pdfselect/pdfselect

   1 #! /bin/sh
   2 ##
   3 ##  pdfselect: A shell program to pages subset of a PDF file
   4 ##
   5 ##  Author ezzetabi, based on the work of
   6 ##  Author David Firth (http://www.warwick.ac.uk/go/dfirth)
   7 ##
   8 version=1.1
   9 echo "This is pdfselect version ""$version"
  10 ##
  11 ##  Relies on pdflatex and the 'pdfpages' package (version 0.2e
  12 ##  or later).
  13 ##
  14 #######################################################################
  15 ##  CONFIGURATION: change this block as necessary
  16 ##
  17 ##  THESE SETTINGS WILL BE OVER-RIDDEN by any found at
  18 ##     /etc/pdfselect.conf
  19 ##     /usr/share/etc/pdfselect.conf
  20 ##     /usr/local/share/pdfselect.conf
  21 ##     /usr/local/etc/pdfselect.conf
  22 ##     ~/.pdfselect.conf
  23 ##  (which are read in that order)
  24 ##
  25 ##  First say where your "pdflatex" program lives:
  26 ##
  27 pdflatex=pdflatex
  28 #pdflatex="pdflatex.exe"    ## this for Windows computers
  29 ##
  30 ##  Next pdfinfo name:
  31 ##
  32 pdfinfo=pdfinfo
  33 ##
  34 ##  Next a permitted location for temporary files on your system:
  35 ##
  36 tempfileDir="/var/tmp" ## /var/tmp is standard on many unix systems
  37 #tempfileDir="C:/tmp"  ## use something like this under Windows
  38 ##
  39 ##  Now specify the default settings for pdfselect:
  40 ##
  41 paper=a4paper          ## alternatives are other LaTeX paper sizes
  42 pages=all
  43 evenonly=false
  44 oddonly=false
  45 reversed=false
  46 orient=portrait        ## alternatives are landscape and portrait
  47 tidy=true              ## delete all temporary files immediately
  48 subset=false           ## internal, true if evenonly or oddonly is true
  49 pagesknown=false       ## internal, if true the page range is the same for
  50                        ## every file. (e.i. pages=- or pages=last,1)
  51 ##
  52 ##  END OF CONFIGURATION
  53 #######################################################################
  54 ##
  55 ##  Read the configuration file(s) if such exist:
  56 ##
  57 for d in /etc /usr/share/etc /usr/local/share /usr/local/etc
  58 do if test -f $d/pdfselect.conf; then
  59      echo "Reading site configuration from $d/pdfselect.conf"
  60      source $d/pdfselect.conf
  61    fi
  62 done
  63 if test -f ~/.pdfselect.conf; then
  64    echo "Reading user defaults from ~/.pdfselect.conf";
  65    source ~/.pdfselect.conf;
  66 fi
  67 #######################################################################
  68 ##
  69 ##  Define the output of "pdfselect --help"
  70 ##
  71 helptext="
  72 Usage: pdfselect args
  73 where args must include source pdf filename(s) and optionally also
  74 * a list or range of pages to be included, for example
  75                            --pages 3-6
  76                            --pages 2,8,4,5
  77                            --pages all
  78 * a LaTeX papersize, for example --paper a4paper
  79                                  --paper letterpaper
  80 * one of --evenonly true
  81          --evenonly false
  82   if true the outfile will contains only even pages.
  83 * one of --oddonly true
  84          --oddnonly false
  85   if true the outfile will contains only odd pages.
  86 * one of --reversed true
  87          --reversed false
  88   if true the outfile will contains only odd pages.
  89 * the output page orientation, one of --orient landscape
  90                                       --orient portrait
  91 * a specific name for the output file, e.g. --outfile printme.pdf
  92 * one of --tidy true
  93          --tidy false
  94   according to whether or not temporary files should be deleted immediately.
  95   If --tidy false is used, such files are left in $tempfileDir.
  96
  97 Default arguments for you at this site are
  98   --paper $paper --pages $pages --tidy $tidy --evenonly $evenonly --oddonly
  99   $oddonly --reversed $reversed --orient $orient
 100 "
 101 ##
 102 ##  Check that necessary LaTeX packages are installed
 103 ##
 104 PATH=`dirname "$pdflatex"`:$PATH
 105 export PATH
 106 case `kpsewhich pdfpages.sty` in
 107   "") echo "pdfselect: pdfpages.sty not installed"; exit 1;;
 108 esac
 109 case `kpsewhich eso-pic.sty` in
 110   "") echo \
 111       "pdfselect: eso-pic.sty not installed (see the pdfpages manual)"
 112       exit 1;;
 113 esac
 114 case `kpsewhich everyshi.sty` in
 115   "") echo \
 116       "pdfselect: everyshi.sty not installed (see the pdfpages manual)"
 117       exit 1;;
 118 esac
 119 ##
 120 ##  Now do the argument loop...
 121 ##
 122 sourcePath=
 123 outFile=
 124 inFiles=0
 125 newline='
 126 '
 127 while test -n "${1}"; do
 128   case "${1}" in
 129     *.pdf) inFiles=`expr $inFiles + 1`;
 130            sourcePath="$sourcePath$newline${1}";;
 131     --help) echo "$helptext";
 132             exit 0;;
 133     --pages) pages="${2}"
 134              shift;;
 135     --paper) paper="${2}"
 136              shift;;
 137     --outfile) outFile="${2}"
 138                case "$outFile" in
 139                 *".pdf");;
 140                 *) echo "pdfselect: outfile name must end in .pdf";
 141                    exit 1;;
 142                esac
 143                shift;;
 144     --tidy) tidy="${2}"
 145             shift;;
 146     --evenonly) evenonly="${2}"
 147             if [ $oddonly == true ] ; then
 148               echo "pdfselect: oddonly and evenonly must not appear together."
 149               exit 1
 150             fi
 151             subset=true;
 152             shift;;
 153     --oddonly) oddonly="${2}"
 154             if [ $evenonly == true ] ; then
 155               echo "pdfselect: oddonly and evenonly must not appear together."
 156               exit 1
 157             fi
 158             subset=true;
 159             shift;;
 160     --reversed) reversed="${2}"
 161             shift;;
 162     --orient) orient="${2}"
 163             shift;;
 164     *) echo "pdfselect: unrecognised argument ${1}"; exit 1;;
 165   esac
 166   shift
 167 done
 168 if test $inFiles -gt 1 ; then
 169   if test "$outFile" != "" ; then
 170     echo "pdfselect: --outfile cannot be used with multiple input files";
 171     echo "pdfselect: no files processed"
 172     exit 1;
 173   fi
 174 fi
 175 case "$sourcePath" in
 176   "") echo "pdfselect: no pdf source file specified
 177 For information on usage try \"pdfselect --help\""; exit 1;;
 178 esac
 179 OIFS=IFS
 180 IFS="$newline"
 181 for k in $sourcePath
 182 do
 183   if test -f "$k"; then :;
 184     else echo "pdfselect: ""$k"" does not exist, no files were processed";
 185     exit 1;
 186   fi
 187 done
 188 IFS=OIFS
 189 if [ $subset == false ] ; then
 190   if [ $pages == all ] ; then
 191     pagesknown=true
 192     if [ $reversed == true ] ; then
 193       pages='last-1'
 194     else
 195       pages=-
 196     fi
 197   fi
 198 fi
 199 if [ $oddonly == true ] ; then
 200   oddonly=o
 201 else
 202   oddonly=''
 203 fi
 204 if [ $evenonly == true ] ; then
 205   evenonly=e
 206 else
 207   evenonly=''
 208 fi
 209 if [ $reversed == true ] ; then
 210   reversed=r
 211 else
 212   reversed=''
 213 fi
 214 ##
 215 ##  That's the arguments done.
 216 ##
 217 pwd=`pwd | sed 's/ /\\ /'`
 218 IFS="$newline"
 219 ##
 220 ##  Now work on the input file (or files in turn)
 221 ##
 222 counter=0
 223 opages=$pages
 224 for inFile in $sourcePath
 225 do
 226   echo "Processing $inFile..."
 227   counter=`expr $counter + 1`
 228   cd "$pwd"
 229   pdfName=`basename "$inFile"`
 230   sourceDir=`dirname "$inFile"` ; cd "$sourceDir" ; sourceDir=`pwd`
 231   sourceFullPath="$sourceDir"/"$pdfName"
 232   #if needed generate the pages range
 233   if [ $pagesknown == false ] ; then
 234       pages=$opages
 235       # detect the number of pages
 236       lastpage=`$pdfinfo "$sourceFullPath" | sed -n -e '/^Pages: / s_^Pages: *__ p'`
 237       # replaces 'all' and 'last' with explicit ranges.
 238       pages=`echo -n $pages | \
 239         sed -e 's_^  *\|  *$__' -e 's_last_'$lastpage'_g' -e 's_all_1-'$lastpage'_g'`
 240       # enumerate the page list explicitly
 241       pages=`pdfselect_i $lastpage $pages $reversed $oddonly $evenonly`
 242       # split the page numbers with commas
 243       pages=`echo -n $pages | sed -e 's_^  *\|  *$__' -e 's_ _,_g'`
 244       # add surrounding brackets
 245       pages={$pages}
 246       if [ $pages == '{}' ] ; then
 247         echo Selected an empty/invalid range for $pdfName. Skipping...
 248         pages=''
 249       fi
 250   fi
 251   if [ x$pages != x ] ; then
 252     cd "$pwd"
 253     case "$outFile" in
 254       "") ## no --outfile argument supplied
 255           outfileNameRoot=`echo "$pdfName" | \
 256             sed 's/\.pdf$//'`-$evenonly$oddonly$reversed;
 257           outfileDir="$sourceDir";
 258           outFile="$outfileNameRoot"".pdf";;
 259        *) ## --outfile argument was supplied
 260           outfileNameRoot=`basename "$outFile" | sed 's/\.pdf$//';`
 261           outfileDir=`dirname "$outFile"` ;
 262           cd "$outfileDir" ;
 263           outfileDir=`pwd` ;;
 264     esac
 265     case "$outfileDir"/"$outfileNameRoot"".pdf" in
 266       $sourceFullPath) echo "pdfselect: outfile and source cannot be the same";
 267                   exit 1;;
 268       //"$outfileNameRoot"".pdf") outfileDir="";;  ## in case of output
 269     esac                                           ## to the root directory!
 270     ##
 271     ##  Now edit the temporary LaTeX input file
 272     ##
 273     uniqueName="$$"-"$counter"
 274     ln -s "$sourceFullPath" "$tempfileDir"/"$uniqueName"source.pdf
 275     texFile="$tempfileDir"/"$uniqueName".tex
 276     msgFile="$tempfileDir"/"$uniqueName".msgs
 277     (sed s*pdfname*"$tempfileDir"/"$uniqueName"source.pdf* <<EndTemplate
 278 \\documentclass[papersize,orientation]{article}
 279 \\usepackage{pdfpages}
 280 \\begin{document}
 281 \\includepdf[fitpaper=true,pages=]{pdfname}
 282 \\end{document}
 283 EndTemplate
 284 ) \
 285       | sed s/pages=/pages="$pages"/ \
 286       | sed s/papersize/"$paper"/ \
 287       | sed s/orientation/"$orient"/ \
 288       > "$texFile"
 289     echo "  Temporary LaTeX file for this job is ""$texFile"
 290     ##
 291     ##  Now run pdflatex and tidy up
 292     ##
 293     echo "  Calling pdflatex..."
 294     cd "$tempfileDir"
 295     "$pdflatex" --interaction batchmode "$texFile" > "$msgFile"
 296     if test -f "$tempfileDir"/"$uniqueName"".aux";
 297           ## ie if LaTeX didn't choke
 298       then if cp "$tempfileDir"/"$uniqueName".pdf \
 299                  "$outfileDir"/"$outfileNameRoot"".pdf"
 300            then echo "  Finished: output is"\
 301                      "$outfileDir"/"$outfileNameRoot"".pdf"
 302            fi
 303            case "$tidy" in
 304              true) rm "$tempfileDir"/"$uniqueName"* ;;
 305            esac
 306            outFile=""
 307       else echo "  Failed: output file not written"
 308     fi
 309   fi
 310 done
 311 IFS=OIFS
 312