commonmark-extensions-0.2.5.2: src/Commonmark/Extensions/PipeTable.hs
{-# LANGUAGE IncoherentInstances #-}
{-# LANGUAGE FlexibleInstances #-}
{-# LANGUAGE FlexibleContexts #-}
{-# LANGUAGE MultiParamTypeClasses #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE DeriveDataTypeable #-}
module Commonmark.Extensions.PipeTable
( HasPipeTable(..)
, ColAlignment(..)
, pipeTableSpec
)
where
import Control.Monad (guard, void, mzero)
import Control.Monad.Trans.Class (lift)
import Commonmark.Syntax
import Commonmark.Types
import Commonmark.Tokens
import Commonmark.TokParsers
import Commonmark.Blocks
import Commonmark.SourceMap
import Commonmark.Html
import Text.Parsec
import Data.Dynamic
import Data.Tree
import Data.Data
data ColAlignment = LeftAlignedCol
| CenterAlignedCol
| RightAlignedCol
| DefaultAlignedCol
deriving (Show, Eq, Data, Typeable)
data PipeTableData = PipeTableData
{ pipeTableAlignments :: [ColAlignment]
, pipeTableColCount :: !Int
, pipeTableRowCount :: !Int
, pipeTableCellCount :: !Int
, pipeTableHeaders :: [[Tok]]
, pipeTableRows :: [[[Tok]]] -- in reverse order
} deriving (Show, Eq, Data, Typeable)
class HasPipeTable il bl where
pipeTable :: [ColAlignment] -> [il] -> [[il]] -> bl
instance HasPipeTable (Html a) (Html a) where
pipeTable aligns headerCells rows =
htmlBlock "table" $ Just $ htmlRaw "\n" <>
(if null headerCells
then mempty
else htmlBlock "thead" $ Just $ htmlRaw "\n" <>
toRow "th" aligns headerCells) <>
(if null rows
then mempty
else htmlBlock "tbody" $ Just $ htmlRaw "\n" <>
mconcat (map (toRow "td" aligns) rows))
where
alignToAttr LeftAlignedCol =
addAttribute ("style","text-align: left;")
alignToAttr CenterAlignedCol =
addAttribute ("style","text-align: center;")
alignToAttr RightAlignedCol =
addAttribute ("style","text-align: right;")
alignToAttr DefaultAlignedCol = id
toRow constructor aligns' cells =
htmlBlock "tr" $ Just $ htmlRaw "\n" <>
mconcat (zipWith (toCell constructor) aligns' cells)
toCell constructor align cell =
(alignToAttr align $ htmlInline constructor $ Just cell)
<> htmlRaw "\n"
instance (HasPipeTable i b, Monoid b)
=> HasPipeTable (WithSourceMap i) (WithSourceMap b) where
pipeTable aligns headerCells rows = do
(pipeTable aligns <$> sequence headerCells <*> mapM sequence rows)
<* addName "pipeTable"
pCells :: Monad m => ParsecT [Tok] s m [[Tok]]
pCells = try $ do
hasPipe <- option False $ True <$ symbol '|'
pipedCells <- many (try $ pCell <* symbol '|')
skipMany $ satisfyTok (hasType Spaces)
unpipedCell <- option [] $ (:[]) <$> pCell
let cells = pipedCells ++ unpipedCell
guard $ not (null cells)
guard $ hasPipe || not (null pipedCells) -- need at least one |
lookAhead blankLine
return $! cells
pCell :: Monad m => ParsecT [Tok] s m [Tok]
pCell = mconcat <$> many1
( try
(do symbol '\\'
tok <- symbol '|'
return $! [tok])
<|> (do tok <- (satisfyTok $ \t -> not (hasType (Symbol '|') t ||
hasType LineEnd t))
return $! [tok])
) <|> ([] <$ lookAhead (symbol '|'))
pDividers :: Monad m => ParsecT [Tok] s m [ColAlignment]
pDividers = try $ do
hasPipe <- option False $ True <$ symbol '|'
pipedAligns <- many (try $ pDivider <* symbol '|')
skipMany $ satisfyTok (hasType Spaces)
unpipedAlign <- option [] $ (:[]) <$> pDivider
let aligns = pipedAligns ++ unpipedAlign
guard $ not (null aligns)
guard $ hasPipe || not (null pipedAligns) -- need at least one |
lookAhead blankLine
return $! aligns
pDivider :: Monad m => ParsecT [Tok] s m ColAlignment
pDivider = try $ do
skipMany $ satisfyTok (hasType Spaces)
align <- choice
[ CenterAlignedCol <$
try (symbol ':' >> many1 (symbol '-') >> symbol ':')
, LeftAlignedCol <$
try (symbol ':' >> many1 (symbol '-'))
, RightAlignedCol <$
try (many1 (symbol '-') >> symbol ':')
, DefaultAlignedCol <$
many1 (symbol '-')
]
skipMany $ satisfyTok (hasType Spaces)
return $! align
-- | Syntax for pipe tables. Note that this should generally be
-- placed AFTER the syntax spec for lists, headings, and other block-level
-- constructs, to avoid bad results when non-table lines contain pipe
-- characters: use @defaultSyntaxSpec <> pipeTableSpec@ rather
-- than @pipeTableSpec <> defaultSyntaxSpec@.
pipeTableSpec :: (Monad m, IsBlock il bl, IsInline il, HasPipeTable il bl)
=> SyntaxSpec m il bl
pipeTableSpec = mempty
{ syntaxBlockSpecs = [pipeTableBlockSpec]
}
getAutoCompletedCellCount :: PipeTableData -> Int
getAutoCompletedCellCount tabledata =
(numrows * numcols) - numcells
where
numrows = pipeTableRowCount tabledata
numcols = pipeTableColCount tabledata
numcells = pipeTableCellCount tabledata
-- This parser is structured as a system that parses the *second* line first,
-- then parses the first line. That is, if it detects a delimiter row as the
-- second line of a paragraph, it converts the paragraph into a table. This seems
-- counterintuitive, but it works better than trying to convert a table into
-- a paragraph, since it might need to be something else.
--
-- See GH-52 and GH-95
pipeTableBlockSpec :: (Monad m, IsBlock il bl, IsInline il,
HasPipeTable il bl)
=> BlockSpec m il bl
pipeTableBlockSpec = BlockSpec
{ blockType = "PipeTable" -- :: Text
, blockStart = try $ do -- :: BlockParser m il bl ()
(cur:rest) <- nodeStack <$> getState
guard $ blockParagraph (bspec cur)
nonindentSpaces
pos <- getPosition
aligns <- pDividers
skipWhile (hasType Spaces)
lookAhead (eof <|> void lineEnd)
st <- getState
let headerLine =
case blockLines $ rootLabel cur of
[onlyLine] -> onlyLine
_ -> []
cellsR <- lift $ runParserT pCells st "" headerLine
case cellsR of
Right cells ->
if length cells /= length aligns
then mzero -- parse fail: not a table
else do
updateState $ \st' -> st'{ nodeStack = rest }
let tabledata = PipeTableData
{ pipeTableAlignments = aligns
, pipeTableColCount = length cells
, pipeTableRowCount = 0
, pipeTableCellCount = 0
, pipeTableHeaders = cells
, pipeTableRows = []
}
addNodeToStack $
Node (defBlockData pipeTableBlockSpec){
blockStartPos = blockStartPos (rootLabel cur) ++ [pos]
, blockData = toDyn tabledata
, blockAttributes = blockAttributes (rootLabel cur)
} []
_ ->
mzero -- parse fail: not a table
return BlockStartMatch
, blockCanContain = \_ -> False -- :: BlockSpec m il bl -> Bool
, blockContainsLines = False -- :: Bool
, blockParagraph = False -- :: Bool
, blockContinue = \(Node ndata children) -> try $ do
nonindentSpaces
notFollowedBy blankLine
let tabledata = fromDyn
(blockData ndata)
PipeTableData{ pipeTableAlignments = []
, pipeTableColCount = 0
, pipeTableRowCount = 0
, pipeTableCellCount = 0
, pipeTableHeaders = []
, pipeTableRows = [] }
pos <- getPosition
cells <- pCells
let cells' = take (pipeTableColCount tabledata) cells
let tabledata' =
tabledata{ pipeTableRows = cells' : pipeTableRows tabledata
, pipeTableRowCount = 1 + pipeTableRowCount tabledata
, pipeTableCellCount = length cells' + pipeTableCellCount tabledata
}
-- Protect against quadratic output size explosion.
--
-- Because the table extension fills in missing table cells,
-- you can, theoretically, force the output to grows as the
-- square of the input by adding one column and one row.
-- This is a side-effect of the extension as specified in GFM,
-- and follows from the geometric definition of "squaring."
--
-- To prevent this, we track the number of Real Cells,
-- and if the number of autocompleted cells exceeds 200,000,
-- stop parsing.
guard $ getAutoCompletedCellCount tabledata <= 200000
return $! (pos, Node ndata{ blockData =
toDyn tabledata' } children)
, blockConstructor = \(Node ndata _) -> do
let tabledata = fromDyn
(blockData ndata)
PipeTableData{ pipeTableAlignments = []
, pipeTableColCount = 0
, pipeTableRowCount = 0
, pipeTableCellCount = 0
, pipeTableHeaders = []
, pipeTableRows = [] }
let aligns = pipeTableAlignments tabledata
headers <- mapM runInlineParser (pipeTableHeaders tabledata)
let numcols = pipeTableColCount tabledata
rows <- mapM (mapM runInlineParser . take numcols . (++ (repeat [])))
(reverse $ pipeTableRows tabledata)
return $! (pipeTable aligns headers rows)
, blockFinalize = \(Node ndata children) parent ->
defaultFinalizer (Node ndata children) parent
}