module Data.Text.ParagraphLayout.Internal.Rich.Paragraph
( Paragraph (..)
, constructParagraph
, paragraphSpanBounds
, paragraphSpanTexts
, paragraphText
)
where
import Data.List.NonEmpty (NonEmpty)
import qualified Data.List.NonEmpty as NonEmpty
import Data.Text.Array (Array)
import Data.Text.Internal (Text (Text))
import Data.Text.ParagraphLayout.Internal.ParagraphOptions
import Data.Text.ParagraphLayout.Internal.Tree
-- | Text to be laid out as a single paragraph.
--
-- May be divided into a hierarchy of boxes and spans.
--
-- The input text must be encoded as UTF-8 in a contiguous byte array.
--
-- You may need to use "Data.Text.Internal" in order to determine the byte
-- array and the necessary offsets to construct the paragraph without copying
-- data.
--
-- Alternatively, you can use `constructParagraph` with piecewise text.
data Paragraph d = Paragraph
Array
-- ^ A byte array containing the whole text to be laid out, in UTF-8.
--
-- This array will be passed to "Data.Text.Glyphize", which passes it to
-- [@hb_buffer_add_utf8()@]
-- (https://harfbuzz.github.io/harfbuzz-hb-buffer.html#hb-buffer-add-utf8).
--
-- In the output, `Data.Text.Glyphize.cluster` will be a byte offset of
-- the corresponding input character from this array.
Int
-- ^ Byte offset of the first text node from the start of the byte array.
-- Any characters preceding this offset will not be shaped, but may still
-- be used to influence the shape of neighbouring characters.
(RootNode Int d)
-- ^ Parts of the text to be laid out, represented as a tree.
-- The in-order walk of this tree corresponds to the logical order
-- of the text.
ParagraphOptions
-- ^ Options applying to the paragraph as a whole.
-- | Construct a `Paragraph` from a tree containing individual pieces of text,
-- plus an optional prefix and suffix, all encoded as UTF-8.
--
-- This internally concatenates all input text into one contiguous byte array.
--
-- Alternatively, you can construct a `Paragraph` directly to avoid copying
-- data.
constructParagraph
:: Text
-- ^ Optional prefix. Will not be laid out but may affect the output.
-> RootNode Text d
-- ^ Parts of the text to be laid out, represented as a tree.
-> Text
-- ^ Optional suffix. Will not be laid out but may affect the output.
-> ParagraphOptions
-- ^ Options to apply to the paragraph as a whole.
-> Paragraph d
-- ^ Constructed paragraph that can be passed to
-- `Data.Text.ParagraphLayout.Rich.layoutRich`.
constructParagraph prefix root suffix = Paragraph arr afterPrefix root'
where
(Text arr beforePrefix _) = txt
afterPrefix = beforePrefix + prefixLen
(txt, prefixLen, root') = glue prefix root suffix
-- | Calculate the offsets into the `Paragraph`'s underlying `Data.Text.Array`
-- where each text node starts and ends, in ascending order. The resulting list
-- will be one larger than the number of text nodes in the input.
--
-- You can use this function to verify that Balkón will slice the input text
-- correctly.
paragraphSpanBounds :: Paragraph d -> NonEmpty Int
paragraphSpanBounds (Paragraph _ initialOffset root _) =
-- TODO: Consider adding checks for array bounds.
NonEmpty.scanl (+) initialOffset $ map len $ flatten root
where
len (TextLeaf _ l _ _) = l
-- | Turn each text node from the input `Paragraph` into a `Text`.
--
-- You can use this function to verify that Balkón will slice the input text
-- correctly.
paragraphSpanTexts :: Paragraph d -> [Text]
paragraphSpanTexts p@(Paragraph arr _ _ _) = zipWith toText sStarts sEnds
where
toText start end = Text arr start (end - start)
sStarts = NonEmpty.init sBounds
sEnds = NonEmpty.tail sBounds
sBounds = paragraphSpanBounds p
-- | Turn all text nodes from the input `Paragraph` into one combined `Text`.
--
-- You can use this function to verify that Balkón will slice the input text
-- correctly.
paragraphText :: Paragraph d -> Text
paragraphText p@(Paragraph arr _ _ _) = Text arr start (end - start)
where
start = NonEmpty.head sBounds
end = NonEmpty.last sBounds
sBounds = paragraphSpanBounds p