From 93d745742ddc06871ed7526e324f01506c7020e2 Mon Sep 17 00:00:00 2001
From: John MacFarlane <jgm@berkeley.edu>
Date: Thu, 3 Oct 2024 21:50:18 -0700
Subject: [PATCH] Docx reader: reset lists after headers in same list numId.

Headings in docx, even ones that do not have a visible number,
can have a numId, and in odd cases can even share a numId with
a list that continues after the header. In this case the list
numbering should be reset by the header.

To accomplish this, we add a Heading constructor to BodyPart and
include on it all the information list items have.

Closes #10258.
---
 src/Text/Pandoc/Readers/Docx.hs       | 40 +++++++++++++++++++--------
 src/Text/Pandoc/Readers/Docx/Parse.hs | 13 +++++++--
 2 files changed, 38 insertions(+), 15 deletions(-)

diff --git a/src/Text/Pandoc/Readers/Docx.hs b/src/Text/Pandoc/Readers/Docx.hs
index 4130a749f7d6..689ee9f7cd0c 100644
--- a/src/Text/Pandoc/Readers/Docx.hs
+++ b/src/Text/Pandoc/Readers/Docx.hs
@@ -669,6 +669,34 @@ normalizeToClassName = T.map go . fromStyleName
              | otherwise = c
 
 bodyPartToBlocks :: PandocMonad m => BodyPart -> DocxContext m Blocks
+bodyPartToBlocks (Heading n style pPr numId lvl mblvlInfo parparts) = do
+    ils <- local (\s-> s{docxInHeaderBlock=True})
+           (smushInlines <$> mapM parPartToInlines parparts)
+    let classes = map normalizeToClassName . delete style
+                $ getStyleNames (pStyle pPr)
+    hasNumbering <- gets docxNumberedHeadings
+    let addNum = if hasNumbering && not (numbered pPr)
+                 then (++ ["unnumbered"])
+                 else id
+    if T.null numId
+       then pure ()
+       else do
+         -- We check whether this current numId has previously been used,
+         -- since Docx expects us to pick up where we left off.
+         listState <- gets docxListState
+         let start = case M.lookup (numId, lvl) listState of
+                       Nothing -> case mblvlInfo of
+                         Nothing -> 1
+                         Just (Level _ _ _ z) -> fromMaybe 1 z
+                       Just z -> z + 1
+         modify $ \st -> st{ docxListState =
+             -- expire all the continuation data for lists of level > this one:
+             -- a new level 1 list item resets continuation for level 2+
+             -- see #10258
+             let notExpired (_, lvl') _ = lvl' <= lvl
+             in M.insert (numId, lvl) start
+                  (M.filterWithKey notExpired listState) }
+    makeHeaderAnchor $ headerWith ("", addNum classes, []) n ils
 bodyPartToBlocks (Paragraph pPr parparts)
   | Just True <- pBidi pPr = do
       let pPr' = pPr { pBidi = Nothing }
@@ -681,18 +709,6 @@ bodyPartToBlocks (Paragraph pPr parparts)
         codeBlock $
         T.concat $
         map parPartToText parparts
-  | Just (style, n) <- pHeading pPr = do
-    ils <- local (\s-> s{docxInHeaderBlock=True})
-           (smushInlines <$> mapM parPartToInlines parparts)
-    let classes = map normalizeToClassName . delete style
-                $ getStyleNames (pStyle pPr)
-
-    hasNumbering <- gets docxNumberedHeadings
-    let addNum = if hasNumbering && not (numbered pPr)
-                 then (++ ["unnumbered"])
-                 else id
-    makeHeaderAnchor $
-      headerWith ("", addNum classes, []) n ils
   | otherwise = do
     ils <- trimSps . smushInlines <$> mapM parPartToInlines parparts
     prevParaIls <- gets docxPrevPara
diff --git a/src/Text/Pandoc/Readers/Docx/Parse.hs b/src/Text/Pandoc/Readers/Docx/Parse.hs
index 3c493022ee7c..ae87fec7a5db 100644
--- a/src/Text/Pandoc/Readers/Docx/Parse.hs
+++ b/src/Text/Pandoc/Readers/Docx/Parse.hs
@@ -283,6 +283,8 @@ defaultParagraphStyle = ParagraphStyle { pStyle = []
 
 
 data BodyPart = Paragraph ParagraphStyle [ParPart]
+              | Heading Int ParaStyleName ParagraphStyle T.Text T.Text (Maybe Level)
+                 [ParPart]
               | ListItem ParagraphStyle T.Text T.Text (Maybe Level) [ParPart]
               | Tbl T.Text TblGrid TblLook [Row]
               | Captioned ParagraphStyle [ParPart] BodyPart
@@ -791,13 +793,15 @@ elemToBodyPart ns element
 elemToBodyPart ns element
   | isElem ns "w" "p" element
   , Just (numId, lvl) <- getNumInfo ns element = do
+    lvlInfo <- lookupLevel numId lvl <$> asks envNumbering
     parstyle <- elemToParagraphStyle ns element
                 <$> asks envParStyles
                 <*> asks envNumbering
     parparts <- mconcat <$> mapD (elemToParPart ns) (elChildren element)
     case pHeading parstyle of
       Nothing -> mkListItem parstyle numId lvl parparts
-      Just _  -> return $ Paragraph parstyle parparts
+      Just (parstylename, lev)
+        -> return $ Heading lev parstylename parstyle numId lvl lvlInfo parparts
 elemToBodyPart ns element
   | isElem ns "w" "p" element
   , [Elem ppr] <- elContent element
@@ -836,8 +840,11 @@ elemToBodyPart ns element
       case pHeading parstyle of
         Nothing | Just (numId, lvl) <- pNumInfo parstyle -> do
                     mkListItem parstyle numId lvl parparts
-        _ -> return $ Paragraph parstyle parparts
-
+        Just (parstylename, lev) -> do
+          let (numId, lvl) = fromMaybe ("","") $ pNumInfo parstyle
+          lvlInfo <- lookupLevel numId lvl <$> asks envNumbering
+          return $ Heading lev parstylename parstyle numId lvl lvlInfo parparts
+        Nothing -> return $ Paragraph parstyle parparts
 elemToBodyPart ns element
   | isElem ns "w" "tbl" element = do
     let tblProperties = findChildByName ns "w" "tblPr" element