Bdtyktl

I have a rather large bunch of files that contains several fields pipe-delimited.

5595340959340|1|MXPYAQWE|870569689456954654|0|0|20181018224126|1212121212121212121212121212|2|0|1000|70|33107||1|Event
5595340959340|1|MXPYAQWE|870569689456954654|0|0|20181018224126|2323232323232323232323232323|2|0|1000|70|33107||1|Event
5595340959340|1|MXPYAQWE|870569689456954654|0|0|20181018224126|3434343434343434343434343434|2|0|1000|70|33107||1|Event
5595340959340|1|MXPYAQWE|870569689456954654|0|0|20181018224126|4545454545454545454545454545|2|0|1000|70|33107||1|Event
5595340959340|1|MXPYAQWE|870569689456954654|0|0|20181018224126|5656565656565656565656565656|2|0|1000|70|33107||1|Event

Notice the eighth field. It currently has 29 characters and I'm supposed to trim it so it has only five characters left.

The only (convoluted) solution I've come up with is this:

Isolate the fields I want to trim:

awk -F "|" '{print $8}' > Original_Fields

Trim the fields

cp Original_Fields Temp

more Temp | cut -c -5 > Trimmed_Fields

Create a susbtitution script with sed

grep -rh -f <file_with_matching_strings> /path/to/files > Original_Strings

vi Original_Strings

:%s/^/grep -rl "/g

:%s/$/" /path/to/file | xargs sed -i 's//g

:wq!

And then edit the Original_Fields and Trimmed_Fields files, so I end up with

grep -rl /path/to/file | xargs sed -i 's/Original_Field/Trimmed_Field/g'

This works, but I strongly suspect there must be a quicker way to accomplish this with AWK and SED, so I can do all of this in just one step.

Yes, you can trim and rebuild each line with AWK:

awk -F'|' 'BEGIN { OFS = FS } { $8 = substr($8, 1, 5); print }'

This sets the input and output separators to “|”, and for each line of input, trims the eighth field to five characters at most, and prints all the fields (including the updated field).

Since your file is fixed width — or at least it seems — you can do it counting columns to keep and discard in sed:

sed -r 's/(.{68}).{23}(.*)/12/' yourfile