gemmula-1.1.1: src/Text/Gemini.hs
{-# LANGUAGE OverloadedStrings #-}
-- |
-- Module : Text.Gemini
-- Copyright : (c) Sena, 2024
-- License : AGPL-3.0-or-later
--
-- Maintainer : Sena <jn-sena@proton.me>
-- Stability : stable
-- Portability : portable
--
-- A tiny gemtext (unofficially @text/gemini@) parser
--
-- Parses gemtext documents from and back to 'Text'.
--
-- See the Gemini hypertext format specification at
-- <https://geminiprotocol.net/docs/gemtext-specification.gmi>.
module Text.Gemini
( -- * Gemtext types
GemDocument
, GemItem (..)
-- * Decoding from text
, decode
, decodeLine
-- * Encoding into text
, encode
, encodeItem
) where
import Data.Bool (bool)
import Data.Char (isSpace)
import Data.Maybe (fromMaybe)
import Data.Text (Text)
import qualified Data.Text as T
-- | A gemtext document, in the form of an ordered list of 'GemItem's
type GemDocument = [GemItem]
-- | A gemtext item
data GemItem
= -- | A regular gemtext line -- @'GemText' \<Text>@
GemText !Text
| -- | A gemtext link -- @'GemLink' \<Link> \[Optional Description]@
GemLink !Text !(Maybe Text)
| -- | A gemtext heading of 3 levels max -- @'GemHeading' \<Level> \<Text>@
GemHeading !Int !Text
| -- | An unordered gemtext list -- @'GemList' \<Lines>@
GemList ![Text]
| -- | A gemtext quote -- @'GemQuote' \<Text>@
GemQuote !Text
| -- | A preformatted gemtext block -- @'GemPre' \<Lines> [Optional Alt Text]@
GemPre ![Text] !(Maybe Text)
deriving (Show, Eq)
-- | Parse a gemtext block as 'GemDocument'.
--
-- The text should be supplied as a 'Text' which uses LF line breaks.
decode :: Text -> GemDocument
decode = parse [] . T.lines
where
-- Default line-by-line recursive parser with 'GemDocument' carried.
-- If a list or preformatting is detected, will continue with
-- @parseList@ and @parsePre@ respectively. Otherwise, will use
-- 'decodeLine' on the current line only.
parse :: GemDocument -> [Text] -> GemDocument
parse doc (l : ls)
| l `hasValueOf` "* " = parseList doc [value 1 l] ls
| "```" `T.isPrefixOf` l = parsePre doc [] (optional $ value 3 l) ls
| otherwise = parse (doc <> [decodeLine l]) ls
parse doc [] = doc
-- List recursive parser. Continues with @parse@ when the line
-- is no longer a list item, and adds the 'GemList' to the 'GemDocument'.
parseList :: GemDocument -> [Text] -> [Text] -> GemDocument
parseList doc glist (l : ls)
| l `hasValueOf` "* " = parseList doc (glist <> [value 1 l]) ls
| otherwise = parse (doc <> [GemList glist]) (l : ls)
parseList doc glist [] = doc <> [GemList glist]
-- Preformatted recursive parser. Continues with @parse@ when the
-- closing delimiter is reached, and adds the 'GemPre' to the 'GemDocument'.
parsePre :: GemDocument -> [Text] -> Maybe Text -> [Text] -> GemDocument
parsePre doc glines alt (l : ls)
| "```" `T.isPrefixOf` l = parse (doc <> [GemPre glines alt]) ls
| otherwise = parsePre doc (glines <> [l]) alt ls
parsePre doc glines alt [] = doc <> [GemPre glines alt]
-- | Parse a /single/ gemtext line as 'GemItem'.
--
-- Preformatted text blocks are regarded as 'GemText's as they are strictly multiline.
decodeLine :: Text -> GemItem
decodeLine line
| line `hasValueOf` "=>" =
let (link, desc) = T.break isSpace $ value 2 line
in GemLink link $ optional $ T.strip desc
| line `hasValueOf` "#" = parseHeading line
| line `hasValueOf` "* " = GemList [value 1 line]
| line `hasValueOf` ">" = GemQuote $ value 1 line
| otherwise = GemText $ T.stripEnd line
where
-- Parse a gemtext heading.
-- The max level of a heading is 3.
parseHeading :: Text -> GemItem
parseHeading l
| T.null $ T.strip text = GemText $ T.strip l
| otherwise = GemHeading (min (T.length pre) 3) $ T.strip text
where
(pre, text) = T.span (== '#') l
-- | Encode a parsed 'GemDocument' into a gemtext block.
--
-- The output 'Text' uses LF line breaks.
--
-- Valid prefixes are escaped before encoding. See the 'encodeItem' function below.
--
-- Empty lists are ignored if given.
encode :: GemDocument -> Text
encode = T.unlines . map encodeItem . filter (not . empty)
where
empty :: GemItem -> Bool
empty (GemList list) = null list
empty _ = False
-- | Encode a /single/ parsed 'GemItem' into gemtext.
--
-- /Beware/ that the output text does /not/ end with a newline.
--
-- The output 'Text' might be multiple lines, in which case it uses LF line breaks.
--
-- Valid prefixes are escaped before encoding.
--
-- The spaces in 'GemLink's are replaced with @%20@ to normalize them.
encodeItem :: GemItem -> Text
encodeItem (GemText line) = escapePrefixes line
encodeItem (GemLink link desc) = "=> " <> T.replace " " "%20" link <> maybe "" (" " <>) desc
encodeItem (GemHeading level text) = T.replicate (min level 3) "#" <> " " <> text
encodeItem (GemList list) = T.intercalate "\n" $ map ("* " <>) list
encodeItem (GemQuote text) = "> " <> text
encodeItem (GemPre text alt) = T.intercalate "\n" $ ["```" <> fromMaybe "" alt] <> text <> ["```"]
-- Escape the line prefixes by adding a whitespace at the beginning for 'GemText'.
escapePrefixes :: Text -> Text
escapePrefixes line = bool line (T.cons ' ' line) escape
where
escape = "```" `T.isPrefixOf` line
|| any (hasValueOf line) ["=>", "* ", ">"]
|| (line `hasValueOf` "#"
&& let (_, text) = T.span (== '#') line
in not (T.null $ T.strip text)
)
-- @True@ if the the line has prefix /and/ a following value.
hasValueOf :: Text -> Text -> Bool
hasValueOf line prefix =
prefix `T.isPrefixOf` line
&& not (T.null $ value (T.length prefix) line)
-- Get the value of a line.
-- Removes the prefix of given length and strips.
value :: Int -> Text -> Text
value prefix = T.strip . T.drop prefix
-- @Nothing@ if the text is empty.
optional :: Text -> Maybe Text
optional text = bool (Just text) Nothing $ T.null text