want to
know where a variable resides in memory. The address (location in memory) of
any variable is obtained by placing the operator ``&'' before its name. Therefore &x is the address of x. C allows us to go one stage further and define
a variable, called a pointer, that contains the address of (i.e. ``points to'') other
variables. For example:
float x;
float* px;
x = 6.5;
px = &x;
defines px to be a pointer to
objects of type float, and sets it equal to the address of x:
Pointer use for a variable
The content of the
memory location referenced by a pointer is obtained using the ``*'' operator (this is called dereferencing the pointer). Thus, *px refers to the value of x.
C allows us to perform
arithmetic operations using pointers, but beware that the ``unit'' in pointer
arithmetic is the size (in bytes) of the object to which the pointer points.
For example, if px is a pointer to a
variable x of typefloat, then the expression px + 1 refers not to the next bit or byte in memory but
to the location of the next float after x (4 bytes away on most workstations); if x were of type double, then px + 1 would refer to a
location 8 bytes (the size of a double)away, and so on. Only
if x is of type char will px + 1 actually refer to the
next byte in memory.
Thus, in
char* pc;
float* px;
float x;
x = 6.5;
px = &x;
pc = (char*) px;
(the (char*) in the last line
is a ``cast'', which converts one data type to another), px and pc both point to the
same location in memory--the address of x--but px + 1 and pc + 1 point to different memory
locations.
Consider the following
simple code.
void main()
{
float x, y; /* x and y are of float
type
*/
float *fp, *fp2; /* fp and fp2 are pointers to
float */
x = 6.5; /* x now contains the
value 6.5 */
/*
print contents and address of x */
printf("Value of x
is %f, address of x %ld\n", x, &x);
fp = &x; /* fp now points to
location of x */
/*
print the contents of fp */
printf("Value in
memory location fp is %f\n", *fp);
/*
change content of memory location */
*fp = 9.2;
printf("New value
of x is %f = %f \n", *fp, x);
/*
perform arithmetic */
*fp = *fp + 1.5;
printf("Final value
of x is %f = %f \n", *fp, x);
/*
transfer values */
y = *fp;
fp2 = fp;
printf("Transfered
value into y = %f and fp2 = %f \n", y, *fp2);
}
Run
this code to see the results of these different operations. Note that, while
the value of a pointer (if you print it out with printf) is typically a large
integer, denoting some particular memory location in the computer, pointers
are not integers--they are a completely different data type.
Arrays
of any type can be formed in C. The syntax is simple:
type name[dim];
In
C, arrays starts at position 0. The elements of the array occupy adjacent
locations in memory. C treats the name of the array as if it were a pointer to
the first element--this is important in understanding how to do arithmetic with
arrays. Thus, if v is an array, *v is the same thing as v[0], *(v+1) is the same thing
as v[1], and so on:
Pointer use for an array
Consider the following
code, which illustrates the use of pointers:
#define SIZE 3
void main()
{
float x[SIZE];
float *fp;
int i;
/*
initialize the array x */
/*
use a "cast" to force i
*/
/*
into the equivalent float */
for (i = 0; i < SIZE;
i++)
x[i] = 0.5*(float)i;
/*
print x */
for (i = 0; i < SIZE;
i++)
printf(" %d %f
\n", i, x[i]);
/*
make fp point to array x */
fp = x;
/*
print via pointer arithmetic */
/*
members of x are adjacent to */
/*
each other in memory */
/*
*(fp+i) refers to content of */
/*
memory location (fp+i) or x[i] */
for (i = 0; i < SIZE;
i++)
printf(" %d %f
\n", i, *(fp+i));
}
(The
expression ``i++'' is C shorthand for ``i = i + 1''.) Since x[i] means the i-th element of the
array x, and fp = x points to the
start of the x array, then *(fp+i) is the content of
the memory address i locations beyond fp, that is, x[i].
A string
constant , such as
"I am a
string"
is
an array of characters. It is represented internally in C by the ASCII
characters in the string, i.e., ``I'', blank, ``a'', ``m'',... for the above string, and terminated by
the special null character ``\0'' so programs can find the end of the string.
String constants are
often used in making the output of code intelligible using printf ;
printf("Hello,
world\n");
printf("The
value of a is: %f\n", a);
String constants can be
associated with variables. C provides the char type variable, which can contain one character--1 byte--at a time.
A character string is stored in an array of character type, one ASCII character
per location. Never forget that, since strings are conventionally terminated by
the null character ``\0'', we require one extra storage location in the array!
C does not provide any
operator which manipulate entire strings at once. Strings are manipulated
either via pointers or via special routines available from the standard string library string.h. Using character
pointers is relatively easy since the name of an array is a just a pointer to
its first element. Consider the following code:
void main()
{
char text_1[100],
text_2[100], text_3[100];
char *ta, *tb;
int i;
/* set
message to be an arrray */
/* of characters;
initialize it */
/* to the
constant string "..." */
/* let
the compiler decide on */
/* its
size by using [] */
char message[] =
"Hello, I am a string; what are you?";
printf("Original
message: %s\n", message);
/* copy
the message to text_1 */
/* the
hard way */
i=0;
while ( (text_1[i] =
message[i]) != '\0' )
i++;
printf("Text_1:
%s\n", text_1);
/* use
explicit pointer arithmetic */
ta=message;
tb=text_2;
while ( ( *tb++ = *ta++
) != '\0' )
;
printf("Text_2:
%s\n", text_2);
}
The
standard ``string'' library contains many useful functions to manipulate
strings; a description of this library can be found in an appendix of the K
& R textbook. Some of the most useful functions are:
char *strcpy(s,ct) ->
copy ct into s, including ``\0''; return
s
char *strncpy(s,ct,n) ->
copy ncharcater of ct into s, return s
char *strncat(s,ct) ->
concatenate ct to end of s; return s
char *strncat(s,ct,n) ->
concatenate n character of ct to end
of s, terminate with ``\0''; return s
int strcmp(cs,ct) ->
compare cs and ct; return 0 if cs=ct,
<0 if cs0 if
cs>ct
char *strchr(cs,c) ->
return pointer to first occurence of c
in cs or NULL if not encountered
size_t strlen(cs) ->
return length of cs
(s and t are char*, cs and ct are const char*, c is an char converted to
type int, and n is an int.)
Consider the following
code which uses some of these functions:
#include < string.h>
void main()
{
char line[100],
*sub_text;
/*
initialize string */
strcpy(line,"hello, I am a string;");
printf("Line:
%s\n", line);
/*
add to end of string */
strcat(line," what
are you?");
printf("Line:
%s\n", line);
/*
find length of string */
/* strlen brings back */
/*
length as type size_t */
printf("Length of
line: %d\n", (int)strlen(line));
/*
find occurence of substrings */
if ( (sub_text = strchr
( line, 'W' ) )!= NULL )
printf("String starting with \"W\" ->%s\n",
sub_text);
if ( ( sub_text =
strchr ( line, 'w' ) )!= NULL )
printf("String starting with \"w\" ->%s\n",
sub_text);
if ( ( sub_text =
strchr ( sub_text, 'u' ) )!= NULL )
printf("String starting with \"w\" ->%s\n",
sub_text);
}
Character level I/O
C
provides (through its libraries) a variety of I/O routines. At the character
level, getchar() reads one
character at a time from stdin, while putchar() writes one character at a time to stdout. For example, consider
#include < stdio.h>
void main()
{
int i, nc;
nc = 0;
i = getchar();
while (i != EOF) {
nc = nc + 1;
i = getchar();
}
printf("Number of
characters in file = %d\n", nc);
}
This
program counts the number of characters in the input stream (e.g. in a file
piped into it at execution time). The code reads characters (whatever they may
be) from stdin (the keyboard),
uses stdout (the X-terminal
you run from) for output, and writes error messages to stderr (usually also your
X-terminal). These streams are always defined at run time. EOF is a special
return value, defined in stdio.h, returned by getchar() when it encounters an end-of-file marker
when reading. Its value is computer dependent, but the C compiler hides this
fact from the user by defining the variable EOF. Thus the program reads
characters from stdin and keeps adding to the counter nc, until it encounters
the ``end of file''.
An experienced C
programmer would probably code this example as:
#include < stdio.h>
void main()
{
int c, nc = 0;
while ( (c = getchar())
!= EOF ) nc++;
printf("Number of
characters in file = %d\n", nc);
}
C
allows great brevity of expression, usually at the expense of readability!
The () in the statement (c = getchar()) says to execute the call to getchar() and assign the result to c before comparing it to EOF; the brackets are
necessary here. Recall that nc++ (and, in fact, also ++nc) is another way of writing nc = nc + 1. (The difference between the prefix and postfix
notation is that in ++nc, nc is incremented before it is used, while in nc++, nc is used before it is incremented. In this particular example,
either would do.) This notation is more compact (not always an advantage, mind
you), and it is often more efficiently coded by the compiler.
The UNIX command wc counts the characters, words and lines in a
file. The program above can be considered as your own wc. Let'sadd a counter for the lines.
#include < stdio.h>
void main()
{
int c, nc = 0, nl = 0;
while ( (c = getchar())
!= EOF )
{
nc++;
if (c == '\n') nl++;
}
printf("Number of
characters = %d, number of lines = %d\n",
nc, nl);
}
Can
you think of a way to count the number of words in the file?
Higher-Level I/O capabilities
We
have already seen that printf handles formatted output to stdout. The counterpart
statement for reading from stdin is scanf. The syntax
scanf("format
string", variables);
resembles
that of printf. The format string may
contain blanks or tabs (ignored), ordinary ASCII characters, which must match
those in stdin, and conversion
specifications as in printf.
Equivalent statements
exist to read from or write to character strings. They are:
sprintf(string,
"format string", variables);
scanf(string,
"format string", variables);
The
``string'' argument is the name of (i.e. a pointer to) the character array into
which you want to write the information.
I/O to and from files
Similar
statements also exist for handling I/O to and from files. The statements are
#include < stdio.h>
FILE *fp;
fp = fopen(name, mode);
fscanf(fp, "format
string", variable list);
fprintf(fp, "format
string", variable list);
fclose(fp );
The
logic here is that the code must
- define a local ``pointer'' of type FILE (note that the uppercase is necessary
here), which is defined in < stdio.h>
- ``open'' the file and associate it with the local
pointer via fopen
- perform the I/O operations using fscanf and fprintf
- disconnect the file from the task with fclose
The
``mode'' argument in the fopen specifies the purpose/positioning in opening the file: ``r'' for reading, ``w'' for writing, and ``a'' for appending to the
file. Try the following:
#include < stdio.h>
void main()
{
FILE *fp;
int i;
fp = fopen("foo.dat",
"w"); /* open foo.dat
for writing */
fprintf(fp,
"\nSample Code\n\n"); /* write
some info */
for (i = 1; i <= 10 ;
i++)
fprintf(fp, "i =
%d\n", i);
fclose(fp); /* close the file */
}
Compile
and run this code; then use any editor to read the file foo.dat.
Functions
are easy to use; they allow complicated programs to be parcelled up into small
blocks, each of which is easier to write, read, and maintain. We have already
encountered the function main and made use of I/O and mathematical routines from the
standard libraries. Now let's look at some other library functions, and how to
write and use our own.
Calling a Function
The
call to a function in C simply entails referencing its name with the appropriate
arguments. The C compiler checks for compatibility between the arguments in the
calling sequence and the definition of the function.
Library functions are
generally not available to us in source form. Argument type checking is
accomplished through the use of header files (like stdio.h) which contain all the necessary information.
For example, as we saw earlier, in order to use the standard mathematical
library you must include math.h via the statement
#include < math.h>
at
the top of the file containing your code. The most commonly used header files
are
< stdio.h> ->
defining I/O routines
< ctype.h> ->
defining character manipulation routines
< string.h> -> defining string manipulation routines
< math.h> ->
defining mathematical routines
< stdlib.h> -> defining number conversion, storage
allocation
and similar tasks
< stdarg.h> -> defining libraries to handle routines with
variable
numbers of arguments
< time.h> ->
defining time-manipulation routines
In
addition, the following header files exist:
< assert.h> -> defining diagnostic routines
< setjmp.h> -> defining non-local function calls
< signal.h> -> defining signal handlers
< limits.h> -> defining constants of the int type
< float.h> ->
defining constants of the float type
Appendix
B in the K & R book describes these libraries in great detail.
Writing Your Own Functions
A
function has the following layout:
return-type function-name ( argument-list-if-necessary )
{
...local-declarations...
...statements...
return return-value;
}
If return-type is omitted, C
defaults to int. The return-value must be of the
declared type.
A function may simply
perform a task without returning any value, in which case it has the following
layout:
void function-name ( argument-list-if-necessary )
{
...local-declarations...
...statements...
}
As an example of
function calls, consider the following code:
/*
include headers of library */
/*
defined for all routines */
/* in
the file */
#include < stdio.h>
#include < string.h>
/*
prototyping of functions */
/* to
allow type checks by */
/* the
compiler */
void main()
{
int n;
char string[50];
/*
strcpy(a,b) copies string b into a */
/*
defined via the stdio.h header */
strcpy(string,
"Hello World");
/*
call own function */
n = n_char(string);
printf("Length of
string = %d\n", n);
}
/*
definition of local function n_char */
int n_char(char string[])
{
/* local variable in this function */
int n;
/*
strlen(a) returns the length of */
/*
string a */
/*
defined via the string.h header */
n = strlen(string);
if (n > 50)
printf("String
is longer than 50 characters\n");
/*
return the value of integer n */
return n;
}
Arguments are always
passed by value in C function calls. This means that local ``copies'' of the
values of the arguments are passed to the routines. Any change made to the arguments
internally in the function are made only to the local copies of the arguments.
In order to change (or define) an argument in the argument list, this argument
must be passed as an address, thereby forcing C to change the ``real'' argument
in the calling routine.
As an example, consider
exchanging two numbers between variables. First let's illustrate what happen if
the variables are passed by value:
#include < stdio.h>
void exchange(int a, int b);
void main()
{ /*
WRONG CODE */
int a, b;
a = 5;
b = 7;
printf("From main:
a = %d, b = %d\n", a, b);
exchange(a, b);
printf("Back in
main: ");
printf("a = %d, b =
%d\n", a, b);
}
void exchange(int a, int b)
{
int temp;
temp = a;
a = b;
b = temp;
printf(" From
function exchange: ");
printf("a = %d, b =
%d\n", a, b);
}
Run
this code and observe that a and b are NOT exchanged! Only the copies of the
arguments are exchanged. The RIGHT way to do this is of course to use pointers:
#include < stdio.h>
void exchange ( int *a, int *b );
void main()
{ /*
RIGHT CODE */
int a, b;
a = 5;
b = 7;
printf("From main:
a = %d, b = %d\n", a, b);
exchange(&a,
&b);
printf("Back in
main: ");
printf("a = %d, b =
%d\n", a, b);
}
void exchange ( int *a, int *b )
{
int temp;
temp = *a;
*a = *b;
*b = temp;
printf(" From
function exchange: ");
printf("a = %d, b =
%d\n", *a, *b);
}
The
rule of thumb here is that
- You use regular variables if the function does
not change the values of those arguments
- You MUST use pointers if the function changes the
values of those arguments
It is standard practice in UNIX for information to be passed from the command line directly into a program through the use of one or more command-line arguments, or switches. Switches are typically used to modify the behavior of a program, or to set the values of some internal parameters. You have already encountered several of these--for example, the "ls" command lists the files in your current directory, but when the switch -l is added, "ls -l" produces a so-called ``long'' listing instead. Similarly, "ls -l -a" produces a long listing, including ``hidden'' files, the command "tail -20" prints out the last 20 lines of a file (instead of the default 10), and so on.Conceptually, switches behave very much like arguments to functions within C, and they are passed to a C program from the operating system in precisely the same way as arguments are passed between functions. Up to now, the main() statements in our programs have had nothing between the parentheses. However, UNIX actually makes available to the program (whether the programmer chooses to use the information or not) two arguments to main: an array of character strings, conventionally called argv, and an integer, usually called argc, which specifies the number of strings in that array. The full statement of the first line of the program ismain(int argc, char** argv)(The syntax char** argv declares argv to be a pointer to a pointer to a character, that is, a pointer to a character array (a character string)--in other words, an array of character strings. You could also write this as char* argv[]. Don't worry too much about the details of the syntax, however--the use of the array will be made clearer below.)When you run a program, the array argv contains, in order, all the information on the command line when you entered the command (strings are delineated by whitespace), including the command itself. The integer argcgives the total number of strings, and is therefore equal to equal to the number of arguments plus one. For example, if you typeda.out -i 2 -g -x 3 4the program would receiveargc = 7argv[0] = "a.out"argv[1] = "-i"argv[2] = "2"argv[3] = "-g"argv[4] = "-x"argv[5] = "3"argv[6] = "4"Note that the arguments, even the numeric ones, are all strings at this point. It is the programmer's job to decode them and decide what to do with them.The following program simply prints out its own name and arguments:#include < stdio.h>main(int argc, char** argv){int i;printf("argc = %d\n", argc);for (i = 0; i < argc; i++)printf("argv[%d] = \"%s\"\n", i, argv[i]);}UNIX programmers have certain conventions about how to interpret the argument list. They are by no means mandatory, but it will make your program easier for others to use and understand if you stick to them. First, switches and key terms are always preceded by a ``-'' character. This makes them easy to recognize as you loop through the argument list. Then, depending on the switch, the next arguments may contain information to be interpreted as integers, floats, or just kept as character strings. With these conventions, the most common way to ``parse'' the argument list is with a for loop and a switch statement, as follows:#include < stdio.h>#include < stdlib.h>main(int argc, char** argv){/* Set defaults for all parameters: */int a_value = 0;float b_value = 0.0;char* c_value = NULL;int d1_value = 0, d2_value = 0;int i;/* Start at i = 1 to skip the command name. */for (i = 1; i < argc; i++) {/* Check for a switch (leading "-"). */if (argv[i][0] == '-') {/* Use the next character to decide what to do. */switch (argv[i][1]) {case 'a': a_value = atoi(argv[++i]);break;case 'b': b_value = atof(argv[++i]);break;case 'c': c_value = argv[++i];break;case 'd': d1_value = atoi(argv[++i]);d2_value = atoi(argv[++i]);break;}}}printf("a = %d\n", a_value);printf("b = %f\n", b_value);if (c_value != NULL) printf("c = \"%s\"\n", c_value);printf("d1 = %d, d2 = %d\n", d1_value, d2_value);}Note that argv[i][j] means the j-th character of the i-th character string. The if statement checks for a leading ``-'' (character 0), then the switch statement allows various courses of action to be taken depending on the next character in the string (character 1 here). Note the use of argv[++i] to increase i before use, allowing us to access the next string in a single compact statement. The functions atoi and atof are defined instdlib.h. They convert from character strings to ints and doubles, respectively.A typical command line might be:a.out -a 3 -b 5.6 -c "I am a string" -d 222 111(The use of double quotes with -c here makes sure that the shell treats the entire string, including the spaces, as a single object.)Arbitrarily complex command lines can be handled in this way. Finally, here's a simple program showing how to place parsing statements in a separate function whose purpose is to interpret the command line and set the values of its arguments:/********************************//* *//* Getting arguments from *//* *//* the Command Line *//* *//********************************//* Steve McMillan *//* Written: Winter 1995 */#include < stdio.h>#include < stdlib.h>void get_args(int argc, char** argv, int* a_value, float* b_value){int i;/* Start at i = 1 to skip the command name. */for (i = 1; i < argc; i++) {/* Check for a switch (leading "-"). */if (argv[i][0] == '-') {/* Use the next character to decide what to do. */switch (argv[i][1]) {case 'a': *a_value = atoi(argv[++i]);break;case 'b': *b_value = atof(argv[++i]);break;default: fprintf(stderr,"Unknown switch %s\n", argv[i]);}}}}main(int argc, char** argv){/* Set defaults for all parameters: */int a = 0;float b = 0.0;get_args(argc, argv, &a, &b);printf("a = %d\n", a);printf("b = %f\n", b);}Suppose you don't want to deal with command line interpretation, but you still want your program to be able to change the values of certain variables in an interactive way. You could simply program in a seriesprintf/scanf lines to quiz the user about their preferences:...printf("Please enter the value of n: ");scanf("%d", &n);printf("Please enter the value of x: ");scanf("%f", &x);...and so on, but this won't work well if your program is to be used as part of a pipeline (see the UNIX primer), for example using ther graphics program plot_data, since the questions and answers will get mixed up with the data stream.A convenient alternative is to use a simple graphical interface which generates a dialog box, offering you the option of varying key parameters in your program. Our graphics package provides a number of easy-to-use tools for constructing and using such boxes. The simplest way to set the integer variable n and the float variable x (i.e. to perform the same effect as the above lines of code) using a dialog box is as follows:/* Simple program to illustrate use of a dialog box */main(){/* Define default values: */int n = 0;float x = 0.0;/* Define contents of dialog window */create_int_dialog_entry("n", &n);create_float_dialog_entry("x", &x);/* Create window with name "Setup" and top-left corner at (0,0) */set_up_dialog("Setup", 0, 0);/* Display the window and read the results */read_dialog_window();/* Print out the new values */printf("n = %d, x = %f\n", n, x);}Compile this program using the alias Cgfx (see the page on compilation) to link in all necessary libraries.The two create lines define the entries in the box and the variables to be associated with them, set_up_dialog names the box and defines its location. Finally, read_dialog_window pops up a window and allows you to change the values of the variables. When the program runs, you will see a box that looks something like this:Modify the numbers shown, click "OK" (or just hit carriage return), and the changes are made. That's all there is to it! The great advantage of this approach is that it operates independently of the flow of data throughstdin/stdout. In principle, you could even control the operation of every stage in a pipeline of many chained commands, using a separate dialog box for each.
No comments:
Post a Comment