Stripping substring from string in filepath

Question

I have the path to a directory, like this:

DIRNAME=/path/to/projects/proj1/dir1

I want strip everything after projects, but keep projects in the file path. However, this code does not work:

DIRNAME=/path/to/projects/proj1/dir1
My_Dir=projects

echo ${DIRNAME%$My_Dir*}

This returns /path/to/; I want to return /path/to/projects instead. Nothing after projects in the file path is static and thus cannot be used in the substring match as per this example.

Obligatory sidenote: I have read over a dozen different pages with multiple examples each of substring manipulation but have not been able to find one that shows how to do this. I think sed might be able to do something like this but I haven't seen any examples that do this.

My usual solution in this kind of scenario is to just do something like this: ${DIRNAME%$My_Dir*}/$My_Dir — Julian
– Julian, Commented Dec 16, 2015 at 0:47
Maybe you can use ${DIRNAME%$proj1*}, so this would return /path/to/projects/ — Ferit
– Ferit, Commented Dec 16, 2015 at 0:50
...by the way -- this may or may not be entirely under your control, but using all-caps names for your own variables is contrary to POSIX-established conventions (pubs.opengroup.org/onlinepubs/009695399/basedefs/…, see fourth paragraph), which reserve all-caps names for system use and suggest that application-controlled names should have at least one lower-case character. — Charles Duffy
– Charles Duffy, Commented Dec 16, 2015 at 1:11

Julian · Accepted Answer · 2015-12-16 00:54:32Z

2

This is generally how I would pull apart directories in this kind of situation:

DIRNAME=/path/to/projects/proj1/dir1
My_Dir=projects

PROJROOT=${DIRNAME%$My_Dir*}$My_Dir
PROJDIR=${DIRNAME#*$PROJROOT}

echo $DIRNAME
echo $PROJROOT
echo $PROJDIR

The output for this is:

/path/to/projects/proj1/dir1
/path/to/projects
/proj1/dir1

answered Dec 16, 2015 at 0:54

Julian

2,90520 silver badges15 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Jonathan Leffler · Accepted Answer · 2015-12-16 01:15:44Z

You need to be careful, but you can make the shell substitution work. For example:

keep="projects"
base="${DIRNAME%/$keep/*}/$keep"

Note that the /$keep/ part ensures that you get the right path with /path/to/projects/with/remote_projects_here/to/confuse/you; your original would mismanage that string.

You can also do it with sed:

base=$(sed "s%\(.*/$keep\)/.*%\1%" <<< "$DIRNAME")

using a here string; or you could do it the portable and classic way with echo:

base=$(echo "$DIRNAME" | sed "s%\(.*/$keep\)/.*%\1%")

Of course, you can still run into problems with what's in the sub-directory names:

/path/to/projects/with/remote_projects_here/projects/select

Test code:

#!/bin/bash

My_Dir=projects
keep="projects"
for dir in \
    /path/to/projects/proj1/dir1 \
    /path/to/projects/with/remote_projects_here/to/confuse/you
do
    DIRNAME="$dir"

    echo "Data"
    echo "$DIRNAME"

    echo "Original"
    echo "${DIRNAME%$My_Dir*}"

    echo "Variant 1"
    base="${DIRNAME%/$keep/*}/$keep"
    echo "$base"

    echo "Variant 2"
    base=$(sed "s%\(.*/$keep\)/.*%\1%" <<< "$DIRNAME")
    echo "$base"

    echo "Variant 3"
    base=$(echo "$DIRNAME" | sed "s%\(.*/$keep\)/.*%\1%")
    echo "$base"
done

Example output:

Data
/path/to/projects/proj1/dir1
Original
/path/to/
Variant 1
/path/to/projects
Variant 2
/path/to/projects
Variant 3
/path/to/projects
Data
/path/to/projects/with/remote_projects_here/to/confuse/you
Original
/path/to/projects/with/remote_
Variant 1
/path/to/projects
Variant 2
/path/to/projects
Variant 3
/path/to/projects

The output could be better formatted, but it makes the point.

Ferit · Accepted Answer · 2015-12-16 01:22:41Z

0

DIR=/path/to/projects/proj1/dir1 

dirname "$DIR"
basename "$DIR"

outputs

/path/to/projects/proj1
dir1

so you can concatenate these with a slash, and you're done. I think the simplest way is this.

edited Dec 16, 2015 at 1:22

answered Dec 16, 2015 at 1:09

Ferit

9,8278 gold badges40 silver badges62 bronze badges

12 Comments

Charles Duffy Over a year ago

Quotes -- basename "$DIR", dirname "$DIR".

Charles Duffy Over a year ago

Try testing again with directory names that contain spaces.

Charles Duffy Over a year ago

...and that's only one pathological case out of many. For instance, one can run mkdir '*' to create a directory named *; you might look at how your code works (without the quotes) with that, too.

Ferit Over a year ago

I didn't understand your last example but, I tried this DIR=/path/to/projects\ /proj1/dir1, after projects it has a space, but it's working. I really couldn't see why it's not working with spaces, can you please give an example input? So I can learn.

Charles Duffy Over a year ago

...and when I run DIR=/path/to/projects\ /proj1/dir1; dirname $DIR on a Linux system with GNU tools, I get two separate lines of output, /path/to and /proj1. Which is what I expect, since the shell is breaking the expanded value into two arguments and passing them separately.

|

Collectives™ on Stack Overflow

Stripping substring from string in filepath

3 Answers 3

Comments

Comments

12 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

12 Comments

Your Answer

Sign up or log in

Post as a guest

Related