bugfix: multi-line block quote inside list item (bd-vet6)

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-minus.qmd

@@ -0,0 +1,2 @@
+- > a
+  > b

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-ordered-dot.qmd

@@ -0,0 +1,2 @@
+1. > a
+   > b

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-ordered-paren.qmd

@@ -0,0 +1,2 @@
+1) > a
+   > b

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-plus.qmd

@@ -0,0 +1,2 @@
++ > a
+  > b

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-star.qmd

@@ -0,0 +1,2 @@
+* > a
+  > b

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-then-paragraph.qmd

@@ -0,0 +1,4 @@
+- > a
+  > b
+
+c

crates/pampa/tests/pandoc-match-corpus/markdown/bq-multiline-in-list-three-lines.qmd

@@ -0,0 +1,3 @@
+- > a
+  > b
+  > c

crates/tree-sitter-qmd/tree-sitter-markdown/src/scanner.c

@@ -2037,7 +2037,23 @@ static bool scan(Scanner *s, TSLexer *lexer, const bool *valid_symbols) {
         return false;
     }
 
-    if (s->state & STATE_MATCHING) { // we are in the state of trying to match all currently open blocks
+    // bd-vet6: when we re-enter STATE_MATCHING after a SOFT_LINE_ENDING
+    // and the current lookahead is the trailing \n / \r of the same
+    // logical line, do NOT call match_line here. The LIST_ITEM match()
+    // returns case 2 on \n (see line ~537) and advances past it; the
+    // line-ending gate at line ~2233 then has nothing to match against
+    // and scan() returns false. The state changes get rolled back, but
+    // tree-sitter retries with another lex_external state and emits
+    // CLOSE_BLOCK, which the parser cannot shift in this position.
+    // Bypassing match_line here lets the line-ending gate run and emit
+    // the LINE_ENDING / SOFT_LINE_ENDING the parser actually expects.
+    // (EOF is handled above at line ~2027 and never reaches here.)
+    bool at_soft_break_line_end =
+        (s->state & STATE_MATCHING) &&
+        (s->state & STATE_WAS_SOFT_LINE_BREAK) &&
+        (lexer->lookahead == '\n' || lexer->lookahead == '\r');
+
+    if ((s->state & STATE_MATCHING) && !at_soft_break_line_end) { // we are in the state of trying to match all currently open blocks
         DEBUG_PRINT("scan() while STATE_MATCHING\n");
         int match_line_return = match_line(s, lexer);
         // bool might_be_soft_break = match_line_return & 2;

crates/tree-sitter-qmd/tree-sitter-markdown/test/corpus/list.txt

@@ -425,4 +425,129 @@ b
                 (pandoc_str)
                 (block_continuation)))
             (list_item
-              (list_marker_star))))))
+              (list_marker_star))))))
+================================================================================
+24 - multi-line block quote inside list item, followed by paragraph (minus)
+================================================================================
+- > a
+  > b
+
+c
+--------------------------------------------------------------------------------
+(document
+  (section
+    (pandoc_list
+      (list_item
+        (list_marker_minus)
+        (pandoc_block_quote
+          (block_quote_marker)
+          (pandoc_paragraph
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)))))
+    (pandoc_paragraph
+      (pandoc_str))))
+================================================================================
+25 - multi-line block quote inside list item, followed by paragraph (star)
+================================================================================
+* > a
+  > b
+
+c
+--------------------------------------------------------------------------------
+(document
+  (section
+    (pandoc_list
+      (list_item
+        (list_marker_star)
+        (pandoc_block_quote
+          (block_quote_marker)
+          (pandoc_paragraph
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)))))
+    (pandoc_paragraph
+      (pandoc_str))))
+================================================================================
+26 - multi-line block quote inside list item, followed by paragraph (plus)
+================================================================================
++ > a
+  > b
+
+c
+--------------------------------------------------------------------------------
+(document
+  (section
+    (pandoc_list
+      (list_item
+        (list_marker_plus)
+        (pandoc_block_quote
+          (block_quote_marker)
+          (pandoc_paragraph
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)))))
+    (pandoc_paragraph
+      (pandoc_str))))
+================================================================================
+27 - multi-line block quote inside list item, followed by paragraph (ordered, dot)
+================================================================================
+1. > a
+   > b
+
+c
+--------------------------------------------------------------------------------
+(document
+  (section
+    (pandoc_list
+      (list_item
+        (list_marker_dot)
+        (pandoc_block_quote
+          (block_quote_marker)
+          (pandoc_paragraph
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)))))
+    (pandoc_paragraph
+      (pandoc_str))))
+================================================================================
+28 - multi-line block quote inside list item, followed by paragraph (ordered, paren)
+================================================================================
+1) > a
+   > b
+
+c
+--------------------------------------------------------------------------------
+(document
+  (section
+    (pandoc_list
+      (list_item
+        (list_marker_parenthesis)
+        (pandoc_block_quote
+          (block_quote_marker)
+          (pandoc_paragraph
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)))))
+    (pandoc_paragraph
+      (pandoc_str))))
+================================================================================
+29 - three-line block quote inside list item
+================================================================================
+- > a
+  > b
+  > c
+--------------------------------------------------------------------------------
+(document
+  (section
+    (pandoc_list
+      (list_item
+        (list_marker_minus)
+        (pandoc_block_quote
+          (block_quote_marker)
+          (pandoc_paragraph
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)
+            (pandoc_soft_break)
+            (pandoc_str)))))))