aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorJohn MacFarlane <fiddlosopher@gmail.com>2012-03-24 15:53:57 -0700
committerJohn MacFarlane <fiddlosopher@gmail.com>2012-03-24 15:53:57 -0700
commit9352750cb24b672e869e7351df9875d5fa8bf7f9 (patch)
tree5d979b63659c8c30b98c69d186cada19a7846a44
parent4c2249ea8905d708fb61c82fa86fd67a503d7832 (diff)
downloadpandoc-9352750cb24b672e869e7351df9875d5fa8bf7f9.tar.gz
RST writer: Better handling of inline formatting.
* `hi*there*hi` now gets rendered properly as `hi\ *there*\ hi` * Unnecessary '\ ' are avoided around :math:, :sub:, :sup: * Implemented most of the rules in http://docutils.sourceforge.net/docs/ref/rst/restructuredtext.html#inline-markup-recognition-rules, though some unnecessary '\ ' may be inserted when unicode quotes are used.
-rw-r--r--src/Text/Pandoc/Writers/RST.hs55
-rw-r--r--tests/writer.rst6
2 files changed, 54 insertions, 7 deletions
diff --git a/src/Text/Pandoc/Writers/RST.hs b/src/Text/Pandoc/Writers/RST.hs
index d6e5b5c9e..e6c84760e 100644
--- a/src/Text/Pandoc/Writers/RST.hs
+++ b/src/Text/Pandoc/Writers/RST.hs
@@ -38,6 +38,7 @@ import Data.List ( isPrefixOf, intersperse, transpose )
import Text.Pandoc.Pretty
import Control.Monad.State
import Control.Applicative ( (<$>) )
+import Data.Char (isSpace)
type Refs = [([Inline], Target)]
@@ -253,7 +254,52 @@ blockListToRST blocks = mapM blockToRST blocks >>= return . vcat
-- | Convert list of Pandoc inline elements to RST.
inlineListToRST :: [Inline] -> State WriterState Doc
-inlineListToRST lst = mapM inlineToRST lst >>= return . hcat
+inlineListToRST lst = mapM inlineToRST (insertBS lst) >>= return . hcat
+ where insertBS :: [Inline] -> [Inline] -- insert '\ ' where needed
+ insertBS (x:y:z:zs)
+ | isComplex y && surroundComplex x z =
+ x : y : RawInline "rst" "\\ " : insertBS (z:zs)
+ insertBS (x:y:zs)
+ | isComplex x && not (okAfterComplex y) =
+ x : RawInline "rst" "\\ " : insertBS (y : zs)
+ | isComplex y && not (okBeforeComplex x) =
+ x : RawInline "rst" "\\ " : insertBS (y : zs)
+ | otherwise =
+ x : insertBS (y : zs)
+ insertBS (x:ys) = x : insertBS ys
+ insertBS [] = []
+ surroundComplex :: Inline -> Inline -> Bool
+ surroundComplex (Str s@(_:_)) (Str s'@(_:_)) =
+ case (last s, head s') of
+ ('\'','\'') -> True
+ ('"','"') -> True
+ ('<','>') -> True
+ ('[',']') -> True
+ ('{','}') -> True
+ _ -> False
+ surroundComplex _ _ = False
+ okAfterComplex :: Inline -> Bool
+ okAfterComplex Space = True
+ okAfterComplex LineBreak = True
+ okAfterComplex (Str (c:_)) = isSpace c || c `elem` "-.,:;!?\\/'\")]}>–—"
+ okAfterComplex _ = False
+ okBeforeComplex :: Inline -> Bool
+ okBeforeComplex Space = True
+ okBeforeComplex LineBreak = True
+ okBeforeComplex (Str (c:_)) = isSpace c || c `elem` "-:/'\"<([{–—"
+ okBeforeComplex _ = False
+ isComplex :: Inline -> Bool
+ isComplex (Emph _) = True
+ isComplex (Strong _) = True
+ isComplex (SmallCaps _) = True
+ isComplex (Strikeout _) = True
+ isComplex (Superscript _) = True
+ isComplex (Subscript _) = True
+ isComplex (Link _ _) = True
+ isComplex (Image _ _) = True
+ isComplex (Code _ _) = True
+ isComplex (Math _ _) = True
+ isComplex _ = False
-- | Convert Pandoc inline element to RST.
inlineToRST :: Inline -> State WriterState Doc
@@ -268,10 +314,10 @@ inlineToRST (Strikeout lst) = do
return $ "[STRIKEOUT:" <> contents <> "]"
inlineToRST (Superscript lst) = do
contents <- inlineListToRST lst
- return $ "\\ :sup:`" <> contents <> "`\\ "
+ return $ ":sup:`" <> contents <> "`"
inlineToRST (Subscript lst) = do
contents <- inlineListToRST lst
- return $ "\\ :sub:`" <> contents <> "`\\ "
+ return $ ":sub:`" <> contents <> "`"
inlineToRST (SmallCaps lst) = inlineListToRST lst
inlineToRST (Quoted SingleQuote lst) = do
contents <- inlineListToRST lst
@@ -286,11 +332,12 @@ inlineToRST (Str str) = return $ text $ escapeString str
inlineToRST (Math t str) = do
modify $ \st -> st{ stHasMath = True }
return $ if t == InlineMath
- then ":math:`" <> text str <> "`" <> beforeNonBlank "\\ "
+ then ":math:`" <> text str <> "`"
else if '\n' `elem` str
then blankline $$ ".. math::" $$
blankline $$ nest 3 (text str) $$ blankline
else blankline $$ (".. math:: " <> text str) $$ blankline
+inlineToRST (RawInline "rst" x) = return $ text x
inlineToRST (RawInline _ _) = return empty
inlineToRST (LineBreak) = return cr -- there's no line break in RST
inlineToRST Space = return space
diff --git a/tests/writer.rst b/tests/writer.rst
index 09fd8dcb3..8d7c7915c 100644
--- a/tests/writer.rst
+++ b/tests/writer.rst
@@ -563,7 +563,7 @@ This is code: ``>``, ``$``, ``\``, ``\$``, ``<html>``.
[STRIKEOUT:This is *strikeout*.]
-Superscripts: a\ :sup:`bc`\ d a\ :sup:`*hello*`\ a\ :sup:`hello there`\ .
+Superscripts: a\ :sup:`bc`\ d a\ :sup:`*hello*` a\ :sup:`hello there`.
Subscripts: H\ :sub:`2`\ O, H\ :sub:`23`\ O, H\ :sub:`many of them`\ O.
@@ -602,12 +602,12 @@ LaTeX
- :math:`x \in y`
- :math:`\alpha \wedge \omega`
- :math:`223`
-- :math:`p`\ -Tree
+- :math:`p`-Tree
- Here’s some display math:
.. math:: \frac{d}{dx}f(x)=\lim_{h\to 0}\frac{f(x+h)-f(x)}{h}
-- Here’s one that has a line break in it: :math:`\alpha + \omega \times x^2`\ .
+- Here’s one that has a line break in it: :math:`\alpha + \omega \times x^2`.
These shouldn’t be math: