Add dir-to-pdfv3.sh, for use alongside bookpipeline

1 files changed, 118 insertions, 0 deletions

diff --git a/dir-to-pdfv3.sh b/dir-to-pdfv3.sh
new file mode 100755
index 0000000..4248be9
--- /dev/null
+++ b/dir-to-pdfv3.sh
@@ -0,0 +1,118 @@
+#!/bin/sh
+usage="Usage: $0 [-c] indir [pdf]
+
+Creates a PDF from image and hocr files in indir, saving it to pdf, or
+indir.pdf if not specified.
+
+The necessary files are first copied to a temporary directory where
+they are renamed and reformatted for the hocr-pdf tool from hocr-tools.
+
+The PDF is then created with hocr-pdf, and the temporary directory is
+removed.
+
+This is designed to work with files which have gone through the
+rescribe.xyz/bookpipeline process, with a 'best' file which lists the
+best hocr files for each page.
+
+-c: colour output"
+
+# Set image compression and dpi
+QUALITY=20
+DPI=600
+
+# Set resize ratio
+RESIZEPERC=25%
+RESIZEDIV=4
+DPI=`expr $DPI / $RESIZEDIV`
+
+colour=0
+test $# -gt 1 && test "$1" = "-c" && colour=1 && shift
+
+test $# -lt 1 -o $# -gt 2 && echo "$usage" && exit 1
+
+if command -v gm > /dev/null ; then
+	convert="gm convert"
+elif command -v convert > /dev/null ; then
+	convert="convert"
+else
+	echo "Error: no graphicksmagick or imagemagick found"
+	exit 1
+fi
+
+if ! command -v hocr-pdf > /dev/null ; then
+	echo "Error: no hocr-pdf tool found"
+	exit 1
+fi
+
+if ! test -d "$1"; then
+	echo "Error: $1 does not exist"
+	exit 1
+fi
+
+if test $# -eq 2 ; then
+	outfile="$2"
+else
+	o=`echo "$1" | sed 's/\/$//'`
+	outfile="$o.pdf"
+fi
+
+tmpdir=`mktemp -d`
+if test $? -ne 0 ; then
+	echo "Error: Failed to create temporary directory"
+	exit 1
+fi
+
+mkdir -p "$tmpdir" || exit 1
+
+echo "Copying hocrs and converting jpgs from $1 to $tmpdir"
+while read i; do
+	if ! test -f "$1/$i"; then
+		echo "Warning: no hocr file found for $i, skipping"
+		continue
+	fi
+
+	n=`echo "$i" | sed 's/_bin.*//'`
+
+	b=`basename "$i" .hocr`
+	if test $colour -eq 1; then
+		img="$1/$n.jpg"  # colour
+	else
+		img="$1/$b.png" # binarised
+	fi
+	if ! test -f "$img"; then
+		echo "Warning: no image found for hocr file $i, skipping"
+		continue
+	fi
+
+	$convert -quality $QUALITY -geometry ${RESIZEPERC}x${RESIZEPERC} -density ${DPI}x${DPI} "$img" "$tmpdir/$n.jpg" || exit 1
+
+	# Adjust the bounding boxes to match the new geometry after resizing
+	cat "$1/$i" | while read line; do
+		if `echo "$line" | grep -q -v "title=['\"]bbox"`; then
+	                printf "%s\n" "$line"
+	                continue
+	        fi
+		# get original values
+	        bbox=`echo "$line" | sed "s/.*title=['\"]bbox \(.[^;']*\)[;'\"].*/\1/"`
+	        x1=`echo "$bbox"|awk '{print $1}'`
+	        y1=`echo "$bbox"|awk '{print $2}'`
+	        x2=`echo "$bbox"|awk '{print $3}'`
+	        y2=`echo "$bbox"|awk '{print $4}'`
+
+	        # first halve all values
+	        x1=`echo "$x1 / $RESIZEDIV" | bc`
+	        y1=`echo "$y1 / $RESIZEDIV" | bc`
+	        x2=`echo "$x2 / $RESIZEDIV" | bc`
+	        y2=`echo "$y2 / $RESIZEDIV" | bc`
+
+		newbbox="$x1 $y1 $x2 $y2"
+	        newline=`echo "$line" | sed "s/$bbox/$newbbox/"`
+	        printf "%s\n" "$newline"
+	done > "$tmpdir/$n.hocr"
+done < "$1/best"
+
+echo "Creating PDF"
+hocr-pdf --savefile "$outfile" "$tmpdir" || exit 1
+
+echo "Created a PDF at $outfile"
+rm -rf "$tmpdir"