--- Example for a Curry program which extracts email addresses
--- from a HTML page.
--- Since this program requires a few libraries, it must be executed
--- as follows:
---
---     > cypm add -d setfunctions # add dependencies
---     > cypm add -d html2
---     > cypm install             # install required dependencies
---     > pakcs :l EmailsInHtml :eval main :quit

import Control.SetFunctions
import HTML.Base
import HTML.Parser

-- Representation of HTML documents defined in HTML.Base as follows:
-- data StaticHtml = HText String | HStruct String Attrs [StaticHtml]

-- Gets some HTML document containing the argument HTML element.
withHtmlElem :: StaticHtml -> StaticHtml
withHtmlElem helem = helem ? HStruct _ _ (withElem (withHtmlElem helem))

-- Gets some list containing a given element.
withElem :: Data a => a -> [a]
withElem e = e:_ ? _:withElem e

-- Extracts some email address in an HTML document.
anyEmailAddress :: StaticHtml -> String
anyEmailAddress (withHtmlElem
                   (HStruct _ 
                      (withElem ("href","mailto:" ++ name)) _)) = name

-- collect and print all email addresses in an HTML document:
main :: IO ()
main = do
  elems <- readHtmlFile "syllabus.html"
  printValues (set1 anyEmailAddress (head elems))