#!/usr/bin/awk -f
#
#	Take the blog-avatars collection page.
#	Pass anchor tags through.
#	Turn images into images and HTML to include the images
#	into blog entries.
#
#	Image SRC/ALT/HREF must all be on one line in the input file.
#
#	This does only the middle of the resultant new web page.
#	The head and tail are handled by another script.
#	Input lines that are not anchors or images have been filtered out.
#
#	Brent Chivers				2005 July 8
#
#	For the images that need to be tables with artists'	BDC 2005/07/11
#	names, put the HTML in the 2nd column of a table.
#	(Trying to make it easier to highlight/copy the HTML.  It worked.)
#
#	Put them ALL in 2-column tables -- they format better for grabbing.
#
#	Add support for <TITLE> tags.				BDC 2005/11/12
#	If there was no <ALT> attribute in the source, make an empty one.
#
#	<ALT> and <TITLE> strings can be delimited by ``	BDC 2005/11/13
#	(backticks) as well as "" or '' -- adding support for this.
#
#	Was creating output file with "" as the string delimiter always.
#	Note and preserve the original delimiter characters -- there may
#	have been a reason for using something other than "".
#	(That got ugly fast.)
#
#	Look for (by) strings if there is ALT or TITLE (not just ALT).
#
#	Stop generating output after reading "</HTML>".		BDC 2005/11/24
#	(Doesn't work, but stopping at '<A NAME="bottom">' does work.)
#
#	Add BORDER="0" to copy/paste output.			BDC 2006/12/16
#
#	Didn't handle nested (by) strings with nested parens	BDC 2009/04/30
#	properly.  So far, all end with "))", so look for those.
#
#	Creating the pasteable links, ALIGN=RIGHT instead of	BDC 2018/Mar/17
#	LEFT. (I put more icons on the right in my journal entries.
#	I sHould have changed this LONG ago.)


/<\/[Hh][Tt][Mm][Ll]>/			{exit}

/<[Aa] [Nn][Aa][Mm][Ee]="bottom">/	{print ; exit}

/<[Aa] [Nn][Aa][Mm][Ee]="+.*"+>/	{print ; next}
#					All else are images
{
#					Look for IMG SRC tags....
src1 = index ( $0, "<IMG SRC=\"" )
if (src1) {	src1 += 10
		src2 = index ( substr ($0, src1), "\"" )
		src2 -= 1
		src = substr ($0, src1, src2)
		src_delim = "\""
	}
if (src1 == 0) {
	src1 = index ( $0, "<img src=\"" )
	if (src1) {	src1 += 10
			src2 = index ( substr ($0, src1), "\"" )
			src2 -= 1
			src = substr ($0, src1, src2)
			src_delim = "\""
		}
	}
if (src1 == 0) {
	src1 = index ( $0, "<IMG SRC='" )
	if (src1) {	src1 += 10
			src2 = index ( substr ($0, src1), "'" )
			src2 -= 1
			src = substr ($0, src1, src2)
			src_delim = "'"
		}
	}
if (src1 == 0) {
	src1 = index ( $0, "<img src='" )
	if (src1) {	src1 += 10
			src2 = index ( substr ($0, src1), "'" )
			src2 -= 1
			src = substr ($0, src1, src2)
			src_delim = "'"
		}
	}
if (src1 == 0) {
	src1 = index ( $0, "<IMG SRC=`" )
	if (src1) {	src1 += 10
			src2 = index ( substr ($0, src1), "`" )
			src2 -= 1
			src = substr ($0, src1, src2)
			src_delim = "`"
		}
	}
if (src1 == 0) {
	src1 = index ( $0, "<img src=`" )
	if (src1) {	src1 += 10
			src2 = index ( substr ($0, src1), "`" )
			src2 -= 1
			src = substr ($0, src1, src2)
			src_delim = "`"
		}
	}
#					Look for ALT tags....
alt1 = index ( $0, "ALT=\"" )
if (alt1) {	alt1 += 5
		alt2 = index ( substr ($0, alt1), "\"" )
		alt2 -= 1
		alt = substr ($0, alt1, alt2)
		alt_delim = "\""
	}
if (alt1 == 0) {
	alt1 = index ( $0, "alt=\"" )
	if (alt1) {	alt1 += 5
			alt2 = index ( substr ($0, alt1), "\"" )
			alt2 -= 1
			alt = substr ($0, alt1, alt2)
			alt_delim = "\""
		}
	}
if (alt1 == 0) {
	alt1 = index ( $0, "ALT='" )
	if (alt1) {	alt1 += 5
			alt2 = index ( substr ($0, alt1), "'" )
			alt2 -= 1
			alt = substr ($0, alt1, alt2)
			alt_delim = "'"
		}
	}
if (alt1 == 0) {
	alt1 = index ( $0, "alt='" )
	if (alt1) {	alt1 += 5
			alt2 = index ( substr ($0, alt1), "'" )
			alt2 -= 1
			alt = substr ($0, alt1, alt2)
			alt_delim = "'"
		}
	}
if (alt1 == 0) {
	alt1 = index ( $0, "ALT=`" )
	if (alt1) {	alt1 += 5
			alt2 = index ( substr ($0, alt1), "`" )
			alt2 -= 1
			alt = substr ($0, alt1, alt2)
			alt_delim = "`"
		}
	}
if (alt1 == 0) {
	alt1 = index ( $0, "alt=`" )
	if (alt1) {	alt1 += 5
			alt2 = index ( substr ($0, alt1), "`" )
			alt2 -= 1
			alt = substr ($0, alt1, alt2)
			alt_delim = "`"
		}
	}
#					Look for TITLE tags....
title1 = index ( $0, "TITLE=\"" )
if (title1) {	title1 += 7
		title2 = index ( substr ($0, title1), "\"" )
		title2 -= 1
		title = substr ($0, title1, title2)
		title_delim = "\""
	}
if (title1 == 0) {
	title1 = index ( $0, "title=\"" )
	if (title1) {	title1 += 7
			title2 = index ( substr ($0, title1), "\"" )
			title2 -= 1
			title = substr ($0, title1, title2)
			title_delim = "\""
		}
	}
if (title1 == 0) {
	title1 = index ( $0, "TITLE='" )
	if (title1) {	title1 += 7
			title2 = index ( substr ($0, title1), "'" )
			title2 -= 1
			title = substr ($0, title1, title2)
			title_delim = "'"
		}
	}
if (title1 == 0) {
	title1 = index ( $0, "title='" )
	if (title1) {	title1 += 7
			title2 = index ( substr ($0, title1), "'" )
			title2 -= 1
			title = substr ($0, title1, title2)
			title_delim = "'"
		}
	}
if (title1 == 0) {
	title1 = index ( $0, "TITLE=`" )
	if (title1) {	title1 += 7
			title2 = index ( substr ($0, title1), "`" )
			title2 -= 1
			title = substr ($0, title1, title2)
			title_delim = "`"
		}
	}
if (title1 == 0) {
	title1 = index ( $0, "title=`" )
	if (title1) {	title1 += 7
			title2 = index ( substr ($0, title1), "`" )
			title2 -= 1
			title = substr ($0, title1, title2)
			title_delim = "`"
		}
	}
#					Look for (by artist content)
if ((alt1) || (title1)) {
	by1 = index ( $0, " (by " )
	if (by1) {	by1 += 5
			by2 = index ( substr ($0, by1), "))" )
			if (by2 == 0) {	by2 = index ( substr ($0, by1), ")" )
					by2 -= 1
					}
			by = substr ($0, by1, by2)
		}
	}
else	{ by1 = 0 }
#					Look for Anchor Link tags....
href1 = index ( $0, "<A HREF=\"" )
if (href1) {	href1 += 9
		href2 = index ( substr ($0, href1), "\"" )
		href2 -= 1
		href = substr ($0, href1, href2)
		href_delim = "\""
	}
if (href1 == 0) {
	href1 = index ( $0, "<a href=\"" )
	if (href1) {	href1 += 9
			href2 = index ( substr ($0, href1), "\"" )
			href2 -= 1
			href = substr ($0, href1, href2)
			href_delim = "\""
		}
	}
if (href1 == 0) {
	href1 = index ( $0, "<A HREF='" )
	if (href1) {	href1 += 9
			href2 = index ( substr ($0, href1), "'" )
			href2 -= 1
			href = substr ($0, href1, href2)
			href_delim = "'"
		}
	}
if (href1 == 0) {
	href1 = index ( $0, "<a href='" )
	if (href1) {	href1 += 9
			href2 = index ( substr ($0, href1), "'" )
			href2 -= 1
			href = substr ($0, href1, href2)
			href_delim = "'"
		}
	}
if (href1 == 0) {
	href1 = index ( $0, "<A HREF=`" )
	if (href1) {	href1 += 9
			href2 = index ( substr ($0, href1), "`" )
			href2 -= 1
			href = substr ($0, href1, href2)
			href_delim = "`"
		}
	}
if (href1 == 0) {
	href1 = index ( $0, "<a href=`" )
	if (href1) {	href1 += 9
			href2 = index ( substr ($0, href1), "`" )
			href2 -= 1
			href = substr ($0, href1, href2)
			href_delim = "`"
		}
	}
if (src1) {
	if (by1) {
		printf ("<table align=left><tr><td>")
		if (href1) { printf ("<a href=%s%s%s>", href_delim, href, href_delim) }
		printf ("<img src=%s%s%s alt=%s%s%s", src_delim, src, src_delim, alt_delim, alt, alt_delim)
		if (title1) { printf (" title=%s%s%s", title_delim, title, title_delim) }
		printf (">")
		if (href1) { printf ("</a>") }
		printf ("</td><td rowspan=2>")
		if (href1) { printf ("&lt;A HREF=%s%s%s&gt;", href_delim, href, href_delim) }
		printf ("&lt;TABLE ALIGN=RIGHT&gt;")
		printf ("&lt;TR&gt;&lt;TD&gt;&lt;IMG SRC=%s%s%s BORDER=\"0\"", src_delim, src, src_delim)
		if (alt1) { printf (" ALT=%s%s%s", alt_delim, alt, alt_delim) } else { printf (" ALT=\"\"") }
		if (title1) { printf (" TITLE=%s%s%s", title_delim, title, title_delim) }
		printf ("&gt;&lt;/TD&gt;&lt;/TR&gt;")
		printf ("&lt;TR&gt;&lt;TD&gt;&lt;SMALL&gt;&lt;SMALL&gt;&lt;SMALL&gt;%s&lt;/SMALL&gt;&lt;/SMALL&gt;&lt;/SMALL&gt;", by)
		printf ("&lt;/TD&gt;&lt;/TR&gt;&lt;/TABLE&gt;")
		if (href1) { printf ("&lt;/A&gt;") }
		printf ("</td></tr><tr><td>")
		if (href1) { printf ("<a href=%s%s%s>", href_delim, href, href_delim) }
		printf ("<small><small><small>%s</small></small></small>", by)
		if (href1) { printf ("</a>") }
		printf ("</td></tr></table><br clear=all>\n")
		}
	else	{
		printf ("<table align=left><tr><td>")
		if (href1) { printf ("<a href=%s%s%s>", href_delim, href, href_delim) }
		printf ("<img src=%s%s%s", src_delim, src, src_delim)
		if (alt1) { printf (" alt=%s%s%s", alt_delim, alt, alt_delim) }
		if (title1) { printf (" title=%s%s%s", title_delim, title, title_delim) }
		printf (" align=left>")
		if (href1) { printf ("</a>") }
		printf ("</td>\n<td>")
		if (href1) { printf ("&lt;A HREF=%s%s%s&gt;", href_delim, href, href_delim) }
		printf ("&lt;IMG SRC=%s%s%s BORDER=\"0\"", src_delim, src, src_delim)
		if (alt1) { printf (" ALT=%s%s%s", alt_delim, alt, alt_delim) } else { printf (" ALT=\"\"") }
		if (title1) { printf (" TITLE=%s%s%s", title_delim, title, title_delim) }
		printf (" ALIGN=RIGHT&gt;")
		if (href1) { printf ("&lt;/A&gt;") }
		printf ("</td></tr></table><br clear=all>\n")
		}
	}
#    else { printf ("SKIPPING:	%s\n", $0) }
# print
# printf ("%d	%d	%s\n", src1, src2, src)
# printf ("%d	%d	%s\n", alt1, alt2, alt)
# printf ("%d	%d	%s\n", title1, title2, title)
# printf ("%d	%d	%s\n", href1, href2, href)
}