Reduce file size of PDF documents with embedded images

pdfcrunch

#!/bin/bash
##
##  Reduce a PDFs byte size by using either grayscale images and/or
##  the '/ebook' output target of 'gs' (Ghostscript)
##
##  Requires: Ghostscript probably 7.something or higher
##
##  Author:   Kevin Ernst
##  Date:     ages ago
##
set -u

ME=$(basename $BASH_SOURCE)
GIST='https://gist.github.com/ernstki/d94052852e1ee0826d05c1a63de2e4a8'

# don't print ANSI color codes unless stdout is a terminal
if [[ -t 1 ]]; then
    BOLD=$(tput bold)
    UL=$(tput sgr 0 1)
    BLUE=$(tput setaf 4)
    YELLOW=$(tput setaf 3)
    RESET=$(tput sgr0)
else
    BOLD=;UL=;BLUE=;YELLOW=;RESET=
fi
WARNING="$BOLD${YELLOW}WARNING$RESET:"

suf='_crunched'
serial=
grayscale=

# '-h' or '--help'
if [[ $1 =~ ^--?h ]]; then
    echo "
  $BOLD$BLUE$ME$RESET - reduce file size of PDFs with embedded images

    ${UL}usage$RESET:  $ME [-h|--help] [-g|--grayscale] FILE [FILE ...]

    ${UL}where$RESET:  -h / --help       you're looking at it
            -g / --grayscale  make PDFs a tiny bit smaller by converting to B&W
                              ('--greyscale' also accepted)

    ${UL}source$RESET: $GIST
"
    exit
fi

# '-g', '--grayscale', or '--greyscale'
if [[ $1 =~ ^--?g ]]; then
    # Source http://stackoverflow.com/a/20133915
    grayscale="-sProcessColorModel=DeviceGray \
               -sColorConversionStrategy=Gray \
               -dOverrideICC"
    shift
fi

while (( $# )); do
    infile=$1
    outfile="${infile%.pdf}${suf}.pdf"

    if [[ ! -f $1 ]]; then
        echo "$WARNING Input file '$1' not found. Skipping." >&2
        shift; continue
    fi

    # increment the serial each time so we're not clobbering existing files
    # not because I really care, but just because I wondered how to do it
    while [[ -f $outfile ]]; do
        serial=$(printf "%03d" $(( $serial + 1 )))
        outfile="${infile%.pdf}${suf}-${serial}.pdf"
    done

    gs -q -dBATCH -dNOPAUSE -sDEVICE=pdfwrite -dPDFSETTINGS=/ebook \
        $grayscale \
        -o "$outfile" \
        "$infile"

    # -c "/osetcolor {/setcolor} bind def /setcolor {pop [0 0 0] osetcolor} def" \
    #   -sOutputFile="$outfile" "$infile"
    # (didn't seem to work for PDF; maybe try again someday)
    # source: http://superuser.com/a/249038

    if [[ $grayscale ]]; then
        echo "Finished writing grayscale output to '$outfile'."
    else
        echo "Finished writing output to '$outfile'."
    fi

    # work on the next file
    shift
done

See also: https://gist.github.com/ahmed-musallam/27de7d7c5ac68ecbd1ed65b6b48416f9