Getting your
department account

The instructions are at

Creating a CS account

Files

A normal "flat" file is a collection of information. It's usually stored somewhere reasonably non-volatile — when you turn off your power, it's nice to still have your files around when you power back up.

You can reasonably view a Unix file as a stream of bytes; as Kernighan and Pike put it in the The Unix Programming Environment, "A file is a sequence of bytes. ... No structure is imposed on a file by the system, and no meaning is attached to its contents — the meaning of the bytes depends solely on the programs that interpret the file. ... Magnetic tapes, mail messages, characters typed on the keyboard, line printer output, data flowing in pipes — each of these files is just a sequence of bytes as far as the system and the programs in it are concerned." (page 41)

Although this seems quite normal to us these days, back in the early days of Unix, such a simple view of files was not the norm; instead, many files were structured in some manner, generally by some sort of "record" scheme, though there were many other ideas floating around.

So what are some of the common ways that we interpret the contents of a file? The simplest of course for human being is simply as a text file. The simplest for a computer might be as a binary program, where its contents are loaded into memory and directly executed by a processor. Other types include PDF files, spreadsheets, SVG files, database files, and a host of other types.

But isn't there some way to indicate what the purpose of a file is? Some operating systems, for instance, have used the idea of indicating types in the file names: "letter.pdf", for instance, in such a scheme, would be a PDF file. But Unix doesn't use that sort of scheme. Instead, it has the idea of a "magic number" at the beginning of a file to indicate the file type.

We aggregate these files into what we call "directories". Directories are recursive structures, since they can also contain other directories.

Windows has generally called the concept "Folders" rather than "directories".

Files and Directories

Pathnames

We need to be able to name the locations of our files.

The "root" is the starting point, referred to with "/"

Sub-directories identified by name, separated by "/"

Absolute pathnames always start with the "root" (i.e., the slash character)

Relative pathnames are from current directory

Going up the directory tree

".." refers to directory above

"." refers to the current working directory

Referencing user directories:

~USER represents the absolute path to USER's home directory

"~/" represents the absolute path to your own home directory

Directory Structure

  • The leading item is called the "root", written simply as "/"
  • Items in a subgraph are "contained" by the directories in the path leading back to the root

Pathnames

Absolute pathnames simply give directions on how to get to a node

Examples of absolute pathnames:

"/bin/emacs"

"/home/carol/Mail/"

 

Relative pathnames are relative to the current working directory

Examples of relative pathnames

"Mail/"

"../../bin/emacs"

Commands

When you type an initial element at a shell prompt, you might be doing one of three things:
  • Starting some sort of aggregation, such as a loop or an "if/then" structure
  • Invoking a "built-in", such as cd
  • Calling a program, such as emacs
Typical command structure: 
COMMANDNAME [FLAGS] [PARAMETERS]

FLAGS: Commands often accept one or more flags after command name.
       Each flag starts with "-" or with "--", and is separated 
       from other flags by spaces.
       Individual flags often may be combined with a single "-".

ls -l -a
ls -la

PARAMETERS: Commands often accept one or more parameters.
            Parameters are often pathnames representing files
            or directories.
            Parameters are separated by spaces.

cp -a /home/carol/Mail /home/carol/Mail-Backup

Finding commands

Generally, you don't want to type the entire pathname for a program like Perl: /usr/bin/perl

Instead, it would be nice to use the "relative" syntax of simply perl — but you don't want to have to also first cd to /usr/bin either.

How do we solve this dilemma? A fundamental construct of most shells is the idea of a PATH variable. The PATH variable simply lists a number of directories that we will search for commands that have no explicit path information. 

$ echo $PATH
/usr/kerberos/bin:/usr/local/bin:/bin:/usr/bin

This PATH means that each of the paths /usr/kerberos/bin, /usr/local/bin, /bin, and /usr/bin will be searched when an unadorned command name is invoked.

Operating on files

Command Purpose
ls List files and directories
cat Output a bytestream
less Paginate a bytestream
pr Paginate a bytestream
touch Update the timestamp on a file; if the file does not exist, it is created
cd Change the shell's current working directory (built-in)
Command Purpose
cp Copy files and directories
mv Move files and directories
rm Remove files and directories
rmdir Remove empty directories
mkdir Create directories
Command Purpose
wc Count lines, words, and characters in a file
grep Search a file
sort Sort a file
head Show the initial lines of a file
tail Show the terminal lines of a file
cmp Find the first difference in two files
diff Find the differences in two files

Three very useful
additional commands

Command Purpose
man Read the fine manual
date Find date and time
who List the people logged in

Permissions

Every file and directory has a set of permissions associated with it. 

$ ls -l /etc/hosts
-rw-r--r-- 1 root root 905 Jun 16  2010 /etc/hosts 
$ ls -dl /etc
drwxr-xr-x 105 root root 12288 Jan 15 14:24 /etc
$ ls -l /usr/bin/emacs-21.4-x 
-rwxr-xr-x 2 root root 6649776 Apr 28  2011 /usr/bin/emacs-21.4-x
$ ls -l /usr/bin/emacs
lrwxrwxrwx 1 root root 19 Feb  8  2012 /usr/bin/emacs -> /usr/bin/emacs-21.4

The first component tells us about the item in the filesystem; an unadorned "-" indicates a regular file; "d" indicates a directory; "l" indicates a soft link; "s" indicates a socket; "c" is a character device file; "b" is a block device file; "p" is a named pipe. 

$ ls -l /etc/hosts
-rw-r--r-- 1 root root 905 Jun 16  2010 /etc/hosts 
$ ls -dl /etc
drwxr-xr-x 105 root root 12288 Jan 15 14:24 /etc
$ ls -l /usr/bin/emacs-21.4-x 
-rwxr-xr-x 2 root root 6649776 Apr 28  2011 /usr/bin/emacs-21.4-x
$ ls -l /usr/bin/emacs
lrwxrwxrwx 1 root root 19 Feb  8  2012 /usr/bin/emacs -> /usr/bin/emacs-21.4
The next 3 characters indicate whether the owner of a file has (1) read (2) write or (3) execute permission (execute in the case of a directory means being able to cd into the directory). The second set of three characters indicate whether users in the same group as this item have read/write/execute permission; the last group of three indicates if all users (the "world") have read/write/execute permission.

Command Reference

ls [-a][-l][-p][-r][-R][-x] [pathname]

Description: Lists the files in a directory.

Options:
 [-a] Display all files
 [-l] Displays all information
 [-r] Reverses order
 [-R] Includes sub-directories

Examples:
  ls
  ls -al
  ls -al *.exe

touch filename

Description: Immediately creates an empty file, or updates the
             time stamp on an existing file.

cp [-i][-R][-r] source destination

Description: Copies the contents of a file or directory to another
             file or directory.

Options:
 [-a] Archive mode; preserve ownership, permissions, and any special attributes
 [-i] Ask before you replace
 [-r] Recursive copy

Parameters:
 source - What you want to copy
 target - New location

Example:
 cp .plan .plan.backup
 cp -r ~/public_html/* /tmp
 cp -a /home/carol /home/carol-backup

mv [-i] source target

Description: Renames or moves a file from one directory to
     another, either with the same name or a different one. Note the
     original file (and name) will no longer exist. This is not a copy.

Options:
  [-i] Prompt you before replacing a file

Parameters:
  source - what you want to copy
  target - where to put it

Examples:
  mv x y
  mv x dir/
  mv -i /etc/passwd-old /etc/passwd

rm [-i][-r][-f] NAMES

Description: Delete files.

Options:
  [-i] Prompt you before replacing a file
  [-r] Recursive, deletes an entire directory and all contents
       and subdirectories. 
  [-f] Forcible removal. Don't give any error messages or ask questions
       even if files don't exist or permissions are awkward.

Parameters:
  NAMES - the names of what to remove


Example:
  rm -rf /
  rm -i this.*
  rm -r /home/carol

less [filename]

Description: paginates a file or stdin.

wc [-c][-l][-w] [NAMES]
Description: Counts characters, lines, and/or words in files or stdin.

Options:
 [-c] Number of characters
 [-l] Number of lines
 [-w] Number of words

pwd

Description: Displays the current working directory.

cd [directory]

Description: Changes current working directory.

Examples:
  cd /
  cd 
  cd /etc
  cd ../public_html/classes

mkdir DIRECTORIES

Description: Creates new directories.


Examples:
  mkdir classes
  mkdir ../classes/new

rmdir DIRECTORIES

Description: Removes the specified empty directories. Will not
  remove a directory that has any contents.

Examples:
  rmdir classes/
  rmdir /home/carol/

man command

Description: Displays the fine manual page for a command.