API tools faq
paste
Advertisement
Guest User

pdf2epub

a guest
Jul 30th, 2014
225
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Bash 8.46 KB | None | 0 0
raw download clone embed print report
  1. SOURCE=${1%/}
  2. DESTINATION=${2%/}
  3. FILETYPE='*.pdf'
  4.  
  5. #pdf to html(smallest file first :)
  6.  
  7. for i in $(ls -Sr $SOURCE/$FILETYPE)
  8.   do
  9.       pdf_file=$(basename $i)
  10.       echo "Converting $pdf_file to html"
  11.  
  12.      #Create supporting directory structure
  13.      dirname=$(echo $pdf_file | awk -F'.pdf' '{print $1}')
  14.      $(mkdir -p $DESTINATION/$dirname)
  15.      $(mkdir -p $DESTINATION/$dirname/META-INF)
  16.      $(mkdir -p $DESTINATION/$dirname/OEBPS)
  17.      $(mkdir -p $DESTINATION/$dirname/Images)
  18.      $(mkdir -p $DESTINATION/$dirname/Style)
  19.      $(mkdir -p $DESTINATION/$dirname/Audio)
  20.      $(mkdir -p $DESTINATION/$dirname/Video)
  21.      $(mkdir -p $DESTINATION/$dirname/Misc)
  22.      $(mkdir -p $DESTINATION/$dirname/Fonts)
  23.      $(mkdir -p $DESTINATION/$dirname/html)
  24.  
  25.      pdftohtml -q -c -enc UTF-8 -nodrm -fmt png $i $DESTINATION/$dirname/book_
  26.      
  27.      sleep 2
  28.        
  29.     #seperate png & html(i.e. pre xhtml)
  30.  
  31.     echo "Moving images to appropriate directories "
  32.     $(mv $DESTINATION/$dirname/*.png  $DESTINATION/$dirname/Images/)
  33.     $(mv $DESTINATION/$dirname/*.html  $DESTINATION/$dirname/html/)
  34.  
  35.     #update html pages with new image location
  36.     echo "Updating html with new image location"
  37.     $(sed -i "s/<IMG width=\"\(.*\)\" height=\"\(.*\)\" src=\"\(.*png\)\" alt=\"background image\"\/>/<IMG width=\"\1\" height=\"\2\" src=\"\.\.\/Images\/\3\" alt=\"backgroundimage\"\/>/Ig" $DESTINATION/$dirname/html/*.html)
  38.  
  39.   #convert html to xhtml
  40.     echo "Converting html to xhtml"
  41.    
  42.     for file in $(ls $DESTINATION/$dirname/html/*.html)
  43.       do
  44.         dst_file=$(basename $file)
  45.         $(html2xhtml $file -o $DESTINATION/$dirname/OEBPS/$dst_file)  
  46.     done
  47.    
  48.  
  49.     # Fixing the html header to make it valid xhtml
  50.     $(sed -i "/<!DOCTYPE html/{n;N;d}" $DESTINATION/$dirname/OEBPS/*.html)
  51.     $(sed -i "s/<!DOCTYPE html/<!DOCTYPE html>/" $DESTINATION/$dirname/OEBPS/*.html)
  52.    
  53.     #fix css in <body> to make it valid xhtml
  54.     for file in $(ls $DESTINATION/$dirname/OEBPS/*.html)  
  55.       do  
  56.         j=$(grep '<body' $file)
  57.         bgcolour=$(echo $j | cut -d'=' -f2 | cut -d'"' -f2)
  58.         vlink=$(echo $j | cut -d'=' -f3 | cut -d'"' -f2)
  59.         $(sed -i "s/<\/style>/\ta:link {color: $vlink}\n\ta:visited {color: $vlink}\n\ta:active {color: $vlink}\n\tbody{color: $bgcolour}\n\t<\/style>/" $file)
  60.         $(sed -i "s/<body.*\"blue\">/<body>/" $file)
  61.       done
  62.  
  63.    
  64.     sleep 2
  65.    
  66.     echo "Creating mime types & container"
  67.    
  68.     #Creating mime types
  69.     echo -n 'application/epub+zip' > $DESTINATION/$dirname/mimetype
  70.    
  71.     ##container.xml
  72.     echo -e "<?xml version=\"1.0\"?>\n<container version=\"1.0\" xmlns=\"urn:oasis:names:tc:opendocument:xmlns:container\">\n\t<rootfiles>\n\t\t<rootfile full-path=\"package.opf\" media-type=\"application/oebps-package+xml\"/>\n\t</rootfiles>\n</container>" > $DESTINATION/$dirname/META-INF/container.xml
  73.  
  74.     echo "Generating uuid for book"
  75.     book_uuid=$(uuidgen)
  76.    
  77.     echo "Populate OPF file"
  78.    
  79.     echo -e "<?xml version='1.0' encoding='utf-8'?>\n<package xmlns=\"http://www.idpf.org/2007/opf\" xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:dcterms=\"http://purl.org/dc/terms/\" version=\"3.0\" xml:lang=\"en\" unique-identifier=\"book_id\">\n<metadata>\n\t<dc:identifier id=\"book_id\">urn::uuid::$book_uuid</dc:identifier>\n\t<dc:language>en</dc:language>\n\t<dc:title>$dirname</dc:title>\n\t<meta property=\"dcterms:modified\">$(date +%Y-%m-%d)T$(date +%H:%M:%S)Z</meta>\n</metadata>\n" > $DESTINATION/$dirname/package.opf
  80.    
  81.  
  82.     echo "<manifest>" >> $DESTINATION/$dirname/package.opf
  83.  
  84.     #Populating OPF file with Html files
  85.     last_file=$(ls  $DESTINATION/$dirname/OEBPS/*.html | awk -F'-' '{print $2}' | awk -F'.html' '{print $1}' | sort -n | tail -n 1)
  86.     pagename=$(basename $(ls $DESTINATION/$dirname/OEBPS/*.html| head -1 | awk -F'-' '{print $1}'))    
  87.  
  88.    
  89.     for j in $(seq 1 $last_file)
  90.       do
  91.         echo -e "<item href=\"OEBPS/$pagename-$j.html\"\tid=\"html_$j\"\tmedia-type=\"application/xhtml+xml\"/>" >> $DESTINATION/$dirname/package.opf
  92.       done    
  93.    
  94.    #Populating OPF file with Image files
  95.    last_png=$(ls $DESTINATION/$dirname/Images/*.png | awk -F'book_' '{print $2}' | awk -F'.png' '{print $1}' | sort -n | tail -n 1)    
  96.    
  97.    for j in $(seq -w 1 $last_png)
  98.      do
  99.        echo -e "<item href=\"Images/$pagename$j.png\"\tid=\"png_$j\"\tmedia-type=\"image/png\"/>" >> $DESTINATION/$dirname/package.opf
  100.      done
  101.  
  102.    #Populating OPF File with toc and, nav file  
  103.    echo  -e "<item href=\"OEBPS/toc.ncx\"\tid=\"ncx\"\tmedia-type=\"application/x-dtbncx+xml\"/>\n<item properties=\"nav\" id=\"nav\" href=\"OEBPS/nav.html\" media-type=\"application/xhtml+xml\"/>\n</manifest>\n<spine toc=\"ncx\">" >> $DESTINATION/$dirname/package.opf
  104.  
  105.   #Populating OPF file with spine tree
  106.  
  107.   for j in $(cut -f2 $DESTINATION/$dirname/package.opf | grep  'id="html_' | cut -d'=' -f2 | cut -d'"' -f2)
  108.     do
  109.       echo -e "\t<itemref\tidref=\"$j\"/>" >> $DESTINATION/$dirname/package.opf
  110.   done  
  111.  
  112.  
  113.   echo -e "\t<itemref idref=\"nav\" linear=\"no\"/>\n</spine>\n</package>" >> $DESTINATION/$dirname/package.opf
  114.  
  115.   sleep 2
  116.  
  117.   #Fix outline Before using it to generate TOC
  118.   $(awk '/<a href=".*html"\>.*$/ { printf("%s ",$0); next } 1' $DESTINATION/$dirname/OEBPS/$pagename-outline.html | sed -e "s/ \+/ /g" | sed -e "s/<\/a> <\/li>/<\/a>\n<\/li>/g" > $DESTINATION/$dirname/OEBPS/outline.html)
  119.  
  120.   #Creating Index Data for TOC
  121.   $(sed -e "s/<a href=\"\(.*html\)\">\(.*\)<\/a>/#\"\1\"#\"\2\"/g" $DESTINATION/$dirname/OEBPS/outline.html | grep "#" | sed -e "s/\(.*\)\W<.*>/\1/g" | sed "s/ /_/g" | sed "s/\"//g" > $DESTINATION/$dirname/OEBPS/index_data.txt)
  122.  
  123.  
  124.   #Generating TOC(ncx)
  125.   echo "Generating TOC(ncx)"
  126.  
  127.   echo -e "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n<ncx xmlns=\"http://www.daisy.org/z3986/2005/ncx/\" version=\"2005-1\" xml:lang=\"eng\">\n\t<head>\n\t\t<meta name=\"dtb:uid\"\tcontent=\"urn::uuid::$book_uuid\"/>\n\t\t<meta name=\"dbt:depth\"\tcontent=\"0\"/>\n\t\t<meta name=\"dbt:totalPageCount\"\tcontent=\"0\"/>\n\t\t<meta name=\"dbt:maxPageNumber\"\tcontent=\"0\"/>\n\t</head>\n<docTitle>\n\t<text>$dirname</text>\n</docTitle>\n<navMap>" > $DESTINATION/$dirname/OEBPS/toc.ncx
  128.  
  129.   #create Nav Data from index_data.txt
  130.   count=1
  131.   for j in $(cat $DESTINATION/$dirname/OEBPS/index_data.txt)
  132.     do
  133.       link=$(echo $j | cut -d'#' -f2)
  134.       title=$(echo $j | cut -d'#' -f3)
  135.       title=$(echo $title | sed -e 's/"//g' | sed -e 's/_/ /g')
  136.       echo -e "\t<navPoint class=\"title\" id=\"title_$count\" playOrder=\"$count\">\n\t\t<navLabel>\n\t\t\t<text>$title</text>\n\t\t</navLabel>\n\t\t<content src=\"$link#ref$count\"/>\n\t</navPoint>" >> $DESTINATION/$dirname/OEBPS/toc.ncx
  137.       count=$(expr $count + 1)
  138.     done  
  139.  
  140.   echo -e "</navMap>\n</ncx>" >> $DESTINATION/$dirname/OEBPS/toc.ncx
  141.  
  142.   #fixing the toc.ncx ref error
  143.   for j in $(grep '<text>\|src=' $DESTINATION/$dirname/OEBPS/toc.ncx | grep '#'  | sed 's/^[ \t]*//;s/[ \t]*$//' | awk -F'"' '{print $2}')
  144.     do  
  145.       page=$(echo $j| cut -d'#' -f1)
  146.       ref=$(echo $j | cut -d'#' -f2)  
  147.       $(sed -i "s/<body>/<body>\n\t<a id=\"$ref\"\><\/a>/" $DESTINATION\/$dirname\/OEBPS\/$page)
  148.     done
  149.  
  150.   #Generating  Navigation html(nav.html)
  151.   echo "Generating  Navigation html(nav.html)"
  152.  
  153.   $(awk '/<a href=".*html"\>.*$/ { printf("%s ",$0); next } 1' $DESTINATION/$dirname/OEBPS/$pagename-outline.html | sed -e 's/ \+/ /g' | sed -e 's/<\/a> <\(.*\)>/<\/a>\n<\1>/g' > $DESTINATION/$dirname/OEBPS/nav.html)
  154.  
  155.  
  156.   $(sed -i 's/<li>/\t<li>/g' $DESTINATION/$dirname/OEBPS/nav.html)
  157.   $(sed -i 's/<\/li>/\t<\/li>/g' $DESTINATION/$dirname/OEBPS/nav.html)
  158.   $(sed -i 's/<a href=/\t\t<a href=/g' $DESTINATION/$dirname/OEBPS/nav.html)
  159.   $(sed -i "s/<body>/<body>\n<nav id=\"toc\" epub:type=\"toc\">/" $DESTINATION/$dirname/OEBPS/nav.html)
  160.   $(sed -i "s/<\/body>/<\/nav>\n<\/body>/" $DESTINATION/$dirname/OEBPS/nav.html)
  161.   $(sed -i "s/^[ \t]*//;s/[ \t]*$//" $DESTINATION/$dirname/OEBPS/nav.html)
  162.  
  163.  
  164.   #Delete the unwanted files and, directory
  165.  
  166.   echo "Deleting Unwanted files and directories"
  167.  
  168.   delete_base=$DESTINATION/$dirname
  169.   $(rm -f  $delete_base/OEBPS/$pagename\_ind.html)
  170.   $(rm -f  $delete_base/OEBPS/$pagename.html)
  171.   $(rm -f  $delete_base/OEBPS/$pagename-outline.html)
  172.   $(rm -f  $delete_base/OEBPS/outline.html)
  173.   $(rm -f  $delete_base/OEBPS/index_data.txt)
  174.   $(rm -f $DESTINATION/$dirname/*.html)
  175.   $(rm -rf $delete_base/html)
  176.  
  177.   echo "Done with $i"
  178.  
  179. done
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!