From 12cce1758ff82cced76fa1961076f99176ea2689 Mon Sep 17 00:00:00 2001
From: Igor Pashev <pashev.igor-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Date: Thu, 13 Jan 2022 19:11:50 +0200
Subject: LaTeX: parse thebibliography

---
 src/Text/Pandoc/Readers/LaTeX.hs         | 32 +++++++++++++++++++++
 src/Text/Pandoc/Readers/LaTeX/Parsing.hs |  2 ++
 test/command/latex-thebibliography.md    | 48 ++++++++++++++++++++++++++++++++
 3 files changed, 82 insertions(+)
 create mode 100644 test/command/latex-thebibliography.md

diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs
index 20a2db76b..e4a3aaa58 100644
--- a/src/Text/Pandoc/Readers/LaTeX.hs
+++ b/src/Text/Pandoc/Readers/LaTeX.hs
@@ -741,6 +741,14 @@ looseItem = do
   skipopts
   return mempty
 
+looseBibItem :: PandocMonad m => LP m Blocks
+looseBibItem = do
+  inListItem <- sInListItem <$> getState
+  guard $ not inListItem
+  skipopts
+  void braced
+  return mempty
+
 epigraph :: PandocMonad m => LP m Blocks
 epigraph = do
   p1 <- grouped block
@@ -886,6 +894,7 @@ blockCommands = M.fromList
    , ("strut", pure mempty)
    , ("rule", rule)
    , ("item", looseItem)
+   , ("bibitem", looseBibItem)
    , ("documentclass", skipopts *> braced *> preamble)
    , ("centerline", para . trimInlines <$> (skipopts *> tok))
    , ("caption", mempty <$ setCaption inline)
@@ -975,6 +984,7 @@ environments = M.union (tableEnvironments blocks inline) $
    , ("togglefalse", braced >>= setToggle False)
    , ("iftoggle", try $ ifToggle >> block)
    , ("CSLReferences", braced >> braced >> env "CSLReferences" blocks)
+   , ("thebibliography", theBibliography)
    ]
 
 filecontents :: PandocMonad m => LP m Blocks
@@ -1211,6 +1221,28 @@ descItem = do
   bs <- blocks
   return (ils, [bs])
 
+bibItem :: PandocMonad m => LP m (Inlines, [Blocks])
+bibItem = do
+  blocks
+  controlSeq "bibitem"
+  sp
+  lbl <- opt <|> nextCite
+  cite_key <- untokenize <$> braced
+  bs <- blocks
+  return (lbl, [divWith (cite_key, [], []) bs])
+  where
+    nextCite = do
+      st <- getState
+      let n = sTheBibItemNum st + 1
+      setState st {sTheBibItemNum = n}
+      return . singleton . Str . T.pack . show $ n
+
+theBibliography :: PandocMonad m => LP m Blocks
+theBibliography =
+  divWith ("", ["thebibliography"], []) . definitionList <$>
+  listenv "thebibliography" (many bibItem)
+
+
 listenv :: PandocMonad m => Text -> LP m a -> LP m a
 listenv name p = try $ do
   oldInListItem <- sInListItem `fmap` getState
diff --git a/src/Text/Pandoc/Readers/LaTeX/Parsing.hs b/src/Text/Pandoc/Readers/LaTeX/Parsing.hs
index 9eb4a0cbc..8fb6bd5bc 100644
--- a/src/Text/Pandoc/Readers/LaTeX/Parsing.hs
+++ b/src/Text/Pandoc/Readers/LaTeX/Parsing.hs
@@ -172,6 +172,7 @@ data LaTeXState = LaTeXState{ sOptions       :: ReaderOptions
                             , sFileContents  :: M.Map Text Text
                             , sEnableWithRaw :: Bool
                             , sRawTokens     :: IntMap.IntMap [Tok]
+                            , sTheBibItemNum :: Int
                             }
      deriving Show
 
@@ -199,6 +200,7 @@ defaultLaTeXState = LaTeXState{ sOptions       = def
                               , sFileContents  = M.empty
                               , sEnableWithRaw = True
                               , sRawTokens     = IntMap.empty
+                              , sTheBibItemNum = 0
                               }
 
 instance PandocMonad m => HasQuoteContext LaTeXState m where
diff --git a/test/command/latex-thebibliography.md b/test/command/latex-thebibliography.md
new file mode 100644
index 000000000..153fbfc13
--- /dev/null
+++ b/test/command/latex-thebibliography.md
@@ -0,0 +1,48 @@
+# The bibliography
+
+```
+% pandoc -f latex -t native
+\begin{thebibliography}{100}
+  \bibitem[One1990]{one} The First.
+  \bibitem{two} The Second.
+  \bibitem[Three 1998]{three} The Third.
+  \bibitem{four} The Fourth.
+\end{thebibliography}
+^D
+[ Div
+    ( "" , [ "thebibliography" ] , [] )
+    [ DefinitionList
+        [ ( [ Str "One1990" ]
+          , [ [ Div
+                  ( "one" , [] , [] )
+                  [ Para [ Str "The" , Space , Str "First." ] ]
+              ]
+            ]
+          )
+        , ( [ Str "1" ]
+          , [ [ Div
+                  ( "two" , [] , [] )
+                  [ Para [ Str "The" , Space , Str "Second." ] ]
+              ]
+            ]
+          )
+        , ( [ Str "Three" , Space , Str "1998" ]
+          , [ [ Div
+                  ( "three" , [] , [] )
+                  [ Para [ Str "The" , Space , Str "Third." ] ]
+              ]
+            ]
+          )
+        , ( [ Str "2" ]
+          , [ [ Div
+                  ( "four" , [] , [] )
+                  [ Para [ Str "The" , Space , Str "Fourth." ] ]
+              ]
+            ]
+          )
+        ]
+    ]
+]
+
+```
+