UNIX for Dummies Questions & Answers

I have a text file that is about 90,000 lines long. How would I delete lines 64-89, 152-177, 240-265, 328-353... etc? The sections I would like to delete are 26 lines long and the number of lines between the sections I would like to delete is 62 lines. Thanks very much in advance.

Suggestion - there must be an observable pattern to this, not just line numbers. We would have to have a huge list of line numbers, or try to calculate offsets to attempt what you ask as it stands.

Using patterns allows for things like, 'Oops, the first line to delete is not 62 but 63 lines from the top'. In other words, you applied some kind of logic to your data. To do this in a reasonable way we need to understand the layout and logic behind it.

Alternatively, you need to generate a file that is made of ranges of line numbers and verify there are no Oops moments in there. Otherwise it is likely this exercise will turn your file into hash. You don't want to have to verify 50000 lines by hand. And if verification is simple - that means there is a pattern we can use.

This is one way:

Code:

$ wc -l 90000_records.txt
90000 90000_records.txt

$ sed '64,89d;152,177d;240,265d;328,353d;' 90000_records.txt > new

$ wc -l new
89896 new

Thanks for the reply Jim. I understand what you mean about needing a pattern. Ive edited my file a bit. Now the first line in the section I want to delete always begins with the term "New_matrix." Would it be possible to search the file for lines that begin with New_matrix and then delete it and the 24 lines that follow it?

The way I interpret the description of the problem, the 1st line to be deleted is line 64. Then there is a loop where 26 lines are deleted and 62 lines are copied until we run out of input lines. The following does this:

Code:

#!/bin/ksh
CopyCount=62    # # of lines to copy
FirstSkip=64    # Line # of 1st line to skip
SkipCount=26    # # of lines to skip
awk -v fs="$FirstSkip" -v cc="$CopyCount" -v sc="$SkipCount" '
BEGIN { i = fs - 1;j = sc }
i     { i--;print;next }
j     { j--;next }
      { i = cc - 1;j = sc;print }' file

I use the Korn shell, but this will work with any shell that recognizes basic Bourne shell syntax. If you are using a Solaris/SunOS system, use /usr/xpg4/bin/awk, /usr/xpg6/bin/awk, or nawk instead of awk.

I think Jim did not want you to edit the file. Instead you should look for existing patterns: an empty line, a different indentation, ...
It is unusual that a file has fixed line numbers.

Thank you Don. That worked perfectly! Thank you all who replied.