getMixedWords

#!/bin/bash
# Script to find mixed words in killer_cronicas.txt

INPUT=input.txt
WORDS=words.txt
COUNT=count.txt
MIXED=mixed_words.txt

if [[ $# -ne 1 ]]; then
    echo "Usage: $0 Filename ";
    exit 1;
else
    cp "$1" "$INPUT"; # Preserve original copy
fi

aspell -C list < $INPUT > $WORDS
aspell -C -des list < $WORDS > temp && mv temp $WORDS
aspell -C -dit list < $WORDS > temp && mv temp $WORDS
aspell -C -dfr list < $WORDS > temp && mv temp $WORDS
aspell -C -dpt_BR list < $WORDS > temp && mv temp $WORDS
aspell -C -dpt_PT list  < $WORDS > temp && mv temp $WORDS
aspell -C -daf list  < $WORDS > temp && mv temp $WORDS
    # Filter out valid words in English, Spanish, Italian, French,
    # Portuguese, and Afrikaans

tr '[:upper:]' '[:lower:]' < $WORDS | sort > temp && mv temp $WORDS 

grep -v -E "killer|crónic[a|as]|xx|arg" $WORDS > temp && mv temp $WORDS
grep -v -E "maría|pizarnik" $WORDS > temp && mv temp $WORDS
grep -v -E "sh[o|a]|voh|dra|th|pomo|ion|ex|line" $WORDS > temp && mv temp $WORDS
grep -v -E "of|hasn|isn|hadn|didn|couldn|wouldn|doesn|weren|wasn|ness|nne|liz" $WORDS > temp && mv temp $WORDS
grep -v -E "^..$" $WORDS > temp && mv temp $WORDS
grep -v -E "^...$" $WORDS > temp && mv temp $WORDS
grep -v -E "h$" $WORDS > temp && mv temp $WORDS

uniq -i -c $WORDS | sort -rn > $COUNT

uniq -i $WORDS > temp && mv temp $WORDS

wc $WORDS

uniq -i -c $WORDS | sort -rn > $COUNT
grep -v -E "ao|ía|ht|zh|sh|ñ|hc|ó|hp" $WORDS > temp && mv temp $WORDS

wc $WORDS

uniq -i -c $WORDS | sort -rn > $COUNT

awk '{ print $2 }' $COUNT > temp && mv temp $COUNT

for i in `cat $COUNT`; do
    read -p "Is \"$i\" a word or name (y/n)? " choice
    case "$choice" in 
        y|Y ) grep -i -v "$i" $WORDS > temp && mv temp $WORDS;;
        n|N ) ;;
        * ) echo "invalid";;
    esac 
done