Diff for "BashFAQ/100"

How do I do string manipulations in bash?

Bash can do string operations. LOTS of string operations. This is an introduction to bash string operations for those new to Bash's special tool/feature called Parameter Expansion (PE), with a focus on typical string operations. (Note: much of this material is in FAQ #73. It is being presented here with an approach intended to make it more approachable by people who have not already learned bash PEs.)

Note Bash's PE capability is a lot more powerful than the typical string manipulation calls you may be used to. There are some twists in the road up ahead.

Here is a list of some typical string manipulation functions/subroutines from other languages that you may already be familiar with:

Two that you may not have heard of but would want to use all the time when scripting on *NIX:

This article will cover how to do all of these using the Bash and will introduce the more powerful actions available with Bash's PEs. Please note there is a BashFAQ about PEs already. FAQ #73 covers a larger scope of PE capabilities, where this one focuses on string operations.

Filename manipulation

Lest say we have a bash variable named fullpath that contains /usr/home/JosephBaldwin/Its_only_Rock_and_Roll.mp3

Often in scripting we want to manipulate certain pieces of the path, like just the file name, which is the last component of the full path. So lets get just the filename from the full path: In *NIX we have the command basename which does this very nicely for us: basename "$fullpath" returns "Its_only_Rock_and_Roll.mp3". In Bash we can do that much faster with this command: echo ${fullpath##*/}

"WHAT? What the heck is that? That's not a command! That's just a bunch of garbage someone made by whacking some of the stranger keys on the keyboard! I mean really! dollar curly pound pound star slash curly What IS that?"

Um, OK - uh, just calm down for a moment. I know it doesn't look like the typical programming language keyword or library call, but consider a language like Perl. See? To shoehorn new features into Bash, you have to find ways to do it without creating keywords (or anything else) that might cause older scripts to break, so all these string manipulation functions got placed inside the syntax from an old sh feature where there just happened to be room for them: "Parameter Expansion".

For a basename we use the PE expression: ${fullpath##*/} which returns "Its_only_Rock_and_Roll.mp3".

To find the dirname we use the PE expression: ${fullpath%/*} which produces "/usr/home/JosephBaldwin".

To drop the filename extesnion, we use the PE expression: ${fullpath%.*} giving out "/usr/home/JosephBaldwin/Its_only_Rock_and_Roll"

To get the filename's extension we use the PE expression: ${fullpath##*.} generating only "mp3".

To find the strlen, the PE expression: ${#fullpath} finds it, and it's 49.

To get a leftstr, the PE expression: ${fullpath:0:20} grabs the first 20 chars of fullpath to make "usr/home/JosephBaldw".

To perform a rightstr in bash we use the following PE expression: ${fullpath:(-20)} which gets the last 20 chars, "ly_Rock_and_Roll.mp3". The parentheses are needed, although there are a couple other ways to write it that also work.

To perform a midstr in bash we use the following PE expression: ${fullpath:10:20} making "osephBaldwin/Its_onl".

To perform a substr in bash we use the following PE expression: ${fullpath//Rock/Roll} rolling it into "usr/home/JosephBaldwin/Its_only_Roll_and_Roll.mp3".

Why aren't the PE things named more nicely

Can't I just have these a regular functions with nice names?

Not totally generalizable.

What can I do with PE's that I couldn't do with the string functions above?

Todo: offset code examples in code boxes as done in the rest of the wiki.

CategoryShell