aboutsummaryrefslogtreecommitdiff
path: root/src/Text/Pandoc
diff options
context:
space:
mode:
Diffstat (limited to 'src/Text/Pandoc')
-rw-r--r--src/Text/Pandoc/Asciify.hs4
-rw-r--r--src/Text/Pandoc/Error.hs4
-rw-r--r--src/Text/Pandoc/Highlighting.hs4
-rw-r--r--src/Text/Pandoc/ImageSize.hs4
-rw-r--r--src/Text/Pandoc/MIME.hs4
-rw-r--r--src/Text/Pandoc/Options.hs8
-rw-r--r--src/Text/Pandoc/PDF.hs4
-rw-r--r--src/Text/Pandoc/Parsing.hs4
-rw-r--r--src/Text/Pandoc/Pretty.hs4
-rw-r--r--src/Text/Pandoc/Process.hs4
-rw-r--r--src/Text/Pandoc/Readers/Docx.hs30
-rw-r--r--src/Text/Pandoc/Readers/Docx/Parse.hs94
-rw-r--r--src/Text/Pandoc/Readers/EPUB.hs6
-rw-r--r--src/Text/Pandoc/Readers/HTML.hs55
-rw-r--r--src/Text/Pandoc/Readers/Markdown.hs80
-rw-r--r--src/Text/Pandoc/Readers/MediaWiki.hs2
-rw-r--r--src/Text/Pandoc/Readers/Odt.hs4
-rw-r--r--src/Text/Pandoc/Readers/Org.hs161
-rw-r--r--src/Text/Pandoc/SelfContained.hs4
-rw-r--r--src/Text/Pandoc/Shared.hs4
-rw-r--r--src/Text/Pandoc/Slides.hs4
-rw-r--r--src/Text/Pandoc/Templates.hs4
-rw-r--r--src/Text/Pandoc/UTF8.hs4
-rw-r--r--src/Text/Pandoc/UUID.hs4
-rw-r--r--src/Text/Pandoc/Writers/ConTeXt.hs12
-rw-r--r--src/Text/Pandoc/Writers/Docbook.hs21
-rw-r--r--src/Text/Pandoc/Writers/Docx.hs4
-rw-r--r--src/Text/Pandoc/Writers/DokuWiki.hs5
-rw-r--r--src/Text/Pandoc/Writers/EPUB.hs3
-rw-r--r--src/Text/Pandoc/Writers/HTML.hs5
-rw-r--r--src/Text/Pandoc/Writers/LaTeX.hs117
-rw-r--r--src/Text/Pandoc/Writers/Org.hs2
-rw-r--r--src/Text/Pandoc/XML.hs4
33 files changed, 405 insertions, 268 deletions
diff --git a/src/Text/Pandoc/Asciify.hs b/src/Text/Pandoc/Asciify.hs
index c183458e4..8eb1ba663 100644
--- a/src/Text/Pandoc/Asciify.hs
+++ b/src/Text/Pandoc/Asciify.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2013-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2013-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Asciify
- Copyright : Copyright (C) 2013-2015 John MacFarlane
+ Copyright : Copyright (C) 2013-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Error.hs b/src/Text/Pandoc/Error.hs
index 0a4e08175..792098b35 100644
--- a/src/Text/Pandoc/Error.hs
+++ b/src/Text/Pandoc/Error.hs
@@ -1,6 +1,6 @@
{-# LANGUAGE DeriveDataTypeable, DeriveGeneric #-}
{-
-Copyright (C) 2006-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2006-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
-}
{- |
Module : Text.Pandoc.Error
- Copyright : Copyright (C) 2006-2015 John MacFarlane
+ Copyright : Copyright (C) 2006-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Highlighting.hs b/src/Text/Pandoc/Highlighting.hs
index ecfef1832..1b9e92ae2 100644
--- a/src/Text/Pandoc/Highlighting.hs
+++ b/src/Text/Pandoc/Highlighting.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2008-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2008-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Highlighting
- Copyright : Copyright (C) 2008-2015 John MacFarlane
+ Copyright : Copyright (C) 2008-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/ImageSize.hs b/src/Text/Pandoc/ImageSize.hs
index 571fdd665..90dfbb5fb 100644
--- a/src/Text/Pandoc/ImageSize.hs
+++ b/src/Text/Pandoc/ImageSize.hs
@@ -1,7 +1,7 @@
{-# LANGUAGE OverloadedStrings, ScopedTypeVariables, CPP #-}
{-# OPTIONS_GHC -fno-warn-type-defaults #-}
{-
- Copyright (C) 2011-2015 John MacFarlane <jgm@berkeley.edu>
+ Copyright (C) 2011-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -20,7 +20,7 @@
{- |
Module : Text.Pandoc.ImageSize
-Copyright : Copyright (C) 2011-2015 John MacFarlane
+Copyright : Copyright (C) 2011-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/MIME.hs b/src/Text/Pandoc/MIME.hs
index 6fd9ac373..1164e04b3 100644
--- a/src/Text/Pandoc/MIME.hs
+++ b/src/Text/Pandoc/MIME.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2011-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2011-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.MIME
- Copyright : Copyright (C) 2011-2015 John MacFarlane
+ Copyright : Copyright (C) 2011-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Options.hs b/src/Text/Pandoc/Options.hs
index 333f499fb..701cd8bd1 100644
--- a/src/Text/Pandoc/Options.hs
+++ b/src/Text/Pandoc/Options.hs
@@ -1,6 +1,6 @@
{-# LANGUAGE DeriveDataTypeable, DeriveGeneric #-}
{-
-Copyright (C) 2012-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2012-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -19,7 +19,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Options
- Copyright : Copyright (C) 2012-2015 John MacFarlane
+ Copyright : Copyright (C) 2012-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
@@ -264,6 +264,7 @@ data ReaderOptions = ReaderOptions{
, readerDefaultImageExtension :: String -- ^ Default extension for images
, readerTrace :: Bool -- ^ Print debugging info
, readerTrackChanges :: TrackChanges
+ , readerFileScope :: Bool -- ^ Parse before combining
} deriving (Show, Read, Data, Typeable, Generic)
instance Default ReaderOptions
@@ -280,6 +281,7 @@ instance Default ReaderOptions
, readerDefaultImageExtension = ""
, readerTrace = False
, readerTrackChanges = AcceptChanges
+ , readerFileScope = False
}
--
@@ -355,6 +357,7 @@ data WriterOptions = WriterOptions
, writerSourceURL :: Maybe String -- ^ Absolute URL + directory of 1st source file
, writerUserDataDir :: Maybe FilePath -- ^ Path of user data directory
, writerCiteMethod :: CiteMethod -- ^ How to print cites
+ , writerDocbook5 :: Bool -- ^ Produce DocBook5
, writerHtml5 :: Bool -- ^ Produce HTML5
, writerHtmlQTags :: Bool -- ^ Use @<q>@ tags for quotes in HTML
, writerBeamer :: Bool -- ^ Produce beamer LaTeX slide show
@@ -401,6 +404,7 @@ instance Default WriterOptions where
, writerSourceURL = Nothing
, writerUserDataDir = Nothing
, writerCiteMethod = Citeproc
+ , writerDocbook5 = False
, writerHtml5 = False
, writerHtmlQTags = False
, writerBeamer = False
diff --git a/src/Text/Pandoc/PDF.hs b/src/Text/Pandoc/PDF.hs
index da4ee4e33..4dbe1f000 100644
--- a/src/Text/Pandoc/PDF.hs
+++ b/src/Text/Pandoc/PDF.hs
@@ -1,6 +1,6 @@
{-# LANGUAGE OverloadedStrings, CPP, ScopedTypeVariables #-}
{-
-Copyright (C) 2012-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2012-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -19,7 +19,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.PDF
- Copyright : Copyright (C) 2012-2015 John MacFarlane
+ Copyright : Copyright (C) 2012-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Parsing.hs b/src/Text/Pandoc/Parsing.hs
index 325231846..7bf827019 100644
--- a/src/Text/Pandoc/Parsing.hs
+++ b/src/Text/Pandoc/Parsing.hs
@@ -5,7 +5,7 @@
, MultiParamTypeClasses
, FlexibleInstances #-}
{-
-Copyright (C) 2006-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2006-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -24,7 +24,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Parsing
- Copyright : Copyright (C) 2006-2015 John MacFarlane
+ Copyright : Copyright (C) 2006-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Pretty.hs b/src/Text/Pandoc/Pretty.hs
index 88b7dd09e..f3ef0ef10 100644
--- a/src/Text/Pandoc/Pretty.hs
+++ b/src/Text/Pandoc/Pretty.hs
@@ -1,6 +1,6 @@
{-# LANGUAGE GeneralizedNewtypeDeriving, CPP #-}
{-
-Copyright (C) 2010-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2010-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -19,7 +19,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111(-1)307 USA
{- |
Module : Text.Pandoc.Pretty
- Copyright : Copyright (C) 2010-2015 John MacFarlane
+ Copyright : Copyright (C) 2010-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Process.hs b/src/Text/Pandoc/Process.hs
index e5245638d..bc71f1392 100644
--- a/src/Text/Pandoc/Process.hs
+++ b/src/Text/Pandoc/Process.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2013-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2013-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Process
- Copyright : Copyright (C) 2013-2015 John MacFarlane
+ Copyright : Copyright (C) 2013-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Readers/Docx.hs b/src/Text/Pandoc/Readers/Docx.hs
index eb71d8dd8..9c7c3b264 100644
--- a/src/Text/Pandoc/Readers/Docx.hs
+++ b/src/Text/Pandoc/Readers/Docx.hs
@@ -50,8 +50,7 @@ implemented, [-] means partially implemented):
* Inlines
- [X] Str
- - [X] Emph (From italics. `underline` currently read as span. In
- future, it might optionally be emph as well)
+ - [X] Emph (italics and underline both read as Emph)
- [X] Strong
- [X] Strikeout
- [X] Superscript
@@ -62,16 +61,16 @@ implemented, [-] means partially implemented):
- [X] Code (styled with `VerbatimChar`)
- [X] Space
- [X] LineBreak (these are invisible in Word: entered with Shift-Return)
- - [ ] Math
+ - [X] Math
- [X] Link (links to an arbitrary bookmark create a span with the target as
id and "anchor" class)
- - [-] Image (Links to path in archive. Future option for
- data-encoded URI likely.)
+ - [X] Image
- [X] Note (Footnotes and Endnotes are silently combined.)
-}
module Text.Pandoc.Readers.Docx
- ( readDocx
+ ( readDocxWithWarnings
+ , readDocx
) where
import Codec.Archive.Zip
@@ -98,14 +97,23 @@ import qualified Data.Sequence as Seq (null)
import Text.Pandoc.Error
import Text.Pandoc.Compat.Except
+readDocxWithWarnings :: ReaderOptions
+ -> B.ByteString
+ -> Either PandocError (Pandoc, MediaBag, [String])
+readDocxWithWarnings opts bytes
+ | Right archive <- toArchiveOrFail bytes
+ , Right (docx, warnings) <- archiveToDocxWithWarnings archive = do
+ (meta, blks, mediaBag) <- docxToOutput opts docx
+ return (Pandoc meta blks, mediaBag, warnings)
+readDocxWithWarnings _ _ =
+ Left (ParseFailure "couldn't parse docx file")
+
readDocx :: ReaderOptions
-> B.ByteString
-> Either PandocError (Pandoc, MediaBag)
-readDocx opts bytes =
- case archiveToDocx (toArchive bytes) of
- Right docx -> (\(meta, blks, mediaBag) -> (Pandoc meta blks, mediaBag))
- <$> (docxToOutput opts docx)
- Left _ -> Left (ParseFailure "couldn't parse docx file")
+readDocx opts bytes = do
+ (pandoc, mediaBag, _) <- readDocxWithWarnings opts bytes
+ return (pandoc, mediaBag)
data DState = DState { docxAnchorMap :: M.Map String String
, docxMediaBag :: MediaBag
diff --git a/src/Text/Pandoc/Readers/Docx/Parse.hs b/src/Text/Pandoc/Readers/Docx/Parse.hs
index eec8b12c9..7265ef8dd 100644
--- a/src/Text/Pandoc/Readers/Docx/Parse.hs
+++ b/src/Text/Pandoc/Readers/Docx/Parse.hs
@@ -50,6 +50,7 @@ module Text.Pandoc.Readers.Docx.Parse ( Docx(..)
, Row(..)
, Cell(..)
, archiveToDocx
+ , archiveToDocxWithWarnings
) where
import Codec.Archive.Zip
import Text.XML.Light
@@ -60,6 +61,7 @@ import Data.Bits ((.|.))
import qualified Data.ByteString.Lazy as B
import qualified Text.Pandoc.UTF8 as UTF8
import Control.Monad.Reader
+import Control.Monad.State
import Control.Applicative ((<|>))
import qualified Data.Map as M
import Text.Pandoc.Compat.Except
@@ -81,16 +83,20 @@ data ReaderEnv = ReaderEnv { envNotes :: Notes
}
deriving Show
+data ReaderState = ReaderState { stateWarnings :: [String] }
+ deriving Show
+
+
data DocxError = DocxError | WrongElem
deriving Show
instance Error DocxError where
noMsg = WrongElem
-type D = ExceptT DocxError (Reader ReaderEnv)
+type D = ExceptT DocxError (ReaderT ReaderEnv (State ReaderState))
-runD :: D a -> ReaderEnv -> Either DocxError a
-runD dx re = runReader (runExceptT dx) re
+runD :: D a -> ReaderEnv -> ReaderState -> (Either DocxError a, ReaderState)
+runD dx re rs = runState (runReaderT (runExceptT dx) re) rs
maybeToD :: Maybe a -> D a
maybeToD (Just a) = return a
@@ -257,7 +263,10 @@ type Author = String
type ChangeDate = String
archiveToDocx :: Archive -> Either DocxError Docx
-archiveToDocx archive = do
+archiveToDocx archive = fst <$> archiveToDocxWithWarnings archive
+
+archiveToDocxWithWarnings :: Archive -> Either DocxError (Docx, [String])
+archiveToDocxWithWarnings archive = do
let notes = archiveToNotes archive
numbering = archiveToNumbering archive
rels = archiveToRelationships archive
@@ -265,8 +274,12 @@ archiveToDocx archive = do
(styles, parstyles) = archiveToStyles archive
rEnv =
ReaderEnv notes numbering rels media Nothing styles parstyles InDocument
- doc <- runD (archiveToDocument archive) rEnv
- return $ Docx doc
+ rState = ReaderState { stateWarnings = [] }
+ (eitherDoc, st) = runD (archiveToDocument archive) rEnv rState
+ case eitherDoc of
+ Right doc -> Right (Docx doc, stateWarnings st)
+ Left e -> Left e
+
archiveToDocument :: Archive -> D Document
@@ -576,12 +589,14 @@ elemToBodyPart ns element
sty <- asks envParStyles
let parstyle = elemToParagraphStyle ns element sty
parparts <- mapD (elemToParPart ns) (elChildren element)
- case pNumInfo parstyle of
- Just (numId, lvl) -> do
- num <- asks envNumbering
- let levelInfo = lookupLevel numId lvl num
- return $ ListItem parstyle numId lvl levelInfo parparts
- Nothing -> return $ Paragraph parstyle parparts
+ -- Word uses list enumeration for numbered headings, so we only
+ -- want to infer a list from the styles if it is NOT a heading.
+ case pHeading parstyle of
+ Nothing | Just (numId, lvl) <- pNumInfo parstyle -> do
+ num <- asks envNumbering
+ let levelInfo = lookupLevel numId lvl num
+ return $ ListItem parstyle numId lvl levelInfo parparts
+ _ -> return $ Paragraph parstyle parparts
elemToBodyPart ns element
| isElem ns "w" "tbl" element = do
let caption' = findChild (elemName ns "w" "tblPr") element
@@ -646,14 +661,14 @@ elemToParPart ns element
| isElem ns "w" "r" element =
elemToRun ns element >>= (\r -> return $ PlainRun r)
elemToParPart ns element
- | isElem ns "w" "ins" element
+ | isElem ns "w" "ins" element || isElem ns "w" "moveTo" element
, Just cId <- findAttr (elemName ns "w" "id") element
, Just cAuthor <- findAttr (elemName ns "w" "author") element
, Just cDate <- findAttr (elemName ns "w" "date") element = do
runs <- mapD (elemToRun ns) (elChildren element)
return $ Insertion cId cAuthor cDate runs
elemToParPart ns element
- | isElem ns "w" "del" element
+ | isElem ns "w" "del" element || isElem ns "w" "moveFrom" element
, Just cId <- findAttr (elemName ns "w" "id") element
, Just cAuthor <- findAttr (elemName ns "w" "author") element
, Just cDate <- findAttr (elemName ns "w" "date") element = do
@@ -702,36 +717,58 @@ elemToExtent drawingElem =
getDim at = findElement (QName "extent" (Just wp_ns) (Just "wp")) drawingElem
>>= findAttr (QName at Nothing Nothing) >>= safeRead
-elemToRun :: NameSpaces -> Element -> D Run
-elemToRun ns element
- | isElem ns "w" "r" element
- , Just drawingElem <- findChild (elemName ns "w" "drawing") element =
+
+childElemToRun :: NameSpaces -> Element -> D Run
+childElemToRun ns element
+ | isElem ns "w" "drawing" element =
let a_ns = "http://schemas.openxmlformats.org/drawingml/2006/main"
- drawing = findElement (QName "blip" (Just a_ns) (Just "a")) drawingElem
+ drawing = findElement (QName "blip" (Just a_ns) (Just "a")) element
>>= findAttr (QName "embed" (lookup "r" ns) (Just "r"))
in
case drawing of
Just s -> expandDrawingId s >>=
- (\(fp, bs) -> return $ InlineDrawing fp bs $ elemToExtent drawingElem)
+ (\(fp, bs) -> return $ InlineDrawing fp bs $ elemToExtent element)
Nothing -> throwError WrongElem
-elemToRun ns element
- | isElem ns "w" "r" element
- , Just ref <- findChild (elemName ns "w" "footnoteReference") element
- , Just fnId <- findAttr (elemName ns "w" "id") ref = do
+childElemToRun ns element
+ | isElem ns "w" "footnoteReference" element
+ , Just fnId <- findAttr (elemName ns "w" "id") element = do
notes <- asks envNotes
case lookupFootnote fnId notes of
Just e -> do bps <- local (\r -> r {envLocation=InFootnote}) $ mapD (elemToBodyPart ns) (elChildren e)
return $ Footnote bps
Nothing -> return $ Footnote []
-elemToRun ns element
- | isElem ns "w" "r" element
- , Just ref <- findChild (elemName ns "w" "endnoteReference") element
- , Just enId <- findAttr (elemName ns "w" "id") ref = do
+childElemToRun ns element
+ | isElem ns "w" "endnoteReference" element
+ , Just enId <- findAttr (elemName ns "w" "id") element = do
notes <- asks envNotes
case lookupEndnote enId notes of
Just e -> do bps <- local (\r -> r {envLocation=InEndnote}) $ mapD (elemToBodyPart ns) (elChildren e)
return $ Endnote bps
Nothing -> return $ Endnote []
+childElemToRun _ _ = throwError WrongElem
+
+elemToRun :: NameSpaces -> Element -> D Run
+elemToRun ns element
+ | isElem ns "w" "r" element
+ , Just altCont <- findChild (elemName ns "mc" "AlternateContent") element =
+ do let choices = findChildren (elemName ns "mc" "Choice") altCont
+ choiceChildren = map head $ filter (not . null) $ map elChildren choices
+ outputs <- mapD (childElemToRun ns) choiceChildren
+ case outputs of
+ r : _ -> return r
+ [] -> throwError WrongElem
+elemToRun ns element
+ | isElem ns "w" "r" element
+ , Just drawingElem <- findChild (elemName ns "w" "drawing") element =
+ childElemToRun ns drawingElem
+elemToRun ns element
+ | isElem ns "w" "r" element
+ , Just ref <- findChild (elemName ns "w" "footnoteReference") element =
+ childElemToRun ns ref
+elemToRun ns element
+ | isElem ns "w" "r" element
+ , Just ref <- findChild (elemName ns "w" "endnoteReference") element =
+ childElemToRun ns ref
elemToRun ns element
| isElem ns "w" "r" element = do
runElems <- elemToRunElems ns element
@@ -940,3 +977,4 @@ elemToRunElems _ _ = throwError WrongElem
setFont :: Maybe Font -> ReaderEnv -> ReaderEnv
setFont f s = s{envFont = f}
+
diff --git a/src/Text/Pandoc/Readers/EPUB.hs b/src/Text/Pandoc/Readers/EPUB.hs
index 07d282708..144ba9ca2 100644
--- a/src/Text/Pandoc/Readers/EPUB.hs
+++ b/src/Text/Pandoc/Readers/EPUB.hs
@@ -19,7 +19,7 @@ import Text.Pandoc.Compat.Except (MonadError, throwError, runExcept, Except)
import Text.Pandoc.Compat.Monoid ((<>))
import Text.Pandoc.MIME (MimeType)
import qualified Text.Pandoc.Builder as B
-import Codec.Archive.Zip ( Archive (..), toArchive, fromEntry
+import Codec.Archive.Zip ( Archive (..), toArchiveOrFail, fromEntry
, findEntryByPath, Entry)
import qualified Data.ByteString.Lazy as BL (ByteString)
import System.FilePath ( takeFileName, (</>), dropFileName, normalise
@@ -39,7 +39,9 @@ import Text.Pandoc.Error
type Items = M.Map String (FilePath, MimeType)
readEPUB :: ReaderOptions -> BL.ByteString -> Either PandocError (Pandoc, MediaBag)
-readEPUB opts bytes = runEPUB (archiveToEPUB opts $ toArchive bytes)
+readEPUB opts bytes = case toArchiveOrFail bytes of
+ Right archive -> runEPUB $ archiveToEPUB opts $ archive
+ Left _ -> Left $ ParseFailure "Couldn't extract ePub file"
runEPUB :: Except PandocError a -> Either PandocError a
runEPUB = runExcept
diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs
index 69df13aac..8ee5da543 100644
--- a/src/Text/Pandoc/Readers/HTML.hs
+++ b/src/Text/Pandoc/Readers/HTML.hs
@@ -63,7 +63,7 @@ import Debug.Trace (trace)
import Text.TeXMath (readMathML, writeTeX)
import Data.Default (Default (..), def)
import Control.Monad.Reader (Reader,ask, asks, local, runReader)
-import Network.URI (isURI)
+import Network.URI (URI, parseURIReference, nonStrictRelativeTo)
import Text.Pandoc.Error
import Text.Pandoc.CSS (foldOrElse, pickStyleAttrProps)
import Text.Pandoc.Compat.Monoid ((<>))
@@ -103,7 +103,7 @@ data HTMLState =
HTMLState
{ parserState :: ParserState,
noteTable :: [(String, Blocks)],
- baseHref :: Maybe String,
+ baseHref :: Maybe URI,
identifiers :: Set.Set String,
headerMap :: M.Map Inlines String
}
@@ -145,15 +145,9 @@ pHead = pInTags "head" $ pTitle <|> pMetaTag <|> pBaseTag <|> (mempty <$ pAnyTag
return mempty
pBaseTag = do
bt <- pSatisfy (~== TagOpen "base" [])
- let baseH = fromAttrib "href" bt
- if null baseH
- then return mempty
- else do
- let baseH' = case reverse baseH of
- '/':_ -> baseH
- _ -> baseH ++ "/"
- updateState $ \st -> st{ baseHref = Just baseH' }
- return mempty
+ updateState $ \st -> st{ baseHref =
+ parseURIReference $ fromAttrib "href" bt }
+ return mempty
block :: TagParser Blocks
block = do
@@ -610,9 +604,9 @@ pLink = try $ do
tag <- pSatisfy $ tagOpenLit "a" (const True)
mbBaseHref <- baseHref <$> getState
let url' = fromAttrib "href" tag
- let url = case (isURI url', mbBaseHref) of
- (False, Just h) -> h ++ url'
- _ -> url'
+ let url = case (parseURIReference url', mbBaseHref) of
+ (Just rel, Just bs) -> show (rel `nonStrictRelativeTo` bs)
+ _ -> url'
let title = fromAttrib "title" tag
let uid = fromAttrib "id" tag
let cls = words $ fromAttrib "class" tag
@@ -624,9 +618,9 @@ pImage = do
tag <- pSelfClosing (=="img") (isJust . lookup "src")
mbBaseHref <- baseHref <$> getState
let url' = fromAttrib "src" tag
- let url = case (isURI url', mbBaseHref) of
- (False, Just h) -> h ++ url'
- _ -> url'
+ let url = case (parseURIReference url', mbBaseHref) of
+ (Just rel, Just bs) -> show (rel `nonStrictRelativeTo` bs)
+ _ -> url'
let title = fromAttrib "title" tag
let alt = fromAttrib "alt" tag
let uid = fromAttrib "id" tag
@@ -945,7 +939,7 @@ htmlInBalanced f = try $ do
(TagClose _ : TagPosition er ec : _) -> do
let ls = er - sr
let cs = ec - sc
- lscontents <- concat <$> count ls anyLine
+ lscontents <- unlines <$> count ls anyLine
cscontents <- count cs anyChar
(_,closetag) <- htmlTag (~== TagClose tn)
return (lscontents ++ cscontents ++ closetag)
@@ -977,11 +971,20 @@ htmlTag :: Monad m
htmlTag f = try $ do
lookAhead (char '<')
inp <- getInput
- let (next : rest) = canonicalizeTags $ parseTagsOptions
- parseOptions{ optTagWarning = True } inp
+ let (next : _) = canonicalizeTags $ parseTagsOptions
+ parseOptions{ optTagWarning = False } inp
guard $ f next
+ let handleTag tagname = do
+ -- <www.boe.es/buscar/act.php?id=BOE-A-1996-8930#a66>
+ -- should NOT be parsed as an HTML tag, see #2277
+ guard $ not ('.' `elem` tagname)
+ -- <https://example.org> should NOT be a tag either.
+ -- tagsoup will parse it as TagOpen "https:" [("example.org","")]
+ guard $ not (null tagname)
+ guard $ last tagname /= ':'
+ rendered <- manyTill anyChar (char '>')
+ return (next, rendered ++ ">")
case next of
- TagWarning _ -> fail "encountered TagWarning"
TagComment s
| "<!--" `isPrefixOf` inp -> do
count (length s + 4) anyChar
@@ -989,13 +992,9 @@ htmlTag f = try $ do
char '>'
return (next, "<!--" ++ s ++ "-->")
| otherwise -> fail "bogus comment mode, HTML5 parse error"
- _ -> do
- -- we get a TagWarning on things like
- -- <www.boe.es/buscar/act.php?id=BOE-A-1996-8930#a66>
- -- which should NOT be parsed as an HTML tag, see #2277
- guard $ not $ hasTagWarning rest
- rendered <- manyTill anyChar (char '>')
- return (next, rendered ++ ">")
+ TagOpen tagname _attr -> handleTag tagname
+ TagClose tagname -> handleTag tagname
+ _ -> mzero
mkAttr :: [(String, String)] -> Attr
mkAttr attr = (attribsId, attribsClasses, attribsKV)
diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs
index 587726084..e43714526 100644
--- a/src/Text/Pandoc/Readers/Markdown.hs
+++ b/src/Text/Pandoc/Readers/Markdown.hs
@@ -122,9 +122,6 @@ inList = do
ctx <- stateParserContext <$> getState
guard (ctx == ListItemState)
-isNull :: F Inlines -> Bool
-isNull ils = B.isNull $ runF ils def
-
spnl :: Parser [Char] st ()
spnl = try $ do
skipSpaces
@@ -188,31 +185,38 @@ charsInBalancedBrackets openBrackets =
-- document structure
--
-titleLine :: MarkdownParser (F Inlines)
-titleLine = try $ do
+rawTitleBlockLine :: MarkdownParser String
+rawTitleBlockLine = do
char '%'
skipSpaces
- res <- many $ (notFollowedBy newline >> inline)
- <|> try (endline >> whitespace)
- newline
+ first <- anyLine
+ rest <- many $ try $ do spaceChar
+ notFollowedBy blankline
+ skipSpaces
+ anyLine
+ return $ trim $ unlines (first:rest)
+
+titleLine :: MarkdownParser (F Inlines)
+titleLine = try $ do
+ raw <- rawTitleBlockLine
+ res <- parseFromString (many inline) raw
return $ trimInlinesF $ mconcat res
authorsLine :: MarkdownParser (F [Inlines])
authorsLine = try $ do
- char '%'
- skipSpaces
- authors <- sepEndBy (many (notFollowedBy (satisfy $ \c ->
- c == ';' || c == '\n') >> inline))
- (char ';' <|>
- try (newline >> notFollowedBy blankline >> spaceChar))
- newline
- return $ sequence $ filter (not . isNull) $ map (trimInlinesF . mconcat) authors
+ raw <- rawTitleBlockLine
+ let sep = (char ';' <* spaces) <|> newline
+ let pAuthors = sepEndBy
+ (trimInlinesF . mconcat <$> many
+ (try $ notFollowedBy sep >> inline))
+ sep
+ sequence <$> parseFromString pAuthors raw
dateLine :: MarkdownParser (F Inlines)
dateLine = try $ do
- char '%'
- skipSpaces
- trimInlinesF . mconcat <$> manyTill inline newline
+ raw <- rawTitleBlockLine
+ res <- parseFromString (many inline) raw
+ return $ trimInlinesF $ mconcat res
titleBlock :: MarkdownParser ()
titleBlock = pandocTitleBlock <|> mmdTitleBlock
@@ -1354,16 +1358,18 @@ pipeTable = try $ do
nonindentSpaces
lookAhead nonspaceChar
(heads,(aligns, seplengths)) <- (,) <$> pipeTableRow <*> pipeBreak
+ let heads' = take (length aligns) <$> heads
lines' <- many pipeTableRow
+ let lines'' = map (take (length aligns) <$>) lines'
let maxlength = maximum $
- map (\x -> length . stringify $ runF x def) (heads : lines')
+ map (\x -> length . stringify $ runF x def) (heads' : lines'')
numColumns <- getOption readerColumns
let widths = if maxlength > numColumns
then map (\len ->
fromIntegral (len + 1) / fromIntegral numColumns)
seplengths
else replicate (length aligns) 0.0
- return $ (aligns, widths, heads, sequence lines')
+ return $ (aligns, widths, heads', sequence lines'')
sepPipe :: MarkdownParser ()
sepPipe = try $ do
@@ -1372,25 +1378,27 @@ sepPipe = try $ do
-- parse a row, also returning probable alignments for org-table cells
pipeTableRow :: MarkdownParser (F [Blocks])
-pipeTableRow = do
+pipeTableRow = try $ do
+ scanForPipe
skipMany spaceChar
openPipe <- (True <$ char '|') <|> return False
- let cell = mconcat <$>
- many (notFollowedBy (blankline <|> char '|') >> inline)
- first <- cell
- rest <- many $ sepPipe *> cell
+ -- split into cells
+ let chunk = void (code <|> rawHtmlInline <|> escapedChar <|> rawLaTeXInline')
+ <|> void (noneOf "|\n\r")
+ let cellContents = ((trim . snd) <$> withRaw (many chunk)) >>=
+ parseFromString pipeTableCell
+ cells <- cellContents `sepEndBy1` (char '|')
-- surrounding pipes needed for a one-column table:
- guard $ not (null rest && not openPipe)
- optional (char '|')
+ guard $ not (length cells == 1 && not openPipe)
blankline
- let cells = sequence (first:rest)
- return $ do
- cells' <- cells
- return $ map
- (\ils ->
- case trimInlines ils of
- ils' | B.isNull ils' -> mempty
- | otherwise -> B.plain $ ils') cells'
+ return $ sequence cells
+
+pipeTableCell :: MarkdownParser (F Blocks)
+pipeTableCell = do
+ result <- many inline
+ if null result
+ then return mempty
+ else return $ B.plain . mconcat <$> sequence result
pipeTableHeaderPart :: Parser [Char] st (Alignment, Int)
pipeTableHeaderPart = try $ do
diff --git a/src/Text/Pandoc/Readers/MediaWiki.hs b/src/Text/Pandoc/Readers/MediaWiki.hs
index 950497992..d3cee08e2 100644
--- a/src/Text/Pandoc/Readers/MediaWiki.hs
+++ b/src/Text/Pandoc/Readers/MediaWiki.hs
@@ -225,7 +225,7 @@ table = do
Nothing -> 1.0
caption <- option mempty tableCaption
optional rowsep
- hasheader <- option False $ True <$ (lookAhead (char '!'))
+ hasheader <- option False $ True <$ (lookAhead (skipSpaces *> char '!'))
(cellspecs',hdr) <- unzip <$> tableRow
let widths = map ((tableWidth *) . snd) cellspecs'
let restwidth = tableWidth - sum widths
diff --git a/src/Text/Pandoc/Readers/Odt.hs b/src/Text/Pandoc/Readers/Odt.hs
index a925c1d84..68e89263c 100644
--- a/src/Text/Pandoc/Readers/Odt.hs
+++ b/src/Text/Pandoc/Readers/Odt.hs
@@ -59,7 +59,9 @@ readOdt _ bytes = case bytesToOdt bytes of
--
bytesToOdt :: B.ByteString -> Either PandocError Pandoc
-bytesToOdt bytes = archiveToOdt $ toArchive bytes
+bytesToOdt bytes = case toArchiveOrFail bytes of
+ Right archive -> archiveToOdt archive
+ Left _ -> Left $ ParseFailure "Couldn't parse odt file."
--
archiveToOdt :: Archive -> Either PandocError Pandoc
diff --git a/src/Text/Pandoc/Readers/Org.hs b/src/Text/Pandoc/Readers/Org.hs
index 7dd611be3..5a50a8f34 100644
--- a/src/Text/Pandoc/Readers/Org.hs
+++ b/src/Text/Pandoc/Readers/Org.hs
@@ -35,6 +35,7 @@ import Text.Pandoc.Builder ( Inlines, Blocks, HasMeta(..),
trimInlines )
import Text.Pandoc.Definition
import Text.Pandoc.Compat.Monoid ((<>))
+import Text.Pandoc.Error
import Text.Pandoc.Options
import qualified Text.Pandoc.Parsing as P
import Text.Pandoc.Parsing hiding ( F, unF, askF, asksF, runF
@@ -49,7 +50,7 @@ import qualified Text.TeXMath.Readers.MathML.EntityMap as MathMLEntityMap
import Control.Arrow (first)
import Control.Monad (foldM, guard, liftM, liftM2, mplus, mzero, when)
import Control.Monad.Reader (Reader, runReader, ask, asks, local)
-import Data.Char (isAlphaNum, toLower)
+import Data.Char (isAlphaNum, isSpace, toLower)
import Data.Default
import Data.List (intersperse, isPrefixOf, isSuffixOf)
import qualified Data.Map as M
@@ -57,8 +58,6 @@ import qualified Data.Set as Set
import Data.Maybe (fromMaybe, isJust)
import Network.HTTP (urlEncode)
-import Text.Pandoc.Error
-
-- | Parse org-mode string and return a Pandoc document.
readOrg :: ReaderOptions -- ^ Reader options
-> String -- ^ String to parse (assuming @'\n'@ line endings)
@@ -391,6 +390,9 @@ lookupBlockAttribute key =
type BlockProperties = (Int, String) -- (Indentation, Block-Type)
+updateIndent :: BlockProperties -> Int -> BlockProperties
+updateIndent (_, blkType) indent = (indent, blkType)
+
orgBlock :: OrgParser (F Blocks)
orgBlock = try $ do
blockProp@(_, blkType) <- blockHeaderStart
@@ -407,11 +409,23 @@ orgBlock = try $ do
_ -> withParsed (fmap $ divWithClass blkType)
blockHeaderStart :: OrgParser (Int, String)
-blockHeaderStart = try $ (,) <$> indent <*> blockType
+blockHeaderStart = try $ (,) <$> indentation <*> blockType
where
- indent = length <$> many spaceChar
blockType = map toLower <$> (stringAnyCase "#+begin_" *> orgArgWord)
+indentation :: OrgParser Int
+indentation = try $ do
+ tabStop <- getOption readerTabStop
+ s <- many spaceChar
+ return $ spaceLength tabStop s
+
+spaceLength :: Int -> String -> Int
+spaceLength tabStop s = (sum . map charLen) s
+ where
+ charLen ' ' = 1
+ charLen '\t' = tabStop
+ charLen _ = 0
+
withRaw' :: (String -> F Blocks) -> BlockProperties -> OrgParser (F Blocks)
withRaw' f blockProp = (ignHeaders *> (f <$> rawBlockContent blockProp))
@@ -450,7 +464,8 @@ codeBlock blkProp = do
skipSpaces
(classes, kv) <- codeHeaderArgs <|> (mempty <$ ignHeaders)
id' <- fromMaybe "" <$> lookupBlockAttribute "name"
- content <- rawBlockContent blkProp
+ leadingIndent <- lookAhead indentation
+ content <- rawBlockContent (updateIndent blkProp leadingIndent)
resultsContent <- followingResultsBlock
let includeCode = exportsCode kv
let includeResults = exportsResults kv
@@ -472,7 +487,7 @@ rawBlockContent (indent, blockType) = try $
unlines . map commaEscaped <$> manyTill indentedLine blockEnder
where
indentedLine = try $ ("" <$ blankline) <|> (indentWith indent *> anyLine)
- blockEnder = try $ indentWith indent *> stringAnyCase ("#+end_" <> blockType)
+ blockEnder = try $ skipSpaces *> stringAnyCase ("#+end_" <> blockType)
parsedBlockContent :: BlockProperties -> OrgParser (F Blocks)
parsedBlockContent blkProps = try $ do
@@ -758,9 +773,13 @@ data OrgTableRow = OrgContentRow (F [Blocks])
| OrgAlignRow [Alignment]
| OrgHlineRow
+-- OrgTable is strongly related to the pandoc table ADT. Using the same
+-- (i.e. pandoc-global) ADT would mean that the reader would break if the
+-- global structure was to be changed, which would be bad. The final table
+-- should be generated using a builder function. Column widths aren't
+-- implemented yet, so they are not tracked here.
data OrgTable = OrgTable
- { orgTableColumns :: Int
- , orgTableAlignments :: [Alignment]
+ { orgTableAlignments :: [Alignment]
, orgTableHeader :: [Blocks]
, orgTableRows :: [[Blocks]]
}
@@ -776,7 +795,7 @@ table = try $ do
orgToPandocTable :: OrgTable
-> Inlines
-> Blocks
-orgToPandocTable (OrgTable _ aligns heads lns) caption =
+orgToPandocTable (OrgTable aligns heads lns) caption =
B.table caption (zip aligns $ repeat 0) heads lns
tableStart :: OrgParser Char
@@ -787,18 +806,19 @@ tableRows = try $ many (tableAlignRow <|> tableHline <|> tableContentRow)
tableContentRow :: OrgParser OrgTableRow
tableContentRow = try $
- OrgContentRow . sequence <$> (tableStart *> manyTill tableContentCell newline)
+ OrgContentRow . sequence <$> (tableStart *> many1Till tableContentCell newline)
tableContentCell :: OrgParser (F Blocks)
tableContentCell = try $
- fmap B.plain . trimInlinesF . mconcat <$> many1Till inline endOfCell
-
-endOfCell :: OrgParser Char
-endOfCell = try $ char '|' <|> lookAhead newline
+ fmap B.plain . trimInlinesF . mconcat <$> manyTill inline endOfCell
tableAlignRow :: OrgParser OrgTableRow
-tableAlignRow = try $
- OrgAlignRow <$> (tableStart *> manyTill tableAlignCell newline)
+tableAlignRow = try $ do
+ tableStart
+ cells <- many1Till tableAlignCell newline
+ -- Empty rows are regular (i.e. content) rows, not alignment rows.
+ guard $ any (/= AlignDefault) cells
+ return $ OrgAlignRow cells
tableAlignCell :: OrgParser Alignment
tableAlignCell =
@@ -813,65 +833,61 @@ tableAlignCell =
where emptyCell = try $ skipSpaces *> endOfCell
tableAlignFromChar :: OrgParser Alignment
-tableAlignFromChar = try $ choice [ char 'l' *> return AlignLeft
- , char 'c' *> return AlignCenter
- , char 'r' *> return AlignRight
- ]
+tableAlignFromChar = try $
+ choice [ char 'l' *> return AlignLeft
+ , char 'c' *> return AlignCenter
+ , char 'r' *> return AlignRight
+ ]
tableHline :: OrgParser OrgTableRow
tableHline = try $
OrgHlineRow <$ (tableStart *> char '-' *> anyLine)
+endOfCell :: OrgParser Char
+endOfCell = try $ char '|' <|> lookAhead newline
+
rowsToTable :: [OrgTableRow]
-> F OrgTable
-rowsToTable = foldM (flip rowToContent) zeroTable
- where zeroTable = OrgTable 0 mempty mempty mempty
-
-normalizeTable :: OrgTable
- -> OrgTable
-normalizeTable (OrgTable cols aligns heads lns) =
- let aligns' = fillColumns aligns AlignDefault
- heads' = if heads == mempty
- then mempty
- else fillColumns heads (B.plain mempty)
- lns' = map (`fillColumns` B.plain mempty) lns
- fillColumns base padding = take cols $ base ++ repeat padding
- in OrgTable cols aligns' heads' lns'
+rowsToTable = foldM rowToContent emptyTable
+ where emptyTable = OrgTable mempty mempty mempty
+normalizeTable :: OrgTable -> OrgTable
+normalizeTable (OrgTable aligns heads rows) = OrgTable aligns' heads rows
+ where
+ refRow = if heads /= mempty
+ then heads
+ else if rows == mempty then mempty else head rows
+ cols = length refRow
+ fillColumns base padding = take cols $ base ++ repeat padding
+ aligns' = fillColumns aligns AlignDefault
-- One or more horizontal rules after the first content line mark the previous
-- line as a header. All other horizontal lines are discarded.
-rowToContent :: OrgTableRow
- -> OrgTable
- -> F OrgTable
-rowToContent OrgHlineRow t = maybeBodyToHeader t
-rowToContent (OrgAlignRow as) t = setLongestRow as =<< setAligns as t
-rowToContent (OrgContentRow rf) t = do
- rs <- rf
- setLongestRow rs =<< appendToBody rs t
-
-setLongestRow :: [a]
- -> OrgTable
- -> F OrgTable
-setLongestRow rs t =
- return t{ orgTableColumns = max (length rs) (orgTableColumns t) }
-
-maybeBodyToHeader :: OrgTable
- -> F OrgTable
-maybeBodyToHeader t = case t of
- OrgTable{ orgTableHeader = [], orgTableRows = b:[] } ->
- return t{ orgTableHeader = b , orgTableRows = [] }
- _ -> return t
-
-appendToBody :: [Blocks]
- -> OrgTable
+rowToContent :: OrgTable
+ -> OrgTableRow
-> F OrgTable
-appendToBody r t = return t{ orgTableRows = orgTableRows t ++ [r] }
+rowToContent orgTable row =
+ case row of
+ OrgHlineRow -> return singleRowPromotedToHeader
+ OrgAlignRow as -> return . setAligns $ as
+ OrgContentRow cs -> appendToBody cs
+ where
+ singleRowPromotedToHeader :: OrgTable
+ singleRowPromotedToHeader = case orgTable of
+ OrgTable{ orgTableHeader = [], orgTableRows = b:[] } ->
+ orgTable{ orgTableHeader = b , orgTableRows = [] }
+ _ -> orgTable
-setAligns :: [Alignment]
- -> OrgTable
- -> F OrgTable
-setAligns aligns t = return $ t{ orgTableAlignments = aligns }
+ setAligns :: [Alignment] -> OrgTable
+ setAligns aligns = orgTable{ orgTableAlignments = aligns }
+
+ appendToBody :: F [Blocks] -> F OrgTable
+ appendToBody frow = do
+ newRow <- frow
+ let oldRows = orgTableRows orgTable
+ -- NOTE: This is an inefficient O(n) operation. This should be changed
+ -- if performance ever becomes a problem.
+ return orgTable{ orgTableRows = oldRows ++ [newRow] }
--
@@ -1565,14 +1581,14 @@ inlineLaTeX = try $ do
parseAsMathMLSym :: String -> Maybe Inlines
parseAsMathMLSym cs = B.str <$> MathMLEntityMap.getUnicode (clean cs)
- -- dropWhileEnd would be nice here, but it's not available before base 4.5
- where clean = reverse . dropWhile (`elem` ("{}" :: String)) . reverse . drop 1
+ -- drop initial backslash and any trailing "{}"
+ where clean = dropWhileEnd (`elem` ("{}" :: String)) . drop 1
state :: ParserState
state = def{ stateOptions = def{ readerParseRaw = True }}
- texMathToPandoc inp = (maybeRight $ readTeX inp) >>=
- writePandoc DisplayInline
+ texMathToPandoc :: String -> Maybe [Inline]
+ texMathToPandoc cs = (maybeRight $ readTeX cs) >>= writePandoc DisplayInline
maybeRight :: Either a b -> Maybe b
maybeRight = either (const Nothing) Just
@@ -1582,11 +1598,18 @@ inlineLaTeXCommand = try $ do
rest <- getInput
case runParser rawLaTeXInline def "source" rest of
Right (RawInline _ cs) -> do
- let len = length cs
+ -- drop any trailing whitespace, those are not be part of the command as
+ -- far as org mode is concerned.
+ let cmdNoSpc = dropWhileEnd isSpace cs
+ let len = length cmdNoSpc
count len anyChar
- return cs
+ return cmdNoSpc
_ -> mzero
+-- Taken from Data.OldList.
+dropWhileEnd :: (a -> Bool) -> [a] -> [a]
+dropWhileEnd p = foldr (\x xs -> if p x && null xs then [] else x : xs) []
+
smart :: OrgParser (F Inlines)
smart = do
getOption readerSmart >>= guard
diff --git a/src/Text/Pandoc/SelfContained.hs b/src/Text/Pandoc/SelfContained.hs
index 390a7a21a..d08d636df 100644
--- a/src/Text/Pandoc/SelfContained.hs
+++ b/src/Text/Pandoc/SelfContained.hs
@@ -1,6 +1,6 @@
{-# LANGUAGE OverloadedStrings #-}
{-
-Copyright (C) 2011-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2011-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -19,7 +19,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.SelfContained
- Copyright : Copyright (C) 2011-2015 John MacFarlane
+ Copyright : Copyright (C) 2011-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Shared.hs b/src/Text/Pandoc/Shared.hs
index 075d76847..d6b088338 100644
--- a/src/Text/Pandoc/Shared.hs
+++ b/src/Text/Pandoc/Shared.hs
@@ -2,7 +2,7 @@
FlexibleContexts, ScopedTypeVariables, PatternGuards,
ViewPatterns #-}
{-
-Copyright (C) 2006-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2006-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -21,7 +21,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Shared
- Copyright : Copyright (C) 2006-2015 John MacFarlane
+ Copyright : Copyright (C) 2006-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Slides.hs b/src/Text/Pandoc/Slides.hs
index 1a27ab5ac..e19dba3e2 100644
--- a/src/Text/Pandoc/Slides.hs
+++ b/src/Text/Pandoc/Slides.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2012-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2012-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Slides
- Copyright : Copyright (C) 2012-2015 John MacFarlane
+ Copyright : Copyright (C) 2012-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Templates.hs b/src/Text/Pandoc/Templates.hs
index a010433fa..925925872 100644
--- a/src/Text/Pandoc/Templates.hs
+++ b/src/Text/Pandoc/Templates.hs
@@ -1,7 +1,7 @@
{-# LANGUAGE TypeSynonymInstances, FlexibleInstances, CPP,
OverloadedStrings, GeneralizedNewtypeDeriving #-}
{-
-Copyright (C) 2009-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2009-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -20,7 +20,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.Templates
- Copyright : Copyright (C) 2009-2015 John MacFarlane
+ Copyright : Copyright (C) 2009-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/UTF8.hs b/src/Text/Pandoc/UTF8.hs
index de3314a0d..87ed5312b 100644
--- a/src/Text/Pandoc/UTF8.hs
+++ b/src/Text/Pandoc/UTF8.hs
@@ -1,6 +1,6 @@
{-# LANGUAGE CPP #-}
{-
-Copyright (C) 2010-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2010-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -19,7 +19,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.UTF8
- Copyright : Copyright (C) 2010-2015 John MacFarlane
+ Copyright : Copyright (C) 2010-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/UUID.hs b/src/Text/Pandoc/UUID.hs
index 463be044c..5d05fa303 100644
--- a/src/Text/Pandoc/UUID.hs
+++ b/src/Text/Pandoc/UUID.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2010-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2010-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.UUID
- Copyright : Copyright (C) 2010-2015 John MacFarlane
+ Copyright : Copyright (C) 2010-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>
diff --git a/src/Text/Pandoc/Writers/ConTeXt.hs b/src/Text/Pandoc/Writers/ConTeXt.hs
index 498e2d10f..8d54d62bd 100644
--- a/src/Text/Pandoc/Writers/ConTeXt.hs
+++ b/src/Text/Pandoc/Writers/ConTeXt.hs
@@ -279,7 +279,17 @@ blockListToConTeXt lst = liftM vcat $ mapM blockToConTeXt lst
-- | Convert list of inline elements to ConTeXt.
inlineListToConTeXt :: [Inline] -- ^ Inlines to convert
-> State WriterState Doc
-inlineListToConTeXt lst = liftM hcat $ mapM inlineToConTeXt lst
+inlineListToConTeXt lst = liftM hcat $ mapM inlineToConTeXt $ addStruts lst
+ -- We add a \strut after a line break that precedes a space,
+ -- or the space gets swallowed
+ where addStruts (LineBreak : s : xs) | isSpacey s =
+ LineBreak : RawInline (Format "context") "\\strut " : s :
+ addStruts xs
+ addStruts (x:xs) = x : addStruts xs
+ addStruts [] = []
+ isSpacey Space = True
+ isSpacey (Str ('\160':_)) = True
+ isSpacey _ = False
-- | Convert inline element to ConTeXt
inlineToConTeXt :: Inline -- ^ Inline to convert
diff --git a/src/Text/Pandoc/Writers/Docbook.hs b/src/Text/Pandoc/Writers/Docbook.hs
index 2aaebf99f..9acfe289a 100644
--- a/src/Text/Pandoc/Writers/Docbook.hs
+++ b/src/Text/Pandoc/Writers/Docbook.hs
@@ -112,10 +112,15 @@ elementToDocbook opts lvl (Sec _ _num (id',_,_) title elements) =
else elements
tag = case lvl of
n | n == 0 -> "chapter"
- | n >= 1 && n <= 5 -> "sect" ++ show n
+ | n >= 1 && n <= 5 -> if writerDocbook5 opts
+ then "section"
+ else "sect" ++ show n
| otherwise -> "simplesect"
- in inTags True tag [("id", writerIdentifierPrefix opts ++ id') |
- not (null id')] $
+ idAttr = [("id", writerIdentifierPrefix opts ++ id') | not (null id')]
+ nsAttr = if writerDocbook5 opts && lvl == 0 then [("xmlns", "http://docbook.org/ns/docbook")]
+ else []
+ attribs = nsAttr ++ idAttr
+ in inTags True tag attribs $
inTagsSimple "title" (inlinesToDocbook opts title) $$
vcat (map (elementToDocbook opts (lvl + 1)) elements')
@@ -227,9 +232,11 @@ blockToDocbook opts (OrderedList (start, numstyle, _) (first:rest)) =
blockToDocbook opts (DefinitionList lst) =
let attribs = [("spacing", "compact") | isTightList $ concatMap snd lst]
in inTags True "variablelist" attribs $ deflistItemsToDocbook opts lst
-blockToDocbook _ (RawBlock f str)
+blockToDocbook opts (RawBlock f str)
| f == "docbook" = text str -- raw XML block
- | f == "html" = text str -- allow html for backwards compatibility
+ | f == "html" = if writerDocbook5 opts
+ then empty -- No html in Docbook5
+ else text str -- allow html for backwards compatibility
| otherwise = empty
blockToDocbook _ HorizontalRule = empty -- not semantic
blockToDocbook opts (Table caption aligns widths headers rows) =
@@ -344,7 +351,9 @@ inlineToDocbook opts (Link attr txt (src, _))
| otherwise =
(if isPrefixOf "#" src
then inTags False "link" $ ("linkend", drop 1 src) : idAndRole attr
- else inTags False "ulink" $ ("url", src) : idAndRole attr ) $
+ else if writerDocbook5 opts
+ then inTags False "link" $ ("xlink:href", src) : idAndRole attr
+ else inTags False "ulink" $ ("url", src) : idAndRole attr ) $
inlinesToDocbook opts txt
inlineToDocbook opts (Image attr _ (src, tit)) =
let titleDoc = if null tit
diff --git a/src/Text/Pandoc/Writers/Docx.hs b/src/Text/Pandoc/Writers/Docx.hs
index 150e19043..a841e1b66 100644
--- a/src/Text/Pandoc/Writers/Docx.hs
+++ b/src/Text/Pandoc/Writers/Docx.hs
@@ -1103,7 +1103,7 @@ inlineToOpenXML opts (Link _ txt (src,_)) = do
M.insert src i extlinks }
return i
return [ mknode "w:hyperlink" [("r:id",id')] contents ]
-inlineToOpenXML opts (Image attr alt (src, tit)) = do
+inlineToOpenXML opts (Image attr alt (src, _)) = do
-- first, check to see if we've already done this image
pageWidth <- gets stPrintWidth
imgs <- gets stImages
@@ -1154,7 +1154,7 @@ inlineToOpenXML opts (Image attr alt (src, tit)) = do
mknode "wp:inline" []
[ mknode "wp:extent" [("cx",show xemu),("cy",show yemu)] ()
, mknode "wp:effectExtent" [("b","0"),("l","0"),("r","0"),("t","0")] ()
- , mknode "wp:docPr" [("descr",tit),("id","1"),("name","Picture")] ()
+ , mknode "wp:docPr" [("descr",stringify alt),("id","1"),("name","Picture")] ()
, graphic ]
let imgext = case mt >>= extensionFromMimeType of
Just x -> '.':x
diff --git a/src/Text/Pandoc/Writers/DokuWiki.hs b/src/Text/Pandoc/Writers/DokuWiki.hs
index f1088b158..56e2b9027 100644
--- a/src/Text/Pandoc/Writers/DokuWiki.hs
+++ b/src/Text/Pandoc/Writers/DokuWiki.hs
@@ -452,8 +452,11 @@ inlineToDokuWiki _ (Code _ str) =
inlineToDokuWiki _ (Str str) = return $ escapeString str
-inlineToDokuWiki _ (Math _ str) = return $ "$" ++ str ++ "$"
+inlineToDokuWiki _ (Math mathType str) = return $ delim ++ str ++ delim
-- note: str should NOT be escaped
+ where delim = case mathType of
+ DisplayMath -> "$$"
+ InlineMath -> "$"
inlineToDokuWiki _ (RawInline f str)
| f == Format "dokuwiki" = return str
diff --git a/src/Text/Pandoc/Writers/EPUB.hs b/src/Text/Pandoc/Writers/EPUB.hs
index 804dbb926..90f502f6f 100644
--- a/src/Text/Pandoc/Writers/EPUB.hs
+++ b/src/Text/Pandoc/Writers/EPUB.hs
@@ -667,7 +667,8 @@ writeEPUB opts doc@(Pandoc meta _) = do
]
]
else []
- let navData = renderHtml $ writeHtml opts'
+ let navData = renderHtml $ writeHtml
+ opts'{ writerVariables = ("navpage","true"):vars }
(Pandoc (setMeta "title"
(walk removeNote $ fromList $ docTitle' meta) nullMeta)
(navBlocks ++ landmarks))
diff --git a/src/Text/Pandoc/Writers/HTML.hs b/src/Text/Pandoc/Writers/HTML.hs
index c5b6a6db2..d8b8384e7 100644
--- a/src/Text/Pandoc/Writers/HTML.hs
+++ b/src/Text/Pandoc/Writers/HTML.hs
@@ -855,13 +855,12 @@ inlineToHtml opts inline =
(Note contents)
| writerIgnoreNotes opts -> return mempty
| otherwise -> do
- st <- get
- let notes = stNotes st
+ notes <- gets stNotes
let number = (length notes) + 1
let ref = show number
htmlContents <- blockListToNote opts ref contents
-- push contents onto front of notes
- put $ st {stNotes = (htmlContents:notes)}
+ modify $ \st -> st {stNotes = (htmlContents:notes)}
let revealSlash = ['/' | writerSlideVariant opts
== RevealJsSlides]
let link = H.a ! A.href (toValue $ "#" ++
diff --git a/src/Text/Pandoc/Writers/LaTeX.hs b/src/Text/Pandoc/Writers/LaTeX.hs
index 4e4279ec5..804e0febc 100644
--- a/src/Text/Pandoc/Writers/LaTeX.hs
+++ b/src/Text/Pandoc/Writers/LaTeX.hs
@@ -39,8 +39,10 @@ import Text.Pandoc.Templates
import Text.Printf ( printf )
import Network.URI ( isURI, unEscapeString )
import Data.Aeson (object, (.=), FromJSON)
-import Data.List ( (\\), isInfixOf, stripPrefix, intercalate, intersperse, nub, nubBy )
-import Data.Char ( toLower, isPunctuation, isAscii, isLetter, isDigit, ord )
+import Data.List ( (\\), isInfixOf, stripPrefix, intercalate, intersperse,
+ nub, nubBy, foldl' )
+import Data.Char ( toLower, isPunctuation, isAscii, isLetter, isDigit,
+ ord, isAlphaNum )
import Data.Maybe ( fromMaybe, isJust, catMaybes )
import qualified Data.Text as T
import Control.Applicative ((<|>))
@@ -223,7 +225,7 @@ pandocToLaTeX options (Pandoc meta blocks) = do
++ poly ++ "}{##2}}}\n"
else "\\newcommand{\\text" ++ poly ++ "}[2][]{\\foreignlanguage{"
++ babel ++ "}{#2}}\n" ++
- "\\newenvironment{" ++ poly ++ "}[1]{\\begin{otherlanguage}{"
+ "\\newenvironment{" ++ poly ++ "}[2][]{\\begin{otherlanguage}{"
++ babel ++ "}}{\\end{otherlanguage}}\n"
)
-- eliminate duplicates that have same polyglossia name
@@ -403,25 +405,28 @@ blockToLaTeX (Div (identifier,classes,kvs) bs) = do
blockToLaTeX (Plain lst) =
inlineListToLaTeX $ dropWhile isLineBreakOrSpace lst
-- title beginning with fig: indicates that the image is a figure
-blockToLaTeX (Para [Image attr txt (src,'f':'i':'g':':':tit)]) = do
+blockToLaTeX (Para [Image attr@(ident, _, _) txt (src,'f':'i':'g':':':tit)]) = do
inNote <- gets stInNote
modify $ \st -> st{ stInMinipage = True, stNotes = [] }
capt <- inlineListToLaTeX txt
notes <- gets stNotes
modify $ \st -> st{ stInMinipage = False, stNotes = [] }
+
-- We can't have footnotes in the list of figures, so remove them:
captForLof <- if null notes
then return empty
else brackets <$> inlineListToLaTeX (walk deNote txt)
img <- inlineToLaTeX (Image attr txt (src,tit))
let footnotes = notesToLaTeX notes
+ lab <- labelFor ident
+ let caption = "\\caption" <> captForLof <> braces capt <> lab
+ figure <- hypertarget ident (cr <>
+ "\\begin{figure}[htbp]" $$ "\\centering" $$ img $$
+ caption $$ "\\end{figure}" <> cr)
return $ if inNote
-- can't have figures in notes
then "\\begin{center}" $$ img $+$ capt $$ "\\end{center}"
- else "\\begin{figure}[htbp]" $$ "\\centering" $$ img $$
- ("\\caption" <> captForLof <> braces capt) $$
- "\\end{figure}" $$
- footnotes
+ else figure $$ footnotes
-- . . . indicates pause in beamer slides
blockToLaTeX (Para [Str ".",Space,Str ".",Space,Str "."]) = do
beamer <- writerBeamer `fmap` gets stOptions
@@ -468,23 +473,27 @@ blockToLaTeX (CodeBlock (identifier,classes,keyvalAttr) str) = do
st <- get
let params = if writerListings (stOptions st)
then (case getListingsLanguage classes of
- Just l -> [ "language=" ++ l ]
+ Just l -> [ "language=" ++ mbBraced l ]
Nothing -> []) ++
[ "numbers=left" | "numberLines" `elem` classes
|| "number" `elem` classes
|| "number-lines" `elem` classes ] ++
[ (if key == "startFrom"
then "firstnumber"
- else key) ++ "=" ++ attr |
+ else key) ++ "=" ++ mbBraced attr |
(key,attr) <- keyvalAttr ] ++
(if identifier == ""
then []
else [ "label=" ++ ref ])
else []
+ mbBraced x = if not (all isAlphaNum x)
+ then "{" <> x <> "}"
+ else x
printParams
| null params = empty
- | otherwise = brackets $ hcat (intersperse ", " (map text params))
+ | otherwise = brackets $ hcat (intersperse ", "
+ (map text params))
return $ flush ("\\begin{lstlisting}" <> printParams $$ text str $$
"\\end{lstlisting}") $$ cr
let highlightedCodeBlock =
@@ -505,7 +514,8 @@ blockToLaTeX (RawBlock f x)
blockToLaTeX (BulletList []) = return empty -- otherwise latex error
blockToLaTeX (BulletList lst) = do
incremental <- gets stIncremental
- let inc = if incremental then "[<+->]" else ""
+ beamer <- writerBeamer `fmap` gets stOptions
+ let inc = if beamer && incremental then "[<+->]" else ""
items <- mapM listItemToLaTeX lst
let spacing = if isTightList lst
then text "\\tightlist"
@@ -571,18 +581,21 @@ blockToLaTeX (Header level (id',classes,_) lst) = do
blockToLaTeX (Table caption aligns widths heads rows) = do
headers <- if all null heads
then return empty
- else ($$ "\\midrule\n") `fmap`
- (tableRowToLaTeX True aligns widths) heads
+ else do
+ contents <- (tableRowToLaTeX True aligns widths) heads
+ return ("\\toprule" $$ contents $$ "\\midrule")
let endhead = if all null heads
then empty
else text "\\endhead"
+ let endfirsthead = if all null heads
+ then empty
+ else text "\\endfirsthead"
captionText <- inlineListToLaTeX caption
let capt = if isEmpty captionText
then empty
- else text "\\caption" <> braces captionText
- <> "\\tabularnewline\n\\toprule\n"
- <> headers
- <> "\\endfirsthead"
+ else text "\\caption" <> braces captionText <> "\\tabularnewline"
+ $$ headers
+ $$ endfirsthead
rows' <- mapM (tableRowToLaTeX False aligns widths) rows
let colDescriptors = text $ concat $ map toColDescriptor aligns
modify $ \s -> s{ stTable = True }
@@ -590,7 +603,7 @@ blockToLaTeX (Table caption aligns widths heads rows) = do
braces ("@{}" <> colDescriptors <> "@{}")
-- the @{} removes extra space at beginning and end
$$ capt
- $$ "\\toprule"
+ $$ (if all null heads then "\\toprule" else empty)
$$ headers
$$ endhead
$$ vcat rows'
@@ -662,8 +675,7 @@ tableCellToLaTeX header (width, align, blocks) = do
AlignDefault -> "\\raggedright"
return $ ("\\begin{minipage}" <> valign <>
braces (text (printf "%.2f\\columnwidth" width)) <>
- (halign <> "\\strut" <> cr <> cellContents <> cr) <>
- "\\strut\\end{minipage}") $$
+ (halign <> cr <> cellContents <> cr) <> "\\end{minipage}") $$
notesToLaTeX notes
notesToLaTeX :: [Doc] -> Doc
@@ -712,10 +724,9 @@ sectionHeader :: Bool -- True for unnumbered
-> Int
-> [Inline]
-> State WriterState Doc
-sectionHeader unnumbered ref level lst = do
+sectionHeader unnumbered ident level lst = do
txt <- inlineListToLaTeX lst
- lab <- text `fmap` toLabel ref
- plain <- stringToLaTeX TextString $ foldl (++) "" $ map stringify lst
+ plain <- stringToLaTeX TextString $ concatMap stringify lst
let noNote (Note _) = Str ""
noNote x = x
let lstNoNotes = walk noNote lst
@@ -737,16 +748,6 @@ sectionHeader unnumbered ref level lst = do
book <- gets stBook
opts <- gets stOptions
let level' = if book || writerChapters opts then level - 1 else level
- internalLinks <- gets stInternalLinks
- let refLabel x = (if ref `elem` internalLinks
- then text "\\hypertarget"
- <> braces lab
- <> braces x
- else x)
- let headerWith x y = refLabel $ text x <> y <>
- if null ref
- then empty
- else text "\\label" <> braces lab
let sectionType = case level' of
0 | writerBeamer opts -> "part"
| otherwise -> "chapter"
@@ -762,16 +763,34 @@ sectionHeader unnumbered ref level lst = do
-- needed for \paragraph, \subparagraph in quote environment
-- see http://tex.stackexchange.com/questions/169830/
else empty
+ lab <- labelFor ident
+ stuffing' <- hypertarget ident $ text ('\\':sectionType) <> stuffing <> lab
return $ if level' > 5
then txt
- else prefix $$
- headerWith ('\\':sectionType) stuffing
+ else prefix $$ stuffing'
$$ if unnumbered
then "\\addcontentsline{toc}" <>
braces (text sectionType) <>
braces txtNoNotes
else empty
+hypertarget :: String -> Doc -> State WriterState Doc
+hypertarget ident x = do
+ ref <- text `fmap` toLabel ident
+ internalLinks <- gets stInternalLinks
+ return $
+ if ident `elem` internalLinks
+ then text "\\hypertarget"
+ <> braces ref
+ <> braces x
+ else x
+
+labelFor :: String -> State WriterState Doc
+labelFor "" = return empty
+labelFor ident = do
+ ref <- text `fmap` toLabel ident
+ return $ text "\\label" <> braces ref
+
-- | Convert list of inline elements to LaTeX.
inlineListToLaTeX :: [Inline] -- ^ Inlines to convert
-> State WriterState Doc
@@ -1019,7 +1038,7 @@ citationsToNatbib (c:cs) | citationMode c == AuthorInText = do
citationsToNatbib cits = do
cits' <- mapM convertOne cits
- return $ text "\\citetext{" <> foldl combineTwo empty cits' <> text "}"
+ return $ text "\\citetext{" <> foldl' combineTwo empty cits' <> text "}"
where
combineTwo a b | isEmpty a = b
| otherwise = a <> text "; " <> b
@@ -1068,7 +1087,7 @@ citationsToBiblatex (one:[])
citationsToBiblatex (c:cs) = do
args <- mapM convertOne (c:cs)
- return $ text cmd <> foldl (<>) empty args
+ return $ text cmd <> foldl' (<>) empty args
where
cmd = case citationMode c of
AuthorInText -> "\\textcites"
@@ -1112,7 +1131,7 @@ toPolyglossiaEnv l =
-- Takes a list of the constituents of a BCP 47 language code and
-- converts it to a Polyglossia (language, options) tuple
--- http://mirrors.concertpass.com/tex-archive/macros/latex/contrib/polyglossia/polyglossia.pdf
+-- http://mirrors.ctan.org/macros/latex/contrib/polyglossia/polyglossia.pdf
toPolyglossia :: [String] -> (String, String)
toPolyglossia ("ar":"DZ":_) = ("arabic", "locale=algeria")
toPolyglossia ("ar":"IQ":_) = ("arabic", "locale=mashriq")
@@ -1140,17 +1159,21 @@ toPolyglossia ("en":"UK":_) = ("english", "variant=british")
toPolyglossia ("en":"US":_) = ("english", "variant=american")
toPolyglossia ("grc":_) = ("greek", "variant=ancient")
toPolyglossia ("hsb":_) = ("usorbian", "")
+toPolyglossia ("la":"x":"classic":_) = ("latin", "variant=classic")
toPolyglossia ("sl":_) = ("slovenian", "")
toPolyglossia x = (commonFromBcp47 x, "")
-- Takes a list of the constituents of a BCP 47 language code and
-- converts it to a Babel language string.
--- http://mirrors.concertpass.com/tex-archive/macros/latex/required/babel/base/babel.pdf
--- Note that the PDF unfortunately does not contain a complete list of supported languages.
+-- http://mirrors.ctan.org/macros/latex/required/babel/base/babel.pdf
+-- List of supported languages (slightly outdated):
+-- http://tug.ctan.org/language/hyph-utf8/doc/generic/hyph-utf8/hyphenation.pdf
toBabel :: [String] -> String
toBabel ("de":"1901":_) = "german"
toBabel ("de":"AT":"1901":_) = "austrian"
toBabel ("de":"AT":_) = "naustrian"
+toBabel ("de":"CH":"1901":_) = "swissgerman"
+toBabel ("de":"CH":_) = "nswissgerman"
toBabel ("de":_) = "ngerman"
toBabel ("dsb":_) = "lowersorbian"
toBabel ("el":"polyton":_) = "polutonikogreek"
@@ -1164,6 +1187,7 @@ toBabel ("fr":"CA":_) = "canadien"
toBabel ("fra":"aca":_) = "acadian"
toBabel ("grc":_) = "polutonikogreek"
toBabel ("hsb":_) = "uppersorbian"
+toBabel ("la":"x":"classic":_) = "classiclatin"
toBabel ("sl":_) = "slovene"
toBabel x = commonFromBcp47 x
@@ -1172,12 +1196,15 @@ toBabel x = commonFromBcp47 x
-- https://tools.ietf.org/html/bcp47#section-2.1
commonFromBcp47 :: [String] -> String
commonFromBcp47 [] = ""
-commonFromBcp47 ("pt":"BR":_) = "brazilian"
+commonFromBcp47 ("pt":"BR":_) = "brazilian"
+commonFromBcp47 ("sr":"Cyrl":_) = "serbianc"
+commonFromBcp47 ("zh":"Latn":"pinyin":_) = "pinyin"
commonFromBcp47 x = fromIso $ head x
where
fromIso "af" = "afrikaans"
fromIso "am" = "amharic"
fromIso "ar" = "arabic"
+ fromIso "as" = "assamese"
fromIso "ast" = "asturian"
fromIso "bg" = "bulgarian"
fromIso "bn" = "bengali"
@@ -1201,12 +1228,13 @@ commonFromBcp47 x = fromIso $ head x
fromIso "fur" = "friulan"
fromIso "ga" = "irish"
fromIso "gd" = "scottish"
+ fromIso "gez" = "ethiopic"
fromIso "gl" = "galician"
fromIso "he" = "hebrew"
fromIso "hi" = "hindi"
fromIso "hr" = "croatian"
- fromIso "hy" = "armenian"
fromIso "hu" = "magyar"
+ fromIso "hy" = "armenian"
fromIso "ia" = "interlingua"
fromIso "id" = "indonesian"
fromIso "ie" = "interlingua"
@@ -1214,6 +1242,7 @@ commonFromBcp47 x = fromIso $ head x
fromIso "it" = "italian"
fromIso "jp" = "japanese"
fromIso "km" = "khmer"
+ fromIso "kmr" = "kurmanji"
fromIso "kn" = "kannada"
fromIso "ko" = "korean"
fromIso "la" = "latin"
@@ -1229,6 +1258,7 @@ commonFromBcp47 x = fromIso $ head x
fromIso "no" = "norsk"
fromIso "nqo" = "nko"
fromIso "oc" = "occitan"
+ fromIso "pa" = "panjabi"
fromIso "pl" = "polish"
fromIso "pms" = "piedmontese"
fromIso "pt" = "portuguese"
@@ -1245,6 +1275,7 @@ commonFromBcp47 x = fromIso $ head x
fromIso "ta" = "tamil"
fromIso "te" = "telugu"
fromIso "th" = "thai"
+ fromIso "ti" = "ethiopic"
fromIso "tk" = "turkmen"
fromIso "tr" = "turkish"
fromIso "uk" = "ukrainian"
diff --git a/src/Text/Pandoc/Writers/Org.hs b/src/Text/Pandoc/Writers/Org.hs
index 20086ed19..e57a6fc11 100644
--- a/src/Text/Pandoc/Writers/Org.hs
+++ b/src/Text/Pandoc/Writers/Org.hs
@@ -170,7 +170,7 @@ blockToOrg (Table caption' _ _ headers rows) = do
map ((+2) . numChars) $ transpose (headers' : rawRows)
-- FIXME: Org doesn't allow blocks with height more than 1.
let hpipeBlocks blocks = hcat [beg, middle, end]
- where h = maximum (map height blocks)
+ where h = maximum (1 : map height blocks)
sep' = lblock 3 $ vcat (map text $ replicate h " | ")
beg = lblock 2 $ vcat (map text $ replicate h "| ")
end = lblock 2 $ vcat (map text $ replicate h " |")
diff --git a/src/Text/Pandoc/XML.hs b/src/Text/Pandoc/XML.hs
index 70d8efba6..4cc2141b4 100644
--- a/src/Text/Pandoc/XML.hs
+++ b/src/Text/Pandoc/XML.hs
@@ -1,5 +1,5 @@
{-
-Copyright (C) 2006-2015 John MacFarlane <jgm@berkeley.edu>
+Copyright (C) 2006-2016 John MacFarlane <jgm@berkeley.edu>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -18,7 +18,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
{- |
Module : Text.Pandoc.XML
- Copyright : Copyright (C) 2006-2015 John MacFarlane
+ Copyright : Copyright (C) 2006-2016 John MacFarlane
License : GNU GPL, version 2 or above
Maintainer : John MacFarlane <jgm@berkeley.edu>