#! /bin/sh # file fig2pdf.sh # This file is part of LyX, the document processor. # Licence details can be found in the file COPYING. # # author Angus Leeming # # Full author contact details are available in file CREDITS # This script converts an XFIG image to something that pdflatex can process # into high quality PDF. # Usage: sh fig2pdf.sh ${base}.xfig # to generate ${base}.pdftex_t # Thereafter, you need only '\input{${base}.pdftex_t}' in your latex document. # modern_xfig() and legacy_xfig() are the functions that do all the work. # Modern versions of xfig can output the image without "special" text as # a PDF file ${base}.pdf and place the text in a LaTeX file # ${base}.pdftex_t for typesetting by pdflatex itself. modern_xfig() { input=$1.fig pdftex=$1.pdf pdftex_t=$1.pdftex_t fig2dev -Lpdftex ${input} ${pdftex} fig2dev -Lpdftex_t -p$1 ${input} ${pdftex_t} exit 0; } # Older versions of xfig cannot do this, so we emulate the behaviour using # pstex and pstex_t output. legacy_xfig() { input=$1.fig pstex=$1.pstex png=$1.png pdftex_t=$1.pdftex_t fig2dev -Lpstex ${input} ${pstex} fig2dev -Lpstex_t -p$1 ${input} ${pdftex_t} # Convert the ${pstex} EPS file (free of "special" text) to PDF format # using gs. # gs is extremely fussy about the EPS files it converts, so ensure it is # "clean" first. clean=${pstex}.$$ eps2eps ${pstex} ${clean} rm -f ${pstex} # Extract the width and height of the image using gs' bbox device. # Ie, take output that includes line "%%BoundingBox: 0 0 " # and rewrite it as "-gx" geometry=`gs -q -dSAFER -dNOPAUSE -dBATCH -sDEVICE=bbox ${clean} 2>&1 | \ sed '/^%%BoundingBox/! d' | cut -d' ' -f4,5 | \ sed 's/^\([0-9]\{1,\}\) \([0-9]\{1,\}\)$/-g\1x\2/'` # Generate a PNG file using the -g option to ensure the size is the same # as the original. # If we're using a version of gs that does not have a bbox device, then # ${geometry} = "", so there are no unwanted side effects. gs -q -dSAFER -dBATCH -dNOPAUSE ${geometry} -sDEVICE=png16m \ -sOutputFile=${png} ${clean} rm -f ${clean} exit 0; } # The main logic of the script is below. # All it does is ascertain which of the two functions above to call. # We expect a single arg, the name of the input file. test $# -eq 1 || exit 1 # Remove the .fig extension input=`basename $1` base=`echo ${input} | sed 's/\.fig$//'` # Ascertain whether fig2dev is "modern enough". # Here "modern" means "fig2dev Version 3.2 Patchlevel 4" version_info=`fig2dev -h | sed '/^fig2dev/! d'` # If no line begins "fig2dev" then default to legacy_xfig test "x${version_info}" = "x" && legacy_xfig ${base} version=`echo ${version_info} | cut -d' ' -f3` patchlevel=`echo ${version_info} | cut -d' ' -f5` # If we cannot extract the version of patchlevel info # then default to legacy_xfig test "x${version}" = "x" -o "x${patchlevel}" = "x" && legacy_xfig ${base} echo ${version} ${patchlevel} | grep '[0-9]!' -o && legacy_xfig ${base} # So, is it am old version? test ${version} != "3.2" -o ${patchlevel} -lt 4 && legacy_xfig ${base} # I guess not ;-) modern_xfig ${base} # The end