#!/bin/bash
#
## Wizard Kit: MS Word content search tool
TMP_FILE="$(mktemp)"

IFS=$'\n'
for s in $*; do
    REGEX="$s"
    REGEX=$(echo "$REGEX" | sed -r 's/\s+/\\s\*/g')
    
    # Word Doc
    for d in *doc; do
        if antiword "$d" | grep -iqsP "($REGEX)"; then
            echo "Possible match: $d"
            echo "$d" >> "$TMP_FILE"
        fi
    done
    
    # Word Docx
    for d in *docx; do
        if unzip -p "$d" word/document.xml | grep -iqsP "($REGEX)"; then
            echo "Possible match: $d"
            echo "$d" >> "$TMP_FILE"
        fi
    done

done

# Cleanup results
if [[ -s "$TMP_FILE" ]]; then
    sort -u "$TMP_FILE" >> "$HOME/msword-matches.txt"
fi
rm "$TMP_FILE"

# Done
if [[ -s "$HOME/msword-matches.txt" ]]; then
    echo "Found $(wc -l "$HOME/msword-matches.txt") possible matches"
    echo "The results have been saved to $HOME"
fi