sed Tip - #6

[Trevor]

I've used the sed command more than a few times, but I've never paid attention to the use of the semicolon (;) in the script.  You'll recall that the generic syntax of the sed command is:

           sed  [options]  'script'  filename

The command that brought this to my attention was the following:

          sed  's/^[ \t]*//;s/[ \t]*$//'  filename

I'll comment on what the actual command does in a moment, but for now, I want to focus on what caught my attention on the command - that's the use of 2 sed commands in the script.  In the above command, the script is   's/^[ \t]*//;s/[ \t]*$//'

That script is comprised of 2 sed commands - the substitute command (s).  Now, it's nothing new about the use of 2 sed commands in a sed script.  What is not-so-common is the use of the "same" sed command in the sed script - in this case, the 's' (substitute) command is used twice.  This is made possible by the use of the semicolon.  

I know, you know, there's many wonderful, exotic constructs that can be composed within a sed script, thanks to the use of RegEx characters.  I've seen some that seemed as though they were constructed by a mad scientist   However, this use of the same sed command, used multiple times in the same sed script, is a new addition to my knowledgebase.  And like with so many other discoveries that I've made, the need to achieve a particular task led me to this find.  Okay, that's enough with my fascination!

Now, as far as an explanation of that sed command that I mentioned above, what it does is remove the presence of spaces or tabs, at the beginning and/or end of each line in a file.

Here's a little breakdown of the sed script syntax:

s/^[ \t'*//
-  this removes all leading whitespace (spaces or tabs)
-  ^ matches the start of a line
-  [ \t]* refers to zero or more spaces or tabs

s/[ \t]*$//
-  this removes all trailing whitespace 
-  $ anchors the pattern (i.e.[ \t]*) to the end of the line

The magic piece that makes this entire construct possible is the semicolon (;) - this allows both substitutions to be run in a single sed command; a single sed script.  

Just one more something to let me know that I'll never be able to say that I know it all

[Chetan_Tiwary_]

@Trevor awsome post ! Love this! The semicolon is basically sed’s backstage pass andthat is how it lets multiple commands perform in one act ! 

[Trevor]

"... sed's backstage pass ..."   Wow!  Well, let's pin one more title ribbon on you:  Master of Analogies

I'd like your permission to refer to the semicolon in this manner, please!!!

As always Chetan, thank you for your continued support!!!

[Chetan_Tiwary_]

@Trevor you dont need my permission   This is your community !!