#! lua-5.0.exe
-- Create an HTML table of content (TOC) from an HTML file. Made mostly for the Lua manual...
--
-- Take two parameters: the file to process as input, the resulting file as output.
--
-- The program extracts the hierarchical titles (<Hn> tags) and store them with their level.
-- It will then output the list of titles in a hierarchical HTML list.
--
-- Actually, it will also take the latest named link before each title, so the above list
-- can be used to jump to the chosen level.
--
-- Note that the input HTML file must follow this structure for the output to be accurate.
-- I had to make minor, invisible edits to the original Lua manuals,
-- to create anchors or to move them before the title.
--
-- The result must be inserted by hand in the HTML file. If used in a separated page,
-- you must add the destination page name to the links.
--
-- by Philippe Lhoste <PhiLho(a)GMX.net> http://Phi.Lho.free.fr
-- v. 1.2 -- 2004/04/22 -- Conversion to Lua 5.0
-- v. 1.1 -- 2003/05/20 -- Output of TOC
-- v. 1.0 -- 2002/07/26 -- Initial code, extraction of data

filenameIn  = (arg and arg[1]) or "-"	-- If absent, use standard input
filenameOut = (arg and arg[2]) or "-"	-- If absent, use standard output

tocList = {}
--lineNumber = 0
refNumber = 0
prevRef = ""
prevLevel = 0

-- Extract useful information from the given line.
function ProcessLine(line)
	if line == nil then
		return nil
	end
--	lineNumber = lineNumber + 1

	-- Remove tags other than those we are looking for.
	-- Replace the interesting tags by control chars, remove all other tags,
	-- and restore the control chars to the corresponding tags.
	line = string.gsub(line, "<[Hh]", "\1")
	line = string.gsub(line, "</[Hh]", "\2")
	line = string.gsub(line, "<[Aa] ", "\3")
	line = string.gsub(line, "</[Aa]>", "\4")
	line = string.gsub(line, "<[^>]->", "")
	line = string.gsub(line, "\1", "<H")
	line = string.gsub(line, "\2", "</H")
	line, nn = string.gsub(line, "\3", "<A ")
	line = string.gsub(line, "\4", "</A>")

	local startRef, endRef, level, ref, title
	local tocData = {}

	-- Capture latest referenced string (by a <A NAME="..."> tag) of the line
	endRef = 0
	repeat
		startRef, endRef, ref = string.find(line, [[<A [Nn][Aa][Mm][Ee]=['"]([^'"]*)['"]>.*</A>]], endRef + 1)
		if ref ~= nil then
			prevRef = ref
		end
	until startRef == nil
	-- Remove the anchors
	line = string.gsub(line, "<A [^>]+>", "")
	line = string.gsub(line, "</A>", "")

	-- Search if the line has a title (<Hn> tag, up to level 4)
	_, _, level, title = string.find(line, [[<[Hh]([1234])>([^<]*)</[Hh][1234]>]], 1)
	if level ~= nil then
		-- No need to loop to search another title in the line,
		-- unless the HTML is very sloppy...
		refNumber = refNumber + 1
		tocData =
		{
			level = level,
			ref   = prevRef,
			title = title
		}
		tocList[refNumber] = tocData
	end
	return true
end

-- Output an element of table of content.
function OutputTOC(el)
	local level, nLevel
	level = el.level
	nLevel = tonumber(level)
	while nLevel > prevLevel do
		io.write"<UL>\n"
		nLevel = nLevel - 1
	end
	while nLevel < prevLevel do
		io.write"</UL>\n"
		nLevel = nLevel + 1
	end
	prevLevel = tonumber(level)
	io.write([[<LI><A HREF="#]] .. el.ref .. [[">]] .. el.title .. [[</A>]] .. "\n")
end

-- Process the input file
function ProcessFile()
	local fI, fO

	if filenameIn ~= '-' then
		fI = io.open(filenameIn, "r")
		if fI == nil then
			return false, "open r " .. filenameIn
		end
	else
		fI = io.stdin
	end

	if filenameOut ~= '-' then
		fO = io.open(filenameOut, "w")
		if fO == nil then
			return false, "open w " .. filenameOut
		end
		io.output(fO)
	end

	-- Loop on the lines and add any <Hn> tag to the tocList table
	-- Note that these titles must be on the same line, no <Hn> on one line and the title on the next...
	for line in fI:lines() do
		ProcessLine(line)
	end

	-- Build the table of content
	io.write([[
<P>
<A NAME="contents"></A>
<H1>0 - Contents</H1>
]])
	for i, elements in ipairs(tocList) do
		OutputTOC(elements)
	end
	while prevLevel > 0 do
		io.write"</UL>\n"
		prevLevel = prevLevel - 1
	end
	io.write"\n<HR>\n"

	if filenameIn ~= '-' then
		fI:close()
	end
	if filenameOut ~= '-' then
		io.output()
		fO:close()
	end

	return true, nil
end

result, op = ProcessFile()
if not result then
	io.stderr:write("Error on operation : " .. (op or 'nil') .. '\n')
else
	io.stderr:write("Done !\n")
end