4.2. Back to command line.

Section quotes:

• "If there's more than one way to do a job, and one of those ways will result in disaster, then somebody will do it that way".
      -- Murphy's law (Edward A. Murphy)

Section contents:

• 4.2.1. Working with files.
• 4.2.2. Working with file systems.
• 4.2.3. Working with archives.
• 4.2.4. Managing processes.
• 4.2.5. Text filters.
• 4.2.6. Miscellaneous tools.

Working with files.

We have discussed how to create a new directory 'mkdir' remove an empty one with 'rmdir', and how to copy, move (also rename) and remove file(s) with 'cp', 'mv', 'rm' respectively, all in section '3.2. Commandline does NOT scares' but we did not give examples, he they go. Let's assume that you have a file called 'test.txt' on your desktop, do the following:

bash$ pwd
 /home/ahmad
bash$ ls Desktop/*.txt
 test.txt
bash$ mv Desktop/*.txt ./
bash$ ls *.txt
 test.txt
bash$ ls Desktop/*.txt
ls: Desktop/*.txt: No such file or directory
bash$ mv test.txt temp.txt
bash$ ls *.txt
 temp.txt
bash$ rm -i *.txt
rm: remove 'temp.txt'? y

Tip

Recall that '.' mean the current working directory, '..' te parent directory, '-' the previous directory and at last '~' the home directory.

bash$ pwd
 /home/ahmad
bash$ mkdir test1 test2
bash$ ls -F
 Desktop/ test1/ test2/ oldfile link@ exec.sh*
bash$ cd test1
bash$ echo "file one" > test1.txt
bash$ less test1.txt
bash$ cp -vi test1.txt test2.txt
 test1.txt -> test2.txt
bash$ cp -vi test2.txt test1.txt
 overwrite test1.txt [yn] ? y
 test2.txt -> test1.txt
bash$ ls *.txt
 test1.txt test2.txt
bash$ cd ..
bash$ rm -R test1

Warning

When you execute a command as root user, even an extra space could case catastrophic results, for example if 'rm -R test1/test2' is written as 'rm -R test1 / test2' this will remove 'test1' and every thing and 'test2'. Double read what your type before you press ENTER.

To know which program is executed when you type some command type 'which' followed by the command name, for example 'which lilo' could give '/sbin/lilo' this is useful is you have two or more programs having the same name. But if you are searching for the location of a program, a manual or a library use 'whereis' followed by the file named without any prefixes or suffixes, it will search for it only in specific directories, for example

bash$ whereis lilo
/sbin/lilo
/usr/share/man/man8/lilo.8.gz
/usr/share/man/man5/lilo.conf.5.gz

bash$ locate '*lilo*'
/sbin/lilo
/etc/lilo.conf
/usr/share/man/man8/lilo.8.gz
/usr/share/man/man5/lilo.conf.5.gz
/usr/share/doc/Lilo-Doc/lilo-readme.html
/home/ahmad/using lilo.html
bash$ locate '*.swp' -i -d /mnt/win_c
/mnt/win_c/windows/win386.swp

bash# export PRUNEPATHES="/mnt /tmp /proc $PRUNEPATHES"
bash# updatedb

To search for files directly on a device (not on a cached database of it) use 'find' followed by directories to be searched followed by some specifiers (or constraints) like filename pattern: '-name PAT', age of last modification in days: '-mtime DAYS', change in file permissions: '-ctime DAYS' and the last access to it (eg. reading) '-atime DAYS', and if you want execute a command of the resulted files add "-exec 'COMMAND {} ;'" where it will replace '{}' with the filename of the file, for example "find ~/tmp/ -atime +3 -exec 'rm {} ;'" will search for all files in '~/tmp/' that have not been accessed 3 (or more) days ago.

The following tools change some file properties:

as we have discussed before file permissions are execution permission 'x' (being able to run the file or search the directory contents), writing permission 'w' and reading permission 'r'. Those permissions are controled by 'chmod' tool for example 'chmod +x file1' case the file 'file1' to be executable by every body, while 'chmod -x file1' will deny execution from every body (including it's owner). You could specify the change to be for the owning user 'u', owning group 'g' and others 'o' (comma separated) for example 'chmod g-w,o-w file1' will deny the group and others from writing leaving the permission for the user unchanged. Another method of specifing the permissions of a file is to use the 0-7 octet notation instead of the 'rwx' notation where we give 1 for 'x', 2 for 'w' and 4 for 'r' (from right to left) and the summation (logical OR) for other, owning group and owning user (respectively from right to left) will form the mode digits, for example if we want the owner to read+write+execute then that digit will be 4+2+1=7, and the group to read and execute then that digit will be 4+1=5, and others do nothing their digit will be 0, put them all to gather to have '750', so the command for that 'chmod 750 file1'. There are more attributes discussed later. It's much more simple than it looks.

user group others rwx rwx rwx 421 421 421 \ / \ / \ / 7 7 7 chmod 777 file1

user group others rwx r-x r-x 421 4 1 4 1 \ / \ / \ / 7 5 5 chmod 755 file2

user group others rw- r-- r-- 42 4 4 \ / \ / \ / 6 4 4 chmod 644 file3

Sticky 's' value is 1 on the 4th digit, for example 'chmod 1755 mydir'. To change the permissions of a directory and its contents recursively use '-R' for example 'chmod -R 750 myfolder'

Tip

It's very useful to use '-X' instead of '-x' to deny execution of directory contents recursively with '-R' so it only affects non-directory items, eg. 'chmod -R -X myfolder'.

4.2.2. Working with file systems.

We have discussed 'mount' tool in section '4.1. Hardware configuration.', which maps devices to the Linux root file system, we have mentioned that this process won't copy, it's just a virtual process where an empty directory (actually it need not be empty, it's preevious contents will be hidden) appears to contian files on that device, The unmounting tool 'umount' will cancel this mapping. What you will know on this section that this is not only about real devices, but also for pseudo file systems like 'proc' (refere to section '3.1 File System Hierarchy.') which contian informations provided by the kernel as files like 'cpuinfo', 'uptime', 'loadavg', 'mounts', 'filesystems'. We have said that such devices and more are automatically mounted when you boot Linux, the configuration file that specify this is '/etc/fstab' it's name is short for file system table. When you want to mount a device listed on that table you don't have to provide any detials, just type 'mount' followed by the device or the mount point directory. Removable devices that should not be mounted automatically on each boot should have 'noauto' option.

bash$ cat /proc/cpuinfo
 ..you see it yourself
bash$ mount
 ..tells you what is mounted now
bash$ man mount
 ..tells about mount
bash$ man fstab
 ..tells about after boot mount table
bash$ cat /etc/fstab

The tool 'eject' will unmount and open the door of the CD drive, if called without arguments it will eject the first CD, you could specify which CD you like to eject by passing the device filename with or without '/dev' prefix.

# try this joke
bash$ eject ; mount /mnt/cdrom ; eject ; mount /mnt/cdrom
# or this, and wait a minute
bash$ { sleep 60; eject } &

If you try to unmount a device (or through its mount point directory) and get 'device is busy' error then it could be because you are in that directory, or there is a program trying to read from it or write on it but the most nontrivial case is that you have mounted a device to a mount point somewhere inside that directory (this is common in 'chroot' environments)

bash$ pwd
/mnt/floppy
bash$ umount floppy
umount: /mnt/floppy: device is busy

Do you have two installed Linux distributions ? or maybe you have a Linux rescue floppy or a Linux live-CD distribution, can you run an application from one distribution other than the one you are running ? let's assume that the other distribution is mounted on '/mnt/linux' (the one you are running of course is mounted to '/'). Let's try to run '/mnt/linux/usr/games/gnuchess', it may work but more likely it won't, because of missing files,missing libraries, bad library version, or files do not exist where it expect them to be for exmample '/usr/share' becomes '/mnt/linux/usr/share'. What is the solution ? maybe you are not yet interested, it does not look necessary, but when it's about maintaining a broken system from a limited rescue environment, like fixing LiLo, it becomes very importent. What we want is to create a virtual environment with with different root file system, in the example we want to change it from '/' to '/mnt/linux' (without copying files), the change root tools 'chroot' do that, it takes the new root directory as first argument, for example 'chroot /mnt/linux' will try to execute '/bin/sh from the new environment, which will give you a shell to work in the new environment until the shell ends with 'exit', this mean that the new environment should have a working 'sh' and it's dependency.

bash# ls -F
bin/	sbin/	etc/	mnt/	lib/	tmp/
bash# mount /dev/hda5 /mnt/linux
bash# chroot /mnt/linux
sh# ls -F
bin/	sbin/	etc/	mnt/	lib/	usr/
tmp/	var/
sh# lilo -vv
sh# exit
bash# 

bash# ls /mnt/linux/usr/games/
fortune		gnuchess
bash# /mnt/linux/usr/games/fortune
file not found
bash# chroot /mnt/linux /usr/games/fortune
Error float point unit overflow
bash# 

4.2.3. Working with archives.

GNU ZIP 'gzip' is the most used comperession method in GNU/Linux systems, because it's very fast an effective, this tool compresses files one by one giving each one '.gz' suffix (the original file disappear), to decompress you have to pass '-d' option or just use 'gunzip' tool. You could specify compression level for 'gzip' from 1 to 9 after '-', use '-1' for best speed and '-9' for best compression. Another compression method is BZip2 'bzip2' it's based on a new algorithm called block-sorting, it compresses much better to give smaller files but takes few more time. It works just like 'gzip' and have similar options but uses '.bz2' suffix. Both tools are programmed to compress standard input and send results to standard output if no files are given, look at the following examples:

bash$  ls -lh delme.* 
-rw-r--r-- 1	ali users	 1M May 28 19:20 delme.bmp
bash$  gzip -9 delme.bmp 
bash$  ls -lh delme.* 
-rw-r--r-- 1	ali users	12K May 28 19:20 delme.bmp.gz
# notice that no delme.bmp
bash$  gzip -d delme.bmp.gz 
bash$  ls -lh delme.* 
-rw-r--r-- 1	ali users	 1M May 28 19:20 delme.bmp
bash$  gzip -9 < delme.bmp > delme.bmp.gz 
bash$  ls -lh delme.* 
-rw-r--r-- 1	ali users	 1M May 28 19:20 delme.bmp
-rw-r--r-- 1	ali users	12K May 28 19:20 delme.bmp.gz
# notice that the delme.bmp is not deleted
bash$  bzip2 -9 < delme.bmp > delme.bmp.bz2 
bash$  ls -lh delme.* 
-rw-r--r-- 1	ali users	 1M May 28 19:20 delme.bmp
-rw-r--r-- 1	ali users	12K May 28 19:20 delme.bmp.gz
-rw-r--r-- 1	ali users	 9K May 28 19:20 delme.bmp.bz2

Tip

Archive then compress (general compression) is more efficient than compress (file by file) then archive (like WinZip) because the tool in first case could find relations between segments in different files.

bash$  ls 
 delme.tar.bz2
bash$  bzip2 -d < delme.tar.bz2 | tar -xvf -  
# here we have decompressed it to have uncompressed TAR archive
# then we have extracted files from it.
bash$  ls 
 delme1.bmp  delme2.bmp  delme3.bmp delme.tar.bz2
bash$  rm *.tar* && ls
 delme1.bmp  delme2.bmp  delme3.bmp
# to archive and compress all non-hidden files:
bash$  tar -cvf - * | gzip -9 > delme.tar.gz 
bash$  ls 
 delme1.bmp  delme2.bmp  delme3.bmp delme.tar.gz

bash$ tar -cvzf delme.tar.gz *

bash$ tar -xvjf delme.tar.bz2

bash$ tar -xjvf delme.tar.bz2 -C ~/mynew/dir 

Tip

Wildcards ('*' and '?') substitution by shell does not include the leading dot of hidden files, this is very useful to exclude temporary (or backup) files, configuration files, file managers caching files, to create an archive of all files in the directory including hidden files use the diretory name, for example: tar -cvzf delme.tar.gz ./.

Tip

When you create an archive for a directory from itself with
'tar -cvzf delme.tar.gz ./' file will be added directely in the archive root, but typing the following from it's parent:
'tar -cvzf delme.tar.gz myfolder/' thhe archive will contain 'myfolder' and it's contents, you could pass many folders and many files to be included:
'tar -cvzf delme.tar.gz myfile1.txt myfolder1/ myfolder2/'.

There are many other archiving tools for GNU/Linux, like 'cpio', 'afio' and 'ar', they are less common for general purpose archiving. And there are many tools for common but non-standard GNU/Linux archives they could come with your distribution, like 'unarc', 'unarj', 'unrar', 'zip', 'unzip' and 'cabextract'. For example to extract 'file.zip' in the current directory type: 'unzip /path/to/file.zip'

You could split a file into smaller many files with specific size limit (to put each on a floppies) with 'split' tool, for example 'split -b 1440k mybigfile' which will produce many files named something like 'xaa','xab','xac'...etc. then you could join them together with 'cat'. There are many tools to provide an almost unique number (called checksum) for different files, this number changes alot if a single byte in the file is changed, it's used to detect modifications (eg. due to transmission interference) just by comparing the checksums of the two files (before and after) those tools are (from the weekest to the strongest): 'sum', 'chksum', 'md5sum' and 'sha1sum' the most common is 'md5sum' but now most sites are moving to use GPG digital signatures see section '4.10. Digital signatures and privacy.'. To do dumb comparison between the corresponding lines of two files use 'comm' tool, but GNU provide much smarter tool called 'diff' to find the difference between two file, for exmaple it detect adding a line to the first of a file, while other tools thinks that all lines after also differ. It could be used to create a script called patch file that automatically do the change on unchanged (old) files, go back to subsection '3.5.3. Patching.'.

4.2.4. Managing processes.

Each running program is called a 'process', when a program calls another we say that the parent process forks a child process. The child process gets the privileges of it's parent, so when you run a program as root user and it forks another program, the last program runs as root too. If you close the parent process the child process will also terminated, this is very useful but becomes annoying when you run a GUI program like 'gedit' from one of the X-terminal then by closing the terminal, the child process will be closed ('gedit' in the example). If you don't want the child process to be closed use 'nohup' to call the child, then if the parent is closed the child keep running as an orphan.

There are many tools to manage processes in many ways like terminating, killing a process and raising or lowing it's priority. To make it easy a unique process identifier (PID) is given to each process, the tool 'ps' lists processes currently running on the same terminal by the same user with it's corresponding PID, pass '-a' argument to list those of all terminals, the are many more options for specify what details to show like the user, memory and CPU usage. A tool called 'top' (a friend of mine told me that it looks better when called from 'xterm') show an automatically updated list of system wide processes ordered by load and resources consuming, it's an interactive tool waits to be exited by pressing 'q' (or CTRL+C), other keys are used to killing a process and raising or lowing it's priority. There are some similar GUI tools like 'gnome-system-monitor'.

To terminate a process use 'killall' tool followed by the the name of the program for example 'killall wine', you could specify how it would be killed by specifying a signal like the normal termination signal '-TERM' (or by number '-15'), this signal gives the process a chance to handle it for example by asking the user to save his work,

Tip

Many GNU documentations suggest that you should use symbolic signal names instead of numerical because they could differ in other Unices.

You could specify different priorities to different processes, priority is denoted by number from -20 to 19, the more the negativity the higher the priority, to run a program with non default priority type 'nice' followed by the priority followed by the command and it's arguments, the default priority is 0 and regular users could not have higher priorities. To change the priority of a running process type 'renice' followed by the new priority then the PID of the process. Here a sammary:

killall [-SEGNAL] PROCNAME	terminate a running process by name (or send it a signal)
kill [-SEGNAL] PID	terminate a running process by PID (or send it a signal)
nice PRIORITY COMMAND	execute COMMAND with specific priority.
renice PRIORITY PID	change the priority of a running process by it's PID.

Many people think that multitasking is for GUI systems only, this is not true, even when you are running Linux without X you have many pseudo-terminals/virtual-terminals you could use CTRL+ALT+F1, CTRL+ALT+F2... and so on to move between them, and even with one terminal you could run many programs. When you are using the GUI, you minimize a window and activate another, you could do similar thing with console programs, press CTRL+Z to suspend the program you are running. Let's assume you are running VIM on terminal to edit a source code and you want to compile it, press CTRL+Z, VIM will be suspended and you will see something like:

[1] 171 stoped		vim
bash$

[2] 175 stoped		cp
bash$

To know which process is using certain file type 'fuser' followed by filename(s), you may use '-v' option (eg. 'fuser -v myfile') to have output in the following syntax:

FILE	USER	PID	ACCESS	COMMAND

4.2.5. Text filters.

As you have noticed before, you could edit files without a text editor, simply with 'cat' and output redirection, for example:

bash$  echo "create a file" > delme.txt 
bash$  echo "add this line to the end" >> delme.txt 
bash$  {echo "add this line at the top" ; cat delme.txt } > delme.txt 
bash$  cat delme.txt 
add this line at the top
create a file
add this line to the end
bash$  rm delme.txt 
bash$  cat <<EOF > delme.txt 
blah blah
blah
EOF
bash$  cat delme.txt 
blah blah
blah
bash$ 

You could display first some lines of a file with 'head', and last with 'tail'. To make text look better use 'pr' which justify text between left and right margins, for example 'pr -o 5 --width=70 FILE | less' will add 5 characters left margin and justify the text (assuming a screen width of 80, you will have 5 chars right margin too). Word counter 'wc' could tell how many lines, works, characters in a file. 'tr' replace members from the first string with the corresponding one from the second, for exmaple to convert all upper case letters to lower one could use: 'tr A-Z a-z'. Numbering lines could be done with 'nl' (which skip empty lines), while 'sort' is used to sort lines of file alphabetically in ascending order, with '-n' it sorts them as numbers (alphabetically, '1' is greater than '10'), with '-r' the order is reversed. 'uniq' is used to remove repeated consecutive lines, with '<>-c' it prefix lines with number of occurrences (beside the removing). There are tools capable of much more complicated things, like 'grep', 'sed', 'cut' and 'awk'. Those are relatively complicted tools, read about them quickly for now and revisit this section when writing Bash shell scripts. 'grep' for example, searches for lines matching its argument on given files (or else on the standard input) and them, for example 'grep vfat /etc/fstab' will print all lines containing 'vfat' on 'fstab', in other words it tell you about windows partitions, with '-i' it becomes case insensitive, for example 'grep -i "VfAt" /etc/fstab'.

Tip

in the last example it could match a line like this '# NON vfat goes here' which we don't want, we need to learn more.

POSIX REs are extended with more special characters those are '? + { } ( ) | ', and for compatibility with the standard POSIX RE (where those chracters represent themselves) they need to be escaped with '\' to represent the new special meaning.

Tip

Some programs (like 'mc' and 'perl') do not follow this rule, they use unescaped extended REs (without '\') to represent the special meaning, and escaped to represent themselves.

	alpha	[A-Za-z]
	upper	[A-Z]
	lower	[a-z]
	digit	[0-9]
	alnum	[A-Za-z0-9]
	space	any whitespace (space ,tab ,new-line)

'grep' could be used to learn English , for example to list all words that ends with 'tion' type "grep -ie 'tion$' /usr/dict/words /usr/share/dict/words | less". You will notice how useful REs are when you apply them in real life.

Tip

You could test your RE skills with something like "echo 'SOMETHING' | egrep 'PAT'" if it's printed then it matches.

Tip

You could searh for files containing some REs by passing '-l' to 'grep' which case 'grep' to print filenames only.

In the following examples we practice "greping" REs on input concesting of "bo", "bb", "bob", "bobob" and "boob" lines.

# lines starting with "b" followed with zero or more "o" followed by ending "b"
bash$ echo -e "bo\nbb\nbob\nbobob\nboob" | grep '^bo*b$'
bb
bob
boob
# lines starting with zero or more "bo" followed by ending "b"
bash$ echo -e "bo\nbb\nbob\nbobob\nboob" | grep -e '^\(bo\)*b$'
bob
bobob
# lines containing "b" followed with zero or more "o" followed by ending "b"
bash$ echo -e "bo\nbb\nbob\nbobob\nboob" | grep -e 'bo*b'
bb
bob
bobob
boob

bash$ echo -e "bo\nbb\nbob\nbobob\nboob" | grep -e 'o\|x$'
bo
bob
bobob
boob

bash$ echo -e "bo\nbb\nbob\nbobob\nboob" | grep -e '\(o\|x\)$'
bo

To deal with fields we use 'cut' tool, fields are parts of a line separated by a given delimiter (TAB by default). The syntax of 'cut' is like: 'cut -d Delimiter -f FieldNo[,FieldNo]...[FieldNo-FieldNo]' which mean "give me the those fields of each line considering 'Delimiter' to be fields separator" for example Delimiter could be '|' and fields could be '2-5' or '1,6'. Let's take an example from real life, the file '/etc/passwd' contain informations about users in the form: his login name then his password (usually omitted) then his UID ... then his default shell, fields are separated by ':', the commands needed to find the UID of 'ali' (the UID field is the third):

# let's look at the file
bash# cat /etc/passwd
...
ali:x:1000:100:users:/home/ali:/bin/bash
ahmad:x:1001:100:users:/home/ahmad:/bin/bash
...
bash# cat /etc/passwd | grep ^ali | cut -d ':' -f 3
1000

Exercise

Type commands needed to find the device(s) that the CDROM is connected to (assumng that 'fstab' is outdated), you could use 'dmesg'

'awk' is very powerful tool, it's a complete programming language that natively support RE, also 'awk' could be used as fields filter but unlike 'cut' it supports more than one signle delimiter, by default it uses whitespaces (multiple spaces, TABs, newlines or any combnations of them), the delimiters are specified by IFS enironment variable (before you change it, you should save the old value to be able restore it), the simple syntax to do so is like: "awk '{print $FieldNo [$FieldNo]...}'" for example "IFS=":" awk '{print $3}' /etc/passwd" to print the third field of each line of the 'passwd' file, here we have used a BASH feature to change an evironment variable just to execute a command. Awk support loops, arrays and much complicated thigs, the following awk script find the longest English word

bash$ awk '
{ if (length($0) > max) { max = length($0); r=$0 }
END { print "the longest word is" r }
' /usr/share/dict/words

Stream Text Editor (SED) is a very useful tool, it's a noninterative text editor, act on the specifyed range (by default all the file) which could be line numbers or a line that match some RE. SED will by the default print all input lines after change unless -n option is used, the following table shows very small part of SED capabilities:

p	print the range
d	delete the range
s/PAT1/PAT2	substitute, replace PAT1 with PAT2 once
s/PAT1/PAT2/g	global substitution, all occurrences of PAT1
y/SET1/SET2	replace members of SET1 with the corresponding member of SET2 (SET1 and SET2 should have the same size)

where patterns are in Regular Expresions. Range is a prefix having line numbers or RE contained by two '/', for example "cat file | sed -e '/^Ali/s/ahmad/Ahmad/g'" means to replace all 'ahmad' with 'Ahmad' only on lines starting with 'Ali'. Another example 'cat file | sed -e '/^$/d'' will remove empty lines. Notice the following trivial session:

# replace 1st occurrence of "el" with "le" in each line
bash$ echo "elelel" | sed -e 's/el/le/'
leelel
# replace every occurrence of "el" with "le" in each line
bash$ echo "elelel" | sed -e 's/el/le/g'
lelele
bash$ echo "angel" | sed -e 's/el/le/g'
angle
# replace every occurrence of ending "el" with "le" in each line
bash$ echo -e "angel\nelelel" | sed -e 's/el$/le/g'
angle
elelle

bash$ echo "foobar" | sed -e 's/o*/x/g'
xfxxbxaxrx
bash$ echo "foobar" | sed -e 's/\(o*\)/[\1]/g'
[]f[oo][]b[]a[]r[]
bash$ echo "foobar" | sed -e 's/o\+/x/g'
fxbar

bash$ echo "<image src="x.png">foobar" | sed -e 's/<image.*>/[IMG]/g'
[IMG]
bash$ echo "<image src="x.png"> <b>foobar</b>" | sed -e 's/<image.*>/[IMG]/g'
[IMG]
bash$ echo "<image src="x.png"> <b>foobar</b>" | sed -e 's/<image[^>]*>/[IMG]/g'
[IMG] <b>foobar</b>

# flip user and domain
bash$ echo "<ahmad@somwehere.com>" |
> sed -e 's/<\([^>]*\)@\([^>]*\)>/<\2@\1>/g'
<somewhere.com@ahmad>
# SPAM protection
bash$ echo "<someone@somewhere.com>" |
> sed -e 's/\(<[^>]*\)@\([^>]*\)\.\([^>]*>\)/\1[AT]\2[DOT]\3/g'
<someone[AT]somewhere[DOT]com>
# email to web
bash$ echo "<ahmad@somwehere.com>" |
> sed -e 's/<\([^>]*\)@\([^>]*\)>/www.\2\/~\1\//g'
www.somwehere.com/~ahmad/
# web to email
bash$ echo "www.somwehere.com/~ahmad/" |
> sed -e 's/www.\([^/ ]*\)\/~\([^/ ]*\)\/\?/<\2@\1>/g'
<ahmad@somwehere.com>
# first name goes first
bash$ cat <<EOS | sed -e 's/ *\([^,]\+\), *\([^.<]\+\)\.\?/\2 \1/'
> Lastname, Firstname. <someone@somewhere.com>
> Ali, Muhammad
> Alsadi, Muayyad.
> Hacker, Jack Random.
> EOS
Firstname Lastname <someone@somewhere.com>
Muhammad Ali
Muayyad Alsadi
Jack Random Hacker

To convert from an encoding (aka. charcter set) to another you could use internation convert tool 'iconv', it has a very simple syntax 'iconv -f FROMCODE -t TOCODE INPUT > OUTPUT' for example

bash$ iconv -f WINDOWS-1256 -t UTF-8 arabic.html > arabic-u.html

4.2.6. Miscellaneous tools.

To display a calender of the current month type 'cal', it has options to display the year calender or to jump to a specefic month. Date and time could be displayed by typing 'date', if you don't like that format you could specify any format you like by using + sign followed by the format you like in a 'strftime' style refere to subsection 7.2.6 of section '7.2. Using standard C library (libc)' (or date manul), for example "date '+%Y-%m-%d'" or with time "date '+%Y-%m-%d %I:%M:%S%p'". Date could do complex time operations with '-d' option, for example to know the date before some hours or after some days(seems simple, but take the case of the day before the first of January), for example to display the date two days ago "date '+%Y-%m-%d' -d '-2 days'" or after two days "date '+%Y-%m-%d' -d '+2 days'". You could use '-s' to change the date and time, the default syntax is 'date -s MMDDhhmmYYYY', for example 'date -s 072820452004' sets the date to be 2004-07-28 and sets the time to be '20:45' (ie. '08:45pm'), it could be more readable if you specify a custom format like
"date -s '2004-07-28 08:45:00pm' '+%Y-%m-%d %I:%M:%S%P'"

'uptime' tool displays some useful system statistics, it displays the current time, followed by the uptime in seconds (how long the system has been running, since last boot), number of logged users and load averages for the past 1, 5, and 15 minutes, the higher load average the more busy the system is. Similarly 'cat /proc/loadavg' prints the same three load average percentages followed by a ratio of number of non-sleeping processes to the number of all processes and followed by the PID of last process, it could be something like '0.20 0.18 0.12 1/80 1120'. 'cat /proc/uptime' will display two times in seconds, the first is the uptime and the other is the idle time (summation of all times the system was doing nothing), the closer those numbers the smother the system goes, you could calculate the ratio with

bash$ ( cat /proc/uptime; echo "2 k r / 100 * p q" ) | dc
91.05

bash$ ( cat /proc/uptime; echo "- p q" ) | dc
1052
bash$ ( cat /proc/uptime; echo "2 k r / 1 r - p q" ) | dc
0.12 

To send a message that shows up in all terminal (pseudo terminal VTs and real serial terminals) the same way you get 'Message to all users: the system is going down' when you type 'halt' or in case of power failur indicated by UPS unit, type 'wall' then at the prompt type the message then press 'CTRL+D', to send t to specific user use 'write' tool similarly, if those messages were annoying, disable them with 'mesg n' to enable them type 'mesg y'. Those messages are not mail they are not saved, they are displayed just the time they arrive, you may use 'talk' (like chatting) or usual email to send messages to users locally or through a network.