CODEWITHSUNDEEP
bash
Bowi
Bowi

Reputation: 1528

Read input multiple times

Consider this small example:

2020-05-11--karli.txt:

Let me tell you what my sister's best friend told her last week about that Mr. Smith living in the little house next to the car park in that little village, what's it called, Br... no, Berekre... I don't know the name, but she said that Mr. Smith bought a budgie last week and it already flew away last Sunday! Isn't that funny?

shortenKarli.sh:

#!/bin/bash
inputFile=$1
shortenWords=4

numWords=`wc -w $inputFile | sed "s/^\([0-9]\+\).*/\1/g"`
introText=`cat $inputFile | cut -d' ' -f 1-$shortenWords`
outroText=`cat $inputFile | cut -d' ' -f $(expr $numWords - $shortenWords + 1)-$numWords`

echo "That were $numWords words, let me shorten it for you:"
echo "$introText... ...blah blah... ...$outroText"

Calling 

shortenKarli.sh 2020-05-11--karli.txt

brings 

That were 63 words, let me shorten it for you: Let me tell you... ...blah blah... ...Sunday! Isn't that funny?

as expected.

But how could I modify the script to read from stdin, like 

phoneBot --callee=Karli --interruptions="Oh!;Mmh.;Well...;Really?" --speechrecognize=on | shortenKarli.sh

? It would need to read the same input 3 times.

I could write it to a temporary file, but is there something more elegant?

Upvotes: 2

Views: 358

Answers (1)

Barmar
Barmar

Reputation: 780929

You could read it into a variable instead of a temporary file.

if [[ $# = 1 ]]
then
    inputData=$(cat "$1")
else
    inputData=$(cat)
fi

Then in place of $inputFile in the rest of the script, use <<<"$inputData"

numWords=$(wc -w <<<"$inputData" | sed 's/^\([0-9]\+\).*/\1/g')
introText=$(cut -d' ' -f 1-$shortenWords <<<"$inputData")
outroText=$(cut -d' ' -f $(expr $numWords - $shortenWords + 1)-$numWords <<<"$inputData")

Upvotes: 4

Related Questions