A script to filter out repeated partial slides in PDF presentations (especially lecture slides made in LaTeX Beamer) and generate a printer-friendly version.

pdf-autoselect.sh

#!/bin/bash

# Autor: Mikołaj Dądela <[email protected]>

inputPdf="$1"

threshold="10"

if [ "$inputPdf" == "" ]; then
    echo "Usage: $0 input.pdf [output.pdf] [<threshold>]"
    echo ""
    echo "Default output filename for foo.pdf is foo_s.pdf."
    echo ""
    echo "Threshold is the minimal file size (in KB) of difference PNG. After"
    echo "running this script have a look at diff files to decide what value"
    echo "would be ok for you. The default is $threshold."
    exit 1
fi
if [ ! -r "$inputPdf" ]; then
    echo "Unable to open $inputPdf."
    exit 1
fi

basename="${inputPdf%.pdf}"
basename="${basename%.PDF}"

outputPdf="${basename}_s.pdf"

if [ "$2" != "" ]; then
    echo "$2" | grep -q "^[0-9]*$"
    if [ "$?" == "0" ]; then #if it's a number
        threshold="$2"
    elif [ "$2" != "" ]; then
        outputPdf="$2"
    fi
fi

if [ "$3" != "" ] ; then
    echo "$3" | grep -q "^[0-9]*$"
    if [ "$?" == "0" ]; then #if it's a number
        threshold="$3"
    fi
fi

# ----------------------------------------------------------------------


numPages=`pdftk "$inputPdf" dump_data | grep NumberOfPages | cut -d' ' -f2`
echo "$inputPdf contains $numPages pages."

# 1st phase: thumbnails
thumbsDir="${basename}_thumbs"

if [ ! -f "$thumbsDir/p-0.png" ]; then
    mkdir -p "$thumbsDir"
    for i in `seq 0 $((numPages/10))`; do
        start=$((i*10))
        end=$((((i+1)*10)-1))
        if [ "$end" -ge "$numPages" ]; then
            end=$((numPages-1))
        fi
        echo -ne "\rGenerating thumbnails... $start-$end"
        convert -resize 200x200 "$inputPdf[$start-$end]" $thumbsDir/p.png || exit
    done
    echo -e "\rGenerating thumbnails... done. (saved to $thumbsDir/p-*.png)"
else
    echo "Using found $thumbsDir/p-*.png files."
fi


# 2nd phase: diffs
if [ ! -f "$thumbsDir/diff-0.png" ]; then
    i=0
    while true; do
        if [ ! -f "$thumbsDir/p-$((i+1)).png" ]; then
            break
        fi
        echo -ne "\rComparing pages... $i "
        # this blending mode catches only things that disappear
        # on the next slide - this what appears or doesn't change
        # comes out as white on the diff image.
        convert "$thumbsDir/p-$((i+1)).png" -negate \
            "$thumbsDir/p-$i.png" \
            -compose Plus -composite \
            "$thumbsDir/diff-$i.png" || exit
        i=$((i+1))
    done
    echo -e "\rComparing pages... done. (saved to $thumbsDir/diff-*.png)"
else
    echo "Using found $thumbsDir/diff-*.png files."
fi


# 3rd phase: selection
i=1
pages="1"
echo -n "Selecting pages (with size threshold = ${threshold}KB)... $pages"
while [ -f "$thumbsDir/diff-$i.png" ]; do
    if [ "`stat -c%s "$thumbsDir/diff-$i.png"`" -gt $((threshold*1024)) ]; then
        echo -n " $((i+1))"
        pages="$pages $((i+1))"
    fi
    i=$((i+1))
done
pages="$pages $((i+1))"
echo " $((i+1)) done."


# 4th phase: output
echo -n "Writing $outputPdf... "
pdftk "$inputPdf" cat $pages output "$outputPdf" || exit
echo "done."


echo "Finished. You can now remove $thumbsDir/."