Language/TCT/Write/XML.hs

   1 {-# LANGUAGE FlexibleContexts #-}
   2 {-# LANGUAGE FlexibleInstances #-}
   3 {-# LANGUAGE OverloadedLists #-}
   4 {-# LANGUAGE OverloadedStrings #-}
   5 {-# LANGUAGE ViewPatterns #-}
   6 {-# OPTIONS_GHC -fno-warn-orphans #-}
   7 module Language.TCT.Write.XML where
   8
   9 import Control.Monad (Monad(..))
  10 import Data.Bool
  11 import Data.Default.Class (Default(..))
  12 import Data.Eq (Eq(..))
  13 import Data.Foldable (Foldable(..))
  14 import Data.Function (($), (.))
  15 import Data.Functor ((<$>), (<$), ($>))
  16 import Data.Maybe (Maybe(..))
  17 import Data.Monoid (Monoid(..))
  18 import Data.Ord (Ord(..))
  19 import Data.Semigroup (Semigroup(..))
  20 import Data.Sequence (Seq, ViewL(..), ViewR(..), (<|))
  21 import Data.Set (Set)
  22 import Data.TreeSeq.Strict (Tree(..), tree0)
  23 import Data.Tuple (uncurry)
  24 import Prelude (Num(..), undefined)
  25 import qualified Data.Char as Char
  26 import qualified Data.List as List
  27 import qualified Data.Sequence as Seq
  28 import qualified Data.Text.Lazy as TL
  29 import qualified Language.TCT.Write.Plain as Plain
  30 import qualified System.FilePath as FP
  31
  32 -- import Language.TCT.Debug
  33 import Language.TCT.Utils
  34 import Language.TCT hiding (Parser)
  35 import Language.XML
  36 import Text.Blaze.XML ()
  37
  38 -- | Main entry point
  39 --
  40 -- NOTE: 'XmlNode' are still annotated with 'Cell',
  41 --       but nothing is done to preserve any ordering amongst them,
  42 --       because 'Node's sometimes need to be reordered
  43 --       (eg. about/title may have a title from the section before,
  44 --       hence outside of about).
  45 --       Still holding: @'cell_begin' < 'cell_end'@ and 'Cell' nesting.
  46 document :: Roots -> XMLs
  47 document doc =
  48         -- (`S.evalState` def) $
  49         case Seq.viewl doc of
  50          sec@(Tree (unCell -> NodeHeader HeaderSection{}) _body) :< foot ->
  51                 let (titles, content) = partitionSection sec in
  52                 case Seq.viewl titles of
  53                  (unTree -> Cell bt et _) :< _ ->
  54                         xmlify def
  55                          { inh_titles = titles
  56                          , inh_figure = True
  57                          } contentWithAbout <>
  58                         xmlify def foot
  59                         where
  60                         contentWithAbout =
  61                                 case Seq.findIndexL isAbout content of
  62                                  Nothing -> Tree (Cell bt et $ NodeHeader $ HeaderColon "about" "") mempty <| content
  63                                  Just{} -> content
  64                         isAbout = \case
  65                          (unTree -> (unCell -> NodeHeader (HeaderColon "about" _wh))) -> True
  66                          _ -> False
  67                  _ -> xmlify def doc
  68          _ -> xmlify def doc
  69
  70 partitionSection :: Root -> (Roots, Roots)
  71 partitionSection (Tree (unCell -> NodeHeader (HeaderSection lvlPar)) body) =
  72         case Seq.viewl body of
  73          EmptyL -> mempty
  74          title@(unTree -> Cell _bt et NodePara) :< rest ->
  75                 let (subtitles, content) = spanlSubtitles et rest in
  76                 (title <| (subtitles >>= subTrees), content)
  77                 where
  78                 spanlSubtitles ep ts =
  79                         case Seq.viewl ts of
  80                          sub@(unTree -> Cell bs es (NodeHeader (HeaderSection lvlSub))) :< rs
  81                                 | lvlSub <= lvlPar
  82                                 , pos_line bs - pos_line ep <= 1 ->
  83                                 let (subs, ts') = spanlSubtitles es rs in
  84                                 (sub <| subs, ts')
  85                          _ -> (mempty, ts)
  86          _ -> (mempty, body)
  87 partitionSection _ = mempty
  88
  89 -- * Type 'Inh'
  90 data Inh
  91  =   Inh
  92  {   inh_figure :: Bool
  93  ,   inh_para   :: [Inh -> Root -> XML]
  94  ,   inh_titles :: Roots
  95  }
  96 instance Default Inh where
  97         def = Inh
  98          { inh_figure = False
  99          , inh_para   = List.repeat elementPara
 100          , inh_titles = mempty
 101          }
 102
 103 -- ** 'inh_para'
 104 elementPara :: Inh -> Root -> XML
 105 elementPara inh (Tree c ts) = Tree (XmlElem "para" <$ c) $ xmlify inh ts
 106
 107 elementTitle :: Inh -> Root -> XML
 108 elementTitle inh (Tree c ts) = Tree (XmlElem "title" <$ c) $ xmlify inh ts
 109
 110 elementName :: Inh -> Root -> XML
 111 elementName inh (Tree c ts) = Tree (XmlElem "name" <$ c) $ xmlify inh ts
 112
 113 attributeName :: Inh -> Root -> XML
 114 attributeName _inh (Tree c ts) = tree0 (XmlAttr "name" (Plain.document ts) <$ c)
 115
 116 -- * Class 'Xmlify'
 117 class Xmlify a where
 118         xmlify :: Inh -> a -> XMLs
 119 instance Xmlify Roots where
 120         xmlify inh roots =
 121                 case Seq.viewl roots of
 122                  EmptyL -> mempty
 123                  r@(Tree cr@(Cell _br _er nr) ts) :< rs ->
 124                         case nr of
 125                         ----------------------
 126                          -- NOTE: HeaderColon becomes parent
 127                          -- of any continuous following-sibling HeaderBar or HeaderGreat
 128                          NodeHeader (HeaderColon n _wh)
 129                           | (span, rest) <- spanlHeaderColon rs
 130                           , not $ null span ->
 131                                 xmlify inh (Tree cr (ts<>span)) <>
 132                                 xmlify inh rest
 133                                 where
 134                                 spanlHeaderColon :: Roots -> (Roots, Roots)
 135                                 spanlHeaderColon =
 136                                         Seq.spanl $ \case
 137                                          Tree (unCell -> NodeHeader (HeaderBar   m _)) _ -> m == n
 138                                          Tree (unCell -> NodeHeader (HeaderGreat m _)) _ -> m == n
 139                                          _ -> False
 140                         ----------------------
 141                          -- NOTE: gather HeaderBrackets
 142                          NodeHeader HeaderBrackets{}
 143                           | (span,rest) <- spanlBrackets roots
 144                           , not (null span) ->
 145                                 (<| xmlify inh rest) $
 146                                 element "references" $
 147                                         span >>= xmlify inh
 148                                 where
 149                                 spanlBrackets :: Roots -> (Roots, Roots)
 150                                 spanlBrackets =
 151                                         Seq.spanl $ \case
 152                                          Tree (unCell -> NodeHeader HeaderBrackets{}) _ -> True
 153                                          _ -> False
 154                         ----------------------
 155                          -- NOTE: merge adjacent NodeText-s, shouldn't be needed, but just in case.
 156                          NodeText x
 157                           | Tree (cy@(unCell -> NodeText y)) ys :< rs' <- Seq.viewl rs ->
 158                                 xmlify inh $ Tree (NodeText <$> (x <$ cr) <> (y <$ cy)) (ts <> ys) <| rs'
 159                         ----------------------
 160                          -- NOTE: detect [some text](http://some.url) or [SomeRef]
 161                          NodePair PairParen
 162                           | Tree (Cell bb eb (NodePair PairBracket)) bracket :< rs' <- Seq.viewl rs ->
 163                                 (<| xmlify inh rs') $
 164                                 case bracket of
 165                                  (toList -> [unTree -> Cell bl el (NodeToken (TokenLink lnk))]) ->
 166                                         element "eref" $
 167                                                 xmlAttrs [Cell bl el ("to",lnk)] <>
 168                                                 xmlify inh ts
 169                                  _ ->
 170                                         element "rref" $
 171                                                 xmlAttrs [Cell bb eb ("to",Plain.document bracket)] <>
 172                                                 xmlify inh ts
 173                         ----------------------
 174                          -- NOTE: gather HeaderDash
 175                          _ | (span, rest) <- spanlItems (==HeaderDash) roots
 176                            , not $ null span ->
 177                                 (<| xmlify inh rest) $
 178                                 element "ul" $
 179                                         span >>= xmlify inh{inh_para=List.repeat elementPara}
 180                         ----------------------
 181                          -- NOTE: gather HeaderDot
 182                            | (span,rest) <- spanlItems (\case HeaderDot{} -> True; _ -> False) roots
 183                            , not $ null span ->
 184                                 (<| xmlify inh rest) $
 185                                 element "ol" $
 186                                         span >>= xmlify inh{inh_para=List.repeat elementPara}
 187                                 where
 188                                 spanlItems :: (Header -> Bool) -> Roots -> (Roots, Roots)
 189                                 spanlItems liHeader =
 190                                         Seq.spanl $ \(unTree -> (unCell -> nod)) ->
 191                                                 case nod of
 192                                                  NodeHeader (HeaderColon "li" _wh) -> True
 193                                                  NodeHeader hdr -> liHeader hdr
 194                                                  NodePair (PairElem "li" _as) -> True
 195                                                  _ -> False
 196                         ----------------------
 197                          NodePara | para:inh_para <- inh_para inh ->
 198                                 para inh r <|
 199                                 -- para (() <$ cr) (xmlify inh ts) <|
 200                                 xmlify inh{inh_para} rs
 201                         ----------------------
 202                          -- NOTE: context-free Root
 203                          _ ->
 204                                 xmlify inh r <>
 205                                 xmlify inh rs
 206                         where
 207                         element :: XmlName -> XMLs -> XML
 208                         element n = Tree (XmlElem n <$ cr)
 209 instance Xmlify Root where
 210         xmlify inh tn@(Tree (Cell bn en nod) ts) =
 211                 case nod of
 212                  NodeGroup -> xmlify inh ts
 213                 ----------------------
 214                  NodePara ->
 215                         case inh_para inh of
 216                          [] -> xmlify inh ts
 217                          para:_ -> Seq.singleton $ para inh tn -- para (() <$ cn) $ xmlify inh ts
 218                 ----------------------
 219                  NodeHeader hdr ->
 220                         case hdr of
 221                         --
 222                          HeaderSection{} ->
 223                                 Seq.singleton $
 224                                 element "section" $ head <> xmlify inh' body
 225                                 where
 226                                 (titles, content) = partitionSection tn
 227                                 (attrs, body)     = partitionAttrs content
 228                                 head =
 229                                         case Seq.viewl titles of
 230                                          EmptyL -> mempty
 231                                          title@(unTree -> ct) :< subtitles ->
 232                                                 xmlAttrs (attrs `defaultAttr` (ct $> ("id",getAttrId title))) <>
 233                                                 xmlify inh{inh_para=List.repeat elementTitle} title <>
 234                                                 aliases
 235                                                 where
 236                                                 aliases =
 237                                                         subtitles >>= \subtitle@(unTree -> cs) ->
 238                                                                 return $
 239                                                                 Tree (cs $> XmlElem "alias") $
 240                                                                         xmlAttrs [cs $> ("id",getAttrId subtitle)]
 241                                 inh' = inh
 242                                  { inh_para   = List.repeat elementPara
 243                                  , inh_figure = True
 244                                  }
 245                         --
 246                          HeaderColon n _wh ->
 247                                 let (attrs,body) = partitionAttrs ts in
 248                                 case n of
 249                                 -- NOTE: insert titles into <about>.
 250                                  "about" ->
 251                                         Seq.singleton $
 252                                         element "about" $
 253                                                 xmlify inh' (inh_titles inh) <>
 254                                                 xmlAttrs attrs <>
 255                                                 xmlify inh' body
 256                                 -- NOTE: in <figure> mode, unreserved elements become <figure>
 257                                  _ | inh_figure inh && n`List.notElem`elems || TL.null n ->
 258                                         Seq.singleton $
 259                                         element "figure" $
 260                                                 -- xmlAttrs (setAttr (Cell en en ("type",n)) attrs) <>
 261                                                 xmlAttrs (attrs `defaultAttr` Cell bn bn ("type", n)) <>
 262                                                 case toList body of
 263                                                  [Tree0{}] -> xmlify inh'{inh_para = List.repeat elementPara} body
 264                                                  _         -> xmlify inh'{inh_para = elementTitle : List.repeat elementPara} body
 265                                 -- NOTE: reserved elements
 266                                  _ ->
 267                                         Seq.singleton $
 268                                         element (xmlLocalName n) $
 269                                                 xmlAttrs attrs <>
 270                                                 xmlify inh' body
 271                                 where
 272                                 inh' = inh
 273                                  { inh_para =
 274                                         case n of
 275                                          "about"     -> List.repeat elementTitle
 276                                          "reference" -> elementTitle : List.repeat elementPara
 277                                          "serie"     -> List.repeat attributeName
 278                                          "author"    -> List.repeat attributeName
 279                                          "editor"    -> List.repeat attributeName
 280                                          "org"       -> List.repeat attributeName
 281                                          "note"      -> List.repeat elementPara
 282                                          _           -> []
 283                                  }
 284                         --
 285                          HeaderBar n _wh ->
 286                                 Seq.singleton $
 287                                 element "artwork" $
 288                                         xmlAttrs (Seq.singleton $ Cell bn bn ("type", n)) <>
 289                                         xmlify inh{inh_para=[]} ts
 290                         --
 291                          HeaderGreat n _wh ->
 292                                 Seq.singleton $
 293                                 let (attrs,body) = partitionAttrs ts in
 294                                 element "quote" $
 295                                         xmlAttrs (attrs `defaultAttr` Cell bn bn ("type", n)) <>
 296                                         xmlify inh{inh_para=List.repeat elementPara} body
 297                         --
 298                          HeaderEqual n _wh ->
 299                                 Seq.singleton $
 300                                 Tree0 $ cell $ XmlAttr (xmlLocalName n) $
 301                                         Plain.text (Plain.setStart ts def{Plain.state_escape = False}) ts
 302                         --
 303                          HeaderDot _n   -> Seq.singleton $ element "li" $ xmlify inh ts
 304                         --
 305                          HeaderDash     -> Seq.singleton $ element "li" $ xmlify inh ts
 306                         --
 307                          HeaderDashDash -> Seq.singleton $ Tree0 $ cell $ XmlComment $
 308                                         -- debug1_ ("TS", ts) $
 309                                         -- debug1_ ("RS", (S.evalState (Plain.rackUpLeft ts) Nothing)) $
 310                                         Plain.document ts
 311                                         -- Plain.text def $ S.evalState (Plain.rackUpLeft ts) Nothing
 312                                         {-
 313                                         TreeSeq.mapAlsoNode
 314                                          (cell1 . unCell)
 315                                          (\_k -> fmap $
 316                                                 TreeSeq.mapAlsoNode
 317                                                  (cell1 . unCell)
 318                                                  (\_k' -> cell1 . unCell)) <$> ts
 319                                         -}
 320                         --
 321                          HeaderBrackets ident ->
 322                                 let (attrs,body) = partitionAttrs ts in
 323                                 Seq.singleton $
 324                                 element "reference" $
 325                                         xmlAttrs (setAttr (Cell en en ("id",ident)) attrs) <>
 326                                         xmlify inh'{inh_para = elementTitle : elementTitle : List.repeat elementPara} body
 327                                 where
 328                                 inh' = inh{inh_figure = False}
 329                         --
 330                          HeaderDotSlash p ->
 331                                 Seq.singleton $
 332                                 element "include" $
 333                                         xmlAttrs [cell ("href",TL.pack $ FP.replaceExtension p "dtc")] <>
 334                                         xmlify inh ts
 335                 ----------------------
 336                  NodePair pair ->
 337                         case pair of
 338                          PairBracket | to <- Plain.document ts
 339                                      , TL.all (\c -> Char.isAlphaNum c || Char.isSymbol c) to ->
 340                                 Seq.singleton $
 341                                 element "rref" $
 342                                         xmlAttrs [cell ("to",to)]
 343                          PairStar      -> Seq.singleton $ element "b"    $ xmlify inh ts
 344                          PairSlash     -> Seq.singleton $ element "i"    $ xmlify inh ts
 345                          PairBackquote -> Seq.singleton $ element "code" $ xmlify inh ts
 346                          PairFrenchquote ->
 347                                 Seq.singleton $
 348                                 element "q" $
 349                                         xmlify inh ts
 350                                 {-
 351                                 case ts of
 352                                  (Seq.viewl -> Tree0 (Cell bl el (TokenPlain l)) :< ls) ->
 353                                         case Seq.viewr ls of
 354                                          m :> Tree0 (Cell br er (TokenPlain r)) ->
 355                                                 xmlify inh $
 356                                                         Tree0 (Cell bl el (TokenPlain (Text.dropWhile Char.isSpace l)))
 357                                                          <|(m|>Tree0 (Cell br er (TokenPlain (Text.dropWhileEnd Char.isSpace r))))
 358                                          _ ->
 359                                                 xmlify inh $
 360                                                         Tree0 (Cell bl el (TokenPlain (Text.dropAround Char.isSpace l))) <| ls
 361                                  (Seq.viewr -> rs :> Tree0 (Cell br er (TokenPlain r))) ->
 362                                         xmlify inh $
 363                                                 rs |> Tree0 (Cell br er (TokenPlain (Text.dropAround Char.isSpace r)))
 364                                  _ -> xmlify inh ts
 365                                 -}
 366                          PairHash ->
 367                                 Seq.singleton $
 368                                 element "ref" $
 369                                         xmlAttrs [cell ("to",Plain.document ts)]
 370                          PairElem name attrs ->
 371                                 Seq.singleton $
 372                                 element (xmlLocalName name) $
 373                                         xmlAttrs (Seq.fromList $ (\(_wh,ElemAttr{..}) ->
 374                                                 cell (xmlLocalName elemAttr_name,elemAttr_value)) <$> attrs) <>
 375                                         xmlify inh ts
 376                          _ ->
 377                                 Seq.singleton (Tree0 $ Cell bn bn' $ XmlText open) `unionXml`
 378                                 xmlify inh ts `unionXml`
 379                                 Seq.singleton (Tree0 $ Cell en' en $ XmlText close)
 380                                 where
 381                                 (open, close) = pairBorders pair ts
 382                                 bn' = bn{pos_column=pos_column bn + int (TL.length open)}
 383                                 en' = en{pos_column=pos_column bn - int (TL.length close)}
 384                 ----------------------
 385                  NodeText t -> Seq.singleton $ Tree0 $ cell $ XmlText t
 386                 ----------------------
 387                  NodeToken tok ->
 388                         case tok of
 389                          TokenEscape c -> Seq.singleton $ Tree0 $ cell $ XmlText $ TL.singleton c
 390                          TokenText t   -> Seq.singleton $ Tree0 $ cell $ XmlText t
 391                          TokenTag t    -> Seq.singleton $ element "ref"  $ xmlAttrs [cell ("to",t)]
 392                          TokenLink lnk -> Seq.singleton $ element "eref" $ xmlAttrs [cell ("to",lnk)]
 393                 ----------------------
 394                  NodeLower n as ->
 395                         Seq.singleton $
 396                         element "artwork" $
 397                                 xmlify inh ts
 398                 where
 399                 cell :: a -> Cell a
 400                 cell = Cell bn en
 401                 element :: XmlName -> XMLs -> XML
 402                 element n = Tree (cell $ XmlElem n)
 403 instance Xmlify (Seq (Cell (XmlName,TL.Text))) where
 404         xmlify _inh = xmlAttrs
 405
 406 -- * Elements
 407
 408 -- | Reserved elements' name
 409 elems :: Set TL.Text
 410 elems =
 411  [ "about"
 412  , "abstract"
 413  , "address"
 414  , "alias"
 415  , "annotation"
 416  , "area"
 417  , "artwork"
 418  , "aside"
 419  , "audio"
 420  , "author"
 421  , "authors"
 422  , "bcp14"
 423  , "br"
 424  , "call"
 425  , "city"
 426  , "code"
 427  , "comment"
 428  , "comments"
 429  , "country"
 430  , "date"
 431  , "dd"
 432  , "define"
 433  , "del"
 434  , "div"
 435  , "dl"
 436  , "document"
 437  , "dt"
 438  , "editor"
 439  , "email"
 440  , "embed"
 441  , "eref"
 442  , "fax"
 443  , "feed"
 444  , "feedback"
 445  , "figure"
 446  , "filter"
 447  , "format"
 448  , "from"
 449  , "h"
 450  , "hi"
 451  , "html5"
 452  , "i"
 453  , "index"
 454  , "iref"
 455  , "keyword"
 456  , "li"
 457  , "link"
 458  , "name"
 459  , "note"
 460  , "ol"
 461  , "organization"
 462  , "para"
 463  , "postamble"
 464  , "preamble"
 465  , "q"
 466  , "ref"
 467  , "reference"
 468  , "references"
 469  , "region"
 470  , "rref"
 471  , "sc"
 472  , "section"
 473  , "serie"
 474  , "source"
 475  , "span"
 476  , "street"
 477  , "style"
 478  , "sub"
 479  , "sup"
 480  , "table"
 481  , "tbody"
 482  , "td"
 483  , "tel"
 484  , "tfoot"
 485  , "title"
 486  , "th"
 487  , "thead"
 488  , "toc"
 489  , "tof"
 490  , "tr"
 491  , "tt"
 492  , "u"
 493  , "ul"
 494  , "uri"
 495  , "version"
 496  , "video"
 497  , "workgroup"
 498  , "xml"
 499  , "zipcode"
 500  ]
 501
 502 -- * Attributes
 503
 504 -- | Convenient alias, forcing the types
 505 xmlAttrs :: Seq (Cell (XmlName,TL.Text)) -> XMLs
 506 xmlAttrs = (Tree0 . (uncurry XmlAttr <$>) <$>)
 507
 508 -- | Extract attributes
 509 partitionAttrs :: Roots -> (Seq (Cell (XmlName, TL.Text)), Roots)
 510 partitionAttrs ts = (attrs,cs)
 511         where
 512         (as,cs) = (`Seq.partition` ts) $ \case
 513          Tree (unCell -> NodeHeader (HeaderEqual n _wh)) _cs -> not $ TL.null n
 514          _ -> False
 515         attrs = attr <$> as
 516         attr = \case
 517          Tree (Cell bp ep (NodeHeader (HeaderEqual n _wh))) a ->
 518                 Cell bp ep (xmlLocalName n, v)
 519                 where
 520                 v = Plain.text (Plain.setStart a def{Plain.state_escape = False}) a
 521          _ -> undefined
 522
 523 getAttrId :: Root -> TL.Text
 524 getAttrId = Plain.document . Seq.singleton
 525
 526 setAttr ::
 527  Cell (XmlName, TL.Text) ->
 528  Seq (Cell (XmlName, TL.Text)) ->
 529  Seq (Cell (XmlName, TL.Text))
 530 setAttr a@(unCell -> (k, _v)) as =
 531         case Seq.findIndexL (\(unCell -> (n,_v)) -> n == k) as of
 532          Just idx -> Seq.update idx a as
 533          Nothing -> a <| as
 534
 535 defaultAttr ::
 536  Seq (Cell (XmlName, TL.Text)) ->
 537  Cell (XmlName, TL.Text) ->
 538  Seq (Cell (XmlName, TL.Text))
 539 defaultAttr as a@(unCell -> (k, _v)) =
 540         case Seq.findIndexL (\(unCell -> (n,_v)) -> n == k) as of
 541          Just _idx -> as
 542          Nothing -> a <| as
 543
 544 -- * Text
 545
 546 -- | Unify two 'XMLs', merging border 'XmlText's if any.
 547 unionXml :: XMLs -> XMLs -> XMLs
 548 unionXml x y =
 549         case (Seq.viewr x, Seq.viewl y) of
 550          (xs :> x0, y0 :< ys) ->
 551                 case (x0,y0) of
 552                  (  Tree0 (Cell bx ex (XmlText tx))
 553                   , Tree0 (Cell by ey (XmlText ty)) ) ->
 554                         xs `unionXml`
 555                         Seq.singleton (Tree0 $ (XmlText <$>) $ Cell bx ex tx <> Cell by ey ty) `unionXml`
 556                         ys
 557                  _ -> x <> y
 558          (Seq.EmptyR, _) -> y
 559          (_, Seq.EmptyL) -> x
 560
 561 unionsXml :: Foldable f => f XMLs -> XMLs
 562 unionsXml = foldl' unionXml mempty