#! /sed -f
# Thu May 18 12:43:45 2000 by tilmann@bitterberg.de
# Changed Mon Feb 26 12:34:19 2000 by bonzini@gnu.org (removed
# limitations on the number of links)
#
# Description:
# Creates an index of links from a HTML file
# Does something similar like lynx -force_html -dump but
# leaves the document html (generate an index of links)
#
# Example: Input
#
# foo1 Click here foo2
#
#
# Output:
#
# foo1 [1] Click here foo2
#
[1] http://link.org
#
#
# NOTE:
# 1) Will break at links like foo2\n397\nPREVIOUS ENTRIES
# using newline as separator to the 's' command
s
]*>\)\([^\n]*\(\n\)\)\([^\n]*\)\(.*$\)
<||||||=\1[\5] \3\5\4\6\4[\5] \2
# |----------1---------||-------3------| |---5---||--6--|
# |---2----| |--4-|
# Field Contains:
# \1 the link text up to the closing >
# \2 the link itself (http://foo.com)
# \3 the rest of the input line
# \4 a newline (\n)
# \5 the number we would like to use
# \6 everything up to the end of patternspace (i.e. the previous entries)
#
# Now the line looks like:
# foo1 <||||||="blah.html">[1] foo2\n1\n[1] blah.html
h; # save the new entry and the updated counter
s/[^\n]*\n// # remove the HTML line from hold space
x
s/\n.*// # remove what went in hold space from pattern space
b loop; # look if there is another link in that line
}
s/<||||||/ insert index
/<\/[Bb][Oo][Dd][Yy]>/{
G # insert saved stuff
s/\(<\/[Bb][Oo][Dd][Yy]>[^\n]*\)\n[^\n]*\n*\(.*\)/
\2\1/;
}