A photo of Mitesh Shah

Mitesh Shah

Linux Enthusiast
System Administrator

Email Skype Github Twitter Google+ Resume Hire Me Keybase LinkedIn Stackoverflow

Text Processing Tools


Tools for Extracting Text

  1. File Contents: cat and less
  2. File Excerpts: head and tail
  3. Extract by Keyword: grep
  4. Extract by Column or Field: cut

1. Viewing File Contents

cat command
  • cat - Dump one or more file to STDOUT
  • cat command is most useful for viewing the short files
  • Multiple files are concatenated together


  • -A Show all characters, including control characters and non-printing characters
  • -s Squeeze (multiple adjacent blank lines into a single black line)
  • -b Number each (non-blank) line of output

NOTE!: If you dump the content of a binary file with cat to a terminal, you will make it unusable. You can use reset command to clean up your garbled terminal and go on with it.
When you type reset, it won’t be correctly echo-ed.

less command
  • less View file or STDIN one page at a time.
  • less command is more useful for viewing the larger files.

Navigating Text with less

  • Space Moves ahead one full screen
  • b Moves back one full screen

  • Enter Moves ahead one line
  • k Moves back one line

  • g Moves to the top of the file
  • G Moves to the bottom of the file

  • /text Searches for text
  • n Repeats the last search
  • N Repeats the last search, but in the opposite direction

  • v Opens the file in (vi by default)
  • q quits

2. Viewing File Excerpts

head command
  • head: Display the first 10 lines of a file
  • -n: Specify the number of lines to display


[mitesh@Matrix ~]$ head /etc/passwd

[mitesh@Matrix ~]$ head -n 2 /etc/passwd
tail command
  • tail: Display the last 10 lines of a file
  • -n: Specify the number of lines to display
  • -f: Follow subsequent additions to the file
    • Continue to display the file in REAL TIME
    • Very useful for monitoring log files!
    • System Administrators use this feature to keep an eye on the system log.


[root@Matrix ~]# tail -n 2 /var/log/messages
[root@Matrix ~]# tail -f /var/log/messages

3. Extracting Text by Keyword - grep command

  • grep: Prints lines of files or STDIN where the pattern is matched
  • The patterns contain regular expression metacharacters and so it is considered good practice to always quote your regular expressions.


  • -i: Search case insensitively
  • -n: Print line numbers of matches
  • -v: Print lines that does not contain the pattern

  • -Ax: Include x lines after each match
  • -Bx: Include x lines before each match

  • -r: Recursively search a directory
  • -c: Counts the number of lines where the pattern matched
  • -l: Only return the name of the file that have at least one line containing the pattern

  • --color=auto: Highlight the match in color


[mitesh@Matrix ~]$ grep 'bash' /etc/passwd

[mitesh@Matrix ~]$ grep '[Cc]at' pets

[mitesh@Matrix ~]$ ps ax | grep 'init'
1 ?        Ss     0:01 /sbin/init
2701 pts/1    S+     0:00 grep init

[mitesh@Matrix ~]$ date --help | grep 'year'

4. Extract by Column or Field - cut command

  • cut: Display specific columns of file or STDIN


  • -d: Specify the column delimiter (Default is TAB)
  • -f: Specify the column to print
  • -c: Cut by characters


# Display the list of users from /etc/passwd
[mitesh@Matrix ~]$ cut -d: -f1 /etc/passwd
...output truncated...

# Display the Login Shell of root user
[mitesh@Matrix ~]$ grep 'root' /etc/passwd | cut -d: -f7

# Display the list of UID from /etc/passwd
[mitesh@Matrix ~]$ cut -f3 -d: /etc/passwd
...output truncated...

# Cut by characters
[mitesh@Matrix ~]$ cut -c2-5 /usr/share/dict/words

# System's IP Address
[mitesh@Matrix ~]$ ifconfig | grep 'inet addr' | cut -d: -f2 | cut -d' ' -f1

Tools for Analyzing Text

  1. Text Stats: wc
  2. Sorting Text: sort
  3. Comparing Files: diff and patch
  4. Spell Check: aspell

wc command

  • wc command counts the Number of Lines, Words, Bytes and/or Characters in a File or STDIN.
  • On traditional UNIX system every character in a text file took up exactly 1 byte.
  • However, with the advent of internationalization and larger character sets like Unicode some characters can take up to 4 bytes.


  • -l: Only for line count
  • -w: Only for word count
  • -c: Only for byte and/or chatacter count
  • -m: Get an accurate charcter count


[mitesh@Matrix ~]$ wc story.txt
39  237  1901 story.txt

[mitesh@Matrix ~]$ wc .bash*
66  264 1533 .bash_history
2    2   18 .bash_logout
12   27  176 .bash_profile
8   21  124 .bashrc
88  314 1851 total

[mitesh@Matrix ~]$ ls /tmp | wc -l

sort command

  • sort - Sorts Text to STDOUT - Original File Unchanged


sort [OPTION]... [FILE]...


  • -r: Perform a Reverse (Descending) sort
  • -n: Perform a Numerical sort

  • -f: Ignore (Folds) case of character in string
  • -u: Unique (Remove duplicate lines in output)

  • -t: Specify the column delimiter
  • -k: Specify the column to print


[mitesh@Matrix ~]$ grep 'bash' /etc/passwd | sort

# Display the list of sorted UID from /etc/passwd
[mitesh@Matrix ~]$ sort -t : -k 3 -n /etc/passwd
[mitesh@Matrix ~]$ sort -t : -k 3.2 -n /etc/passwd

NOTE!: The argument to the -k option can be two numbers separated by a dot. In this case, The number before the dot is the field number
The number after the dot is the character within that field with which to begin sort

Eliminating Duplicate Lines

  • sort-u: Removes duplicate lines from input
  • uniq: Removes duplicate adjacent lines from input To print only unique line occurrences in a file (Remove all duplicate lines), input to uniq must be first sorted.


  • -c: Produce a frequency listing - count no of occurrences. Each line will be prepended with a number indicating how many times it appears in the input

  • -d: Print one copy of the lines that are repeated in the input.
  • -u: Output only the lines that are truely unique - only occurring once in the input.

  • -fn: Avoid comparing the first n fields in each line.
  • -sn: Avoid comparing the first n characters in each line.


# Use with sort for best result
[mitesh@Matrix ~]$ sort userlist.txt | uniq -c

[mitesh@Matrix ~]$ cut -d: -f7 /etc/passwd | sort | uniq -c
3 /bin/bash
1 /bin/sync
1 /sbin/halt
32 /sbin/nologin
1 /sbin/shutdown

diff command

  • diff: Compare two files for difference.
  • Use gvimdiff for graphical diff - Provided by vim-X11 package.


# Denotes a difference (change) on line 5
[mitesh@Matrix ~]$ diff foo.conf-broken foo.conf-works
< use_widgets = no
> use_widgets = yes
  • Suppose a service on station1 is malfunctioning but the same service works on station2.
  • Thanks to diff and the use of simple, text-based configuration files,
  • We can easily compare the working and non-working configurations.
[mitesh@Matrix ~]$ cat file.conf-station1
Hostname = station1
Setting1 = a
Setting3 = C
Setting4 = D

[mitesh@Matrix ~]$ cat file.conf-station2
Hostname = station2
Setting1 = A
Setting2 = B
Setting3 = C

[mitesh@Matrix ~]$ diff file.conf-station1 file.conf-station2
< Hostname = station1
< Setting1 = a
> Hostname = station2
> Setting1 = A
> Setting2 = B
< Setting4 = D

Duplicating File Changes

  • diff-u: Unified Diff (An alternate way of displaying the same information), Best for patch utility.
  • patch: Duplicate changes in other files (use with care!)


  • -b: Automatically backup changed files.


[mitesh@Matrix ~]$ diff -u file.conf-station1 file.conf-station2
--- file.conf-station1	2011-08-22 12:22:37.648426983 +0530
+++ file.conf-station2	2011-08-22 12:23:36.775147621 +0530
@@ -1,4 +1,4 @@
-Hostname = station1
-Setting1 = a
+Hostname = station2
+Setting1 = A
+Setting2 = B
Setting3 = C
-Setting4 = D
  • To use patch, simply store the output of a diff -u in a file;
  • And run the following command, which would make file.conf-station1 looks like file.conf-station2
[mitesh@Matrix ~]$ diff -u file.conf-station1 file.conf-station2 > file.conf.patch


  • Do you actually want all of the changes above to be made?
  • It would be advisable to first edit file.conf.patch
  • And remove the two lines describing the Hostname variable, since those should remain different between systems.
  • If anything terrible happens, patch -b automatically creates a backup of each file it changes.
  • backups are given the .orig extension.
[mitesh@Matrix ~]$ patch -b file.conf-station1 file.conf.patch

aspell command

  • aspell: A aspell is an interactive spell checker.
  • It offers suggestions for corrections via a simple menu-driven interface.
[mitesh@Matrix ~]$ aspell check file.txt
Some times *peple* type stuff wrong.
1) people			6) peel
2) Pele				7) Pelee
3) Peale			8) peopled
4) purple			9) peoples
5) Peel				0) pep
i) Ignore			I) Ignore all
r) Replace			R) Replace all
a) Add				x) Exit

# A aspell list will non-interactively list the misspelled words in a file read from STDIN.
[mitesh@Matrix ~]$ aspell list < standfast.txt

# A quick spelling dictionary look-up can be performed with the look command.
[mitesh@Matrix ~]$ look exer
...output truncated...

Tools for Manipulating Text

tr command

  • tr: Translate (Alter) Characters.
  • Only reads data from STDIN.
  • Converts characters in one set to corresponding characters in another set.
[mitesh@Matrix ~]$ tr 'a-z' 'A-Z' < lowercase.txt

# This command is commonly used in shell scripts to ensure that data is in an expected case
echo -n "Enter yes or no: "
read answer
answer="$(echo $answer | tr 'A-Z' 'a-z')"

sed command

  • sed: Stream Editor
  • Performs search/replace operations on a stream of data
  • As with grep, it is considered good practice to always quote sed’s search/replace string

  • By Default: sed make maximum one change per line
  • If you want make multiple changes per line then append g (Globle) at the end of search/replace pattern.

  • sed searches are case-sensitive
  • If you want to search case-insensitively then append i (case insensitive) to the pattern.

  • sed operates on all the lines of the file.
  • It is possible to provide sed with address limiting.

  • Normally does not alter the source file
  • use -i.bak to backup and alter the source file


[mitesh@Matrix ~]$ cat pets
cat cat cat Cat CAT
cat cat cow coW COW
Cat cat cat CAR car

[mitesh@Matrix ~]$ sed 's/cat/dog/' pets
dog cat cat Cat CAT
dog cat cow coW COW
Cat dog cat CAR car

[mitesh@Matrix ~]$ sed 's/cat/dog/gi' pets
dog dog dog dog dog
dog dog cow coW COW
dog dog dog CAR car

# sed search/replace pattern starts working on all the lines of the file called pets.
[mitesh@Matrix ~]$ sed 's/cat/dog/g' pets

# sed search/replace pattern starts working between the lines of 1 to 50.
[mitesh@Matrix ~]$ sed '1,50s/cat/dog/g' pets

# sed search/replace pattern starts working on the line that contains the
# string digby and continuing through the line that contain the string duncan.
[mitesh@Matrix ~]$ sed '/digby/,/duncan/s/cat/dog/g' pets

# Multiple sed instruction:
sed -e 's/cat/dog/' -e 's/hi/hello/' pets
sed -f myedits pets

# Delete Last Empty New Line
[mitesh@Matrix ~]$ sed '/^$/d' file.txt

# Insert New Line Above Last Line
[mitesh@Matrix ~]$ sed '$ c\\t\include /etc/nginx/common/*.conf;\n}' file.txt

regex command

  • regex: Regular Expressions
  • For more details see man 7 regex
  • man grep
|	Metachracter	|	Meaning					|
|	^		|	Line Begin				|
|	$		|	Line Ends				|
|	[xyz]		|	A character that is x, y or z		|
|	[^xyz]		|	A character that is not x, y or z	|


[mitesh@Matrix ~]$ grep 'root' /etc/passwd

[mitesh@Matrix ~]$ grep '^root' /etc/passwd

Post Navigation