CODEWITHSUNDEEP
bashawksed
user3639557
user3639557

Reputation: 5291

Converting a text file by replacing words with numbers

I have a cpp code for it. It basically takes a dictionary file formatted as:

blue 1
cat 2
chased 3
dog 4
. 5
....

and takes a text file:

blue cat chased dog .
yellow carrot ate brown fish .

and it converts it into:

1 2 3 4 5
88 90 121 11 133 5
......

Is there a simple one line solution for it in Bash?

Upvotes: 0

Views: 160

Answers (5)

choroba
choroba

Reputation: 241988

Create a sed script from the input file:

sed 's/^/s=/;s/ /=/;s/$/=/' file

And run it on the input:

sed 's/^/s=/;s/ /=/;s/$/=/' file | sed -f- input

This might not work if a word is part of another word, e.g. cat and category.

Perl solution: read the first file into a hash table, then read the second file and replace each word with the corresponding value from the hash table.

perl -lane 'if (! $second) { $h{ $F[0] } = $F[1] }
                else { s/(\S+)/$h{$1}/g; print }
            $second = 1 if eof;' file input

Upvotes: 0

James Brown
James Brown

Reputation: 37424

In awk implementing @karakfa envisioned missing dictionary item:

$ awk 'NR==FNR {
           a[$1]=$2;                 # store dict to a hash
           if($2>m)                  # m is the max number in dict
               m=$2;
           next
       } {
           for(i=1;i<=NF;i++)        # iterate thru all words in record
               if($i in a)           # if a dict match is found
                   $i=a[$i];         # replace it
               else {                # if not
                   a[$i]=++m;        # grow m and make new dictionary entry
                   # print a[$i], m > "new_items" # to store them to a file
                   $i=m              # ... and use it
               }
           } 1' dict text

Upvotes: 0

dawg
dawg

Reputation: 104032

For silliness, here is pure Bash (you should use awk for this IMHO):

declare -A dict
while read k v; do 
    dict[$k]=$v
done < /tmp/f1.txt

while IFS= read -r line || [[ -n $line ]]; do 
    la=($line)
    for word in ${la[@]}; do 
        [[ ${dict[$word]} ]] && printf "%s " ${dict[$word]}; done
    echo
done < /tmp/f2.txt

Upvotes: 0

karakfa
karakfa

Reputation: 67507

awk to the rescue!

$ awk 'NR==FNR {dict[$1]=$2; next} 
               {for(i=1;i<=NF;i++) $i=dict[$i]}1' dict file

perhaps add logic for handling missing items in dictionary

Upvotes: 3

codeforester
codeforester

Reputation: 43039

@choroba's sed solution didn't work for me. I am not sure if there is a one line solution for this. I would do this in Bash:

#!/bin/bash

# read the word values from the first file into an associative array
declare -A map
while IFS=' ' read -r word value; do
  map[$word]=$value
done < 1.txt

# traverse the second file and print out numbers corresponding to each word
# if there is no mapped number, print nothing
while read -r line; do
  read -ra words <<< "$line"
  for word in ${words[@]}; do
    num="${map[$word]}"
    [[ $num ]] && printf "%s " "${map[$word]}"
  done
  printf "\n"
done < 2.txt

Gives the following output for the files in your question:

1 2 3 4 5
5

Upvotes: 0

Related Questions