From 1b19e79650d6bcc6ec45733ee349c3af5c36ab36 Mon Sep 17 00:00:00 2001
From: John MacFarlane <jgm@berkeley.edu>
Date: Sat, 20 Oct 2018 10:00:50 -0700
Subject: [PATCH] Man reader: parse TP as definition lists.
Closes #4981.
---
src/Text/Pandoc/Readers/Man.hs | 61 ++++++++++++++++++++--------------
test/man-reader.man | 11 ++++++
test/man-reader.native | 9 ++++-
3 files changed, 55 insertions(+), 26 deletions(-)
diff --git a/src/Text/Pandoc/Readers/Man.hs b/src/Text/Pandoc/Readers/Man.hs
index 7e5288aa1..01c44066a 100644
--- a/src/Text/Pandoc/Readers/Man.hs
+++ b/src/Text/Pandoc/Readers/Man.hs
@@ -128,6 +128,7 @@ parseMan = do
parseBlock :: PandocMonad m => ManParser m Blocks
parseBlock = choice [ parseList
+ , parseDefinitionList
, parseTitle
, parsePara
, parseSkippedContent
@@ -351,7 +352,9 @@ msatisfy :: (Show t, Stream s m t) => (t -> Bool) -> ParserT s st m t
msatisfy predic = tokenPrim show nextPos testTok
where
testTok t = if predic t then Just t else Nothing
- nextPos pos _x _xs = updatePosString (setSourceColumn (setSourceLine pos $ sourceLine pos + 1) 1) ("")
+ nextPos pos _x _xs = updatePosString
+ (setSourceColumn
+ (setSourceLine pos $ sourceLine pos + 1) 1) ("")
mstr :: PandocMonad m => ManParser m ManToken
mstr = msatisfy isMStr where
@@ -428,25 +431,23 @@ parseInlines = do
let withspaces = intersperse B.space inls
return $ mconcat withspaces
- where
+strInl :: PandocMonad m => ManParser m Inlines
+strInl = do
+ (MStr rstr) <- mstr
+ return $ strToInlines rstr
- strInl :: PandocMonad m => ManParser m Inlines
- strInl = do
- (MStr rstr) <- mstr
- return $ strToInlines rstr
+lineInl :: PandocMonad m => ManParser m Inlines
+lineInl = do
+ (MLine fragments) <- mline
+ return $ mconcat $ strToInlines <$> fragments
- lineInl :: PandocMonad m => ManParser m Inlines
- lineInl = do
- (MLine fragments) <- mline
- return $ mconcat $ strToInlines <$> fragments
-
- linkInl :: PandocMonad m => ManParser m Inlines
- linkInl = do
- (MMaybeLink txt) <- mmaybeLink
- let inls = case runParser linkParser () "" txt of
- Right lnk -> lnk
- Left _ -> strong $ text txt
- return inls
+linkInl :: PandocMonad m => ManParser m Inlines
+linkInl = do
+ (MMaybeLink txt) <- mmaybeLink
+ let inls = case runParser linkParser () "" txt of
+ Right lnk -> lnk
+ Left _ -> strong $ text txt
+ return inls
where
@@ -463,8 +464,8 @@ parseInlines = do
lnkInls = link (manurl mpage [mansect]) mpage (strong $ str mpage)
return $ lnkInls <> strong (str (" ("++[mansect] ++ ")") <> text other)
- comment :: PandocMonad m => ManParser m Inlines
- comment = mcomment >> return mempty
+comment :: PandocMonad m => ManParser m Inlines
+comment = mcomment >> return mempty
bareIP :: PandocMonad m => ManParser m ManToken
bareIP = msatisfy isBareIP where
@@ -497,12 +498,10 @@ parseHeader = do
MMacro name args <- mmacro "SH" <|> mmacro "SS"
contents <- if null args
then do
- MLine ils <- mline
- return $ mconcat $ map strToInlines ils
+ strInl <|> lineInl
else do
- return $ mconcat
- $ intersperse B.space
- $ map strToInlines args
+ return $
+ mconcat $ intersperse B.space $ map strToInlines args
let lvl = if name == "SH" then 1 else 2
return $ header lvl contents
@@ -549,6 +548,18 @@ continuation = do
mmacro "RE"
return bs
+definitionListItem :: PandocMonad m
+ => ManParser m (Inlines, [Blocks])
+definitionListItem = try $ do
+ (MMacro _ _) <- mmacro "TP" -- args specify indent level, can ignore
+ term <- strInl <|> lineInl
+ inls <- parseInlines
+ continuations <- mconcat <$> many continuation
+ return $ (term, [para inls <> continuations])
+
+parseDefinitionList :: PandocMonad m => ManParser m Blocks
+parseDefinitionList = definitionList <$> many1 definitionListItem
+
-- In case of weird man file it will be parsed succesfully
parseSkipMacro :: PandocMonad m => ManParser m Blocks
parseSkipMacro = do
diff --git a/test/man-reader.man b/test/man-reader.man
index e514bc8ca..7a34d5dcb 100644
--- a/test/man-reader.man
+++ b/test/man-reader.man
@@ -144,7 +144,18 @@ Lower alpha with paren
.RE
.RE
.RE
+.SS Ordered
+Definition lists
+.TP
+.B term1
+definition 1
+.RS
.PP
+continued
+.RE
+.TP
+.B term2
+definition 2
* * * * *
.SH Special Characters
AT&T has an ampersand in their name.
diff --git a/test/man-reader.native b/test/man-reader.native
index 811c7c5c8..996ab7cac 100644
--- a/test/man-reader.native
+++ b/test/man-reader.native
@@ -70,7 +70,14 @@ Pandoc (Meta {unMeta = fromList [("date",MetaString "Oct 17, 2018"),("section",M
[[Para [Str "Decimal",Space,Str "start",Space,Str "with",Space,Str "6"]
,OrderedList (3,LowerAlpha,OneParen)
[[Para [Str "Lower",Space,Str "alpha",Space,Str "with",Space,Str "paren"]]]]]]]]]
-,Para [Str "*",Space,Str "*",Space,Str "*",Space,Str "*",Space,Str "*"]
+,Header 2 ("",[],[]) [Str "Ordered"]
+,Para [Str "Definition",Space,Str "lists"]
+,DefinitionList
+ [([Strong [Str "term1"]],
+ [[Para [Str "definition",Space,Str "1"]
+ ,Para [Str "continued"]]])
+ ,([Strong [Str "term2"]],
+ [[Para [Str "definition",Space,Str "2",Space,Str "*",Space,Str "*",Space,Str "*",Space,Str "*",Space,Str "*"]]])]
,Header 1 ("",[],[]) [Str "Special",Space,Str "Characters"]
,Para [Str "AT&T",Space,Str "has",Space,Str "an",Space,Str "ampersand",Space,Str "in",Space,Str "their",Space,Str "name."]
,Para [Str "4",Space,Str "<",Space,Str "5."]