diff options
Diffstat (limited to 'src/Text/Pandoc/Readers/EPUB.hs')
| -rw-r--r-- | src/Text/Pandoc/Readers/EPUB.hs | 17 |
1 files changed, 11 insertions, 6 deletions
diff --git a/src/Text/Pandoc/Readers/EPUB.hs b/src/Text/Pandoc/Readers/EPUB.hs index 5e3326e6d..369c4f0c9 100644 --- a/src/Text/Pandoc/Readers/EPUB.hs +++ b/src/Text/Pandoc/Readers/EPUB.hs @@ -17,7 +17,7 @@ module Text.Pandoc.Readers.EPUB (readEPUB) where -import Codec.Archive.Zip (Archive (..), Entry, findEntryByPath, fromEntry, +import Codec.Archive.Zip (Archive (..), Entry(..), findEntryByPath, fromEntry, toArchiveOrFail) import Control.DeepSeq (NFData, deepseq) import Control.Monad (guard, liftM, liftM2, mplus) @@ -41,9 +41,10 @@ import Text.Pandoc.MIME (MimeType) import Text.Pandoc.Options (ReaderOptions (..)) import Text.Pandoc.Readers.HTML (readHtml) import Text.Pandoc.Shared (addMetaField, collapseFilePath, escapeURI) -import qualified Text.Pandoc.UTF8 as UTF8 (toStringLazy) +import qualified Text.Pandoc.UTF8 as UTF8 (toTextLazy) import Text.Pandoc.Walk (query, walk) import Text.XML.Light +import Text.Pandoc.XMLParser (parseXMLElement) type Items = M.Map String (FilePath, MimeType) @@ -181,7 +182,7 @@ renameMeta s = T.pack s getManifest :: PandocMonad m => Archive -> m (String, Element) getManifest archive = do metaEntry <- findEntryByPathE ("META-INF" </> "container.xml") archive - docElem <- (parseXMLDocE . UTF8.toStringLazy . fromEntry) metaEntry + docElem <- parseXMLDocE metaEntry let namespaces = mapMaybe attrToNSPair (elAttribs docElem) ns <- mkE "xmlns not in namespaces" (lookup "xmlns" namespaces) as <- fmap (map attrToPair . elAttribs) @@ -190,7 +191,7 @@ getManifest archive = do let rootdir = dropFileName manifestFile --mime <- lookup "media-type" as manifest <- findEntryByPathE manifestFile archive - (rootdir,) <$> (parseXMLDocE . UTF8.toStringLazy . fromEntry $ manifest) + (rootdir,) <$> parseXMLDocE manifest -- Fixup @@ -284,8 +285,12 @@ findEntryByPathE :: PandocMonad m => FilePath -> Archive -> m Entry findEntryByPathE (normalise . unEscapeString -> path) a = mkE ("No entry on path: " ++ path) $ findEntryByPath path a -parseXMLDocE :: PandocMonad m => String -> m Element -parseXMLDocE doc = mkE "Unable to parse XML doc" $ parseXMLDoc doc +parseXMLDocE :: PandocMonad m => Entry -> m Element +parseXMLDocE entry = + either (throwError . PandocXMLError fp) return $ parseXMLElement doc + where + doc = UTF8.toTextLazy . fromEntry $ entry + fp = T.pack $ eRelativePath entry findElementE :: PandocMonad m => QName -> Element -> m Element findElementE e x = mkE ("Unable to find element: " ++ show e) $ findElement e x |