#!/bin/ksh
#
# $Header: /afs/northstar/ufac/richard/projects/class-web-builder/RCS/buildhtml,v 1.17 2005/03/19 05:15:48 richard Exp $
#
# Replace some metatag shortcuts with HTML tags, or
# redefine some standard tags with what we'd really
# like them to do.
#
# Arguments are expected to be 1 or more .src files, which are processed and left in
# the corresponding .html file.  If no file names are specified, work with stdin/stdout.
# If an argument is a .html, but the corresponding .src file exists, that will be referenced
# instead.  This allows simplification of the Makefiles - only one list of files is needed.
#
# Use the slide.order file to determine the section ordering and previous/next links
# for the navigation buttons.
#
# If source files are named, and contain DETAIL metatags, we also create .detail.html files
# and appropriate links to them.
#
# Not much error checking yet - assumes the slide.order file exists.
# All sed commands use # as regex delimiters since there are so many "/" in the variables
#
# This script is used in conjunction with buildslidelist and buildframeset
#
# 2001/10/26 Richard Brittain, Dartmouth College.
#
# 2003/10/27 RB Added support for <MCODE> translation - like <CODE> but with a link to
#  online man pages (for introducing new commands).  Changed some of the sed script regex characters to !
#  since # may appear in the replacement strings (as colour specification)
#  Added creation of named anchors for all H[1-3] tags, for use by a more sophisticated indexer.
#  Added buildhtml.conf file option
#
# 2004/04/22 RB Allow for .shtml as well as .html
# 2004/10/19 RB Added option to customize DETAIL tags with fixed entry and exit codes.  Apply these to both
#               screen and printable format files
# 2005/02/03 RB Added css code to make page breaks for each new section in the printed output
#               Omit the slide number at top right of each slide - repeat of navbar just above.
# 2005/02/09 RB Rewrite to use a single navigation plus content page and no more frameset and nav files.
#               Frameless version works much better.
# 2005/03/10 RB Moved the "top" anchor from the very top to the start of the real content - makes
#               class presentations flow much smoother.  Nav buttons at bottom now point to the #top
#               anchor, while the top ones just point to the bare page.
#               Allow for case-independant HTML tags in the input files, and create only lower case tags
#               in output.
# 2005/03/17 RB Make intro page a variable - not hard-coded "welcome.html"

# Solaris has two versions of grep - make sure we get the right one by tweeking $PATH
PATH=/usr/xpg4/bin:$PATH; export PATH

# Make sure we get these only from the .conf file
unset author keywords description

# Read in the config file setting various optional features.  Look only in the current directory
# This can preset several variables used later in this script.
[[ -r ./buildhtml.conf ]] && . ./buildhtml.conf

edit_file()
{
   # Use stdin,stdout.  Let the caller redirect these as needed
   # Uses global variable $mtime (set by caller)
   # Uses preset environment variables for the HTML tag replacements, or hardwired defaults
   # defined here.

   # $1 is the name of the source file
   # $2 is a file ID (page number), to be placed somewhere on the page by these
   #    editing operations.  The string will cause trouble if it confuses the 'sed' parsing.
   # $3 is a filename for the "previous" link 
   # $4 is a filename for the "next" link 
   # $5 is a page title
   src=$1
   fid=$2
   prevf=$3
   nextf=$4
   title=$5

   # HTML replacements
   # Look for variables optionally set in buildhtml.conf, or in pre-existing
   # environment variables.  
   # code_start, code_end, body_fn, h1_start, h1_end, leftarrow, rightarrow, uparrow 
   # man_url, vspace, h2_start, h2_end, h3_start, h3_end

   # <CODE> => Example Code 
   code_start="${code_start:-<code>}"
   code_end="${code_end:-</code>}"

   # URL to use for online man pages (<MCODE> tags).  Pattern \1 is the tag content string
   # If not defined, process as for normal <CODE> tags
   if [[ -z "$man_url" ]]; then
      mancode=$code_start\\1$code_end
   else
      mancode='<a href='$man_url' target=manpage>'$code_start\\1$code_end'</a>'
   fi

   # Default font> (screen only)
   # Use a larger font than normal since mostly this is for class display
   body_fn="${body_fn:-size=+1}"

   # Body start - default top material
   # try to open a new window.
   leftarrow=${leftarrow:-left.gif}
   rightarrow=${rightarrow:-right.gif}
   uparrow=${uparrow:-up.gif}

   # Header generation.
   bodystart='<body '"$bodytag"'>'"\\
"'<font '"$body_fn"'>'"\\
"

   # <H[12]> redefinition
   h1_start="${h1_start:-<h1>}"
   h1_end="${h1_end:-</h1>}"
   h2_start="${h2_start:-<h2>}"
   h2_end="${h2_end:-</h2>}"
   h3_start="${h3_start:-<h3>}"
   h3_end="${h3_end:-</h3>}"

   # Vertical space at end of each page
   vspace=${vspace:-"<br>"}

   # Header and Footer navigation links

   if [[ -n "$prevf" ]]; then
      # Header generation - top navigation links.
      # Note the complicated quoting and newlines are to get escaped newlines
      # into the substitute string for sed, so that the generated HTML is a bit easier to read.
      # Add in a page title and navigation links.
      # Skip the TARGET="mainplusnav" - framed pages will load in the current frame, and frameless pages will not 

      topnavlinks='<table width=100%><tr>'"\\
"' <td align="left"><font size=-1><a href="/cgi-bin/betsie.cgi">Text-only</a></font></td>'"\\
"' <td align="right"><font size=-1>Table of Contents (<a href="index.html" target=_top>frame</a>/'"\\
"'<a href="slide_list_noframes.html" target="_top">no frame</a>)</font></td>'"\\
"' </tr><tr>'"\\
"' <td align=LEFT>'"$fid $title"'</td>'"\\
"' <td align=RIGHT>'"\\
"' <a href="'$prevf'"><img src="'$leftarrow'" border=0 alt="Previous"></a>'"\\
"'<a href=\#top><img src="'"$uparrow"'" border=0 alt="Top"></a> '"\\
"' <a href="'$nextf'"><img src="'$rightarrow'" border=0 alt="Next"></a>'"\\
"' </td>'"\\
"'</tr></table>'"\\
"'<a name="top"></a>'"\\
"

      # Footer generation.
      # For Frameless pages, skip the TARGET=mainplusnav in the nav links
      botnavlinks="\\
"'<p align=CENTER>'"\\
"'<a href='"$prevf\#top"'><img src="'"$leftarrow"'" border=0 alt="Previous"></a> '"\\
"'<a href=\#top><img src="'"$uparrow"'" border=0 alt="Top"></a> '"\\
"'<a href='"$nextf\#top"'><img src="'"$rightarrow"'" border=0 alt="Next"></a>'"\\
"
      # Alternate navigation links with text instead of icons.
      # botnavlinks='<p align=CENTER><a href='"$prevf"'><i>Previous</i></a> <a href=\#top><i>Top</i></a> <a href='"$nextf"'"><i>Next</a></i>'
   else
      topnavlinks=
      botnavlinks=
   fi

   # More footer lines - include a call to the table of contents (slide list) for frames/noframes versions
   # plus the printable version
   lastmod="\\
"'<table width=100%><tr>'"\\
"'<td align="left" width=20% ><font size=-1>'"$src"'\&nbsp;\&nbsp;last modified '"$mtime"'</font></td>'"\\
"'<td align="center" width=20%><a href="'$intro'">Introduction</a></td>'"\\
"'<td align="center" width=20%>Table of Contents<br><font size=-1>(<a href="index.html" target=_top>frame</a>/<a href="slide_list_noframes.html" target="_top">no frame</a>)</font></td>'"\\
"'<td align="center" width=20%><a href="print_pages.shtml" target="_top">Printable<br><font size=-1>(single file)</font></a></td>'"\\
"'<td align="right"><font size=-1>\&copy; Dartmouth College</font></td>'"\\
"'</tr></table>'"\\
"

   if [[ $print = 0 ]]; then
      # process for screen formatting, unless -p was specified
      # <BODY> => Insert background image and default font size.  Insert vertical spacer at end of file
      #           Insert page number (file ID) and navigation links immediately after BODY
      # Place <A NAME="..."> tags around all <H[1-5]> tags.  Leading and trailing whitespace is stripped
      # from the tag content, as it breaks the anchor action.
      # DETAIL metatags are handled separately

      print -R "$doctype"
      print '<html>'
      print '<!-- This was created automatically by buildhtml - do not edit -->'
      print '<head>'
      
      [[ $nometa = 0 && -n "$author" ]]      && print '<meta name="AUTHOR" content="'"$author"'">'
      [[ $nometa = 0 && -n "$keywords" ]]    && print '<meta name="keywords" content="'"$keywords"'">'
      [[ $nometa = 0 && -n "$description" ]] && print '<meta name="description" content="'"$description"'">'
      # Let the TITLE come through as written in the source file, but delete the <HTML> and <HEAD>

      sed \
       -e 's#<[hH][tT][mM][lL]>##' \
       -e 's#<[hH][eE][aA][dD]>##' \
       -e 's!<[cC][oO][dD][eE]>!'"$code_start"'!g' \
       -e 's!</[cC][oO][dD][eE]>!'"$code_end"'!g'  \
       -e 's!<MCODE> *\([^<]*\) *</MCODE>!'"$mancode"'!g' \
       -e 's!^<[bB][oO][dD][yY].*$!'"$bodystart""$topnavlinks"'!' \
       -e 's!^</[bB][oO][dD][yY]>!'"$vspace""$botnavlinks"'<br><hr>'"$lastmod"'</font></body>!' \
       -e 's!<[hH]\([1-5]\)\([^>]*\)>[ 	]*\(.*[^ 	]\)[ 	]*</[hH]\1>!<a name="L\1-\3"><h\1\2>\3</h\1></a>!' \
       -e 's!<[hH]1>!'"$h1_start"'!g' \
       -e 's!</[hH]1>!'"$h1_end"'!g' \
       -e 's!<[hH]2>!'"$h2_start"'!g' \
       -e 's!</[hH]2>!'"$h2_end"'!g' \
       -e 's!<[hH]3>!'"$h3_start"'!g' \
       -e 's!</[hH]3>!'"$h3_end"'!g' \

   else
      # For printable version, use a smaller font and skip spacing at end of pages.
      # Skip background images, and all HTML/BODY tages - we'll supply separate ones
      # at start and end.
      # Skip the internal anchors on <H> tags.
      # Replace DETAIL metatags with the detail_start and detail_end codes, if any.
      # Insert "newpage" code at start of each file, unless the $src is the introduction - quick
      # hack to get pagination right at start of printed notes
      case $src in
      ${intro%.*}.*) class= ;;
      *)         class='class="newpage"' ;;
      esac

      sed \
       -e 's!<[hH][tT][mM][lL]>.*$!!' \
       -e 's!</[hH][tT][mM][lL]>.*$!!' \
       -e 's!<[hH][eE][aA][dD]>.*$!!' \
       -e 's!</[hH][eE][aA][dD]>.*$!!' \
       -e 's!<[tT][iI][tT][lL][eE].*$!!' \
       -e "s!<[cC][oO][dD][eE]>!$code_start!g" \
       -e "s!</[cC][oO][dD][eE]>!$code_end!g"  \
       -e 's!<MCODE> *\([^<]*\) *</MCODE>!'"$mancode"'!g' \
       -e 's!^<[bB][oO][dD][yY].*$!<p align=RIGHT '$class'><font size=-1>'$fid'</font></P>!' \
       -e 's!</[bB][oO][dD][yY]>!!' \
       -e 's!<[hH]1>!'"$h1_start"'!g' \
       -e 's!</[hH]1>!'"$h1_end"'!g' \
       -e 's!<[hH]2>!'"$h2_start"'!g' \
       -e 's!</[hH]2>!'"$h2_end"'!g' \
       -e 's!<[hH]3>!'"$h3_start"'!g' \
       -e 's!</[hH]3>!'"$h3_end"'!g' \
       -e 's#<!-- *DETAIL-.* -->#'"$detail_start"'#g' \
       -e 's#<!-- */DETAIL.* -->#'"$detail_end"'#g' \
      
   fi
}

create_detail()
{
   # Create a "detail" file by processing the DETAIL metatags.
   # We operate on the HTML file already created, so that we don't need to worry 
   # about all the other translations.
   # The non-blank characters following DETAIL- in the tag are used to create anchors
   # for the links between the two versions.

   # $1 is the name of the .[s]html file we need to process.  
   # We will replace $1 with a new version, and create a .detail.[s]html file to go with it.
   src=$1

   # Look for variables optionally set in buildhtml.conf, or in pre-existing
   # environment variables.
   # detail_start, detail_end

   # Link for "detail" page.  The "default" page hides the detail - turns them
   # into an HTML comment, and inserts a link to the page with the details included.
   # For the printable version, we ignore the detail metatags, which leaves the content in the document.
   # A side effect of this is that detail_start and detail_end tags don't appear in the printable
   # copy

   detail=${src%.$html}.detail.$html
   moredetail="\\
"'<a name=\1></a>'"\\
"'<a href="'"$detail"'\#\1"><img src="right.gif" border=0 height=15 width=15 alt="More detail"></a>'"\\
"
   lessdetail="\\
"'<a name=\2></a>'"\\
"'<a href="'"$src"'\#\2"><img src="down.gif" border=0 height=15 width=15 alt="Less detail"></a>'"\\
"
   # First create the modified "standard" file, turning the DETAIL content into comments and inserting a link
   sed < $src > $src.$$ \
   -e 's#<!-- *DETAIL-\([^ ]*\).*-->#'"$moredetail"'<!-- #' \
   -e 's#<!-- */DETAIL.*--># -->#'

   # Now create the "detail" file
   sed < $src > $detail \
   -e 's#\(<!-- *DETAIL-\([^ ]*\).*-->\)#'"$lessdetail"'\1'"$detail_start"'#' \
   -e 's#\(<!-- */DETAIL.*-->\)#'"$detail_end"'\1#'

##   -e 's#\(<!-- *DETAIL-\([^ ]*\).*-->\)#'"$lessdetail"'\1#'

   # overwrite the original input file.
   mv $src.$$ $src
}

print_start()
{
   # print HTML header fluff - this will be stripped out of all component files.
   # Arguments expected are:
   #    1 = the title of the combined document.
   #    2 = base URL (optional)
   # Since we use server side include here, could suck in custom preface for a given class

   print "$doctype"
   print "<html>"
   print "<head>"
   print "<!-- This file is generated automatically by buildhtml.  Do not edit -->"
   [[ $nometa = 0 && -n "$author" ]]      && print '<meta name="AUTHOR" content="'"$author"'">'
   [[ $nometa = 0 && -n "$keywords" ]]    && print '<meta name="keywords" content="'"$keywords"'">'
   [[ $nometa = 0 && -n "$description" ]] && print '<meta name="description" content="'"$description"'">'
   print "<title>$1</title>"
   # Some CSS magic to allow us to force pagebreaks in the printed output
   print "<style type="text/css" media="print"> .newpage { page-break-before: always; }</style>"
   print "<!--#config timefmt=\"%d %B %Y\" -->"
   print "</head>"

   # Leave body at default browser font.  Components will be resized relative to this.
   print "<body bgcolor=WHITE>"
   print "<h1 align=CENTER>$1</h1>"
   print "<h3 align=CENTER> Course Handout: (last update <!--#echo var=\"LAST_MODIFIED\"-->) </h3>"
   print "<br>"
   [[ ! -z $2 ]] && print "<small>These notes may be found at <a href=\"$2\">$2</a>.  The online version has many links to additional"\
                          "information and may be more up to date than the printed notes</small>"

   print "<hr>"
}

print_end()
{
   # print HTML trailer fluff - this has been stripped out of all the component files.
   # Arguments expected are:
   #    1 = the title of the combined document.
   #    2 = base URL (optional)

   # Turn the title into something we can use for a download counterID
   counterid=$(print $1 | sed -es'/ \{1,\}/_/g' -es'/[^A-Za-z0-9_]//g')

   print "<br><br><hr>"
   print "<font size=-1>$1: Course Handout</font><br>"
   # This assumes the simplecounter CGI is available on the system used to deliver the page.
   # Version with visible counter:
   # print '<font size=-1>Download count [ <!--#include virtual="/cgi-bin/cgiwrap/~richard/simplecounter.cgi?counterid='$counterid'" --> ]&nbsp;&nbsp;</font>'
   # Version with invisible counter
   print '<!--#include virtual="/cgi-bin/cgiwrap/~richard/simplecounter.cgi?counterid='$counterid'&silent=true" -->'

   print "<font size=-2>"
   print "(last update&nbsp;&nbsp; <!--#echo var=\"LAST_MODIFIED\"-->)&nbsp;&nbsp;&copy;Dartmouth College"
   [[ ! -z $2 ]] && print "&nbsp;&nbsp;&nbsp;&nbsp;$2"
   print "</font>"

   print "</font>"
   print "</body>"
   print "</html>"
}

# Control starts here

integer i maxslide
print=0
stream=0
nometa=0
baseurl=
doctype=${doctype:-"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\">"}
frametype=${frametype:-"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Frameset//EN\" \"http://www.w3.org/TR/REC-html40/frameset.dtd\">"}
html=html

# Option -p = format for printer-optimized HTML (default is screen)
# Option -s = output to stdout, even if named files are given to us.  Don't create the .html files.
# Option -b = specify a BASE URL
# Option -n = no META tags in header (author, keywords etc.)
# Option -S = create .shtml instead of .html as output (file contains SSI)
while getopts npsSb: o ; do
   case $o in
   n) nometa=1;;
   p) print=1;;
   s) stream=1;;
   S) html=shtml;;
   b) baseurl=$OPTARG;;
   esac
done
shift $OPTIND-1

# Get the intro page name from line 1 of slide.order
intro=$(sed -n -e '1s/^[ 0-9]*//p' slide.order)

# Get the "class" title from the TITLE of the intro page
# We'll use this in the printable version
if [[ -r ${intro%.*}.src ]] ; then
   ctitle=$(grep -i '<TITLE' ${intro%.*}.src | sed -e 's#.*<[tT][iI][tT][lL][eE]>##' -e 's#</[tT][iI][tT][lL][eE]>.*##')
else
   # Oops - well just use a generic title
   ctitle="Class Handout"
fi

# Set up the bodytag variable which is used in a couple of functions
# <BODY BACKGROUND=background> image or colour.  Default is white, special value "none"
# omits the tag and lets the browser default be used
body_bg="${body_bg:-WHITE}"
case $body_bg in
(*.gif|*.jpg) 
   # Assume it is an image
   bodytag="background=$body_bg" ;;
none)
   # We don't want any background
   bodytag= ;;
*)
   # Assume it is a colour
   bodytag="bgcolor=$body_bg" ;;
esac

if [[ $# -gt 0 ]] ; then
   # There are arguments - loop over all the .src files and turn them into .html
   # This code assumes the filenames have the form *.src
   # The '*' part is extracted and used as the file ID tag.

   # Print the headers, for the printable version
   [[ $print = 1 ]] && print_start "$ctitle" "$baseurl"

   # maxslide is the number of the last slide in the sequence
   maxslide=$(tail -1 slide.order|awk '{print $1}')

   for srcfile in $@; do
      # If we were give .[s]html names, but the .src files existed, assume we meant the .src
      # The suffix we want is stored in $html for later use
      case $srcfile in
      *.html)
         html=html
         newfile=${srcfile%.$html}.src
         [[ -r $newfile ]] && srcfile=$newfile
         ;;
      *.shtml)
         html=shtml
         newfile=${srcfile%.$html}.src
         [[ -r $newfile ]] && srcfile=$newfile
         ;;
      esac   
    
      # strip off the .{src|html} to get a base name - use that to look up the
      # file numbering from slide.order, or just use it directly as a fileID
      case $srcfile in
      *.src)
         base=${srcfile%.src} ;;
      *)
         base=$srcfile%.$html}
      esac

      # The slide.order file is [spaces]NN[spaces]filename
      # The counting starts at 0, with the introduction page.
      # $fidno is the number of the file in the slide.order list.
      # $fid   is the "fileID" we will use to label the pages.
      fidno=$(grep " $base\." slide.order | awk '{print $1}')
      fid=$fidno
      # If that didn't work, just use $base
      [[ -z $fid ]] && fid=$base

      # We don't want "page numbers" for the intro and slide list - they look silly so special case it here
      case $fid in
      ${intro%.*}.*) fid= ;;
      (*slide_list*|0) fid= ;;
      esac
  
      # Finally, if fid is non-null, put it in ()
      [[ ! -z $fid ]] && fid="($fid)"

      # Get the modified time for $srcfile, for use in the footers.
      mtime=$(mtime $srcfile)

      # Make the outfile by replacing .src with .$html.  If the given filename
      # wasn't a .src file, this will just append .$html.  We need this for the prev/next links
      # even if we are writing to stdout.
      outfile=${srcfile%.src}.$html

      # Get the "Previous" and "Next" files from the slide.order list for the navigation buttons
      # $i is an integer, so null evaluates to 0
      if [[ -z "$fidno" ]]; then
         # We weren't in the slide.order file, so skip the navigation buttons
         prev=
         next=
      else
         # look up the previous and next files for the nav buttons
         prev=$outfile
         next=$outfile 
         i=$fidno
         [[ $i -gt 0 ]] && prev=$(egrep '^ *'$((i-1))' ' slide.order|awk '{print $2}')
         [[ $i -lt $maxslide ]] && next=$(egrep '^ *'$((i+1))' ' slide.order|awk '{print $2}')

         # for the multiframe model, the next/previous links are not the actual .html files, but the framesets
         # prev=${prev%.*html}.frameset.html
         # next=${next%.*html}.frameset.html
      fi

      ptitle=$(grep -i '<TITLE' $srcfile | sed -e 's#.*<[tT][iI][tT][lL][eE]>##' -e 's#</[tT][iI][tT][lL][eE]>.*##')

      # Call the editing function on this file, with I/O redirected as needed
      # $fid is not optional, but may have a null value, so quote it.
      if [[ $stream = 1 ]]; then
         edit_file < $srcfile $srcfile "$fid" $prev $next "$ptitle"
         # If we are writing to stdout, we skip the "detail" file
      else
         edit_file < $srcfile > $outfile $srcfile "$fid" $prev $next "$ptitle"

         # If the srcfile contains DETAIL metatags, create the *.detail.html file
         # Note that edit_file() leaves DETAIL tags alone, since we need to treat them
         # differently for the "standard" file and the "detail" file.
         if grep -q '<!-- *DETAIL.*-->' $srcfile ; then
            create_detail $outfile
         fi
 
         # Multiframe version needed functions to create *.frameset.html and *.nav.html files
         # per named source file.  See previous version for that code.
      fi
   done
   [[ $print = 1 ]] && print_end  "$ctitle" "$baseurl"
else
   # There are no arguments - do the same thing to stdin, write to stdout.  No FileID
   mtime=$(date '+%d/%m/%Y')
   [[ $print = 1 ]] && print_start "Class Notes" "$baseurl"
   edit_file 
   [[ $print = 1 ]] && print_end "Class Notes" "$baseurl"
fi
exit 0