#!/usr/bin/env lua

--

--	$ bvvcal [bezirksname] [startdatum] [enddatum]

--

--	bezirksname : Default "friedrichshain-kreuzberg"

--	startdatum : in deutscher Notation (z.B. 31.12.2010), Default: heute

--	enddatum : Default: startdatum

--

--	Konvertiert den Berliner BVV-Sitzungskalender 

--	ins iCalendar 2.0 (vCal) Format

--

--	Benötigt: 

--	- htmltidy : http://tidy.sourceforge.net/

--	- wget : http://www.gnu.org/software/wget/

--	- Lua 5.1.x : http://www.lua.org/

--	- LuaExpat : http://www.keplerproject.org/luaexpat/

--

--	Revisionen:

--	- 0.2 : Uhrzeiten waren lokal, nicht GMT. Korrigiert.

--	- 0.1 : initial release

--

--	Autor: Timm S. Müller <timm-pirat@schulze-mueller.de>

--

local lxp = require "lxp"

local district = arg[1] or "friedrichshain-kreuzberg"

local startdate = arg[2] or os.date("%d.%m.%Y")

local enddate = arg[3] or startdate

local uid_uri = district .. ".bvv.berlin.piratenpartei.de"

----- --- --- -- - -  -   -    -

--	Encoding ist laut RFC 2445 UTF-8, dies ist ein Zugeständnis

--	an Kalendertools, die das nicht mitbekommen haben:

local umlconv = { 

	["ä"] = "ae", ["Ä"] = "Ae", 

	["ö"] = "oe", ["Ö"] = "Oe",

	["ü"] = "ue", ["Ü"] = "Ue",

	["ß"] = "ss"

}

local state = "waitcontent"

local record, cell, result = { }, { }, { }

local lnr = 1

local parser = lxp.new {

	StartElement = function(parser, tagname, attr)

		if state == "waitcontent" then

			if tagname == "div" and attr.id == "allrisContent" then

				state = "waittab"

			end

		elseif state == "waittab" then

			if tagname == "table" and attr.class == "tl1" then

				state = "waitrow"

			end

		elseif state == "waitrow" then

			if tagname == "tr" then

				state = "waitcell"

			end

		end

	end,

	EndElement = function(parser, tagname)

		if state == "waitrow" or state == "waittable" then

			if tagname == "table" then

				state = "end"

			end

		elseif state == "waitcell" then

			if tagname == "tr" then

				local day = table.remove(record, 1)

				local sdate = table.remove(record, 1)

				local d, m, y = (sdate or ""):match("^%s*(%d+)%.(%d+)%.(%d+)")

				local time = table.remove(record, 1)

				if d and m and y and time then

					local sH, sM, eH, eM = time:match("^%s*(%d+):(%d+)%s*%-%s*(%d+):(%d+)")

					if not sH then

						sH, sM = time:match("^%s*(%d+):(%d+)")

					end

					if sH then

						-- convert dates to GMT

						local startdate = os.date("!%s", os.time { day=d, month=m, year=y, hour=sH, min=sM })

						local enddate = eH and os.date("!%s", os.time { day=d, month=m, year=y, hour=eH, min=eM })

						local what = table.concat(record, " "):match("^%s*(.-)%s*$")

						what = what:gsub("([\128-\255].)", umlconv)

						table.insert(result, { startdate, enddate, what })

					end

				end

				state = "waitrow"

				record = { }

			elseif tagname == "td" then

				table.insert(record, table.concat(cell))

				cell = { }

			end

		end

	end,

	CharacterData = function(parser, s)

		if state == "waitcell" then

			table.insert(cell, s)

		end

	end

}

parser:setencoding("ISO-8859-1")

----- --- --- -- - -  -   -    -

local cmd = ("wget --quiet --post-data 'kaldatvon=%s&kaldatbis=%s' 'http://www.berlin.de/ba-%s/bvv-online/si010.asp' -O - | tidy -latin1 -asxml -i -w 0 2>/dev/null -q"):format(startdate, enddate, district)

-- print(cmd)

local f = io.popen(cmd)

for line in f:lines() do

-- 	print(lnr .. " : " .. line)

	parser:parse(line)

	lnr = lnr + 1

end

----- --- --- -- - -  -   -    -

print "BEGIN:VCALENDAR"

print "PRODID:BVV2CAL"

print "VERSION:2.0"

print ""

for i = 1, #result do

	local r = result[i]

	print "BEGIN:VEVENT"

	local startdate = os.date("%Y%m%dT%H%M", r[1])

	print("UID:" .. startdate .. "." .. uid_uri)

	print("SUMMARY:" .. r[3])

	print("DTSTART:" .. startdate .. "00Z")

	if r[2] then

		local enddate = r[2] and os.date("%Y%m%dT%H%M", r[2])

		print("DTEND:" .. enddate .. "00Z")

	end

	print "END:VEVENT"

	print ""

end

print "END:VCALENDAR"