How to use sed to replace strings with categories?

Question

I have some categories and their values:

Category: Value: 1*4 15.934934 1*5*3 19.281281 1*8*10*2 78.20912

I would like to replace the categories by numbers that represent the categories.

For example:

1*4 has two numbers then it would be replaced by 2

1*5*3 has three numbers then it would be replaced by 3

1*8*10*2 has four numbers then it would be replaced by 4

I need to do the process for hundreds of files so it is very important to automate the process!

Does it have to be sed? there's a straightforward way in awk for example — steeldriver, CommentedJul 3, 2017 at 18:08
Do a gsub of sequences of digits in $1 and assign the result (number of successful substitutions) back to $1 i.e. $1 = gsub(/[0-9]+/,"",$1) — steeldriver, CommentedJul 3, 2017 at 18:51

Jeff Schaller · Accepted Answer · 2017-07-03 22:20:40Z

Similar to steeldriver's idea, but counting stars:

awk '{ $1 = 1 + gsub(/\*/, "", $1); print; }' < input > output

When run on the sample input, it results in:

2 15.934934 3 19.281281 4 78.20912

Michael Vehrs · Accepted Answer · 2017-07-04 09:35:34Z

As sed has no concept of numbers, this is a bit laborious to do:

sed -E 's/^[0-9]+\*/1*/; s/^1\*[0-9]+/2/; s/^2\*[0-9]+/3/; s/^3\*[0-9]+/4/;' input

2 Answers 2