Lazy indexed local PDF search tool.

pdf-search.sh

#!/usr/env bash

# This script works by using pdftotext and pdfgrep along with standard grep to make a faster PDF searching tool.

# PDF searching is slow, searching through plain text is fast. This tool extracts all plain text and stores it seperately.
# When a search is done it is checked against the plain text first where a list of matching files are stored. 
# It then searches only the matching PDFs to get the page numbers and print the snipit of data.

# TODO - Add loose matching for bad OCR by doing multiple passes replacing visually similar characters and putting white space in query.

key="$1"

dir_text="text"

# Make text dir
if [[ ! -d "$dir_text" ]]
then
    mkdir "$dir_text"
fi

# Index all PDFs
for file in *.pdf
do
    if [[ ! -f "$dir_text/$file.txt" ]]
    then
        pdftotext "$file" "$dir_text/$file.txt"
    fi
done

files=( $(grep -i "$key" text/*.txt | awk -F ":" '{print $1}' | sort -u | sed 's|\.txt||g' | sed 's|text/||g') )

for file in "${files[@]}"
do
	echo "In [$file]"
	pdfgrep -ni "$key" "$file"
done