bpo-42218: Correctly handle errors in left-recursive rules (GH-23065)

Left-recursive rules need to check for errors explicitly, since
even if the rule returns NULL, the parsing might continue and lead
to long-distance failures.

Co-authored-by: Pablo Galindo <[email protected]>

Author: lysnikolaou

Lib/test/test_syntax.py

@@ -972,6 +972,14 @@ def func2():
 """
         self._check_error(code, "invalid syntax")
 
+    def test_invalid_line_continuation_left_recursive(self):
+        # Check bpo-42218: SyntaxErrors following left-recursive rules
+        # (t_primary_raw in this case) need to be tested explicitly
+        self._check_error("A.\u018a\\ ",
+                          "unexpected character after line continuation character")
+        self._check_error("A.\u03bc\\\n",
+                          "unexpected EOF while parsing")
+
 def test_main():
     support.run_unittest(SyntaxTestCase)
     from test import test_syntax

Misc/NEWS.d/next/Core and Builtins/2020-10-31-17-50-23.bpo-42218.Dp_Z3v.rst

@@ -0,0 +1,3 @@
+Fixed a bug in the PEG parser that was causing crashes in debug mode. Now errors are checked
+in left-recursive rules to avoid cases where such errors do not get handled in time and appear
+as long-distance crashes in other places.

Parser/parser.c

@@ -3461,6 +3461,8 @@ dotted_name_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = dotted_name_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -9045,6 +9047,8 @@ bitwise_or_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = bitwise_or_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -9159,6 +9163,8 @@ bitwise_xor_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = bitwise_xor_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -9273,6 +9279,8 @@ bitwise_and_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = bitwise_and_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -9387,6 +9395,8 @@ shift_expr_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = shift_expr_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -9540,6 +9550,8 @@ sum_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = sum_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -9699,6 +9711,8 @@ term_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = term_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -10303,6 +10317,8 @@ primary_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = primary_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;
@@ -13943,6 +13959,8 @@ t_primary_rule(Parser *p)
         }
         p->mark = _mark;
         void *_raw = t_primary_raw(p);
+        if (p->error_indicator)
+            return NULL;
         if (_raw == NULL || p->mark <= _resmark)
             break;
         _resmark = p->mark;

Tools/peg_generator/pegen/c_generator.py

@@ -502,6 +502,9 @@ def _set_up_rule_memoization(self, node: Rule, result_type: str) -> None:
                 )
                 self.print("p->mark = _mark;")
                 self.print(f"void *_raw = {node.name}_raw(p);")
+                self.print("if (p->error_indicator)")
+                with self.indent():
+                    self.print("return NULL;")
                 self.print("if (_raw == NULL || p->mark <= _resmark)")
                 with self.indent():
                     self.print("break;")