CODEWITHSUNDEEP
stringbashshelldivide
JH5774
JH5774

Reputation: 31

How do I split a list of unquoted paths with embedded spaces into individual paths?

I'm not good at bash shell scripting in Ubuntu, so I need your help.

The problem is...

We use Perforce for SCM.

I try to get directories from //Development/ branches.

Until this time, everything was good.

But now I can't split branches by whitespace anymore, because of branches such as the following (note the embedded space):

//Development/graphic/release/Unity Provider

We need each branch on a separate line starting with //Development, but always I get the following result (note the unwanted line break):

//Development/graphic/release/Unity
Provider

How can I fix this?

Please help me. Thank you.

Below is a sample one-line string:

//Development/graphic/release/CM //Development/graphic/release/GManager //Development/graphic/release/Notification //Development/graphic/release/Core //Development/graphic/release/Provider //Development/graphic/release/WH //Development/graphic/release/Accessory //Development/graphic/release/Unity Provider //Development/graphic/release/tipManager

And, I want to get a result string like below (each branch name on its own line):

//Development/graphic/release/CM 
//Development/graphic/release/GManager 
//Development/graphic/release/Notification 
//Development/graphic/release/Core 
//Development/graphic/release/Provider 
//Development/graphic/release/WH 
//Development/graphic/release/Accessory 
//Development/graphic/release/Unity Provider 
//Development/graphic/release/tipManager

I also want to store the results in a list variable.

E.g., list[0] should contain //Development/graphic/release/CM.

Upvotes: 0

Views: 100

Answers (3)

mklement0
mklement0

Reputation: 439317

I'm assuming that:

  • you want to split your input string into individual paths based on substrings starting with //[Development/], either at the start of the string, or, if inside, preceded by a single space.
  • regardless of whether the strings between //[Development/] instances contain spaces or not.
str='//Development/graphic/release/CM //Development/graphic/release/GManager //Development/graphic/release/Notification //Development/graphic/release/Core //Development/graphic/release/Provider //Development/graphic/release/WH //Development/graphic/release/Accessory //Development/graphic/release/Unity Provider //Development/graphic/release/tipManager'    

echo "$str" | sed 's# \(//\)#\'$'\n''\1#g'

The above should work with any POSIX-compatible sed implementation.

To capture the output in a variable, using command substitution:

result=$(echo "$str" | sed 's# \(//\)#\'$'\n''\1#g')

If you then want to process the result line by line:

while read -r path; do echo "$path"; done <<<"$result"

Explanation of the sed command:

  • # was - arbitrarily - chosen as the delimiter for sed's s (string substitution) command so as to make it easier to match / chars (customarily, / is used as the delimiter, which would necessitate \-escaping / instances in the regex and replacement string).
  •  \(//\) matches // if preceded by a space, i.e., inside the string.
  • \'$'\n'' effectively inserts a newline (\n) into the replacement string, using ANSI C quoting (required for OSX compatibility; on Linux, just \n would do).
  • \1 inserts the 1st (and only) capture group from the regex, i.e, //
  • g ensures that matching is global, i.e., that all substrings that match the regex are replaced.

Result:

//Development/graphic/release/CM
//Development/graphic/release/GManager
//Development/graphic/release/Notification
//Development/graphic/release/Core
//Development/graphic/release/Provider
//Development/graphic/release/WH
//Development/graphic/release/Accessory
//Development/graphic/release/Unity Provider
//Development/graphic/release/tipManager

Upvotes: 1

Avinash Raj
Avinash Raj

Reputation: 174786

You could try this awk command also,

$ awk -v RS=" //" '{gsub (/^D/,"//D");}1' file
//Development/graphic/release/CM
//Development/graphic/release/GManager
//Development/graphic/release/Notification
//Development/graphic/release/Core
//Development/graphic/release/Provider
//Development/graphic/release/WH
//Development/graphic/release/Accessory
//Development/graphic/release/Unity Provider
//Development/graphic/release/tipManager

Upvotes: 0

obi1
obi1

Reputation: 314

Use the Internal Field Separator (IFS) Variable to break on newline instead of white space.

This explains it well: http://mindspill.net/computing/linux-notes/using-the-bash-ifs-variable-to-make-for-loops-split-with-non-whitespace-characters/

An example pulled from the link above:

#!/bin/bash
IFS=$','
vals='/mnt,/var/lib/vmware/Virtual Machines,/dev,/proc,/sys,/tmp,/usr/portage,/var/tmp'
for i in $vals; 
     do echo $i; 
done
unset IFS

Upvotes: 0

Related Questions