From 2327e302cb08a34a7326a89c35a12362e016d4b2 Mon Sep 17 00:00:00 2001 From: Tuong Nguyen Manh Date: Fri, 2 Jan 2026 17:43:15 +0100 Subject: RTF reader: Improve hyperlink parsing more (#11370) Both the field instruction and its result may be ungrouped. Closes #10942. --- src/Text/Pandoc/Readers/RTF.hs | 25 ++++++++++++++++--------- 1 file changed, 16 insertions(+), 9 deletions(-) (limited to 'src') diff --git a/src/Text/Pandoc/Readers/RTF.hs b/src/Text/Pandoc/Readers/RTF.hs index fde444854..cb6b99e2b 100644 --- a/src/Text/Pandoc/Readers/RTF.hs +++ b/src/Text/Pandoc/Readers/RTF.hs @@ -790,22 +790,29 @@ handleField bs ts = do let isFieldMod (Tok _ (ControlWord w _)) = w `elem` ["flddirty", "fldedit", "fldlock", "fldpriv"] isFieldMod _ = False + + let instructionTokens (Tok _ (Grouped toks)) = Just toks + instructionTokens unformattedTok@(Tok _ (UnformattedText _)) = Just [unformattedTok] + instructionTokens _ = Nothing case dropWhile isFieldMod ts of [Tok _ (Grouped (Tok _ (ControlSymbol '*') :Tok _ (ControlWord "fldinst" Nothing) - :Tok _ (Grouped instrtoks) + :instrtoks :_)), Tok _ (Grouped (Tok _ (ControlWord "fldrslt" Nothing) - :Tok _ (Grouped resulttoks) : _))] -> do - case getHyperlink instrtoks of - Just linkdest -> do - modifyGroup $ \g -> g{ gHyperlink = Just linkdest } - result <- foldM processTok bs resulttoks - modifyGroup $ \g -> g{ gHyperlink = Nothing } - return result - Nothing -> foldM processTok bs resulttoks + :resulttoks))] -> do + case instructionTokens instrtoks of + Nothing -> pure bs + Just instrtoks' -> + case getHyperlink instrtoks' of + Just linkdest -> do + modifyGroup $ \g -> g{ gHyperlink = Just linkdest } + result <- foldM processTok bs resulttoks + modifyGroup $ \g -> g{ gHyperlink = Nothing } + return result + Nothing -> foldM processTok bs resulttoks _ -> pure bs getHyperlink :: [Tok] -> Maybe Text -- cgit v1.2.3