Split package into library module 'src' and executable 'app'
parent
626eb71097
commit
d63cd98288
@ -1,121 +1,8 @@
|
|||||||
module Main where
|
module Main where
|
||||||
|
|
||||||
import Text.ParserCombinators.ReadP
|
import MdToHTML
|
||||||
import Control.Monad
|
|
||||||
import Control.Applicative
|
|
||||||
import Text.Printf
|
|
||||||
import Debug.Trace
|
|
||||||
import Data.List
|
|
||||||
|
|
||||||
type HeaderLevel = Int
|
|
||||||
|
|
||||||
newtype URL = URL {getUrl :: String}
|
|
||||||
newtype ImgPath = ImgPath {getPath :: String}
|
|
||||||
|
|
||||||
parseMany :: ReadP a -> ReadP [a]
|
|
||||||
parseMany = Text.ParserCombinators.ReadP.many
|
|
||||||
|
|
||||||
data MdToken = Header HeaderLevel MdToken
|
|
||||||
| Para [MdToken]
|
|
||||||
| Linebreak
|
|
||||||
| HorizontalRule
|
|
||||||
| Blockquote MdToken
|
|
||||||
| UnordList [MdToken]
|
|
||||||
| OrdList [MdToken]
|
|
||||||
| Code String
|
|
||||||
| Codeblock String
|
|
||||||
| Link MdToken URL
|
|
||||||
| Image MdToken ImgPath
|
|
||||||
| Bold MdToken
|
|
||||||
| Italic MdToken
|
|
||||||
| Strikethrough MdToken
|
|
||||||
| Unit String
|
|
||||||
|
|
||||||
-- Deriving Show for MdToken
|
|
||||||
instance Show MdToken where
|
|
||||||
show (Header level token) = "<h" ++ show level ++ ">" ++ show token ++ "</h" ++ show level ++ ">"
|
|
||||||
show (Para tokens) = concat(map show tokens)
|
|
||||||
show Linebreak = "\n"
|
|
||||||
show HorizontalRule = "---------"
|
|
||||||
show (Blockquote token) = "BLOCK" ++ show token
|
|
||||||
show (UnordList tokens) = "UNORD" ++ concat(map show tokens)
|
|
||||||
show (OrdList tokens) = "ORD" ++ concat(map show tokens)
|
|
||||||
show (Code code) = show code
|
|
||||||
show (Codeblock code) = show code
|
|
||||||
show (Link txt url) = "<a href=" ++ (getUrl url) ++ ">" ++ show txt ++ "</a>"
|
|
||||||
show (Image txt path) = "<img src=" ++ (getPath path) ++ ">" ++ show txt ++ "</img>"
|
|
||||||
show (Bold token) = "<b>" ++ show token ++ "</b>"
|
|
||||||
show (Italic token) = "<i>" ++ show token ++ "</i>"
|
|
||||||
show (Strikethrough token) = "<s>" ++ show token ++ "</s>"
|
|
||||||
show (Unit unit) = printf "%s" unit
|
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
-- ---------------
|
|
||||||
-- Helpers
|
|
||||||
mustBeHash :: ReadP Char
|
|
||||||
mustBeHash = satisfy (\x -> x == '#')
|
|
||||||
|
|
||||||
leftmostLongest :: (Foldable t) => [(a, t b)] -> (a, t b)
|
|
||||||
leftmostLongest xs =
|
|
||||||
let lastElem = (last xs)
|
|
||||||
filteredLst = (filter (\val -> (length $ snd val) == (length $ snd lastElem)) xs)
|
|
||||||
in head filteredLst
|
|
||||||
|
|
||||||
leftmostLongestParse :: ReadP a -> String -> (a, String)
|
|
||||||
leftmostLongestParse parser input = leftmostLongest $ readP_to_S parser input
|
|
||||||
-- ---------------
|
|
||||||
|
|
||||||
parseHeader :: ReadP MdToken
|
|
||||||
parseHeader = do
|
|
||||||
headers <- many1 mustBeHash
|
|
||||||
when ((length headers) > 6)
|
|
||||||
pfail
|
|
||||||
_ <- string " "
|
|
||||||
text <- munch1 (\x -> x /= '\n') -- Parse until EOL
|
|
||||||
-- traceM text
|
|
||||||
let parsedText = fst $ leftmostLongestParse parseLine text
|
|
||||||
return (Header (length headers) parsedText)
|
|
||||||
|
|
||||||
parseBold :: ReadP MdToken
|
|
||||||
parseBold = do
|
|
||||||
text <- choice[
|
|
||||||
(between (string "__") (string "__") (munch1 (/= '_'))),
|
|
||||||
(between (string "**") (string "**") (munch1 (/= '*')))
|
|
||||||
]
|
|
||||||
-- text <- munch1 (\x -> x /= '_' && x /= '*') -- Parse until first asterisk/underscore
|
|
||||||
-- traceM text
|
|
||||||
-- _ <- char '_' <|> char '*' -- Throw away the second asterisk/underscore
|
|
||||||
let parsedText = fst $ leftmostLongestParse parseLine text
|
|
||||||
return (Bold parsedText)
|
|
||||||
|
|
||||||
parseItalic :: ReadP MdToken
|
|
||||||
parseItalic = do
|
|
||||||
text <- choice[
|
|
||||||
(between (string "_") (string "_") (munch1 (/= '_'))),
|
|
||||||
(between (string "*") (string "*") (munch1 (/= '*')))
|
|
||||||
]
|
|
||||||
let parsedText = fst $ leftmostLongestParse parseLine text
|
|
||||||
return (Italic parsedText)
|
|
||||||
|
|
||||||
parseString :: ReadP MdToken
|
|
||||||
parseString = do
|
|
||||||
firstChar <- get -- Must parse at least one character here
|
|
||||||
text <- munch (\x -> not (elem x "#*_[\n"))
|
|
||||||
return (Unit (firstChar:text))
|
|
||||||
--return (Unit text)
|
|
||||||
|
|
||||||
parseToken :: ReadP MdToken
|
|
||||||
parseToken = choice [parseHeader, parseBold, parseItalic, parseString]
|
|
||||||
|
|
||||||
parseLine :: ReadP MdToken
|
|
||||||
parseLine = do
|
|
||||||
parsed <- parseMany parseToken
|
|
||||||
-- traceM $ show parsed
|
|
||||||
return (Para parsed)
|
|
||||||
|
|
||||||
|
|
||||||
main :: IO ()
|
main :: IO ()
|
||||||
main = do
|
main = do
|
||||||
let res = leftmostLongestParse parseLine "## Hello __world_*"
|
let res = leftmostLongestParse parseLine "## Hello ___world___"
|
||||||
putStrLn (show res)
|
putStrLn (show res)
|
||||||
|
|||||||
@ -0,0 +1,149 @@
|
|||||||
|
module MdToHTML where
|
||||||
|
|
||||||
|
import Text.ParserCombinators.ReadP
|
||||||
|
import Control.Monad
|
||||||
|
import Control.Applicative
|
||||||
|
import Text.Printf
|
||||||
|
import Debug.Trace
|
||||||
|
import Data.List
|
||||||
|
|
||||||
|
type HeaderLevel = Int
|
||||||
|
|
||||||
|
newtype URL = URL {getUrl :: String}
|
||||||
|
newtype ImgPath = ImgPath {getPath :: String}
|
||||||
|
|
||||||
|
parseMany :: ReadP a -> ReadP [a]
|
||||||
|
parseMany = Text.ParserCombinators.ReadP.many
|
||||||
|
|
||||||
|
data MdToken = Header HeaderLevel MdToken
|
||||||
|
| Para MdToken
|
||||||
|
| Line [MdToken]
|
||||||
|
| Linebreak
|
||||||
|
| HorizontalRule
|
||||||
|
| Blockquote MdToken
|
||||||
|
| UnordList [MdToken]
|
||||||
|
| OrdList [MdToken]
|
||||||
|
| Code String
|
||||||
|
| Codeblock String
|
||||||
|
| Link MdToken URL
|
||||||
|
| Image MdToken ImgPath
|
||||||
|
| Bold MdToken
|
||||||
|
| Italic MdToken
|
||||||
|
| Strikethrough MdToken
|
||||||
|
| Unit String
|
||||||
|
|
||||||
|
-- Deriving Show for MdToken
|
||||||
|
instance Show MdToken where
|
||||||
|
show (Header level token) = "<h" ++ show level ++ ">" ++ show token ++ "</h" ++ show level ++ ">"
|
||||||
|
show (Para token) = "<p>" ++ show token ++ "</p>"
|
||||||
|
show (Line tokens) = concat(map show tokens)
|
||||||
|
show Linebreak = "<br>"
|
||||||
|
show HorizontalRule = "---------"
|
||||||
|
show (Blockquote token) = "BLOCK" ++ show token
|
||||||
|
show (UnordList tokens) = "UNORD" ++ concat(map show tokens)
|
||||||
|
show (OrdList tokens) = "ORD" ++ concat(map show tokens)
|
||||||
|
show (Code code) = show code
|
||||||
|
show (Codeblock code) = show code
|
||||||
|
show (Link txt url) = "<a href=" ++ (getUrl url) ++ ">" ++ show txt ++ "</a>"
|
||||||
|
show (Image txt imgPath) = "<img src=" ++ (getPath imgPath) ++ ">" ++ show txt ++ "</img>"
|
||||||
|
show (Bold token) = "<b>" ++ show token ++ "</b>"
|
||||||
|
show (Italic token) = "<i>" ++ show token ++ "</i>"
|
||||||
|
show (Strikethrough token) = "<s>" ++ show token ++ "</s>"
|
||||||
|
show (Unit unit) = printf "%s" unit
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
|
-- ---------------
|
||||||
|
-- Helpers
|
||||||
|
mustBeHash :: ReadP Char
|
||||||
|
mustBeHash = satisfy (\x -> x == '#')
|
||||||
|
|
||||||
|
leftmostLongest :: (Foldable t) => [(a, t b)] -> (a, t b)
|
||||||
|
leftmostLongest xs =
|
||||||
|
let lastElem = (last xs)
|
||||||
|
filteredLst = (filter (\val -> (length $ snd val) == (length $ snd lastElem)) xs)
|
||||||
|
in head filteredLst
|
||||||
|
|
||||||
|
-- Get the first parse returned by readP_to_S that consumed the most input
|
||||||
|
leftmostLongestParse :: ReadP a -> String -> (a, String)
|
||||||
|
leftmostLongestParse parser input = leftmostLongest $ readP_to_S parser input
|
||||||
|
|
||||||
|
-- Parse if the string that's left matches the string comparator function
|
||||||
|
lookaheadParse :: (String -> Bool) -> ReadP Char
|
||||||
|
lookaheadParse stringCmp = do
|
||||||
|
lookahead <- look
|
||||||
|
case stringCmp lookahead of
|
||||||
|
True -> get
|
||||||
|
False -> pfail
|
||||||
|
|
||||||
|
lineToList :: MdToken -> [MdToken]
|
||||||
|
lineToList (Line tokens) = tokens
|
||||||
|
-- ---------------
|
||||||
|
|
||||||
|
-- Parse a markdown header, denoted by 1-6 #'s followed by some text, followed by EOL.
|
||||||
|
parseHeader :: ReadP MdToken
|
||||||
|
parseHeader = do
|
||||||
|
headers <- many1 mustBeHash
|
||||||
|
when ((length headers) > 6)
|
||||||
|
pfail
|
||||||
|
_ <- string " "
|
||||||
|
text <- munch1 (\x -> x /= '\n') -- Parse until EOL
|
||||||
|
-- traceM text
|
||||||
|
let parsedText = fst $ leftmostLongestParse parseLine text
|
||||||
|
return (Header (length headers) parsedText)
|
||||||
|
|
||||||
|
-- Parse bold text
|
||||||
|
parseBold :: ReadP MdToken
|
||||||
|
parseBold = do
|
||||||
|
text <- choice[
|
||||||
|
(between (string "__") (string "__") (many1 (lookaheadParse (/= "__")))),
|
||||||
|
(between (string "**") (string "**") (many1 (lookaheadParse (/= "**"))))
|
||||||
|
]
|
||||||
|
let parsedText = fst $ leftmostLongestParse parseLine text
|
||||||
|
return (Bold parsedText)
|
||||||
|
|
||||||
|
-- Parse italic text
|
||||||
|
parseItalic :: ReadP MdToken
|
||||||
|
parseItalic = do
|
||||||
|
text <- choice[
|
||||||
|
(between (string "_") (string "_") (munch1 (/= '_'))),
|
||||||
|
(between (string "*") (string "*") (munch1 (/= '*')))
|
||||||
|
]
|
||||||
|
let parsedText = fst $ leftmostLongestParse parseLine text
|
||||||
|
return (Italic parsedText)
|
||||||
|
|
||||||
|
-- Parse a linebreak character
|
||||||
|
parseLinebreak :: ReadP MdToken
|
||||||
|
parseLinebreak = do
|
||||||
|
char '\n'
|
||||||
|
return Linebreak
|
||||||
|
|
||||||
|
-- Parse a regular string as a Unit.
|
||||||
|
parseString :: ReadP MdToken
|
||||||
|
parseString = do
|
||||||
|
firstChar <- get -- Must parse at least one character here
|
||||||
|
text <- munch (\x -> not (elem x "#*_[\n"))
|
||||||
|
return (Unit (firstChar:text))
|
||||||
|
|
||||||
|
-- Parse any of the above tokens.
|
||||||
|
parseToken :: ReadP MdToken
|
||||||
|
parseToken = choice [parseHeader, parseLinebreak, parseBold, parseItalic, parseString]
|
||||||
|
|
||||||
|
-- Parse a line, consisting of one or more tokens.
|
||||||
|
parseLine :: ReadP MdToken
|
||||||
|
parseLine = do
|
||||||
|
remaining <- look
|
||||||
|
when (null remaining) pfail
|
||||||
|
parsed <- parseMany parseToken
|
||||||
|
-- traceM $ show parsed
|
||||||
|
return (Line parsed)
|
||||||
|
|
||||||
|
-- Parse a paragraph, which is a 'Line' (can span multiple actual lines), separated by double-newlines.
|
||||||
|
parsePara :: ReadP MdToken
|
||||||
|
parsePara = do
|
||||||
|
parseMany (char '\n')
|
||||||
|
text <- many1 (lookaheadParse (\x -> ((length x) < 2) || (take 2 x) /= "\n\n"))
|
||||||
|
string "\n\n"
|
||||||
|
-- I don't consume the ending double-newline, because the next paragraph will consume it as part of its starting double-newline.
|
||||||
|
let parsedText = fst $ leftmostLongestParse parseLine text
|
||||||
|
return (Para parsedText)
|
||||||
@ -0,0 +1,22 @@
|
|||||||
|
module MdToHtmlTest where
|
||||||
|
|
||||||
|
import MdToHTML
|
||||||
|
import Test.HUnit
|
||||||
|
|
||||||
|
headerTests = TestList
|
||||||
|
[
|
||||||
|
(TestCase (assertEqual "Should convert H1 heading" "<h1>Hello</h1>" (show . fst $ leftmostLongestParse parseLine "# Hello"))),
|
||||||
|
(TestCase (assertEqual "Should convert H2 heading" "<h2>Hello</h2>" (show . fst $ leftmostLongestParse parseLine "## Hello"))),
|
||||||
|
(TestCase (assertEqual "Should convert H3 heading" "<h3>Hello</h3>" (show . fst $ leftmostLongestParse parseLine "### Hello"))),
|
||||||
|
(TestCase (assertEqual "Should convert H4 heading" "<h4>Hello</h4>" (show . fst $ leftmostLongestParse parseLine "#### Hello"))),
|
||||||
|
(TestCase (assertEqual "Should convert H5 heading" "<h5>Hello</h5>" (show . fst $ leftmostLongestParse parseLine "##### Hello"))),
|
||||||
|
(TestCase (assertEqual "Should convert H6 heading" "<h6>Hello</h6>" (show . fst $ leftmostLongestParse parseLine "###### Hello")))
|
||||||
|
]
|
||||||
|
|
||||||
|
|
||||||
|
tests = TestList
|
||||||
|
[
|
||||||
|
headerTests
|
||||||
|
]
|
||||||
|
|
||||||
|
runTests = runTestTT tests
|
||||||
Loading…
Reference in New Issue