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'{inh_figure=False} 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                                 if inh_figure inh && n`List.notElem`elems || TL.null n
 287                                 then
 288                                         Seq.singleton $
 289                                         element "artwork" $
 290                                                 xmlAttrs (Seq.singleton $ Cell bn bn ("type", n)) <>
 291                                                 xmlify inh{inh_para=[]} ts
 292                                 else
 293                                         xmlify inh $
 294                                                 Tree (Cell bn en $ NodeHeader $ HeaderColon n wh) ts
 295                         --
 296                          HeaderGreat n _wh ->
 297                                 Seq.singleton $
 298                                 let (attrs,body) = partitionAttrs ts in
 299                                 element "quote" $
 300                                         xmlAttrs (attrs `defaultAttr` Cell bn bn ("type", n)) <>
 301                                         xmlify inh{inh_para=List.repeat elementPara} body
 302                         --
 303                          HeaderEqual n _wh ->
 304                                 Seq.singleton $
 305                                 Tree0 $ cell $ XmlAttr (xmlLocalName n) $
 306                                         Plain.text (Plain.setStart ts def{Plain.state_escape = False}) ts
 307                         --
 308                          HeaderDot n ->
 309                                 Seq.singleton $
 310                                 element "li" $
 311                                         xmlAttrs (Seq.singleton $ Cell bn bn{pos_column=pos_column bn + int (TL.length n)} ("name", n)) <>
 312                                         xmlify inh ts
 313                         --
 314                          HeaderDash -> Seq.singleton $ element "li" $ xmlify inh ts
 315                         --
 316                          HeaderDashDash ->
 317                                 Seq.singleton $ Tree0 $ cell $
 318                                         XmlComment $ Plain.document ts
 319                         --
 320                          HeaderBrackets ident ->
 321                                 let (attrs,body) = partitionAttrs ts in
 322                                 Seq.singleton $
 323                                 element "reference" $
 324                                         xmlAttrs (setAttr (Cell en en ("id",ident)) attrs) <>
 325                                         xmlify inh'{inh_para = elementTitle : elementTitle : List.repeat elementPara} body
 326                                 where
 327                                 inh' = inh{inh_figure = False}
 328                         --
 329                          HeaderDotSlash p ->
 330                                 Seq.singleton $
 331                                 element "include" $
 332                                         xmlAttrs [cell ("href",TL.pack $ FP.replaceExtension p "dtc")] <>
 333                                         xmlify inh ts
 334                 ----------------------
 335                  NodePair pair ->
 336                         case pair of
 337                          PairBracket | to <- Plain.document ts
 338                                      , TL.all (\c -> Char.isAlphaNum c || Char.isSymbol c) to ->
 339                                 Seq.singleton $
 340                                 element "rref" $
 341                                         xmlAttrs [cell ("to",to)]
 342                          PairStar      -> Seq.singleton $ element "b"    $ xmlify inh ts
 343                          PairSlash     -> Seq.singleton $ element "i"    $ xmlify inh ts
 344                          PairBackquote -> Seq.singleton $ element "code" $ xmlify inh ts
 345                          PairFrenchquote ->
 346                                 Seq.singleton $
 347                                 element "q" $
 348                                         xmlify inh ts
 349                                 {-
 350                                 case ts of
 351                                  (Seq.viewl -> Tree0 (Cell bl el (TokenPlain l)) :< ls) ->
 352                                         case Seq.viewr ls of
 353                                          m :> Tree0 (Cell br er (TokenPlain r)) ->
 354                                                 xmlify inh $
 355                                                         Tree0 (Cell bl el (TokenPlain (Text.dropWhile Char.isSpace l)))
 356                                                          <|(m|>Tree0 (Cell br er (TokenPlain (Text.dropWhileEnd Char.isSpace r))))
 357                                          _ ->
 358                                                 xmlify inh $
 359                                                         Tree0 (Cell bl el (TokenPlain (Text.dropAround Char.isSpace l))) <| ls
 360                                  (Seq.viewr -> rs :> Tree0 (Cell br er (TokenPlain r))) ->
 361                                         xmlify inh $
 362                                                 rs |> Tree0 (Cell br er (TokenPlain (Text.dropAround Char.isSpace r)))
 363                                  _ -> xmlify inh ts
 364                                 -}
 365                          PairHash ->
 366                                 Seq.singleton $
 367                                 element "ref" $
 368                                         xmlAttrs [cell ("to",Plain.document ts)]
 369                          PairElem name attrs ->
 370                                 Seq.singleton $
 371                                 element (xmlLocalName name) $
 372                                         xmlAttrs (Seq.fromList $ (\(_wh,ElemAttr{..}) ->
 373                                                 cell (xmlLocalName elemAttr_name,elemAttr_value)) <$> attrs) <>
 374                                         xmlify inh ts
 375                          _ ->
 376                                 Seq.singleton (Tree0 $ Cell bn bn' $ XmlText open) `unionXml`
 377                                 xmlify inh ts `unionXml`
 378                                 Seq.singleton (Tree0 $ Cell en' en $ XmlText close)
 379                                 where
 380                                 (open, close) = pairBorders pair ts
 381                                 bn' = bn{pos_column=pos_column bn + int (TL.length open)}
 382                                 en' = en{pos_column=pos_column bn - int (TL.length close)}
 383                 ----------------------
 384                  NodeText t -> Seq.singleton $ Tree0 $ cell $ XmlText t
 385                 ----------------------
 386                  NodeToken tok ->
 387                         case tok of
 388                          TokenEscape c -> Seq.singleton $ Tree0 $ cell $ XmlText $ TL.singleton c
 389                          TokenText t   -> Seq.singleton $ Tree0 $ cell $ XmlText t
 390                          TokenTag t    -> Seq.singleton $ element "ref"  $ xmlAttrs [cell ("to",t)]
 391                          TokenLink lnk -> Seq.singleton $ element "eref" $ xmlAttrs [cell ("to",lnk)]
 392                 ----------------------
 393                  NodeLower n as ->
 394                         Seq.singleton $
 395                         element "artwork" $
 396                                 xmlify inh ts
 397                 where
 398                 cell :: a -> Cell a
 399                 cell = Cell bn en
 400                 element :: XmlName -> XMLs -> XML
 401                 element n = Tree (cell $ XmlElem n)
 402 instance Xmlify (Seq (Cell (XmlName,TL.Text))) where
 403         xmlify _inh = xmlAttrs
 404
 405 -- * Elements
 406
 407 -- | Reserved elements' name
 408 elems :: Set TL.Text
 409 elems =
 410  [ "about"
 411  , "abstract"
 412  , "address"
 413  , "alias"
 414  , "annotation"
 415  , "area"
 416  , "artwork"
 417  , "aside"
 418  , "audio"
 419  , "author"
 420  , "authors"
 421  , "bcp14"
 422  , "br"
 423  , "call"
 424  , "city"
 425  , "code"
 426  , "comment"
 427  , "comments"
 428  , "country"
 429  , "date"
 430  , "dd"
 431  , "define"
 432  , "del"
 433  , "div"
 434  , "dl"
 435  , "document"
 436  , "dt"
 437  , "editor"
 438  , "email"
 439  , "embed"
 440  , "eref"
 441  , "fax"
 442  , "feed"
 443  , "feedback"
 444  , "figure"
 445  , "filter"
 446  , "format"
 447  , "from"
 448  , "h"
 449  , "hi"
 450  , "html5"
 451  , "i"
 452  , "index"
 453  , "iref"
 454  , "keyword"
 455  , "li"
 456  , "link"
 457  , "name"
 458  , "note"
 459  , "ol"
 460  , "organization"
 461  , "para"
 462  , "postamble"
 463  , "preamble"
 464  , "q"
 465  , "ref"
 466  , "reference"
 467  , "references"
 468  , "region"
 469  , "rref"
 470  , "sc"
 471  , "section"
 472  , "serie"
 473  , "source"
 474  , "span"
 475  , "street"
 476  , "style"
 477  , "sub"
 478  , "sup"
 479  , "table"
 480  , "tbody"
 481  , "td"
 482  , "tel"
 483  , "tfoot"
 484  , "title"
 485  , "th"
 486  , "thead"
 487  , "toc"
 488  , "tof"
 489  , "tr"
 490  , "tt"
 491  , "u"
 492  , "ul"
 493  , "uri"
 494  , "version"
 495  , "video"
 496  , "workgroup"
 497  , "xml"
 498  , "zipcode"
 499  ]
 500
 501 -- * Attributes
 502
 503 -- | Convenient alias, forcing the types
 504 xmlAttrs :: Seq (Cell (XmlName,TL.Text)) -> XMLs
 505 xmlAttrs = (Tree0 . (uncurry XmlAttr <$>) <$>)
 506
 507 -- | Extract attributes
 508 partitionAttrs :: Roots -> (Seq (Cell (XmlName, TL.Text)), Roots)
 509 partitionAttrs ts = (attrs,cs)
 510         where
 511         (as,cs) = (`Seq.partition` ts) $ \case
 512          Tree (unCell -> NodeHeader (HeaderEqual n _wh)) _cs -> not $ TL.null n
 513          _ -> False
 514         attrs = attr <$> as
 515         attr = \case
 516          Tree (Cell bp ep (NodeHeader (HeaderEqual n _wh))) a ->
 517                 Cell bp ep (xmlLocalName n, v)
 518                 where
 519                 v = Plain.text (Plain.setStart a def{Plain.state_escape = False}) a
 520          _ -> undefined
 521
 522 getAttrId :: Root -> TL.Text
 523 getAttrId = Plain.document . Seq.singleton
 524
 525 setAttr ::
 526  Cell (XmlName, TL.Text) ->
 527  Seq (Cell (XmlName, TL.Text)) ->
 528  Seq (Cell (XmlName, TL.Text))
 529 setAttr a@(unCell -> (k, _v)) as =
 530         case Seq.findIndexL (\(unCell -> (n,_v)) -> n == k) as of
 531          Just idx -> Seq.update idx a as
 532          Nothing -> a <| as
 533
 534 defaultAttr ::
 535  Seq (Cell (XmlName, TL.Text)) ->
 536  Cell (XmlName, TL.Text) ->
 537  Seq (Cell (XmlName, TL.Text))
 538 defaultAttr as a@(unCell -> (k, _v)) =
 539         case Seq.findIndexL (\(unCell -> (n,_v)) -> n == k) as of
 540          Just _idx -> as
 541          Nothing -> a <| as
 542
 543 -- * Text
 544
 545 -- | Unify two 'XMLs', merging border 'XmlText's if any.
 546 unionXml :: XMLs -> XMLs -> XMLs
 547 unionXml x y =
 548         case (Seq.viewr x, Seq.viewl y) of
 549          (xs :> x0, y0 :< ys) ->
 550                 case (x0,y0) of
 551                  (  Tree0 (Cell bx ex (XmlText tx))
 552                   , Tree0 (Cell by ey (XmlText ty)) ) ->
 553                         xs `unionXml`
 554                         Seq.singleton (Tree0 $ (XmlText <$>) $ Cell bx ex tx <> Cell by ey ty) `unionXml`
 555                         ys
 556                  _ -> x <> y
 557          (Seq.EmptyR, _) -> y
 558          (_, Seq.EmptyL) -> x
 559
 560 unionsXml :: Foldable f => f XMLs -> XMLs
 561 unionsXml = foldl' unionXml mempty