-- Parsec parser of tokenized XML obtained from TypEr output (tokenization made by Tagsoup).

-- This parser, although working with different kind of XML than the Edoc XML parser,
-- creates the same data structure as Edoc parser does, thus giving some chance
-- to put them together in the future.

module Language.Edoc.Xml2Hs.TyParser where

import Text.HTML.TagSoup
import Text.HTML.TagSoup.Parser
import Text.HTML.TagSoup.Match
import Text.HTML.TagSoup.Type
import Text.ParserCombinators.Parsec
import Text.ParserCombinators.Parsec.Pos

import Language.Edoc.Xml2Hs.Type

-- Import stuff from the Edoc parser as many of it can be reused.

import Language.Edoc.Xml2Hs.Parser

-- Parse a whole set of specifications. Currently the TypEr output does not
-- contain module names, so assign a name from the argument.
-- The input stream basically consists of type specifications only.
-- So the parser just expects many type specifications each one
-- describing a single function.

pTyMod :: String -> GenParser Tag PState EDocMod

pTyMod mname = do
  skipMany ignore
  fds <- many pSpec
  return $ EDocMod {
    em_name = mname
   ,em_args = []
   ,em_tdefs = []
   ,em_funcs = fds}


-- Parse a function type declaration.

pSpec = do
  fn <- expTag "spec"
  let arity = read $ fromAttrib "arity" fn
  ftyp <- pErlType
  closeTag "spec"
  return $ EFunDecl {
    ef_name = fromAttrib "function" fn
   ,ef_arity = arity
   ,ef_args = map (('_':) . show) (take arity [1 .. ])
   ,ef_tspec = [ETypeSpec {
                  ts_ename = ErlName "" "" ""
                 ,ts_type = ftyp
                 ,ts_ldef = []}]}