From 0d27947b96f7066edeed8a63c196e045ac8822b5 Mon Sep 17 00:00:00 2001
From: Seth Speaks <sspeaks610@gmail.com>
Date: Sat, 16 Sep 2023 09:21:01 -0700
Subject: [PATCH] HTML reader: parse task lists using input elements (#9066)

Allow the HTML reader to parse task lists of the sort produced by pandoc.
Closes #9047
---
 src/Text/Pandoc/Readers/HTML.hs       | 19 ++++++++++-
 src/Text/Pandoc/Readers/HTML/Types.hs |  3 +-
 test/command/9047.md                  | 47 +++++++++++++++++++++++++++
 test/command/tasklist.md              |  2 +-
 4 files changed, 68 insertions(+), 3 deletions(-)
 create mode 100644 test/command/9047.md

diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs
index 5e4c8d3158bd..18441633d945 100644
--- a/src/Text/Pandoc/Readers/HTML.hs
+++ b/src/Text/Pandoc/Readers/HTML.hs
@@ -129,6 +129,10 @@ setInChapter = local (\s -> s {inChapter = True})
 setInPlain :: PandocMonad m => HTMLParser m s a -> HTMLParser m s a
 setInPlain = local (\s -> s {inPlain = True})
 
+-- Some items should be handled differently when in a list item tag, e.g. checkbox
+setInListItem :: PandocMonad m => HTMLParser m s a -> HTMLParser m s a
+setInListItem = local (\s -> s {inListItem = True})
+
 pHtml :: PandocMonad m => TagParser m Blocks
 pHtml = do
   (TagOpen "html" attr) <- lookAhead pAny
@@ -334,7 +338,7 @@ pBulletList = try $ do
   return $ B.bulletList $ map (fixPlains True) items
 
 pListItem :: PandocMonad m => TagParser m Blocks
-pListItem = do
+pListItem = setInListItem $ do
   TagOpen _ attr' <- lookAhead $ pSatisfy (matchTagOpen "li" [])
   let attr = toStringAttr attr'
   let addId ident bs = case B.toList bs of
@@ -344,6 +348,16 @@ pListItem = do
   maybe id addId (lookup "id" attr) <$>
     pInTags "li" block
 
+pCheckbox :: PandocMonad m => TagParser m Inlines
+pCheckbox = do
+  TagOpen _ attr' <- pSatisfy $ matchTagOpen "input" [("type","checkbox")]
+  TagClose _ <- pSatisfy (matchTagClose "input")
+  let attr = toStringAttr attr'
+  let isChecked = isJust $ lookup "checked" attr
+  let escapeSequence = B.str $ if isChecked then "\9746" else "\9744"
+  return $ escapeSequence <> B.space
+
+
 -- | Parses a list item just like 'pListItem', but allows sublists outside of
 -- @li@ tags to be treated as items.
 pListItem' :: PandocMonad m => TagParser m a -> TagParser m Blocks
@@ -673,6 +687,9 @@ inline = pTagText <|> do
         "var" -> pCodeWithClass "var" "variable"
         "span" -> pSpan
         "math" -> pMath False
+        "input" 
+          | lookup "type" attr == Just "checkbox" 
+          -> asks inListItem >>= guard >> pCheckbox
         "script"
           | Just x <- lookup "type" attr
           , "math/tex" `T.isPrefixOf` x -> pScriptMath
diff --git a/src/Text/Pandoc/Readers/HTML/Types.hs b/src/Text/Pandoc/Readers/HTML/Types.hs
index f03454c3bfb6..a16773949d12 100644
--- a/src/Text/Pandoc/Readers/HTML/Types.hs
+++ b/src/Text/Pandoc/Readers/HTML/Types.hs
@@ -60,6 +60,7 @@ data HTMLLocal = HTMLLocal
   { quoteContext :: QuoteContext
   , inChapter    :: Bool -- ^ Set if in chapter section
   , inPlain      :: Bool -- ^ Set if in pPlain
+  , inListItem   :: Bool -- ^ Set if in <li> tag
   }
 
 
@@ -91,7 +92,7 @@ instance HasMeta HTMLState where
   deleteMeta s st = st {parserState = deleteMeta s $ parserState st}
 
 instance Default HTMLLocal where
-  def = HTMLLocal NoQuote False False
+  def = HTMLLocal NoQuote False False False
 
 instance HasLastStrPosition HTMLState where
   setLastStrPos s st = st {parserState = setLastStrPos s (parserState st)}
diff --git a/test/command/9047.md b/test/command/9047.md
new file mode 100644
index 000000000000..84c405e07ffc
--- /dev/null
+++ b/test/command/9047.md
@@ -0,0 +1,47 @@
+tests meant to test the fixes of bug [#9047](https://github.com/jgm/pandoc/issues/9047)
+
+```
+% pandoc -f html -t html
+<ul class="task-list">
+<li><label><input type="checkbox" />foo</label></li>
+<li><label><input type="checkbox" checked="" />bar</label></li>
+<li><label><input type="button" checked="" />foobar</label></li>
+<li><input id="hello" type="checkbox" checked/><label for="hello">hello</label></li>
+</ul>
+^D
+<ul>
+<li><label><input type="checkbox" />foo</label></li>
+<li><label><input type="checkbox" checked="" />bar</label></li>
+<li>foobar</li>
+<li><label><input type="checkbox" checked="" />hello</label></li>
+</ul>
+```
+
+w/ rawHTML extension a checkbox by itself is kept
+```
+% pandoc -f html+raw_html -t html+raw_html
+<input type="checkbox" checked="" />
+^D
+<input type="checkbox" checked>
+</input>
+```
+
+w/ rawHTML extension, a checkbox in an `li` is handled properly
+```
+% pandoc -f html+raw_html -t html+raw_html
+<ul>
+<li><input type="checkbox" checked/>foo</li>
+</ul>
+^D
+<ul class="task-list">
+<li><label><input type="checkbox" checked="" />foo</label></li>
+</ul>
+```
+
+w/o rawHTML extension, a checkbox outside of an `li` is properly ignored
+```
+% pandoc -f html -t html
+<p><input type="checkbox" checked/>foo</p>
+^D
+foo
+```
diff --git a/test/command/tasklist.md b/test/command/tasklist.md
index b1c6491ccb5c..9aaa0a8aaeab 100644
--- a/test/command/tasklist.md
+++ b/test/command/tasklist.md
@@ -102,4 +102,4 @@ round trip:
 ^D
 -   [ ] foo
 -   [x] bar
-```
+```
\ No newline at end of file