mjhjmtu

I have a big text saving lots of file paths and their information

# just for demostration
/a/b/c/d/e
/a/b/c/d
/a/b/c
/a/b
/a
/b/c
/b
/c/d
/c
....

I want to show only those with specific number of specific char, for instance,

1. with only one /
   


2. with no more than two /
   

so that I extract only root folders or only to first-level subfolders.

With an extended regular expression:

grep -E '^([^/]*/[^/]*)1,2$' input

which results in these matches:

/a/b
/a
/b/c
/b
/c/d
/c

The regex syntax says:

• 
  ^ - anchored to the beginning of the line


• 
  ( ... ) - group the following bits together


• 
  [^/]* - any non-forward-slash character, zero or more of them


• 
  / - a forward-slash


• 
  [^/]* - any non-forward-slash character, zero or more of them


• 
  1,2 - one or two of those groups


• 
  $ - anchored to the end of the line

Awk solution:

-- with only one /:

awk -v sep=1 'BEGIN FS=OFS="/"; sep += 1 NF == sep' file

The output:

/a
/b
/c

-- with no more than two /:

awk -v sep=2 'BEGIN FS=OFS="/"; sep += 1 NF <= sep' file

The output:

/a/b
/a
/b/c
/b
/c/d
/c

• 
  sep - variable; stands for directory separator's count


• 
  FS and OFS - are field separator and output field separator respectively


• 
  NF - stands for total number of fields
  

If you need to combine the above approaches into a single awk command:

awk -v sep=1 'BEGIN FS=OFS="/"; sep += 1 (sep==2 && NF==sep) || (sep>2 && NF<=sep)' file