#! d:/mjb/unixutils/bin/sh.exe
# A command line to reformat a text 'documentation' document into
# some kind of html


# D:\mjb\eg\asp\doc>cat fnDisplayRecordset-doc.txt | sed "s/^ *\([A-Z]\{3,\}[A-Z ]*\)/<i>\1<\/i>/g"
# | sed "s/[ ]\{2\}/\&nbsp;\&nbsp;/g" | sed "s/^/<br>/g" | less

# Slightly better.
#D:\mjb\eg\asp\doc>cat fnDisplayRecordset-doc.txt | sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | sed "s/
#^ *\([A-Z]\{3,\}[A-Z ]*\)/<i>\1<\/i>/g" | sed "s/[ ]\{2\}/\&nbsp;\&nbsp;/g" | sed "s/^/<br>/g" > t
#t.html

# This version makes the section headings into 'heading' elements and makes the
# function title or name into an h3 element. This has the disadvantage that unless
# there is a style sheet, browsers put quite a lot of space after the 
# heading elements. However, these heading elements will be usefull if it is
# desired to convert the html into some more 'robust' format such as pdf.
# By using the 'heading' elements and an html2pdf conversion engine, such as 
# for example the 'htmldoc' application, it is possible to generate a table of
# contents with accurate page numbering for a given output device. This is the
# real 'shang-ri-la' of printing as far as I am concerned at the moment.
#
# By placing a 'for' loop around this command line in a 'sh' shell it should be
# possible to batch process plain text documentation files into html and pdf.
# There will be those people who are shaking their heads and muttering the words
# xml, and sgml and document parsing etc, and really I suppose that I am in
# sympathy with those people. But the power of a dodgy plain text format should
# not be underestimated.

#cat fnDisplayRecordset-doc.txt | sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | sed "1,
#/^ *DESC/s/^ *[a-z]\{2\}.*/<h3>&<\/h3>/g" | sed "s/^ *\([A-Z]\{3,\}[A-Z ]*\)/<h4><i>\1<\/i><\/h4>/
#g" | sed "s/[ ]\{2\}/\&nbsp;\&nbsp;/g" | sed "s/^/<br>/g" > tt.html

# This version creates links to url like strings.
#
# The command line below has been run successfully on the Windows 2000
# command shell. (without the line continuation characters. I believe it
# is possible to use the open and close brackets characters for line
# continuation in the Win NT/2000 shell, but I really dont want to go 
# into it.


for f in $(ls *doc.txt)
do
 cat $f | \
 sed -e "s/</\&lt;/g" -e "s/>/\&gt;/g" | \
 sed "1, /^ *DESC/s/^ *[a-z]\{2\}.*/<h3>&<\/h3>/g"  | \
 sed "s/^ *\([A-Z]\{3,\}[A-Z ]*\)$/<h4><i>\1<\/i><\/h4> /g" | \
 sed "s/^ *\([A-Z]\{3,\}[A-Z ]*[: ]\)/<h4><i>\1<\/i><\/h4>/g" | \
 sed "s/[ ]\{2\}/\&nbsp;\&nbsp;/g" | \
 sed "s/\(http:\/\/[^ \n\r]\{3,\}\)/<a href='\1'>\1<\/a>/gi" | \
 sed "s/[^a-zA-Z\/]www\.[^ ]\{2,\}/<a href='&'>&<\/a>/gi" | \
 sed "s/^/<br>/g" 
done