FILE INPUT/OUTPUT
OUTPUT TO A FILE
Example program ------> FORMOUT.C
Load and display the file named FORMOUT.C for your first example of writing data to a file. We begin as before with the include statement for stdio.h, and include the header for the string functions. Then we define some variables for use in the example including a rather strange looking new type.
The type FILE is a structure (we will study structures in the next chapter) and is defined in the stdio.h file. It is used to define a file pointer for use in file operations. The definition of C requires a pointer to a FILE type to access a file, and as usual, the name can be any valid variable name. Many writers use fp for the name of this first example file pointer so I suppose we should start with it too.
OPENING A FILE
Before we can write to a file, we must open it. What this really means is that we must tell the system that we want to write to a file and what the filename is. We do this with the fopen() function illustrated in line 11 of the program. The file pointer, fp in our case, will point to the structure for the file and two arguments are required for this function, the filename first, followed by the file attribute. The filename is any valid filename for your operating system, and can be expressed in upper or lower case letters, or even mixed if you so desire. It is enclosed in double quotes. For this example we have chosen the name TENLINES.TXT. This file should not exist on your disk at this time. If you have a file with this name, you should change its name or move it because when we execute this program, its contents will be overwritten. If you don't have a file by this name, this program will create one and write some data into it.
Note that we are not forced to use a string constant for the file name as we have done here. This is only done here for convenience. We can use a string variable which contains the filename then use any method we wish to fill in the name of the file to open. This will be illustrated later in this chapter.
READING ("r")
The second parameter is the file attribute and can be any of three letters, "r", "w", or "a", and must be lower case. There are actually additional attributes available in C to allow more flexible I/O, and after you complete your study of this chapter, you should check the documentation for your compiler to study the additional file opening attributes. When an "r" is used, the file is opened for reading, a "w" is used to indicate a file to be used for writing, and an "a" indicates that you desire to append additional data to the data already in an existing file. Opening a file for reading requires that the file already exist. If it does not exist, the file pointer will be set to NULL and can be checked by the program. It is not checked in this program, but could be easily checked as follows.
if (fp == NULL) {
printf("File failed to open\n");
exit (1);
}
Good programming practice would dictate that all file pointers be checked to assure proper file opening in a manner similar to the above code. The value of 1 used as the parameter of exit() will be explained shortly.
WRITING ("w")
When a file is opened for writing, it will be created if it does not already exist and it will be reset if it does, resulting in deletion of any data already there. If the file fails to open for any reason, a NULL will be returned so the pointer should be tested as above.
APPENDING ("a")
When a file is opened for appending, it will be created if it does not already exist and it will be initially empty. If it does exist, the data input point will be set to the end of the data already contained in the file so that new data will be added to any data that already exists in the file. Once again, the return value can and should be checked for proper opening.
OUTPUTTING TO THE FILE
The job of actually outputting to the file is nearly identical to the outputting we have already done to the standard output device. The only real differences are the new function names and the addition of the file pointer as one of the function arguments. In the example program, fprintf() replaces our familiar printf() function name, and the file pointer defined earlier is the first argument within the parentheses. The remainder of the statement looks like, and in fact is identical to, the printf() statement.
CLOSING A FILE
To close a file, use the function fclose() with the file pointer in the parentheses. Actually, in this simple program, it is not necessary to close the file because the system will close all open files before returning to the operating system. It would be good programming practice for you to get in the habit of closing all files in spite of the fact that they will be closed automatically, because that would act as a reminder to you of what files are open at the end of each program.
You can open a file for writing, close it, and reopen it for reading, then close it, and open it again for appending, etc. Each time you open it, you could use the same file pointer, or you could use a different one. The file pointer is simply a tool that you use to point to a file and you decide what file it will point to.
Compile and run this program. When you run it, you will not get any output to the monitor because it doesn't generate any. After running it, look in your current directory for a file named TENLINES.TXT and examine it's contents. That is where your output will be. Compare the output with that specified in the program. It should agree. If you add the pointer test code described above, and if the file couldn't be opened for any reason, there will be one line of text on the monitor and the file will be empty.
Do not erase the file named TENLINES.TXT yet. We will use it in some of the other examples in this chapter.
OUTPUTTING A SINGLE CHARACTER AT A TIME
Example program ------> CHAROUT.C
Load the next example file, CHAROUT.C, and display it on your monitor. This program will illustrate how to output a single character at a time.
The program begins with the include statements, then defines some variables including a file pointer. The file pointer is named point this time, but we could have used any other valid variable name. We then define a string of characters to use in the output function using a strcpy() function. We are ready to open the file for appending and we do so with the fopen() function, except this time we use the lower cases for the filename. This is done simply to illustrate that some operating systems don't care about the case of the filename. Some operating systems, including UNIX, are case sensitive for filenames, so you will need to fix the case before compiling and executing this program. Notice that the file will be opened for appending so we will add to the lines inserted during the last program. If the file could not be opened properly, a NULL value is returned by the fopen() function.
Lines 14 through 18 check to see if the file opened properly and returns an error indication to the operating system if it did not. The constant named EXIT_FAILURE is defined in the stdlib.h file and is usually defined to have the value of 1. The constant named EXIT_SUCCESS is also defined in the stdlib.h file and is usually defined to have the value of 0. The operating system can use the returned value to determine if the program operated normally and can take appropriate action if neccessary. For example, if a two part program is to be executed and the first part returns an error indication, there is no need to execute the second part of the program. Your compiler probably executes in several passes with each successive pass depending on successful completion of the previous pass.
The program is actually two nested for loops. The outer loop is simply a count to ten so that we will go through the inner loop ten times. The inner loop calls the function putc() repeatedly until a character in the string named others is detected to be a zero. This is the terminating null for the string.
THE putc() FUNCTION
The part of the program we are interested in is the putc() function in line 23. It outputs one character at a time, the character being the first argument in the parentheses and the file pointer being the second and last argument. Why the designer of C made the pointer first in the fprintf() function, and last in the putc() function is a good question for which there may be no answer. It seems like this would have been a good place to have used some consistency.
When the textline others is exhausted, a newline is needed because a newline was not included in the definition above. A single putc() is then executed which outputs the \n character to return the carriage and do a linefeed.
When the outer loop has been executed ten times, the program closes the file and terminates. Compile and run this program but once again there will be no output to the monitor. You need to assure that TENLINES.TXT is in the current directory prior to execution.
Following execution of the program, examine the contents of the file named TENLINES.TXT and you will see that the 10 new lines were added to the end of the 10 that already existed. If you run it again, yet another 10 lines will be added. Once again, do not erase this file because we are still not finished with it.
READING A FILE
Example program ------> READCHAR.C
Load the file named READCHAR.C and display it on your monitor. This is our first program which can read from a file. This program begins with the familiar include statements, some data definitions, and the file opening statement which should require no explanation except for the fact that an "r" is used here because we want to read from this file. In this program, we check to see that the file exists, and if it does, we execute the main body of the program. If it doesn't exist, we print a message and quit. If the file does not exist, the system will set the pointer equal to NULL which we test in line 12. If the pointer is NULL we display a message and terminate the program.
The main body of the program is one do while loop in which a single character is read from the file and output to the monitor until an EOF (end of file) is detected from the input file. The file is then closed and the program is terminated.
CAUTION CAUTION CAUTION
At this point, we have the potential for one of the most common and most perplexing problems of programming in C. The variable returned from the getc() function is a character, so we can use a char variable for this purpose. There is a problem that could develop here if we happened to use an unsigned char however, because C returns a minus one for an EOF. An unsigned char type variable is not capable of containing a negative value. An unsigned char type variable can only have the values of zero to 255, so it will return a 255 for a minus one which can never compare to the EOF. This is a very frustrating problem to try to find. The program can never find the EOF and will therefore never terminate the loop. This is easy to prevent. Always use a int type variable when the return can be an EOF, because an int is always signed. According to the ANSI-C standard, a char can be implemented as either a signed or an unsigned type by any particular compiler.
Some compilers use a char type that is not 8 bits long. If your compiler uses other than 8 bits for a char type variable, the same arguments apply. Do not use an unsigned type if you need to check for an EOF returned by the function, because an EOF is usually defined as -1 which cannot be returned in an unsigned type variable.
There is yet another problem with this program but we will worry about it when we get to the next program and solve it with the one following that.
After you compile and run this program and are satisfied with the results, it would be a good exercise to change the name of TENLINES.TXT and run the program again to see that the NULL test actually works as stated. Be sure to change the name back because we are still not finished with TENLINES.TXT. In a real production program, you would not actually terminate the program. You would give the user the opportunity to enter another filename for input. We are interested in illustrating the basic file handling techniques here, so we are using a very simple error handling method.
READING A WORD AT A TIME
Example program ------> READTEXT.C
Load and display the file named READTEXT.C for an example of how to read a word at a time. This program is nearly identical to the last except that this program uses the fscanf() function to read in a string at a time. Because the fscanf() function stops reading when it finds a space or a newline character, it will read a word at a time, and display the results one word to a line. You will see this when you compile and run it, but first we must examine a programming problem.
It is left as an exercise for the student to include a check for proper file opening and performing a meaningful response if it does not open. A meaningful response is to simply output an error message and exit to the operating system.
THIS IS A PROBLEM
Inspection of the program will reveal that when we read data in and detect the EOF, we print out something before we check for the EOF resulting in an extra line of printout. What we usually print out is the same thing printed on the prior pass through the loop because it is still in the buffer named oneword. We therefore must check for EOF before we execute the printf() function. This has been done in READGOOD.C, which you will shortly examine, compile, and execute.
Compile and execute the program we have been studying, READTEXT.C and observe the output. If you haven't changed TENLINES.TXT you will end up with "Additional" and "lines." on two separate lines with an extra "lines." displayed at the end of the output because of the printf() before checking for EOF. Note that some compilers apparently clear the buffer after printing so you may get an extra blank line instead of two lines with "lines." on them.
Notice that we failed to check that the file opened properly. This is very poor practice, and it will be left as an exercise for you to add the required code to do so in a fashion similar to that used in the READCHAR.C example program.
NOW LET'S FIX THE PROBLEM
Example program ------> READGOOD.C
Compile and execute READGOOD.C and observe that the extra "lines." does not get displayed because of the extra check for the EOF in the middle of the loop. This was also the problem referred to when we looked at READCHAR.C, but I chose not to expound on it there because the error in the output was not so obvious.
Once again there is no check for the file opening properly, but you know how to fix it by now and you should do so as an exercise.
We should point out that an experienced C programmer would not write the code as given in this example because it compares c to EOF twice during each pass through the loop and this is inefficient. We have been using code that works and is very easy to understand, but as you gain experience with C, you will begin to use more efficient coding methods, even if they tend to become harder to read and understand. An experienced C programmer would code lines 12 through 17 of READGOOD.C in the following manner;
while((c = fscanf(fp1, "%s", oneword) != EOF)
{
printf("%s\n", oneword);
}
There is no question that this code is more difficult to read, but if you spend some time studying it, you will find that it is identical to the code in the example program. Even though is is more efficient, it is not clear whether the slight gain in efficiency is worth the reduced readability. If the program saves ten milliseconds when reading in a file once a day, and it takes a programmer an hour longer to make a modification to the code a year after the program is released, there is not much savings in using such code. Even though the time assumptions are all judgement calls in the above text, it is plain to see that there are often tradeoffs when writing a program. You will make many decisions concerning execution efficiency and readability when you are writing non-trivial programs.
This is a rather contrived example, because most experienced C programmers would not think this code is at all cryptic, but is written in a standard C notation. As you gain experience, you will come to accept this as clearly written C code. The philosophical argument about code complexity and readability has been made however, and should be considered for all software development.
FINALLY, WE READ A FULL LINE
Example program ------> READLINE.C
Load and display the file READLINE.C for an example of reading a complete line. This program is very similar to those we have been studying except that we read a complete line in this example program.
We are using fgets() which reads an entire line, including the newline character, into a buffer. The buffer to be read into is the first argument in the function call, and the maximum number of characters to read is the second argument, followed by the file pointer. This function will read characters into the input buffer until it either finds a newline character, or it reads the maximum number of characters allowed minus one. It leaves one character for the end of string null character. In addition, if it finds an EOF, it will return a value of NULL. In our example, when the EOF is found, the pointer named c will be assigned the value of NULL. NULL is defined as zero in your stdio.h file.
When we find that the pointer named c has been assigned the value of NULL, we can stop processing data, but we must check before we print just like in the last program. Last of course, we close the file.
HOW TO USE A VARIABLE FILENAME
Example program ------> ANYFILE.C
Load and display the program ANYFILE.C for an example of reading from any file. This program asks the user for the filename desired, and reads in the filename, storing it in a string. Then it opens that file for reading. The entire file is then read and displayed on the monitor. It should pose no problems to your understanding so no additional comments will be made.
Compile and run this program. When it requests a filename, enter the name and extension of any text file available, even one of the example C programs.
Enter an invalid file name to see what the system does when it cannot open the file. If you were a user of this program, and possibly not very computer literate, would you prefer that the program gave you a cryptic message of some sort, or would you prefer that the program displayed a neat message such as "The file you have asked for is not available. Would you like to enter another filename?". Of course this is the message that you can emit when you find that the file did not open properly. Your users will appreciate the effort you put into error handling for their program.
HOW DO WE PRINT?
Example program ------> PRINTDAT.C
Load the last example program in this chapter, the one named PRINTDAT.C for an example of how to print. This program should not present any surprises to you, so we will move very quickly through it.
Once again, we open TENLINES.TXT for reading and we open PRN for writing. Printing is identical to writing data to a disk file except that we use a standard name for the filename. Many C compilers use the reserved filename of PRN that instructs the compiler to send the output to the printer. There are other names that are used occasionally such as LPT, LPT1, or LPT2. Check the documentation for your particular compiler. Some of the newest compilers use a predefined file pointer such as stdprn for the print file. Once again, check your documentation.
The program is simply a loop in which a character is read, and if it is not the EOF, it is displayed and printed. When the EOF is found, the input file and the printer output files are both closed. Note that good programming practice includes checking both file pointers to assure that the files opened properly. You can now erase TENLINES.TXT from your disk. We will not be using it in any of the later chapters.
A READING ASSIGNMENT
Spend some time studying the documentation for your compiler and reading about the following functions. You will not understand everything about them but you will get a good idea of how the library functions are documented.
fopen(), fclose(), putc(), putchar(), printf(), fprintf(), scanf(), fgets()
Also spend some time studying stdio.h, looking for prototypes for the above functions and for the declaration of FILE.
PROGRAMMING EXERCISES