Editing n-1th occurrence of a character in a file

[cbres00]

Some of you may remember my posts from last week regarding swapping a slash with a pipe in a file. While I got a successful resolution to that particular problem, I found a new problem with a different file.

In a nutshell, every record in my pipe-delimited file can have one or more slashes, the first one(s)one occurring in the very first field, and the last slash occuring in the last field. I need to preserve the last slash, but convert the next-to-last slash to a pipe.

Ex:
(spaces added for emphasis)

aa/bb/cc /dd| ee|ff|gg|/hh|

I need to change this to:
aa/bb/cc |dd| ee|ff|gg|/hh|

The solution I got from this wonderful community works for the first example but not the second:

while
read input_record
do
start=${input_record%/*}
last=${input_record##*/}
print "$start|$last" >> ${output_file}
done < ${input_file}

Could this code be easily changed to accomodate my needs?

Regards,
Cathy

[curt larson_1 · Accepted solution]

if you just want to modify what you have

while
read input_record
do
Last=${input_record#*|} #everything after the first |
Start=${input_record%%|*} #everything before the first |

start=${Start%/*} # everything before the last / in $Start
last=${Start##*/} # everything after the last / in $Start

print "$start|$last|$Last" >> ${output_file}
done < ${input_file}

[curt larson_1]

but now sed and perl become a bit more useful

sed 's!\([^|]*\)/\(.*|.*\)/\1|\2!'

replaces the last / before the first | with |

[cbres00]

Curt,
Sometimes the solution is SO simple it's easy to miss! I like your first solution simply because it's more readable for the developer who may need to modify this in the future.

Superb work! Many thanks (and points!!).

Cathy

[cbres00]

One minor correction:

I wasn't getting anything out of $Last or $Start. I neutralized the pipe and got results:

Last=${input_record#*\|}
Start=${tactic_state_record%%\|*}

Thanks again!
Cathy