From 910130eff38258fcf3b6a0c48296aebe6da27cfd Mon Sep 17 00:00:00 2001
From: John MacFarlane <jgm@berkeley.edu>
Date: Wed, 9 Sep 2026 02:59:30 +0000
Subject: [PATCH] HTML reader: let the first of duplicate attributes win.

toStringAttr deduplicated attributes with a right fold, so the last
occurrence of a duplicated attribute was kept.  HTML specifies that
the first occurrence wins (and that lang takes precedence over
xml:lang).  Deduplicate left to right with a seen-set instead.

Co-Authored-By: Claude <noreply@anthropic.com>
---
 src/Text/Pandoc/Readers/HTML/Parsing.hs | 27 ++++++++++++++-----------
 1 file changed, 15 insertions(+), 12 deletions(-)

diff --git a/src/Text/Pandoc/Readers/HTML/Parsing.hs b/src/Text/Pandoc/Readers/HTML/Parsing.hs
index faa277be3..a91ad4cfe 100644
--- a/src/Text/Pandoc/Readers/HTML/Parsing.hs
+++ b/src/Text/Pandoc/Readers/HTML/Parsing.hs
@@ -199,20 +199,23 @@ t1 `closes` t2 |
 _ `closes` _ = False
 
 toStringAttr :: [(Text, Text)] -> [(Text, Text)]
-toStringAttr = foldr go []
+toStringAttr = go mempty
   where
-   go :: (Text, Text) -> [(Text, Text)] -> [(Text, Text)]
+   go :: Set.Set Text -> [(Text, Text)] -> [(Text, Text)]
+   go _ [] = []
+   go seen ((x,y):rest)
+     -- prevent duplicate attributes; the first one wins:
+     | x' `Set.member` seen = go seen rest
+     | otherwise = (x', y) : go (Set.insert x' seen) rest
+    where x' = normalizeName x
    -- treat xml:lang as lang
-   go ("xml:lang",y) ats = go ("lang",y) ats
-   -- prevent duplicate attributes
-   go (x,y) ats
-     | any (\(x',_) -> x == x') ats = ats
-     | otherwise      =
-        case T.stripPrefix "data-" x of
-          Just x' | x' `Set.notMember` (html5Attributes <>
-                                        html4Attributes <> rdfaAttributes)
-            -> go (x',y) ats
-          _ -> (x,y):ats
+   normalizeName "xml:lang" = "lang"
+   -- strip data- prefix unless the bare name is a standard attribute
+   normalizeName x =
+     case T.stripPrefix "data-" x of
+       Just x' | x' `Set.notMember` standardAttrs -> x'
+       _ -> x
+   standardAttrs = html5Attributes <> html4Attributes <> rdfaAttributes
 
 -- Unlike fromAttrib from tagsoup, this distinguishes
 -- between a missing attribute and an attribute with empty content.