Compare two PDFs using ImageMagick - provides a visual comaprison and a perceptual hash comparison (numerical)

README.md

PDF tools for comparing PDFs visually (overlaying two PDFs to see changed areas) and using a perceptual hash (numerical value indicating visual difference between the two files).

Useful for command line review of PDFs and de-duplication. Configure git to use these tools for better PDF history / comparison in git.

These scripts require imagemagick and poppler. Both installed from homebrew.

---

Setup git to use a custom diff using:

.gitattributes:

*.pdf binary diff=pdf

.gitconfig:

[diff "pdf"]
    ; textconv = ~/bin/pdf2layout
    command = ~/bin/git-diff-pdf

git-diff-pdf

#!/bin/bash

if [[ -z "$1" || -z "$2" ]]; then
    echo "Usage: $0 <pdf1> <pdf2>"
    exit 1
fi

echo "comparing [$1] and [$2]"

# pdf2layout from poppler on homebrew (brew install poppler)
echo "*** text content"
diff <(~/bin/pdf2layout "$1") <(~/bin/pdf2layout "$2")

echo "*** image perceptual hash"
~/bin/pdf-compare-phash "$1" "$2"

pdf-compare-phash.py

#!/bin/bash

if [[ -z "$1" || -z "$2" ]]; then
    echo "Usage: $0 <pdf1> <pdf2>"
    exit 1
fi

convert -metric phash "$1" null: "$2" -compose Difference -layers composite -format '%[fx:mean]\n' info:

pdf-compare-visual.py

#!/bin/bash

if [[ -z "$1" || -z "$2" ]]; then
    echo "Usage: $0 <pdf1> <pdf2>"
    exit 1
fi

TMP=$(mktemp --suffix=.png)
echo "Comparing [$1] to [$2]"
echo "Saving difference in $TMP"
echo

DENSITY=100

# this supports both simple file names and page indexed file names like:
# file[0] file[1] - will either return one line for each page, or a single
# line if a single page is specified
PAGES1=$(magick identify "$1" | wc -l)
PAGES2=$(magick identify "$2" | wc -l)

if (( PAGES1 != PAGES2 )); then
    echo "Number of pages between documents does not match: $PAGES1 != $PAGES2"
    echo "Only comparing the first page."

    magick compare -density "$DENSITY" -background white "$1[0]" "$2[0]" "$TMP"
    PHASH_DIFF=$(~/bin/pdf-compare-phash "$1[0]" "$2[0]")
elif (( PAGES1 > 5 )); then
    echo "Too many pages ($PAGES1 > 5) to create hyper-image with all pages."
    echo "Only comparing first page."
    magick compare -density "$DENSITY" -background white "$1[0]" "$2[0]" "$TMP"
    PHASH_DIFF=$(~/bin/pdf-compare-phash "$1[0]" "$2[0]")
else
    # convert the PDFs into a single image with the pages vertically stacked
    ALL1=$(mktemp --suffix=.png)
    magick convert -density "$DENSITY" "$1" -append "$ALL1"
    ALL2=$(mktemp --suffix=.png)
    magick convert -density "$DENSITY" "$2" -append "$ALL2"

    magick compare -density "$DENSITY" -background white "$ALL1" "$ALL2" "$TMP"
    PHASH_DIFF=$(~/bin/pdf-compare-phash "$ALL1" "$ALL2")
fi

if [ "$TERM_PROGRAM" = "iTerm.app" ]; then
    echo "Visual difference between images:"
    echo "--------------------------------"
    imgcat-small "$TMP"
    echo "--------------------------------"
else
    open "$TMP"
fi

echo "Perceptual hash difference (0 is exactly the same): $PHASH_DIFF"