function HTML_ToText (text)
-- Declare variables, load the file. Make tags lowercase.
text = string.gsub (text,"(%b<>)",
function (tag)
return tag:lower()
end)
--[[
First we kill the developer formatting (tabs, CR, LF)
and produce a long string with no newlines and tabs.
We also kill repeated spaces as browsers ignore them anyway.
]]
local devkill=
{
["("..string.char(10)..")"] = " ",
["("..string.char(13)..")"] = " ",
["("..string.char(15)..")"] = "",
["(%s%s+)"]=" ",
}
for pat, res in pairs (devkill) do
text = string.gsub (text, pat, res)
end
-- Then we remove the header. We do this by stripping it first.
text = string.gsub (text, "(<%s*head[^>]*>)", "")
text = string.gsub (text, "(<%s*%/%s*head%s*>)", "")
text = string.gsub (text, "(,*<%/head>)", "")
-- Kill all scripts. First we nuke their attribs.
text = string.gsub (text, "(<%s*script[^>]*>)", "")
text = string.gsub (text, "(