Reorder pages in a PDF file to print a booklet

pdf-booklet.sh

#!/usr/bin/env bash
# Reorder pages in a PDF file to print a booklet.
# Then, print the output PDF in duplex mode, short-edge binding, 2 pages per sheet.

SCRIPT=$(basename "$BASH_SOURCE")
INFILE=
OUTFILE=
VERBOSE=false
SYSTEM=$(uname -s)

error() { echo >&2 "$SCRIPT: $*"; exit 1; }
usage() { echo >&2 "usage: $SCRIPT infile [-v] [-o outfile]"; exit 1; }
verbose() { $VERBOSE && echo >&2 "$SCRIPT: $*"; }

# Find various commands. The commands qpdf and jq are required.
# The command pdfinfo is optional, it comes with package poppler-utils (Ubuntu) or xpdf (macOS Homebrew).
[[ -z $(which qpdf 2>/dev/null) ]] && error "command qpdf not installed"
[[ -z $(which jq 2>/dev/null) ]] && error "command jq not installed"
PDFINFO=$(which pdfinfo 2>/dev/null)

# Get file size in bytes of a file.
file-size()
{
    [[ $SYSTEM == Darwin ]] && stat -f %z "$1" || stat -c %s "$1"
}

# Build a PDF file name with a pre-suffix.
out-pdf-file()
{
    local dir=$(dirname "$1")
    [[ $dir == . ]] && dir= || dir+="/"
    echo $dir$(basename "$1" .pdf).$2.pdf
}

# Get page "width height" of a PDF file, in points.
pdf-page-size()
{
    local a4="595 842"
    local file="$1"
    local size=
    if [[ -z $file ]]; then
        # No file provided, use A4 as default.
        size="$a4"
    fi
    if [[ -z $size && -n $PDFINFO ]]; then
        # Command pdfinfo available, try it.
        size=$($PDFINFO "$file" | sed -e '/^Page size:/!d' -e 's/^Page size:[[:space:]]*//' -e 's/ *pts.*$//' -e 's/ *x */ /' | head -1)
    fi
    if [[ -z $size ]]; then
        # Command pdfinfo not found or did not find size, use command qpdf.
        size=$(qpdf "$file" --json | jq -j 'first(.objects[] | select(."/MediaBox") | ."/MediaBox") | .[2]," ",.[3]' 2>/dev/null)
    fi
    if [[ -z $size ]]; then
        # Size not found, use A4 as default.
        size="$a4"
    fi
    echo $size
}

# Create a blank PDF file, with optional reference PDF file for page size.
create-blank-pdf()
{
    local outfile="$1"
    local reffile="$2"
    local size=$(pdf-page-size "$reffile")
    echo '%PDF-1.4' >"$outfile"
    local obj1=$(file-size "$outfile")
    echo '1 0 obj<</Type/Catalog/Pages 2 0 R>>endobj' >>"$outfile"
    local obj2=$(file-size "$outfile")
    echo '2 0 obj<</Type/Pages/Count 1/Kids[3 0 R]>>endobj' >>"$outfile"
    local obj3=$(file-size "$outfile")
    printf '3 0 obj<</Type/Page/MediaBox[0 0 %s]/Parent 2 0 R/Resources<<>>>>endobj\n' "$size" >>"$outfile"
    local xref=$(file-size "$outfile")
    echo 'xref' >>"$outfile"
    echo '0 4' >>"$outfile"
    echo '0000000000 65535 f' >>"$outfile"
    printf '%010d 00000 n\n' $obj1 >>"$outfile"
    printf '%010d 00000 n\n' $obj2 >>"$outfile"
    printf '%010d 00000 n\n' $obj3 >>"$outfile"
    echo 'trailer<</Size 4/Root 1 0 R>>' >>"$outfile"
    echo 'startxref' >>"$outfile"
    printf '%d\n' $xref >>"$outfile"
    echo '%%EOF' >>"$outfile"
}

# Decode command line parameters.
while [[ $# -gt 0 ]]; do
    case "$1" in
        -o)
            [[ $# -gt 1 ]] || usage; shift
            [[ -n "$OUTFILE" ]] && usage
            OUTFILE="$1"
            ;;
        -v)
            VERBOSE=true
            ;;
        -*)
            usage
            ;;
        *)
            [[ -n "$INFILE" ]] && usage
            INFILE="$1"
            ;;
    esac
    shift
done

# Default output file name.
[[ -z "$OUTFILE" ]] && OUTFILE=$(out-pdf-file "$INFILE" booklet)

# Count pages.
INPAGES=$(qpdf --show-npages "$INFILE")
OUTSHEETS=$((($INPAGES + 3) / 4))
OUTPAGES=$(($OUTSHEETS * 4))
[[ "$INPAGES" -gt 0 ]] || error "$INFILE: empty or invalid PDF file"
verbose "$INFILE: $INPAGES pages, rounded to $OUTPAGES pages, $OUTSHEETS sheets"

# If input number of pages is not a multiple of 4, we need a blank page template PDF file.
BLANKFILE=
if [[ $INPAGES -ne $OUTPAGES ]]; then
    BLANKFILE=$(out-pdf-file "$OUTFILE" blank)
    create-blank-pdf "$BLANKFILE" "$INFILE"
fi

# Build page order.
PAGELIST=
for p in $(seq 0 2 $((($OUTPAGES - 4) / 2))); do
    PAGELIST="$PAGELIST $(($OUTPAGES - $p)) $(($p + 1)) $(($p + 2)) $(($OUTPAGES - $p - 1))"
done
PAGELIST=${PAGELIST/# /}

# Build parameters for qpdf --pages (depends on the blank page requirements).
if [[ -z "$BLANKFILE" ]]; then
    # All pages are from the same file.
    PAGEOPT=("$INFILE" ${PAGELIST// /,})
else
    PAGEOPT=()
    CURFILE=
    for p in $PAGELIST; do
        if [[ $p -gt $INPAGES ]]; then
            FILE="$BLANKFILE"
            p=1
        else
            FILE="$INFILE"
        fi
        if [[ $FILE != $CURFILE ]]; then
            PAGEOPT+=("$FILE" $p)
            CURFILE="$FILE"
        else
            PAGEOPT[-1]+=",$p"
        fi
    done
fi

# Generate the output in one pass.
qpdf --empty --pages "${PAGEOPT[@]}" -- "$OUTFILE"

# Final cleanup.
[[ -n "$BLANKFILE" ]] && rm -rf "$BLANKFILE"
verbose "print $OUTFILE, duplex mode, short-edge binding, 2 pages per sheet"
exit 0