b5dd06d · typeable/pandoc

git.iko.soy/typeable/pandoc

Moved bibliography processing into readers.

Previously this was done in src/pandoc.hs, which made it difficult for library users. * Removed readerCitations in ReaderOptions. * Added readerReferences and readerCitationStyle to ReaderOptions. * Moved use of processBiblio from main program to the markdown and LaTeX readers.

commitb5dd06d30374edb7a2756ba58019d7f21f3bbfe5 authorJohn MacFarlane <fiddlosopher@gmail.com> dateSat 13 Oct 2012, 11:33 (14 years ago) parent9a77a3f

4 files changed+43−33

compare with master

unifiedsplit

src/Text/Pandoc/Options.hs+6−3 src/Text/Pandoc/Readers/LaTeX.hs+5−1 src/Text/Pandoc/Readers/Markdown.hs+8−3 src/pandoc.hs+24−26

src/Text/Pandoc/Options.hs+6−3

@@ -44,6 +44,7 @@ import Data.Set (Set)

4444import qualified Data.Set as Set

4545import Data.Default

4646import Text.Pandoc.Highlighting (Style, pygments)

47+import qualified Text.CSL as CSL

4748

4849-- | Individually selectable syntax extensions.

4950data Extension =

@@ -141,11 +142,12 @@ data ReaderOptions = ReaderOptions{

141142 , readerOldDashes :: Bool -- ^ Use pandoc <= 1.8.2.1 behavior

142143 -- in parsing dashes; -- is em-dash;

143144 -- - before numerial is en-dash

144− , readerCitations :: [String] -- ^ List of available citations

145+ , readerReferences :: [CSL.Reference] -- ^ Bibliographic references

146+ , readerCitationStyle :: Maybe CSL.Style -- ^ Citation style

145147 , readerApplyMacros :: Bool -- ^ Apply macros to TeX math

146148 , readerIndentedCodeClasses :: [String] -- ^ Default classes for

147149 -- indented code blocks

148−} deriving (Show, Read)

150+} deriving (Show {- TODO need update in citeproc-hs , Read -})

149151

150152instance Default ReaderOptions

151153 where def = ReaderOptions{

@@ -157,7 +159,8 @@ instance Default ReaderOptions

157159 , readerColumns = 80

158160 , readerTabStop = 4

159161 , readerOldDashes = False

160− , readerCitations = []

162+ , readerReferences = []

163+ , readerCitationStyle = Nothing

161164 , readerApplyMacros = True

162165 , readerIndentedCodeClasses = []

163166 }

src/Text/Pandoc/Readers/LaTeX.hs+5−1

@@ -36,6 +36,7 @@ module Text.Pandoc.Readers.LaTeX ( readLaTeX,

3636import Text.Pandoc.Definition

3737import Text.Pandoc.Shared

3838import Text.Pandoc.Options

39+import Text.Pandoc.Biblio (processBiblio)

3940import Text.Pandoc.Parsing hiding ((<|>), many, optional, space)

4041import qualified Text.Pandoc.UTF8 as UTF8

4142import Data.Char ( chr, ord )

@@ -63,7 +64,10 @@ parseLaTeX = do

6364 let title' = stateTitle st

6465 let authors' = stateAuthors st

6566 let date' = stateDate st

66− return $ Pandoc (Meta title' authors' date') $ toList bs

67+ refs <- getOption readerReferences

68+ mbsty <- getOption readerCitationStyle

69+ return $ processBiblio mbsty refs

70+ $ Pandoc (Meta title' authors' date') $ toList bs

6771

6872type LP = Parser [Char] ParserState

6973

src/Text/Pandoc/Readers/Markdown.hs+8−3

@@ -45,6 +45,8 @@ import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

4545import Text.Pandoc.Readers.HTML ( htmlTag, htmlInBalanced, isInlineTag, isBlockTag,

4646 isTextTag, isCommentTag )

4747import Text.Pandoc.XML ( fromEntities )

48+import Text.Pandoc.Biblio (processBiblio)

49+import qualified Text.CSL as CSL

4850import Data.Monoid (mconcat, mempty)

4951import Control.Applicative ((<$>), (<*), (*>), (<$))

5052import Control.Monad

@@ -216,7 +218,10 @@ parseMarkdown = do

216218 (title, authors, date) <- option (mempty,return [],mempty) titleBlock

217219 blocks <- parseBlocks

218220 st <- getState

219− return $ B.setTitle (runF title st)

221+ mbsty <- getOption readerCitationStyle

222+ refs <- getOption readerReferences

223+ return $ processBiblio mbsty refs

224+ $ B.setTitle (runF title st)

220225 $ B.setAuthors (runF authors st)

221226 $ B.setDate (runF date st)

222227 $ B.doc $ runF blocks st

@@ -1542,7 +1547,7 @@ rawHtmlInline = do

15421547cite :: Parser [Char] ParserState (F Inlines)

15431548cite = do

15441549 guardEnabled Ext_citations

1545− getOption readerCitations >>= guard . not . null

1550+ getOption readerReferences >>= guard . not . null

15461551 citations <- textualCite <|> normalCite

15471552 return $ flip B.cite mempty <$> citations

15481553

@@ -1591,7 +1596,7 @@ citeKey = try $ do

15911596 let internal p = try $ p >>~ lookAhead (letter <|> digit)

15921597 rest <- many $ letter <|> digit <|> internal (oneOf ":.#$%&-_?<>~")

15931598 let key = first:rest

1594− citations' <- getOption readerCitations

1599+ citations' <- map CSL.refId <$> getOption readerReferences

15951600 guard $ key `elem` citations'

15961601 return (suppress_author, key)

15971602

src/pandoc.hs+24−26

@@ -52,7 +52,6 @@ import qualified Control.Exception as E

5252import Control.Exception.Extensible ( throwIO )

5353import qualified Text.Pandoc.UTF8 as UTF8

5454import qualified Text.CSL as CSL

55−import Text.Pandoc.Biblio

5655import Control.Monad (when, unless, liftM)

5756import Network.HTTP (simpleHTTP, mkRequest, getResponseBody, RequestMethod(..))

5857import Network.URI (parseURI, isURI, URI(..))

@@ -924,6 +923,24 @@ main = do

924923 reffiles >>=

925924 return . map unescapeRefId . concat

926925

926+ mbsty <- if citeMethod == Citeproc && not (null refs)

927+ then do

928+ csl <- CSL.parseCSL =<<

929+ case mbCsl of

930+ Nothing -> readDataFile datadir "default.csl"

931+ Just cslfile -> do

932+ exists <- doesFileExist cslfile

933+ if exists

934+ then UTF8.readFile cslfile

935+ else do

936+ csldir <- getAppUserDataDirectory "csl"

937+ print csldir

938+ readDataFile datadir (replaceDirectory

939+ (replaceExtension cslfile "csl") csldir)

940+ abbrevs <- maybe (return []) CSL.readJsonAbbrevFile cslabbrevs

941+ return $ Just csl { CSL.styleAbbrevs = abbrevs }

942+ else return Nothing

943+

927944 let sourceDir = if null sources

928945 then "."

929946 else takeDirectory (head sources)

@@ -935,7 +952,8 @@ main = do

935952 , readerColumns = columns

936953 , readerTabStop = tabStop

937954 , readerOldDashes = oldDashes

938− , readerCitations = map CSL.refId refs

955+ , readerReferences = refs

956+ , readerCitationStyle = mbsty

939957 , readerIndentedCodeClasses = codeBlockClasses

940958 , readerApplyMacros = not laTeXOutput

941959 }

@@ -1004,26 +1022,6 @@ main = do

10041022 then bottomUpM rtfEmbedImage doc0

10051023 else return doc0

10061024

1007− doc2 <- do

1008− if citeMethod == Citeproc && not (null refs)

1009− then do

1010− csl <- CSL.parseCSL =<<

1011− case mbCsl of

1012− Nothing -> readDataFile datadir "default.csl"

1013− Just cslfile -> do

1014− exists <- doesFileExist cslfile

1015− if exists

1016− then UTF8.readFile cslfile

1017− else do

1018− csldir <- getAppUserDataDirectory "csl"

1019− print csldir

1020− readDataFile datadir (replaceDirectory

1021− (replaceExtension cslfile "csl") csldir)

1022− abbrevs <- maybe (return []) CSL.readJsonAbbrevFile cslabbrevs

1023− let csl' = csl { CSL.styleAbbrevs = abbrevs }

1024− return $ processBiblio csl' refs doc1

1025− else return doc1

1026−

10271025 let writeBinary :: B.ByteString -> IO ()

10281026 writeBinary = B.writeFile (UTF8.encodePath outputFile)

10291027

@@ -1033,15 +1031,15 @@ main = do

10331031

10341032 case getWriter writerName' of

10351033 Left e -> err 9 e

1036− Right (IOStringWriter f) -> f writerOptions doc2 >>= writerFn outputFile

1037− Right (IOByteStringWriter f) -> f writerOptions doc2 >>= writeBinary

1034+ Right (IOStringWriter f) -> f writerOptions doc1 >>= writerFn outputFile

1035+ Right (IOByteStringWriter f) -> f writerOptions doc1 >>= writeBinary

10381036 Right (PureStringWriter f)

10391037 | pdfOutput -> do

1040− res <- tex2pdf latexEngine $ f writerOptions doc2

1038+ res <- tex2pdf latexEngine $ f writerOptions doc1

10411039 case res of

10421040 Right pdf -> writeBinary pdf

10431041 Left err' -> err 43 $ UTF8.toStringLazy err'

1044− | otherwise -> selfcontain (f writerOptions doc2 ++

1042+ | otherwise -> selfcontain (f writerOptions doc1 ++

10451043 ['\n' | not standalone'])

10461044 >>= writerFn outputFile . handleEntities

10471045 where htmlFormat = writerName' `elem`