Input/Output

• Input and output are not part of C but are provided by standard library functions in stdio.h.

• Input and output are performed via streams.
  • A stream is a file or a physical device (e.g., printer or monitor) which is manipulated with a pointer to the stream.
  • There is an internal C data structure, FILE, which represents all streams and is defined in stdio.h. A FILE is a struct, defined in stdio.h. Stream IO uses a FILE *.
  • Stream I/O is buffered. A fixed size block is read from or written to a file via some temporary storage area (the buffer). This leads to efficient I/O but data written to a buffer does not appear in a file (or device) until the buffer is flushed. Any abnormal exit of code can cause problems.
  • Streams have to be opened before use, and closed after use.
  • Streams may be considered to transfer either text or binary data.

• Predefined Streams
  • UNIX has 3 predefined streams (in stdio.h):
    • stdin - initially associated with the keyboard
    • stdout - initially associated with the screen
    • stderr - initially associated with the screen
    Their type is FILE *.
  • Predefined streams are automatically open.
  • They all use text as the method of I/O.
  • stdin and stdout can be used with files, programs, and I/O devices such as keyboard, console, etc. stderr normally goes to the console or screen.
  • The stdin and stdout streams may have their input/output devices redirected (> < >>) or piped (|) from the command line, or associated with other devices by the program. This is why stderr exists, so that error messages can continue to be displayed on the screen, even if output is redirected to a file.

Basic stdin and stdout IO

• Character input and output
  • int getchar(void) returns the next input character from the standard input each time it is called. getchar returns EOF (defined in stdio.h) when end of file is encountered.
  • int putchar(int c1) puts c1 onto the standard output stream. putchar returns EOF on error.
  • FileCopy.c - Program to copy a text file character by character
  • Not all characters are printable. C provides library functions, declared in ctype.h, which determine the nature of a character.

• Line input and output
  • char *gets(char s1[]) reads the next input line into the character array s1, replacing the terminating \n with \0. It returns the address of s1[], or NULL if an error occurs.
    • Warning, beep, beep ... you are responsible for making sure that the chcracter array is big enough for the input.
    • Achtung, achtung ... users are not reliable, and may enter more than you expect.
    • Aandag asseblief ... use fgets instead.
  • int puts(char s1[]) writes out s1 and a \n. Returns a non-negative int, or EOF if an error occurs.
  • FileCopyByLine.c - Program to copy a text file line by line

• Formatted output - printf.
  • int printf(char control[],<arg1>,arg2>,.....);
  • printf converts, formats and prints its arguments on the standard output stream under the control of the control argument.
  • printf returns the number of characters printed, or a negative value on error.
  • The control string contains characters to be printed as is and conversion specifications which specify how the following arguments should be printed.
  • Each conversion specification starts with a % and ends with a conversion character.
  • There must be the same number of conversion specifications as following arguments, unless an * is used for width or precision (see below).
  • Between the % and the conversion character there may be (in order) :
    • Flags (in any order)
      • - = left justification.
      • + = always print with a sign (otherwise only - is printed).
      • space = if first character is not a sign, then a space is printed instead. Only one of + or space can be specified.
      • 0 = in numeric output, specifies padding with 0.
      • # = alternate output form.
    • number = minimum field width. A greater width is used if necessary, padded otherwise.
    • . = Separates minimum field width from precision
    • second number = maximum number of characters to be printed from a string, or the number of decimal places for float and double, or the minimum number of digits for an integer.
    • l = long int, h = short int, L = long double.
  • A width or precision may be specified by *, which takes its value from the next argument.
  • The conversion characters are :
    • d,i = int -> decimal
    • o = int -> octal
    • x,X = int -> hexidecimal
    • u = int -> unsigned
    • c = int -> character
    • s = char[] -> character string
    • e,E = double -> real scientific
    • f = double -> real decimal
    • g,G = double -> a combination of f and e
    • p = void * -> pointer
    • n = int * -> puts number of characters written so far into the argument, no argument is converted.
    • % = no argument converted, prints a %.
  • If the character after the % is not a conversion character the behaviour is undefined.

• Formatted input - scanf.
  • int scanf(char control[],<arg1>,<arg2>,....);
  • scanf reads and converts input from the standard input stream under the control of the control argument.
  • scanf returns the number of items read, or EOF on error or end-of-file. Hence it is not immediately known whether end-of-file or an error has occurred.
  • The control string controls input conversion, and may contain blanks, tabs or newlines (white space) which are ignored, characters other than % which are expected to match the next non-white space character on the input stream, and conversion specifications
  • The following arguments must all be addresses.
    • Arrays = give name
    • Other variables = preceed with &.
  • A conversion specification starts with a % and ends with a conversion character.
  • There must be the same number of conversion specifications as following arguments, unless assignment suppression is used (see below).
  • Between the % and the conversion character there may be :
    • * = assignment suppression skips input fields and no variable pointer argument is required.
    • number = maximum field width, or number of characters for c conversion.
    • l = long int or double, h = short int, L = long double.
  • An input field consists of a string of non-white space characters (except for character input), or until the maximum field width (if specified) is exhausted. Leading white space is hence ignored, and scanf will read over line boundaries.
  • The conversion characters are :
  • d = int * <- decimal integer
  • i = int * <- decimal, octal or hex integer
  • o = int * <- octal integer (with or without leading 0)
  • x = int * <- hex integer
  • u = unsigned int * <- unsigned decimal integer
  • c = char * <- character (default one character). White space skip is suppressed. Use %1s to skip white space.
  • s = char * (char []) <- character string. \0 is added to the input.
  • f,e,g = float * <- real
  • % = No assignment, read a %.
  • Example

        int i1;
        float f1;
        char name[50];
    
        scanf("%2d %f %*d %2s",&i1,&f1,name)
              

    with input 56789 1234 45a72 gives i1=56, f1=789.0 , name="45"
  • ReadEchoPeople.c - Program using formatted IO for reading and writing people

File IO

• Opening more streams
  • Files are opened via the standard function fopen which returns a pointer to a FILE.
    • FILE *fopen(char *name,char *mode)
    • The available modes are :
      • "r" - read. The file must exist.
      • "w" - write. The file is re-created destroying any old version.
      • "a" - append. If the file exists the new data is appended, otherwise a new file is created.
      • "r+" - update (read and write)
      • "w+" - update. The file is re-created.
      • "a+" - update, with writes (without positioning) being appended.
      • "b" - binary mode. No conversion of newline sequences, etc. Used in conjunction with one of the above.
    • Returns NULL if an error occurs. An error code is stored and can be retrieved with ferror.
  • Characters are input from a file using :
    • int fgetc(FILE *file)
    • Returns the next character in the file. On end-of-file or error EOF is returned.
    • Has an equivalent macro implementation getc.
    • getchar() is getc(stdin) => same values returned.
  • Characters are output to a file using :
    • int fputc(char c1,FILE *file)
    • Returns c1 if successful, EOF if an error occurs.
    • Has an equivalent macro implementation putc.
    • putchar(c1) is putc(c1,stdout) => same values returned.
  • int fclose(FILE *file) is used to close files. Returns EOF on error, and 0 otherwise.
  • EOF is defined in stdio.h, and is returned when an error occurs on an input operation. End-of-file is one such (common) error, but it is better to use feof to test whether a genuine end-of-file has been reached.
  • FileCopyByName.c - Program that copies named files
  • And don't forget, stdin, stdout, and stderr are all FILEs, so they can be used where ever a FILE is expected.

• Other file output functions.
  • fputs - writes a string to a file
    • int fputs(char *s1,FILE *file)
    • Outputs upto but not including the first \0.
    • Returns non-negative if successful, EOF otherwise.
    • int *puts(char *s1) outputs to stdout.
  • fprintf - writes formatted output to a file.
    • int fprintf(FILE *file,char *control,<arg1>,...)
    • See printf for details
    • Returns number of characters written, negative if an error occurs.
    • printf is fprintf(stdout,...) => same values returned.
  • fwrite - writes data in binary mode to a file.
    • size_t fwrite(void *buffer,size_t size,size_t number,FILE *file)
    • Writes number items of size bytes from memory starting at buffer, out to file.
    • Returns number if successful, number written otherwise.

• Other file input functions.
  • fgets - gets a line of characters from a file.
    • char *fgets(char *s1,int size,FILE *file)
    • Gets a maximum of size-1 characters (upto and including a \n) and appends \0.
    • Returns s1 or NULL if at end-of-file or on error.
    • char *gets(char *s1) reads from stdin.
  • fscanf - reads formatted data from file
    • int fscanf(FILE *file,char *control,<arg1>,...)
    • See scanf for details
    • Returns number of items matched if successful, EOF otherwise.
    • scanf is fscanf(stdin,...) => same values returned.
  • fread - reads data in binary mode from a file.
    • size_t fread(void *buffer,size_t size,size_t number,FILE *file>)
    • Reads number items of size bytes into memory starting at buffer, from file.
    • Returns number if successful, number read otherwise.
  • SavePeople.c - Program that uses file output functions to save people
  • SaveRetrievePeople.c - Program that uses file input functions to retrieve people

• File error handling functions.
  • feof - detects end-of-file condition
    • feof(FILE *file)
    • Returns non-0 on end-of-file, 0 otherwise.
  • ferror - detects i/o errors
    • int ferror(FILE *file)
    • An error indicator is set when an error occurs for each file, and can be retrieved by ferror. Does not clear the error.
    • Returns non-0 if an error has occured, 0 if no error since last fopen or clearerr.
  • clearerr - clears error codes.
    • void clearerr(FILE *file)

• File positioning functions
  • fseek - moves to a random byte in a file.
    • int fseek(FILE *file,long offset,int origin)
    • Moves to the byte in the file offset bytes from origin, where origin can be :
      • SEEK_SET - beginning of file (non negative offset)
      • SEEK_CUR - current position in file
      • SEEK_END - end of file.
    • For text files, the offset must be 0, or the offset must be a value returned by ftell and the origin SEEK_SET.
    • Returns 0 if successful, non-0 otherwise.
    • SaveRetrievePerson.c - Program that uses file seek to retrieve one person
  • rewind - resets a file to its start
    • void rewind(FILE *file)
    • Rewinds both input and output files.
    • rewind(file) is fseek(file,0L,SEEK_SET)
  • ftell - returns current file position
    • long ftell(FILE *file)
    • Returns the current position as an offset from the beginning of the file in bytes, or -1L on error.

• Other file handling functions.
  • ungetc - puts one character into a stream
    • int ungetc(int c1,FILE *file)
    • Only one character (not EOF) may be pushed back onto a stream.
    • Returns c1 if successful, EOF otherwise.
  • fflush - flushes output buffers to OS.
    • int fflush(FILE *file)
    • Returns 0 if successful, EOF otherwise.
    • fflush does not flush OS to disk - use sync() for that.
  • remove - deletes a file
    • int remove(char *file_name)
    • Removes the file, returning non-zero if unsuccessful.
  • rename - renames a file
    • int rename(char *old_name,char *new_name)
    • Changes the name of a file, returning non-zero if unsuccessful.

• Other less common functions are
  • tmpnam, mktemp for creating non-existant file names
  • tmpfile, mkstemp for creating temporary files
  • setvbuf and setbuf for controlling buffering
  • freopen for duplicating access to already open files, especially the std ones.
  • fgetpos for recording the current file position
  • fsetpos for moving to a position recorded by fgetpos

Miscellaneous IO

• In-memory format conversion.
  • int sprintf(char string[],char control[],<arg1>,...) puts output into string.
  • int sscanf(char string[],char control[],<arg1>,...) takes input from string.

• Using a variable number of arguments as arguments
  • int vprintf(char control[],va_list arg)
  • int vsprintf(char string[],char control[], va_list arg)
  • These are the same as printf and sprintf, except that the variable argument list is replaced by arg, which has been initialised with va_start, and possible moved with va_arg.

• Any of the short exercises.
• What are the three predefined IO streams in UNIX, and what devices are they associated with by default?
• Why should fgets be used in preference to gets?
• What is the exact output from the following code? Indicate spaces with an upsidedown triangle.

       int i1 = 27;
       float f1 = 13.13;
       float f2 = -45.45;
       char data[] = "Bonzai!";
  
       printf("%3d %7.3f %-10s %f %o\n",i1,f1,data,f2,i1); 

• What are the values of the variables after the following code is executed with input 56789 1234 45a72?

  int i1;
  float f1;
  char name[50];
  
  scanf("%2d %f %*d %2s",&i1,&f1,name);