# Description:
#   A script to reformat a plain text file document  which contains
#   no particular format. The script recognises some special structures
#   within the plain text document. For example:
#   
#   Where the first non-whitespace character on a line is '=' then
#   all the following text on the line should be formatted as a
#   'heading'. And if the first non-whitespace character is '*' then
#   the following text should be hyperlinked. Also, url style strings should be recognised and given
#   a hyperlink token in from of them, such as '[*]'. I prefer this to underlining
#   the entire url, because I find that the underlining tends to interfer with
#   the readability of the text. Some people would say, "use style-sheets" but to
#   them I would reply that the 'heraldic' visual pattern of the underlined hyperlink
#   is imprinted in many internet users brains, and to change that 'iconography' can
#   lead to unnecessary confusion.
#
#   Lines which consist of only capital letters and numbers (with at least a few
#   capital letters, are interpreted as headings, and constitute the automatically
#   generated table of contents.
#
#   This script, like the linkdoc2html.sh script also accepts the format
#     * Document Title|Html-Url-Or-Path|Text-Url-Or-Path|
#   The script will render this into an emphasised 'document title' with
#   hyper-links to the different formats for the document.
#
#   The script will also format blocks of text between the strings -->> and --<<
#   (where they are the first string on the line) as an HTML <pre> block
#
# Examples:
#   ./plaintext2html.sh mjb-work.txt notran > mjb-work.html
#     This command line, executed in some kind of a bash shell, will
#     transform a plain text file which isn't is any particular format,
#     into an HTML file (that is it will create a new HTML file and
#     leave the original text file unchanged) and will not display the
#     automatic translation links to Google. Also an HTML table of
#     contents (with one entry for each heading) will be inserted in the
#     HTML document.
#
#   ./plaintext2html.sh mjb-work.txt notran notoc > mjb-work.html
#     The text file will be transformed into HTML but no table of contents
#     will be inserted nor any translation links.
#
#   ./plaintext2html.sh mjb-work.txt blah notoc > mjb-work.html
#     If translation links are desired but no table of contents, use a 
#     command line similar to above. The string 'blah' could be anything
#     as long as its not 'notran'. This slighty dodgy 'feature' is owing to the
#     fact that I am not using any 'getopt' style option parsing.
#
# Parameters:
#   textFileName  [required]
#     The name of the text file which is to be transformed from text into html
#   notran        [optional]
#     If the second parameter is the string 'notran' then the javascript links
#     to the google automatic language translation engine will NOT be inserted
#     into the HTML page. This is useful, for example, when the HTML page is 
#     going to be located within a 'password-protected' directory, because
#     the Google translation engine will not be able to access the page, and
#     therefor the translation links will not work.
#   notoc         [optional]
#     If the third parameter is the string "notoc", then no HTML table of
#     contents will be generated.
#    
#
# Notes:
#   This script contains an improved url detection regular expresion, better than that
#   in say txtdoc2html.sh. But the url pattern matcher still has a problem when
#   somebody puts a full stop after a url. It thinks that that dot is part of the 
#   url. There is possibly no reason why you couldn't just use the 'diary2html.sh'
#   filter script, instead of this one. The Html generated is somewhat dodgy but
#   attempts to avoid some of the more heinous html sins, such as <font> tags
#
#   Interestingly, almost all the functionality which this script provides,
#   that is, making an html table of contents, could equally be achieved using
#   the 'htmldoc' program with a line similar to
#      htmldoc -f output.html --book --no-title theTextFile.txt
#   This assumes that the html file contains heading tags in the correct order.
#
# See Also:
#   diary2html.sh, linkdoc2html.sh
# Author:
#   m.j.bishop

 
 if [ "$1" = "" ]
 then
   echo "usage: $0  textFileName"
   cat $0 | sed -n "/^[ ]*#/p" 
   exit 1;
 fi

 #-- The section below creates the table of contents for the diary.
 #-- This line is designed to only number lines which match a pattern
 #-- In theory 'nl -bpPATTERN' should also do this, but it insisted on
 #-- 'double-spacing' the output
 #-- Also the expressions below try and get rid of things like "can't" and "won't"
 #-- because I want to apply some formatting to the content of quotes, and these
 #-- things will get in my way.

 sHeadingPattern='[ A-Z0-9.\/\\]*[A-Z]\{3,\}[ A-Z0-9.\/\\]*'
 
 cat $1 | expand | \
   mawk '/^[ A-Z0-9.\/\\]*[A-Z]+[ A-Z0-9.\/\\]*$/{ii++; print ii $0}!/^[ A-Z0-9.\/\\]*[A-Z]+[ A-Z0-9.\/\\]*$/' | \
   sed "s/\([a-zA-Z]\{2,\}\)n[\"']t/\1nt/g" > $1.temp
   
 (echo "<a name = \"toc\"></a><table align=\"center\" width = \"70%\"><tr><td>"; \
 cat $1.temp | \
    sed "/^[0-9]\{1,\}[ A-Z0-9.\/\\]*[A-Z]\{3,\}[ A-Z0-9.\/\\]*$/!d" | \
    sed "s/^\([0-9]\{1,\}\)\($sHeadingPattern\)$/<br><a href=\"#item\1\">\1. \2<\/a>/g"; \
  echo "</td></tr></table>";) > plain-text-toc.temp

 echo "<html>"
 echo "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 3.2//EN\">"
 echo "<html>"
 echo " <META HTTP-EQUIV=\"Content-Type\" CONTENT=\"text/html; CHARSET=iso-8859-1\">"
 echo " <META HTTP-EQUIV=\"Keywords\""
 echo "          CONTENT=\"transformed-text\">"
 echo " <META HTTP-EQUIV=\"Description\""
 echo "          CONTENT=\"some transformed text\">"
 echo "        <LINK REV=\"made\" HREF=\"mailto:webmaster@ella-associates.org\">"
 echo ""
 echo "<!-- html generated by the \"$(basename $0)\" script         -->"
 echo "<!-- see http://www.ella-associates.org/utils/$(basename $0) -->"
 echo "<head>"
 echo "<script language = \"javascript\">"
 echo "
 <!--
  function redirectToGoogleTranslation(sSourceLanguage, sTargetLanguage)
  {
    var sTranslationUrl = 'http://translate.google.com/translate?u=';

    sTranslationUrl += escape(document.location.href);
    sTranslationUrl += '&langpair=' + sSourceLanguage + '|' + sTargetLanguage;
    sTranslationUrl += '&hl=' + sSourceLanguage;
    // document.testForm.test.value=sTranslationUrl;
    window.location = sTranslationUrl;
  } //-- redirectToGoogleTranslation()
 -->  "
 
 echo "</script>"
 echo "</head>"
 echo "<body>"
 if [ "$2" != "notran" ]
 then
   echo "<center>"
   echo "See this page in (approximate):"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'es');\">Español</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'fr');\">Français</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'it');\">Italiano</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'de');\">Deutsch</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'pt');\">Português</a>"
   echo "</center>"
 fi

#-- Put the page heading before the table of contents
#--
cat $1.temp | \
   sed "/^[ ]*=[ ]*[^=].*/!d" | \
   sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | \
   sed "s/^[ ]*=[ ]*[^=]\(.*\)/<center><h2>\1<\/h2><\/center>/gi" 

echo "<table align=\"center\" width = \"90%\"><tr><td>"
   
#-- Inset the table of contents   
if [ "$3" != "notoc" ]
then
  cat plain-text-toc.temp
fi

#-- Transform the text to HTML, insert anchors
#-- Also delete the heading line which has already been inserted in the HTML
#-- But, the line will also delete lines beginning in == or === etc, which
#-- may not be desirable.
#-- The line below was designed to make the contents of quotes look different
#-- but I think that it made the text less readable
#--
#--  sed "s/\(['\"]\)[^'\"]\{1,\}\1/<tt>&<\/tt>/g" | \
#--
cat $1.temp | \
  expand | \
  sed "/^[ ]*=[ ]*[^=]\(.*\)$/d" | \
  sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | \
  sed -e "s/^[ ]*\-\-\&gt;\&gt;/<pre>/g" -e "s/^[ ]*\-\-\&lt;\&lt;/<\/pre>/g" | \
  sed "s/^\([0-9]\{1,\}\)\([ A-Z0-9\.\/\\]*[A-Z]\{3,\}[ A-Z0-9\.\/\\]*\)$/<strong><a name=\"item\1\">\1. \2<\/a><\/strong> <a href=\"#toc\">[TOC]<\/a>/g" | \
  sed "s/^[ ]*\*[ ]*\([^|]*\)|\([^|]*\)|\([^|]*\)|/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2'>html<\/a> | <a href='\3'>text<\/a>)|/gi" | \
  sed "s/^[ ]*\*[ ]*\([^ ]\{2,\}\)/<a href='\1'>\1<\/a>/gi" | \
  sed "s/[^\"]\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@]\{3,\}\)/<a href='\1'> [*]<\/a><tt>\1<\/tt>/gi" | \
  sed "s/\([^ ]\{2,\}@[^ \"']\{2,\}\)/<a href=\"mailto:\1\">\1<\/a>/g" | \
  sed "s/[^a-zA-Z\/]\(www\.[-a-z\%0-9\~\\\/\"\'\.\@]\{2,\}\)/<a href='http:\/\/\1'> [*]<\/a><tt>\1<\/tt>/gi" | \
  sed "/<pre>/,/<\/pre>/!s/[ ]\{2\}/\&nbsp;\&nbsp;/g" | \
  sed "/<pre>/,/<\/pre>/!s/^/<br>/g" 
echo "</td></tr></table>"


 echo "<br>"
 if [ "$2" != "notran" ]
 then
   echo "<center>"
   echo "See this page in (approximate):"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'es');\">Español</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'fr');\">Français</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'it');\">Italiano</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'de');\">Deutsch</a>|"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('en', 'pt');\">Português</a>"
   echo "</center>"
 fi
 echo "</body>"
 echo "</html>"

 rm -f $1.temp