# Description:
#   A script to reformat a plain text file document  which contains
#   no particular format. The script also generates an HTML form which
#   allows the reader to add a comment or other text to the document.
#   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'.
#   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
#
#   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:
#   ./plaintext2html-forum.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-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.
#
#   ./plaintext2html-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.
#
#   ./plaintext2html-forum.sh stuff.txt notran toc "http://63.105.73.195/cgi-bin/add-comment"
#     This transforms the file stuff.txt omitting translation links, inserting a 
#     hyperlinked table of contents, and setting the target for the comment 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
#     Still to be 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.
#   path-to-style-sheet
#     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:
#   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 'add-comment' 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:
#   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.
#
#   At the moment the script uses special 'stylesheet classes' for particular 
#   elements, such as the <pre> element, although this is probably not really
#   necessary; the style should be attached to the <pre> element itself rather
#   than to a CSS class of the pre element as in <pre class="swish">
#   The second method is probably only necessary when there is more than one
#   type of style which you wish to apply to a particular HTML element in the
#   same document.
#
#   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
#   various Unix tools, a Bash shell
 
 if [ "$1" = "" ]
 then
   echo "usage: $0  textFileName [notran] [notoc] [forum-processor-url]"
   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.

 #-- 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-Z][A-Z][A-Z][A-Z]*[ A-Z0-9.\/\\:]*'
 
#-- I have disable the code below because in a cgi environment, this script doesn't
#-- seem to have permission to create a file. It depends on who originally owns
#-- the $1.temp file. 

#-- Gnu awk (gawk) seems to assume that pattern matching should be case insensitive
#-- by default. The Begin clause below attempts to correct that. Although the script 
#-- says 'mawk' this is just a symbolic link to 'gawk'

#-->-->
#-- I am having all sorts of problems with GNU awk. For some reason it return lower case
#-- lines, even when the regular expression dictates upper case lines. 
#-- One solution to the problem is to use 'nl' instead. For example the line below
#-- almost does the trick
# expand $1 | \
#   sed "s/^[ ]*$//g" | \
#   nl -s" " -bp'^[ A-Z0-9.\/\\:]*[A-Z][A-Z][A-Z]+[ A-Z0-9.\/\\:]*$' | \
#   sed  "/^[ ]\+$/d" | \
#   sed "s/^[ ]*\([1-9][0-9]*\) /\1/g" | \
 
 # The trouble-some gawk line
 #  gawk 'BEGIN{IGNORECASE=0}/^[ A-Z0-9.\/\\]*[A-Z]+[ A-Z0-9.\/\\]*$/{ii++; print ii $0}!/^[ A-Z0-9.\/\\]*[A-Z]+[ A-Z0-9.\/\\]*$/' | \

 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 "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 | \
   sed "/^[ ]*=[ ]*[^=].*/!d" | \
   sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | \
   sed "s/^[ ]*=[ ]*\([^=].*\)/<center><h2>\1<\/h2><\/center>/gi" 

#- This line below is not 'lynx friendly' as style sheet
#- should be used instead.
echo "<table align=\"center\" width = \"90%\"><tr><td>"
echo "
  <strong><center>
    <a href=\"#commentForm\">[make a comment about (or add to) this document]</a> | 
    <a href=\"http://www.ella-associates.org/alexis-info/docs/resources.html\">[Alexis Documentation Home]</a>
  </center></strong>"

#-- Insert the table of contents   
if [ "$3" != "notoc" ]
then
  #-- I put all this code here to get around a problem which arises
  #-- when this script is used from a cgi script (lack of write permissions)
  #-- The problem is not quite so simple as lack of write permissions since
  #-- this script reads and writes a file called $1.temp successfully
  #--
  #-- This is probably faster than the code in 'plaintext2html.sh' because
  #-- no files have to be written.

 echo "<a name = \"toc\"></a><table align=\"center\" width = \"70%\"><tr><td>"
 expand $1 | \
   sed "s/^[ ]*$//g" | \
   nl -s" " -bp'^[ A-Z0-9.\/\\:]*[A-Z][A-Z][A-Z]+[ A-Z0-9.\/\\:]*$' | \
   sed  "/^[ ]\+$/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\" class=\"t\">\1. \2<\/a>/g"
  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.
#

  expand $1 | \
  sed "s/^[ ]*$//g" | \
  #-- Number all lines that are 'section headings'
  nl -s" " -bp'^[ A-Z0-9.\/\\:]*[A-Z][A-Z][A-Z]+[ A-Z0-9.\/\\:]*$' | \
  #-- Get rid of the 'blank' lines which nl puts into the output
  sed  "/^[ ]\+$/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 class = \"sed\">/g" -e "s/^[ ]*\-\-\&lt;\&lt;/<\/pre>/g" | \
  #-- Make each 'section heading' into an HTML anchor to work with the 'Table of Contents'
  sed "s/^\([0-9]\{1,\}\)\([ A-Z0-9.\/\\:]*[A-Z]\{3,\}[ A-Z0-9.\/\\:]*\)$/<strong><a name=\"item\1\">\1. \2<\/a><\/strong> <a href=\"#toc\">[TOC]<\/a>/g" | \
   #-- Example of Format Below: * My Title|/my/path/to/file|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|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|||
   sed "s/^[ ]*\*[ ]*\([^|]*\)|\([^|]*\)|||/<b>\1<\/b> (<em>Formats:<\/em> <a href='\2\.html'>html<\/a> | <a href='\2\.txt'>text<\/a>)/gi" | \
   #-- Example of Format Below: * My Title|/my/path/to/file|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 beginning with http, except between <pre> tags
  sed "/<pre/,/<\/pre>/!s/[^\"]\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@]\{3,\}\)/<a href='\1'> [*]<\/a><tt>\1<\/tt>/gi" | \
  #-- Hyperlink URLs beginning with http, except between <pre> tags
  sed "/<pre/,/<\/pre>/!s/^\(http:\/\/[-a-z\%0-9\~\\\/\"\'\.\@]\{3,\}\)/<a href='\1'> [*]<\/a><tt>\1<\/tt>/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" | \
  #-- 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>/!s/[ ]\{2\}/\&nbsp;\&nbsp;/g" | \
  #-- Turn line breaks into <br> tags unles they are between 'pre' tags
  sed "/<pre/,/<\/pre>/!s/^/<br>/g" 
echo "</td></tr></table>"
  
 echo "<br>"

 #-- Define the cgi program which will handle the adding of 
 #-- comments to a particular text file.
 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/add-comment"
 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 = \"plaintext\">
      
    <hr><a name = \"commentForm\"></a>
    <center>
    <small><strong>
    <a href=\"#toc\">BACK TO THE TABLE OF CONTENTS</a>
    </strong></small>
    <br>
    <em>
      If you wish, you may add a comment, suggestion or other contribution which will appear at the
      end of this document.
      <br>Any input you make is greatly appreciated.
    </em>
    <br><br>

    <strong>Your Comment (or other contribution to this document)</strong><br>
    <textarea name = \"comment\"  cols = \"70\" 
              rows = \"5\"></textarea>
    <br><br>	      
    <strong>Your Name <small>[OPTIONAL BUT NICE]</small></strong><br>
    <textarea name = \"username\"  cols = \"70\" 
              rows = \"1\"></textarea>
    <br><br>
    <input   type = \"submit\"  value = \"S U B M I T   Y O U R   C O M M E N T\">
    </center>
    </form>"
      
 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
 #rm -f plain-text-toc.temp