CMPSC 311, Files and Directories

CMPSC 311, Introduction to Systems Programming

Files, Directories and I/O

Reading

CP:AMA

Ch. 3, scanf, printf
pp. 14-15, 19-20, 22, 130-131, 134, 139-141, scanf, printf, getchar, putchar
Sec. 13.3, scanf, printf, gets, puts
p. 504, second Q&A, FILE
Ch. 21, The Standard Library

Sec. 21.1, Using the Library, getchar
Sec. 21.4, The <stddef.h> Header: Common Definitions

Ch. 22, Input/Output

CS:APP

Ch. 10, System-Level I/O, esp. Sec. 1, 2, 3, 5

APUE, Chapters 3, 4, 5

File I/O	unbuffered	buffered
Which standard defines these functions?	Posix	C, Posix
Which include files do you need?	`<fcntl.h> <unistd.h>`	`<stdio.h>`
How do you refer to an open file?	file descriptor type `int`	file pointer type `FILE *`
How do you open a file?	`open`	`fopen, freopen`
How do you close a file?	`close`	`fclose`
How do you read from an open file?	`read`	`fgetc,` `getc,getchar fgets, gets fread fscanf, scanf`
How do you write to an open file?	`write`	`fputc,` `putc,putchar fputs, puts fwrite fprintf, printf`
Where do read and write operations start?	at the current file offset	at the current stream position
How do you move to another position in an open file (without changing the file)?	`lseek`	`fseek, ftell fgetpos, fsetpos rewind`
Additional types, values, functions	`size_t`	`fpos_t NULL, EOF ungetc feof, ferror clearerrfflushtmpfile, tmpnamsetbuf, setvbufremove, rename`

Note that the C Standard I/O library refers to "streams", not "open files", since a stream need not be connected to an actual file.

A file opened with open() can be associated with a stream by calling fdopen(), and the net effect is as if fopen() had been used. A file opened with fopen() can obtain its file descriptor using fileno().

The Standard I/O library uses three types of buffering:

fully buffered -- writes wait until the buffer is full; reads fill the buffer completely, or up to end-of-file
line buffered -- writes wait until a newline character is encountered; reads fill the buffer up to the next newline character, or up to end-of-file
unbuffered -- writes and reads are done as soon as possible

Use fflush() to force the buffer contents to be written.

Questions

Does the ungetc() function write to a file? If not, does it write to a buffer?
When read() fails, it returns -1 and errno explains the problem. When read() encounters end-of-file, it returns 0. When fgetc() fails, it returns EOF, which is a macro defined as -1, and feof() or ferror() must be called to distinguish end-of-file from error (and then errno explains the problem). Why does fgetc() not return 0 for end-of-file?
Why must you never use gets()?

Excerpts from the Mac OS X man page,

#include <stdio.h>

char *fgets(char *restrict s, int n, FILE *restrict stream);
char *gets(char *s);

The gets() function is equivalent to fgets() with an infinite n and a stream of stdin, except that the newline character (if any) is not stored in the string. It is the caller's responsibility to ensure that the input line, if any, is sufficiently short to fit in the string.

SECURITY CONSIDERATIONS
The gets() function cannot be used securely. Because of its lack of bounds checking, and the inability for the calling program to reliably determine the length of the next incoming line, the use of this function enables malicious users to arbitrarily change a running program's functionality through a buffer overflow attack. It is strongly suggested that the fgets() function be used in all cases.

Directory I/O
Which standard defines these functions?	Posix
Which include files do you need?	`<dirent.h>`
How do you refer to an open directory?	`DIR *`
What data structure describes a directory entry?	`struct dirent`
How do you open a directory?	`opendir`
How do you close an open directory?	`closedir`
How do you read from an open directory?	`readdir`
How do you write to an open directory?	see below
How do you create a new directory?	`mkdir <sys/stat.h>`
How do you delete a directory?	`rmdir <unistd.h>`

Writing to a directory is a side effect of creating a new file.

Note that the C standard does not define any operations on directories.

See APUE Fig. 1.3 for a simple example. (Intro to Unix notes)

File information
Which standard defines these functions?	Posix
Which include files do you need?	`<sys/stat.h> <unistd.h> <utime.h>`
What data structure describes properties of a file system entry?	`struct stat`
How do you discover properties of a file system entry?	`stat`
If the file could be a symbolic link, and if you would then want information about the link and not what it points to,	`lstat`
How do you discover properties of an open file?	`fstat`
How can you tell if a file exists, and what permissions are associated with it?	`access`
How can you change the permissions associated with a file?	`chmod, fchmod`
To limit the default permissions for new files,	`umask`
How can you change the owner of a file?	`chown fchown lchown`
How can you change the times associated with a file?	`utime`

The header <sys/stat.h> defines some macros to determine file types from the struct stat entries. See CS:APP, p. 874, for some examples.

Multiple choice:

The rmdir() function is used to delete a directory. This function should

(1) fail if the directory does not exist.
(2) fail if the directory is not empty.
(3) fail if the directory is not empty (except for the . and .. entries).
(4) always succeed.

Exercise

When you are writing a program that reads a file and writes a file, which part of the program should you write first? Why?

Exercise

Here is the read_line() function from CP:AMA, p. 287. What is wrong with the design of this function?

int read_line(char str[], int
        n)

        {

          int ch, i = 0;

        

          while ((ch = getchar()) != '\n')

            if (i < n)

              str[i++] = ch;

          

          str[i] = '\0';  // terminate the string

          return i;       // number of characters
        stored

        }

Last revised, 18 Feb. 2013