Files

1
Files File Streams

• Chapter 21

2
Files

• Everything in Unix is a file
• Files
• Directories
• Ports
• Drives
• Keyboard
• Monitor
• All attached devices
• Whether you know it or not, youve been using
  file streams since your hello world program on
  the second day of class. Printf and scanf both
  use file streams to do input and output.

3
File Stream

• Making calls to read() and write() which are both
  kernel level operations are very costly to
  performance. When a kernel call is made, the
  process releases its control of the CPU and
  waits for the kernel to provide the service
  needed.
• In addition, reading and writing to secondary
  storage (disk) is thousands of times slower than
  writing to primary storage.
• File streams were created to help reduce the
  number of times read() and write() calls are
  made, by making use of buffers.

4
Buffering

• A program that uses file streams reads data
  from, and writes data to its own memory buffers,
  usually one character at a time. Since memory
  access doesnt require a system call, this type
  of buffered I/O can be done very quickly, and
  without blocking. Only when an input buffer is
  empty or an output buffer is full, do the
  programs issue the read() or write() system
  calls.
• A typical buffer is 1kb or 1024 characters.
• Types of buffering
• Block buffering used by disks, and tapes. A
  buffer is flushed only when it is full.
• Line buffering used for terminals, characters
  are buffered until a newline character is output,
  the buffer is full, or input is requested (which
  makes sure prompts are output before input is
  expected).
• Unbuffered I/O used when characters must be
  transferred immediately.

5
Opening a File

• There is no basic file data type in C, but there
  is a FILE structure defined in stdio.h.
• To open a file, we use the fopen function. fopen
  requires a string with the path and name of the
  file to be opened, and a string to represent the
  mode the file should be opened with.
• The path to the file can be either be the full
  path or a relative path.
• The mode can be one of 6 things
• r read only
• w writing only. If the file exists, it will
  be overwritten.
• a writing only. If the file exists, data is
  appended to the end of the file.
• r reading writing. Data writing starts at
  the beginning of the file, and the file contents
  may be overwritten.
• w reading writing. Existing file contents
  are erased before the first read or write
  operation.
• a reading writing. Like r except
  position is initially the end of the file.

6
Opening a File

• The fopen function will return a pointer to a
  FILE structure, or NULL if the file is unable to
  be opened. Your programs should check if NULL
  was returned from fopen and deal with that
  condition since C wont check for an error or do
  error handling automatically for you.
• ExamplesFILE out
• out fopen(/users/student/student_a/out.txt,w
  )
• out fopen(./out.txt,a)

7
Closing a File

• The fclose function flushes output buffers, then
  closes the file.
• fclose(out)

8
Getting Putting Characters

• There are two options for getting characters from
  a file stream. getc and fgetc both of which take
  the FILE as their argument and return the next
  character read from the file or EOF if end of
  file is reached. getc is implemented as a
  preprocessor macro so it is much faster than
  fgetc, which is implemented as a function.
• There are two options for putting characters out
  to a file stream as well. They are putc and
  fputc and each takes the character to be written
  and the FILE to write to. They will return the
  character written or EOF to indicate an error.
  Again as with getc, putc is implemented as a
  macro and will be much faster than fputc which is
  implemented as a function.

9
Reading Writing Strings

• Fgets get a string from a file stream
• Char fgets(char string, int maxlen, File
  stream)
• Gets a string from the specified file stream and
  stores it in the address pointed at in the first
  argument. A string is defined as ending with a
  new line character or having up to maxlen
  characters.
• Fputs writes a string to a file stream
• Int fputs(char string, FILE stream)
• Writes the string specified as the first argument
  to the stream specified in the second argument.
  Returns 0 on success or EOF on error.

10
Formatted Input Output

• Fscanf reads a formatted string. Identical to
  scanf except it reads from the specified file
  stream instead of from standard input.
• Int fscanf(FILE stream, char format,)
• Returns the number of items converted.
• Fprintf writes formatted numeric data and text
  to a stream. Identical to printf except it
  prints to a stream instead of standard output.
• Int fprintf(FILE stream, char format, )
• Returns the number of characters printed.

11
Checking for EOF

• There is a macro called feof which is meant
  specifically to check for the EOF condition. The
  only drawback is that it is only set after a read
  operation fails due to an end of file condition.
• feof(FILE )
• Will return a non-zero number if an EOF condition
  has been met.
• Because feof doesnt work until after a read
  operation, youve already got the EOF flag back
  from your read operation by the time you can call
  feof and get it to report back an EOF. Because
  youll have to check your data returned from the
  read operation for EOF anyway, theres no use
  using feof. This is a better solution
• While( (chgetc(input)) ! EOF )

12
Temporary Files

• Often times youll need to use a file to
  temporarily store some data. If you hard code a
  name into your program, you can never be sure
  that it will never overwrite another file. C
  provides some functions for creating temporary
  files with names that are guaranteed not to
  interfere with any currently existing files.
• tmpnam(char name) generates a unique file name
  in the system temp directory (/tmp) and stores
  the name in the string that was passed in as an
  argument.
• tempnam(char dir, char prefix) also generates a
  unique file name but allows you to specify the
  directory you want the file in with the first
  argument. The second argument is a prefix for
  the file, so if you pass in a prefix of temp,
  the filename that tempnam returns will begin with
  temp followed by a sequence of random letters
  and/or digits.
• Each of the above functions returns a string with
  the path to a unique temporary file name. You
  can then create the file with fopen.
• A word of warning If you call either of the
  above functions more than once without using the
  name(s) previously generated, they could return a
  duplicate name.

13
Clearing Buffers

• Sometimes its necessary to make sure that your
  data is moved from the buffer to the file or
  screen. You can do this with the fflush
  function. With this function, you dont lose any
  data, it just forces the data to be written.
• fflush(stdout)
• The other option for clearing a buffer is fpurge.
  This function is a destructive one, and will
  delete whatevers in the buffer rather than
  writing it out. This comes in handy when using
  scanf. Some of you have noticed that its
  necessary to put a blank white space at the
  beginning of your format string in a scanf call
  in order to keep your program from skipping the
  part where the user is allowed to enter their
  data. This is because there is a left over white
  space stuck in the stdin stream buffer. A call
  to fpurge(stdin) just prior to your scanf call
  will clean the buffer, removing the white space
  and the removing the need to have a leading white
  space in your scanf format string.
• Linux does not use fpurge, it uses __fpurge
  instead which is provided by stdio_ext.h

14
Error Handling

• Thankfully, the C libraries provide plenty of
  error feedback. Most of that feedback is
  provided by a global integer variable called
  errno. If an I/O error occurs, the resulting
  error code will generally be put in errno. You
  can then use errno to output error messages or
  debug your program.
• There are a couple of functions that will help
  make your life easier with error handling
• ferror(FILE stream) returns a non-zero integer
  if the specified stream has encountered an error.
• perror(char error_string_prefix) Prints the
  error string prefix provided as an argument to
  standard output followed by a colon and textual
  description of the error that occurred.
• strerror(int err_code) - returns a string
  describing the error that corresponds with the
  error code passed in. Can be used as
  follows char error error strerror(errno)

15
Whole File Operations

• Deleting a file
• remove(char path)
• unlink(char path)Both remove and unlink are
  identical in functionality, just the name differs
• Move or rename a file
• rename(char from, char to)
• Making a directory
• mkdir(char path, mode)
• The mode argument in the mkdir call is the
  permissions you want set on the file when it is
  created. The mode number is the same as the
  permissions you would use in the chmod command.
  Ie 755, 700, 644

16
Miscellaneous Stream Functions

• The following functions are helpful for working
  with files, but we likely wont need them in this
  course. I mention them here for completeness.
  If you need to use them or want more information,
  consult the man pages.
• fseek moves the files positional pointer to
  the specified character position within the file.
• ftell returns the current location of the
  files positional pointer within the file.
• rewind allows you to move the positional
  pointer back to the beginning of the file.