# Description:
#   A script to reformat a plain text file document which contains
#   no particular format. The text file stores the content of a web page
#   which is currently at http://poseidonia.ella-associates.org/ . This script
#   is a direct derivation of 'plaintext2html-forum.sh'
#   This current script is designed specifically for a particular webpage or text document.
#
#   The script will work in conjuction with the cgi script '?' to allow the visitor
#   to the web page to edit the page.
#
#   The script also generates an HTML form which allows the reader to edit the text of the document
#   The script recognises some 'cues' within the plain text document. I refer to
#   these cues or 'structures' as 'Invisible Markup Language' (IML) or 
#   Mas o Menos Markup Language (MMML). The basic ideas is to have as little actual
#   'markup' in the text document as possible, and the markup which is present should
#   'look good' in the plain text file. So, instead of using, say, 
#      %^* Section Heading
#    which is valid markup but looks ugly in the text file, we use all capitals which
#    looks better in the text file
#   
#   A line beginning with = is a page title.
#   A line beginning with '*' will be hyperlinked.
#   URLs get automatically hyperlinked in some non-determinate way.
#   All Capitals lines are section headings. These section headings may
#   then be used as a table of contents and hyperlinked in various ways
#
#   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.
#
#   Blocks of text surrounded by '-->>' and '--<<' are not 'formatted' in any way
#
#   The script also formats lines starting in 'added by:' to make those
#   lines stand out from the rest of the text. This is a 'courtesy' to the
#   '/cgi-bin/add-comment' script which added this line to a text file
#   when it inserts a user provided comment in the text file.
#
# Examples:
#   If the scripts are on the system 'path' the the leading './' characters
#   below are not necessary
#
#   ./poseidontext2html-forum.sh concert-details.txt notran > concert-details.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, if there are headings) will be inserted in the
#     HTML document.
#
#   ./poseidontext2html-forum.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.
#
#   ./poseidontext2html-forum.sh mjb-work.txt tran 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.
#
#   ./poseidontext2html-forum.sh stuff.txt notran toc "http://63.105.73.195/cgi-bin/some-weird-script"
#     This transforms the file stuff.txt omitting translation links, inserting a 
#     hyperlinked table of contents, and setting the target for the 'edit document' form
#     to the URL specified in the last parameter.
#
#
# 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.
#   forumProcessorUrl           [optional]
#     This parameter indicates where the processing script is located.
#     If it is omitted, currently the url will default to 
#     http://www.ella-associates.org/cgi-bin/add-comment
#   output-language [optional] {Not implemented}
#     This is the language in which the message on the generated HTML page
#     will appear. For example messages next to the comment boxes and the 
#     translation links.
#   noforum {Not implemented}
#     If this parameter is present no HTML form will be produced in the ouput
#     and therefor the web-visitor will not be able to add comments to the 
#     pages.
#   path-to-style-sheet [optional] {Not implemented}
#     Still to implement
#     This is the full path (relative to the Web Server Document Root)
#     to the style sheet which is to be used by the generated HTML page
#     
#
# Notes:
#   The only difference between this script and the 'poseidontext2html-wiki.sh'
#   is that that script does not have the 'editing form or box' on the same
#   HTML page as the rendered text
#
#   Because of the table used to create a left margin for the table of contents
#   and for the body of the text, this HTML is NOT friendly to 'lynx' which
#   does not support HTML tables. A CSS style-sheet command should be used 
#   instead of the tables.
#
#   This script should also transform quotes into &quot; & into &amp; etc
#   The script appears to be working reasonably well in conjunction with
#   the 'edit-poseidon-forum' cgi script.
#
#   It would be nice to make some kind of 'sub' table of contents for
#   any comments which are present in a document.
#   
#   The translation links wont work from within the 'output' generated 
#   by the 'add-comment' script
#
#   This script has had problems with 'gawk' and different versions of awk. For this reason
#   the 'gawk' or 'awk' code has been removed and replaced with code using the 'nl'
#   program. This program, when used with the -bp option double spaces the object file
#   with lines containing only spaces. Therefore some extra 'sed' lines are necessary
#   to remove these blank lines
#
# See Also:
#   edit-poseidon-forum
#     This is the cgi script which can work in conjunction with the current script
#   poseidontext2html-wiki.sh
#     A very similar script
#   diary2html.sh, 
#     Turns a 'diary' style text file into HTML
#   linkdoc2html.sh,
#     Turns a text file which has a list of URL links and descriptions into HTML
#   linkdoc2html-index.sh
#     As above but also adds an HTML 'table of contents' for possible 'section headings'
#   linkdoc2html-forum.sh
#     Turns a text file with a URL list into an HTML file which has the capability
#     to be contributed to by a web-visitor (using cgi-scripts)
#   plaintext2pdf.sh,
#     Turns a text file into a pdf file with an optional table of contents
#   plaintext2html-simple.sh
#     As below, but doesn't use certain 'bash' tricks
#   plaintext2html.sh
#     Turns a text file with possible section headings and urls into an HTML file
#   glossary2xml.sh
#     Turn a text file which is a sort of 'glossary' into a dodgy xml file
#   alphabetize-glossary.sh
#     Re-arranges a text file which contains a series of definitions of 'items' or 'terms'
#     so that the items are ordered alphabetically.
#   add-comment
#     a cgi-script which can be used in conjuction with some of the 
#     scripts above to add content specified by web-visitors to a web page
#   script-summary.txt
#     contains more short descriptions of scripts and what they do.
# Author:
#   m.j.bishop
#
# Bugs and Ideas
#   See the file linkdoc2html-forum.sh for the beginnings of an attempt to internationalize
#   the output of this script, in the sense that the messages which appear on the 
#   HTML page should be capable of being in various languages, depending on what language
#   the source file is in.
#
#   Add an output-language parameter to this script
#   Also, it would be good to add a 'style-sheet' parameter which would allow
#   this script to change the name or location of the style-sheet which is used
#   by the generated HTML file.
#
#   In Netscape Navigator 4.61, if the style-sheet does not exist at all
#   then the browser is unable to display anything at all. 
#
#   The script could also check if there are translations of the current 
#   HTML or text file, using the standard naming convention of name.file-type.language-code
#   An example of this naming convention is  stuff.html.es  which should
#   be an HTML file which contains Spanish language content. This present
#   script could check for files which have the same name as the source
#   file but which have a different language code extension, and could 
#   therefore automatically add a link to the translated file (in addition,
#   perhaps to the Google translation links). The script would only
#   check in the current directory for these 'translated' files.
#
# Dependencies:
#   iso2html.sed
#   The images used on the 'poseidon site'
#   various Unix tools, a Bash shell
 
 if [ "$1" = "" ]
 then
   echo "usage: $0  textFileName [notran] [notoc] [forum-processor-url] [noforum]"
   echo "PRESS q TO EXIT THIS HELP. PRESS [space-bar] TO SCROLL DOWN, b to SCROLL UP"
   cat $0 | sed -n "/^[ ]*#/p" | less
   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.

 #-- This is the pattern which determines what sort of lines will
 #-- be interpreted as 'section headings'. I cannot use the for the 'awk' line
 #-- because awk does not seem to accept the notation \{n,\}
 
 sHeadingPattern='[ A-Z0-9.\/\\:]*[A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ]*[ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*'
 
 
 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 "<!-- From the file: \"$1\"             -->"
 echo "<!-- On the date: $(date)              -->"
 echo "<!-- see http://www.ella-associates.org/utils/$(basename $0) -->"
 echo "<link   rel = \"stylesheet\"  type = \"text/css\""
 echo "       href = \"/stylesheets/swish-style.css\">"
 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 "Vea este pagina en (aproximado):"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('es', 'en');\">Ingl&eacute;s</a>"
   echo "</center>"
 fi

 echo '
   <center>
   <h2>
     <a href="http://poseidonia.ella-associates.org/index.html">
     <img src="http://poseidonia.ella-associates.org/images/poseidonia-logo-blue-small.jpg" 
        width="40" height="110" border="0" 
	align="middle"></a>'
	
#-- Put the page heading before the table of contents
#--
cat $1 | \
   sed "/^[ ]*=[ ]*[^=].*/!d" | \
   sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | \
   #sed "s/^[ ]*=[ ]*\([^=].*\)/<center><h2>\1<\/h2><\/center>/gi" | \
   sed "s/^[ ]*=[ ]*\([^=].*\)/\1/gi" | \
   #-- line below because the RedHat server uses UTF-8 character set
   iconv --to-code=ISO-8859-1 --from-code=UTF-8 | \
   #-- Try to 'entities' the accented characters
   sed -f /var/www/utils/iso2html.sed

  echo ' </h2></center>'
echo '<table align="center" width = "90%"><tr><td><p><br>'
#- This line below is not 'lynx friendly' as style sheet
#- should be used instead.
echo "<table align=\"center\" width = \"95%\"><tr><td>"
#if [ "$6" != "" ] 
#then
#  echo "
#    <strong><center>
#    <a href=\"#editForm\">[editar este documento]</a> </center></strong>"
#fi
  echo "
    <center>
    <a href=\"#editForm\">[editar este documento]</a> </center>"

#-- Insert the table of contents   
if [ "$3" != "notoc" ]
then

 echo "<a name = \"toc\"></a><table align=\"center\" width = \"70%\"><tr><td>"
 expand $1 | \
   sed "s/^[ ]*$//g" | \
   nl -s" " -bp'^[ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*[A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ]+[ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*$' | \
   sed  "/^[ ]\+$/d" | \
   sed "s/^[ ]*\([1-9][0-9]*\) /\1/g" | \
   sed "s/\([a-zA-Z]\{2,\}\)n[\"']t/\1nt/g" | \
   sed "/^[0-9]\{1,\}$sHeadingPattern$/!d" | \
   sed "s/^\([0-9]\{1,\}\)\($sHeadingPattern\)$/<br><a href=\"#item\1\">\1. \2<\/a>/g" | \
   #-- line below because the RedHat server uses UTF-8 character set
   iconv --to-code=ISO-8859-1 --from-code=UTF-8 | \
   #-- Try to 'entities' the accented characters
   sed -f /var/www/utils/iso2html.sed 
   echo "</td></tr></table>" 

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" | \
#--
#-- I have disabled the line which turns * beginning lines into hyperlinks
#-- since this was not desirable for the netbeans documentation
#-- The version of SED on RedHat linux does not like the syntax "\{,4\}" but "\{0,4\}"
#-- is ok.
#
# What follows below is quite tricky. The order of each of the sed transformation DOES matter
# The tricky bits are allowing for accented european characters, and converting back and forth
# between unicode and iso latin etc
#
# In the context of this 'wiki' script it is reasonably important to display the text 'prettily'
# so I am going to change the presentation of links etc. This allows the user to have
# more control over how the web page is displayed finally
  expand $1 | \
  sed "s/^[ ]*$//g" | \
  #-- Number all lines that are 'section headings', allow for european accented characters
  nl -s" " -bp'^[ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*[A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ][A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ]+[ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*$' | \
  #-- Get rid of the 'blank' lines which nl puts into the output
  sed  "/^[ ]\+$/d" | \
  #-- Reformat the numbered section headings
  sed "s/^[ ]*\([1-9][0-9]*\) /\1/g" | \
  #-- Delete the page title because its already been output
  sed "/^[ ]*=[ ]*\([^=].*\)$/d" | \
  #-- Encode special characters '<>&' as HTML entities
  sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | \
  #-- Do a trick to get the '-->>' and '--<<' blocks of text to work
  sed -e "s/^[ ]*\-\-\&gt;\&gt;/<pre>/g" -e "s/^[ ]*\-\-\&lt;\&lt;/<\/pre>/g" | \
  #-- Make each 'section heading' into an HTML anchor to work with the 'Table of Contents'
  sed "s/^\([0-9]\{1,\}\)\([ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*[A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜÇÑ]\{3,\}[ A-ZÁÉÍÓÚÀÈÌÒÚÄËÏÖÜ·ÇÑ0-9.\/\\:]*\)$/<strong><a name=\"item\1\">\1. \2<\/a> <a href=\"#toc\">[TOC]<\/a><\/strong>/g" | \
  #-- line below because the RedHat server uses UTF-8 character set
  iconv --to-code=ISO-8859-1 --from-code=UTF-8 | \
  #-- Try to 'entitize' the accented characters
  sed -f /var/www/utils/iso2html.sed | \
  #-- Lets deal with image index things. We have to get a few lines into the pattern space
  #-- so that we can hyperlink the image and the first label line
#  sed "/\[IMAGE\-INDEX\-BEGIN\]/,/\[IMAGE\-INDEX\-END\]/ {N;N;s/Image:[ ]*\([^ \n]\{2,\}\)[ ]*\n[ ]*Link:[ ]*\([^ \n]\+\)[ ]*\n\(.*\)/<a href='\2'><img src='\1' border='0' align='left'><\/a><a href='\2'>\3<\/a>/g;}" | \
  sed "/\[IMAGE\-INDEX\-BEGIN\]/,/\[IMAGE\-INDEX\-END\]/ {/^[ ]*Image/{N;N;s/Image:[ ]*\([^ \n]\{2,\}\)[ ]*\n[ ]*Link:[ ]*\([^ \n]*\)[ ]*\n\(.*\)/<a href='\2'><img src='\1' border='0' align='left'><\/a><a href='\2'>\3<\/a>/g;};}" | \
  sed "/\[IMAGE\-INDEX\-BEGIN\]/,/\[IMAGE\-INDEX\-END\]/ {s/^[ ]*$/<p>/;}" | \
   #-- Example of Format Below: [*] My Title|/my/path/to/file-no-extension|html|txt|xml|pdf|
   sed "s/^[ ]*\[\*\][ ]*\([^|]*\)|\([^|]*\)|\([a-zA-Z]\{1,8\}\)|\([a-zA-Z]\{1,8\}\)|\([a-zA-Z]\{1,8\}\)|\([a-zA-Z]\{1,8\}\)|/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2\.\3'>\3<\/a> | <a href='\2\.\4'>\4<\/a> | <a href='\2\.\5'>\5<\/a> | <a href='\2\.\6'>\6<\/a>)/gi" | \
   #-- Example of Format Below: [*] My Title|/my/path/to/file-no-extension|html|txt|pdf|
   sed "s/^[ ]*\[\*\][ ]*\([^|]*\)|\([^|]*\)|\([a-zA-Z]\{1,8\}\)|\([a-zA-Z]\{1,8\}\)|\([a-zA-Z]\{1,8\}\)|/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2\.\3'>\3<\/a> | <a href='\2\.\4'>\4<\/a> | <a href='\2\.\5'>\5<\/a>)/gi" | \
   #-- Example of Format Below: \[*\] My Title|/my/path/to/file-no-extension|pdf|html|
   sed "s/^[ ]*\[\*\][ ]*\([^|]*\)|\([^|]*\)|\([a-zA-Z]\{1,8\}\)|\([a-zA-Z]\{1,8\}\)|/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2\.\3'>\3<\/a> | <a href='\2\.\4'>\4<\/a>)/gi" | \
   #-- Example of Format Below: [*] My Title|/full/path/to/htmlfile|/full/path/to/text/file|/full/path/to/pdffile|
   sed "s/^[ ]*\[\*\][ ]*\([^|]*\)|\([^|]*\)|\([^|]*\)|\([^|]*\)|/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2'>html<\/a> | <a href='\3'>text<\/a> | <a href='\4'>pdf<\/a>)/gi" | \
   #-- Example of Format Below: [*] My Title|/full/path/to/htmlfile|/full/path/to/text/file|
   sed "s/^[ ]*\[\*\][ ]*\([^|]*\)|\([^|]*\)|\([^|]*\)|/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2'>html<\/a> | <a href='\3'>text<\/a>)/gi" | \
   #-- Trick to make 'txt' links into 'text' links for readability
   sed "s/>txt<\/a>/>text<\/a>/gi" | \
   #-- Example of Format Below: [*] My Title|/full/path/to/any-old-file|
   sed "s/^[ ]*\[\*\][ ]*\([^|]*\)|\([^|]*\)|/<b>\1<\/b>(<a href=\"\2\">\2<\/a>)/gi" | \
   #-- Example of Format Below: [*] /full/path/to/any-old-file
   sed "s/^[ ]*\[\*\][ ]*\([^ ]\{2,\}\)/<a href=\"\1\">\1<\/a>/gi" | \
  #-- Hyperlink urls with different display text like: "Some link" http://blah.com
  sed "/<pre>/,/<\/pre>/!s/\"\([^\"]\{1,50\}\)\"[ ]\{0,4\}\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\&\@?=]\{3,\}\)/<a href=\2>\1<\/a>/gi" | \
  #-- Hyperlink URLs beginning with http, except between <pre> tags
  #-- The style immediately below is more 'academic'
  #sed "/<pre>/,/<\/pre>/!s/[^\">]\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@\&?=]\{3,\}\)/<a href=\1>[*]<\/a><tt>\1<\/tt>/gi" | \
  sed "/<pre>/,/<\/pre>/!s/[^\">'=]\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@]\{3,\}\)/<a href='\1'>\1<\/a>/gi" | \
  #-- Hyperlink URLs beginning with http at the beginning of lines, except between <pre> tags
  #sed "/<pre>/,/<\/pre>/!s/^\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@]\{3,\}\)/<a href='\1'>[*]<\/a><tt>\1<\/tt>/gi" | \
  sed "/<pre>/,/<\/pre>/!s/^\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@]\{3,\}\)/<a href='\1'>\1<\/a>/gi" | \
  #-- Hyperlink email addresses with a 'mailto:' link
  sed "/<pre>/,/<\/pre>/!s/\([^ ]\{2,\}@[^ \"']\{2,\}\)/<a href=\"mailto:\1\">\1<\/a>/g" | \
  #-- Hyperlink URLs beginnning with 'www.'
  #sed "/<pre>/,/<\/pre>/!s/[^a-zA-Z\/\">]\(www\.[-a-z\%0-9\~\\\/\"\'\.\@]\{2,\}\)/<a href='http:\/\/\1'>[*]<\/a><tt>\1<\/tt>/gi" | \
  sed "/<pre>/,/<\/pre>/!s/[^a-zA-Z\/\">]\(www\.[-a-z\%0-9\~\\\/\"\'\.\@]\{2,\}\)/<a href='http:\/\/\1'>\1<\/a>/gi" | \
  #-- Format comments added by web-users
   sed "s/^\([ ]*added[ ]\{0,4\}by:\)\([^,]\{1,\}\)\,[ ]*on[ ]*\(.*\)/<u><em>\1<\/em><tt> \2<\/tt><em> on \3<\/em><\/u>/gi" | \
  #-- Turn spaces into non-breaking-spaces unless they are between 'pre' tags
  sed "/<pre>/,/<\/pre>/!{/\[IMAGE-INDEX-BEGIN\]/,/\[IMAGE-INDEX-END\]/! s/[ ]\{2\}/\&nbsp;\&nbsp;/g;}" | \
  #-- Make paragraphs where there are blank lines
  #sed "/<pre>/,/<\/pre>/!s/^[ ]*$/<p>/g" | \
  #-- Make 'fake' headings
  sed "/<pre>/,/<\/pre>/!s/{{/<strong><em><font size='+1'>/g" | \
  sed "/<pre>/,/<\/pre>/!s/}}/<\/font><\/em><\/strong>/g" | \
  sed "/<pre>/,/<\/pre>/!s/[ ]*==[ ]*\(.*\)/<font size=+2><strong><em>\1<\/em><\/strong><\/font>/g" | \
  #-- Turn line breaks into <br> tags unles they are between 'pre' tags. This isn't really
  #-- a good idea since you dont know the width of the target screen
  #sed "/<pre>/,/<\/pre>/!{/\[IMAGE-INDEX-BEGIN\]/,/\[IMAGE-INDEX-END\]/! s/^/<br>/g;}" 
  sed "/<pre>/,/<\/pre>/!{s/^/<br>/g;}" | \
  sed -e "s/\[IMAGE-INDEX-BEGIN\]//g" -e "s/\[IMAGE-INDEX-END\]//g"
echo "</td></tr></table>"
  
 echo "<br>"

 #-- Define the cgi program which will handle the updating of the document 
 #-- according to the contents of the HTML textarea component
 if [ "$4" != "" ]
 then
   sProcessorUrl="$4"
 else
   #-- It would be possible to replace the Domain Name below with
   #-- an IP address, which would mean that the script would still
   #-- work even if the DNS configuration failed. I am not sure if this
   #-- is really a good idea or not.
   #sProcessorUrl="http://www.ella-associates.org/cgi-bin/add-comment"
   sProcessorUrl="http://63.105.73.195/cgi-bin/edit-poseidon-forum"
 fi
 #-- There is a problem in that I need to find the full path 
 #-- name of the $1 variable, but I dont know how to do this. This
 #-- is necessary because the target processor is not in the same
 #-- directory as the source document (the text file)
 #-- For the time being I have used the remedy of seeing if the path
 #-- is relative or absolute. The slightly dodgy path generating code below
 #-- appears to be working. There is almost certainly a much easier way 
 #-- of doing it

 sRelativePath=$(dirname $1)
 sFirstCharacter=$(echo $sRelativePath | sed "s/^\(.\).*$/\1/g")
 if [ "$sRelativePath" = "." ]
 then
   sFullPathName="$(pwd)/$1"
 elif [ "$sFirstCharacter" = "." ]
 then
   sFullPathName="$(pwd)/$1"
 elif [ "$sFirstCharacter" = "/" ]
 then
   sFullPathName="$1"
 else
   sFullPathName="$(pwd)/$1"
 fi
 # echo $sFullPathName
 
 echo "
    <form action = \"$sProcessorUrl\" 
          method = \"post\">
    <input  name = \"filename\" 
            type = \"hidden\"
           value = \"$sFullPathName\">
    <input  name = \"documenttype\" 
            type = \"hidden\"
           value = \"poseidon\">
      
    <hr><a name = \"editForm\"></a>
    <center>
    <br>
    <em><img src = \"http://poseidonia.ella-associates.org/images/uncle-sam.gif\">
      Usted, si USTED, puede cambiar este documento. Edita el texto por el cajillo abajo y haz clic
      en el boton. Gracias por su ayuda.
    </em>
    <br><br>

    <strong>El texto del documento</strong><br>
    <textarea name = \"DocumentText\"  cols = \"80\" 
              rows = \"10\">"
 expand $1 | \
  iconv --to-code=ISO-8859-1 --from-code=UTF-8 
 echo "
	      </textarea>
    <br><br>	      
    <input   type = \"submit\"  value = \"G U A R D A R   L O S   C A M B I O S\">
    </center>
    </form>"
      
 if [ "$2" != "notran" ]
 then
   echo "<center>"
   echo "Vea este pagina en (aproximado):"
   echo "<a  href=\"javascript:redirectToGoogleTranslation('es', 'en');\">Ingl&eacute;s</a>"
   echo "</center>"
 fi

 echo "</body>"
 echo "</html>"

 #rm -f $1.temp
 #rm -f plain-text-toc.temp