Width.hs

-- | Terminal display width for characters and strings.
module Ansiform.Internal.Width (charWidth, textWidth) where

--------------------------------------------------------------------------------
import           Data.Char (GeneralCategory (..), generalCategory, ord)


-- Display Width
--------------------------------------------------------------------------------

-- | Number of terminal columns occupied by a character.
--   Combining marks and other zero width characters return zero,
--   East Asian wide characters and emoji return two.
charWidth :: Char -> Int
charWidth c
    | isZeroWidth c = 0
    | isWide c = 2
    | otherwise = 1

-- | Number of terminal columns occupied by a string.
textWidth :: String -> Int
textWidth = sum . fmap charWidth


-- Support Functions
--------------------------------------------------------------------------------

-- Checks for characters that do not advance the cursor.
isZeroWidth :: Char -> Bool
isZeroWidth c = case generalCategory c of
    NonSpacingMark -> True
    EnclosingMark -> True
    Format -> True
    _ -> inRange 0x1160 0x11FF c       -- Hangul Jamo vowels and finals.
      || inRange 0x200B 0x200B c      -- Zero width space.
      || inRange 0x1F3FB 0x1F3FF c    -- Emoji skin tone modifiers.

-- Checks for characters that occupy two columns.
isWide :: Char -> Bool
isWide c = any (\(lo,hi) -> inRange lo hi c) wideRanges

inRange :: Int -> Int -> Char -> Bool
inRange lo hi c = lo <= ord c && ord c <= hi

-- East Asian wide and fullwidth ranges, plus the main emoji blocks.
wideRanges :: [(Int, Int)]
wideRanges =
    [ (0x1100, 0x115F)    -- Hangul Jamo initials.
    , (0x2329, 0x232A)    -- Angle brackets.
    , (0x2E80, 0x303E)    -- CJK radicals, Kangxi and CJK symbols.
    , (0x3041, 0x33FF)    -- Kana, Bopomofo, Jamo and compatibility.
    , (0x3400, 0x4DBF)    -- CJK extension A.
    , (0x4E00, 0x9FFF)    -- CJK unified ideographs.
    , (0xA000, 0xA4CF)    -- Yi syllables.
    , (0xA960, 0xA97F)    -- Hangul Jamo extended A.
    , (0xAC00, 0xD7A3)    -- Hangul syllables.
    , (0xF900, 0xFAFF)    -- CJK compatibility ideographs.
    , (0xFE10, 0xFE19)    -- Vertical forms.
    , (0xFE30, 0xFE6F)    -- CJK compatibility forms.
    , (0xFF00, 0xFF60)    -- Fullwidth forms.
    , (0xFFE0, 0xFFE6)    -- Fullwidth signs.
    , (0x1F004, 0x1F004)  -- Mahjong red dragon.
    , (0x1F0CF, 0x1F0CF)  -- Playing card joker.
    , (0x1F18E, 0x1F18E)  -- Squared AB.
    , (0x1F191, 0x1F19A)  -- Squared symbols.
    , (0x1F200, 0x1F2FF)  -- Enclosed ideographic supplement.
    , (0x1F300, 0x1F64F)  -- Pictographs and emoticons.
    , (0x1F680, 0x1F6FF)  -- Transport and map symbols.
    , (0x1F780, 0x1F7FF)  -- Geometric shapes extended.
    , (0x1F900, 0x1F9FF)  -- Supplemental pictographs.
    , (0x1FA70, 0x1FAFF)  -- Pictographs extended A.
    , (0x20000, 0x2FFFD)  -- CJK extensions B to F.
    , (0x30000, 0x3FFFD)  -- CJK extensions G and later.
    ]