git.iko.soy/typeable/pandoc
on master help sign in
Docx reader: Make metavalues out of styled paragraphs.

This will make paragraphs styled with `Author`, `Title`, `Subtitle`, `Date`, and `Abstract` into pandoc metavalues, rather than text. The implementation only takes those elements from the beginning of the document (ignoring empty paragraphs). Multiple paragraphs in the `Author` style will be made into a metaList, one paragraph per item. Hard linebreaks (shift-return) in the paragraph will be maintained, and can be used for institution, email, etc.

commit840108a9c11850089a27a3b5458f8561ab1b6a2e authorJesse Rosenthal <jrosenthal@jhu.edu> dateSun 27 Jul 2014, 15:11 (12 years ago) parent3eff378
8 files changed+99−11
compare with master
unifiedsplit
src/Text/Pandoc/Readers/Docx.hs+73−11 tests/Tests/Readers/Docx.hs+15 tests/docx.metadata.docx tests/docx.metadata.native+2 tests/docx.metadata_after_normal.docx tests/docx.metadata_after_normal.native+7 tests/docx.metadata_author_linebreak.docx tests/docx.metadata_author_linebreak.native+2
src/Text/Pandoc/Readers/Docx.hs+73−11
@@ -87,7 +87,7 @@ import Text.Pandoc.Readers.Docx.Reducible
8787import Text.Pandoc.Readers.Docx.TexChar
8888import Text.Pandoc.Shared
8989import Data.Maybe (mapMaybe, fromMaybe)
90−import Data.List (delete, isPrefixOf, (\\), intercalate)
90+import Data.List (delete, isPrefixOf, (\\), intercalate, intersect)
9191import qualified Data.ByteString as BS
9292import qualified Data.ByteString.Lazy as B
9393import Data.ByteString.Base64 (encode)
@@ -101,7 +101,8 @@ readDocx :: ReaderOptions
101101 -> Pandoc
102102readDocx opts bytes =
103103 case archiveToDocx (toArchive bytes) of
104− Right docx -> Pandoc nullMeta (docxToBlocks opts docx)
104+ Right docx -> Pandoc meta blks where
105+ (meta, blks) = (docxToMetaAndBlocks opts docx)
105106 Left _ -> error $ "couldn't parse docx file"
106107
107108data DState = DState { docxAnchorMap :: M.Map String String
@@ -134,6 +135,65 @@ spansToKeep = []
134135divsToKeep :: [String]
135136divsToKeep = ["list-item", "Definition", "DefinitionTerm"]
136137
138+metaStyles :: M.Map String String
139+metaStyles = M.fromList [ ("Title", "title")
140+ , ("Subtitle", "subtitle")
141+ , ("Author", "author")
142+ , ("Date", "date")
143+ , ("Abstract", "abstract")]
144+
145+sepBodyParts :: [BodyPart] -> ([BodyPart], [BodyPart])
146+sepBodyParts = span (\bp -> (isMetaPar bp || isEmptyPar bp))
147+
148+isMetaPar :: BodyPart -> Bool
149+isMetaPar (Paragraph pPr _) =
150+ not $ null $ intersect (pStyle pPr) (M.keys metaStyles)
151+isMetaPar _ = False
152+
153+isEmptyPar :: BodyPart -> Bool
154+isEmptyPar (Paragraph _ parParts) =
155+ all isEmptyParPart parParts
156+ where
157+ isEmptyParPart (PlainRun (Run _ runElems)) = all isEmptyElem runElems
158+ isEmptyParPart _ = False
159+ isEmptyElem (TextRun s) = trim s == ""
160+ isEmptyElem _ = True
161+isEmptyPar _ = False
162+
163+bodyPartsToMeta' :: [BodyPart] -> DocxContext (M.Map String MetaValue)
164+bodyPartsToMeta' [] = return M.empty
165+bodyPartsToMeta' (bp : bps)
166+ | (Paragraph pPr parParts) <- bp
167+ , (c : _)<- intersect (pStyle pPr) (M.keys metaStyles)
168+ , (Just metaField) <- M.lookup c metaStyles = do
169+ inlines <- parPartsToInlines parParts
170+ remaining <- bodyPartsToMeta' bps
171+ let
172+ f (MetaInlines ils) (MetaInlines ils') = MetaBlocks [Para ils, Para ils']
173+ f (MetaInlines ils) (MetaBlocks blks) = MetaBlocks ((Para ils) : blks)
174+ f m (MetaList mv) = MetaList (m : mv)
175+ f m n = MetaList [m, n]
176+ return $ M.insertWith f metaField (MetaInlines inlines) remaining
177+bodyPartsToMeta' (_ : bps) = bodyPartsToMeta' bps
178+
179+bodyPartsToMeta :: [BodyPart] -> DocxContext Meta
180+bodyPartsToMeta bps = do
181+ mp <- bodyPartsToMeta' bps
182+ let mp' =
183+ case M.lookup "author" mp of
184+ Just mv -> M.insert "author" (fixAuthors mv) mp
185+ Nothing -> mp
186+ return $ Meta mp'
187+
188+fixAuthors :: MetaValue -> MetaValue
189+fixAuthors (MetaBlocks blks) =
190+ MetaList $ map g $ filter f blks
191+ where f (Para _) = True
192+ f _ = False
193+ g (Para ils) = MetaInlines ils
194+ g _ = MetaInlines []
195+fixAuthors mv = mv
196+
137197runStyleToContainers :: RunStyle -> [Container Inline]
138198runStyleToContainers rPr =
139199 let spanClassToContainers :: String -> [Container Inline]
@@ -615,24 +675,26 @@ rewriteLink l@(Link ils ('#':target, title)) = do
615675 Nothing -> l
616676rewriteLink il = return il
617677
618−
619−bodyToBlocks :: Body -> DocxContext [Block]
620−bodyToBlocks (Body bps) = do
621− blks <- concatMapM bodyPartToBlocks bps >>=
678+bodyToMetaAndBlocks :: Body -> DocxContext (Meta, [Block])
679+bodyToMetaAndBlocks (Body bps) = do
680+ let (metabps, blkbps) = sepBodyParts bps
681+ meta <- bodyPartsToMeta metabps
682+ blks <- concatMapM bodyPartToBlocks blkbps >>=
622683 walkM rewriteLink
623684 return $
624− blocksToDefinitions $
625− blocksToBullets $ blks
685+ (meta,
686+ blocksToDefinitions $
687+ blocksToBullets $ blks)
626688
627−docxToBlocks :: ReaderOptions -> Docx -> [Block]
628−docxToBlocks opts d@(Docx (Document _ body)) =
689+docxToMetaAndBlocks :: ReaderOptions -> Docx -> (Meta, [Block])
690+docxToMetaAndBlocks opts d@(Docx (Document _ body)) =
629691 let dState = DState { docxAnchorMap = M.empty
630692 , docxInHeaderBlock = False
631693 , docxInTexSubscript = False}
632694 dEnv = DEnv { docxOptions = opts
633695 , docxDocument = d}
634696 in
635− evalDocxContext (bodyToBlocks body) dEnv dState
697+ evalDocxContext (bodyToMetaAndBlocks body) dEnv dState
636698
637699ilToCode :: Inline -> String
638700ilToCode (Str s) = s
tests/Tests/Readers/Docx.hs+15
@@ -164,5 +164,20 @@ tests = [ testGroup "inlines"
164164 "docx.track_changes_deletion.docx"
165165 "docx.track_changes_deletion_all.native"
166166 ]
167+ , testGroup "metadata"
168+ [ testCompareWithOpts def{readerStandalone=True}
169+ "metadata fields"
170+ "docx.metadata.docx"
171+ "docx.metadata.native"
172+ , testCompareWithOpts def{readerStandalone=True}
173+ "linebreak between authors"
174+ "docx.metadata_author_linebreak.docx"
175+ "docx.metadata_author_linebreak.native"
176+ , testCompareWithOpts def{readerStandalone=True}
177+ "stop recording metadata with normal text"
178+ "docx.metadata_after_normal.docx"
179+ "docx.metadata_after_normal.native"
180+ ]
181+
167182 ]
168183
tests/docx.metadata.docxnew file

Binary file not shown.

tests/docx.metadata.nativenew file+2
@@ -0,0 +1,2 @@
1+Pandoc (Meta {unMeta = fromList [("abstract",MetaInlines [Str "This",Space,Str "is",Space,Str "a",Space,Str "test",Space,Str "of",Space,Str "how",Space,Str "this",Space,Str "all",Space,Str "works.",Space,Str "I\8217ve",Space,Str "skipped",Space,Str "lines",Space,Str "here,",Space,Str "which",Space,Str "pandoc",Space,Str "doesn\8217t",Space,Str "do,",Space,Str "but",Space,Str "which",Space,Str "shouldn\8217t",Space,Str "make",Space,Str "a",Space,Str "difference."]),("author",MetaList [MetaInlines [Str "Mary",Space,Str "Ann",Space,Str "Evans"],MetaInlines [Str "Aurore",Space,Str "Dupin"]]),("date",MetaInlines [Str "July",Space,Str "28,",Space,Str "2014"]),("title",MetaInlines [Str "This",Space,Str "Is",Space,Str "the",Space,Str "Title"])]})
2+[Para [Str "And",Space,Str "now",Space,Str "this",Space,Str "is",Space,Str "normal",Space,Str "text."]]
tests/docx.metadata_after_normal.docxnew file

Binary file not shown.

tests/docx.metadata_after_normal.nativenew file+7
@@ -0,0 +1,7 @@
1+Pandoc (Meta {unMeta = fromList [("abstract",MetaInlines [Str "This",Space,Str "is",Space,Str "a",Space,Str "test",Space,Str "of",Space,Str "how",Space,Str "this",Space,Str "all",Space,Str "works.",Space,Str "I\8217ve",Space,Str "skipped",Space,Str "lines",Space,Str "here,",Space,Str "which",Space,Str "pandoc",Space,Str "doesn\8217t",Space,Str "do,",Space,Str "but",Space,Str "which",Space,Str "shouldn\8217t",Space,Str "make",Space,Str "a",Space,Str "difference."]),("author",MetaList [MetaInlines [Str "Mary",Space,Str "Ann",Space,Str "Evans"],MetaInlines [Str "Aurore",Space,Str "Dupin"]]),("date",MetaInlines [Str "July",Space,Str "28,",Space,Str "2014"]),("title",MetaInlines [Str "This",Space,Str "Is",Space,Str "the",Space,Str "Title"])]})
2+[Para [Str "And",Space,Str "now",Space,Str "this",Space,Str "is",Space,Str "normal",Space,Str "text."]
3+,Para [Str "This",Space,Str "Is",Space,Str "the",Space,Str "Title"]
4+,Para [Str "Mary",Space,Str "Ann",Space,Str "Evans"]
5+,Para [Str "Aurore",Space,Str "Dupin"]
6+,Para [Str "July",Space,Str "28,",Space,Str "2014"]
7+,Para [Str "This",Space,Str "is",Space,Str "a",Space,Str "test",Space,Str "of",Space,Str "how",Space,Str "this",Space,Str "all",Space,Str "works.",Space,Str "I\8217ve",Space,Str "skipped",Space,Str "lines",Space,Str "here,",Space,Str "which",Space,Str "pandoc",Space,Str "doesn\8217t",Space,Str "do,",Space,Str "but",Space,Str "which",Space,Str "shouldn\8217t",Space,Str "make",Space,Str "a",Space,Str "difference."]]
tests/docx.metadata_author_linebreak.docxnew file

Binary file not shown.

tests/docx.metadata_author_linebreak.nativenew file+2
@@ -0,0 +1,2 @@
1+Pandoc (Meta {unMeta = fromList [("abstract",MetaInlines [Str "This",Space,Str "is",Space,Str "a",Space,Str "test",Space,Str "of",Space,Str "how",Space,Str "this",Space,Str "all",Space,Str "works.",Space,Str "I\8217ve",Space,Str "skipped",Space,Str "lines",Space,Str "here,",Space,Str "which",Space,Str "pandoc",Space,Str "doesn\8217t",Space,Str "do,",Space,Str "but",Space,Str "which",Space,Str "shouldn\8217t",Space,Str "make",Space,Str "a",Space,Str "difference."]),("author",MetaList [MetaInlines [Str "Mary",Space,Str "Ann",Space,Str "Evans"],MetaInlines [Str "Aurore",Space,Str "Dupin"]]),("date",MetaInlines [Str "July",Space,Str "28,",Space,Str "2014"]),("title",MetaInlines [Str "This",Space,Str "Is",Space,Str "the",Space,Str "Title"])]})
2+[Para [Str "And",Space,Str "now",Space,Str "this",Space,Str "is",Space,Str "normal",Space,Str "text."]]