bpo-46503: Prevent an assert from firing when parsing some invalid \N sequences in f-strings. (GH-30865) (30867)

* bpo-46503: Prevent an assert from firing.  Also fix one nearby tiny PEP-7 nit.

* Added blurb.
(cherry picked from commit 0daf721)

Co-authored-by: Eric V. Smith <[email protected]>

Co-authored-by: Eric V. Smith <[email protected]>

Author: miss-islington

Lib/test/test_fstring.py

@@ -747,12 +747,16 @@ def test_misformed_unicode_character_name(self):
         # differently inside f-strings.
         self.assertAllRaise(SyntaxError, r"\(unicode error\) 'unicodeescape' codec can't decode bytes in position .*: malformed \\N character escape",
                             [r"f'\N'",
+                             r"f'\N '",
+                             r"f'\N  '",  # See bpo-46503.
                              r"f'\N{'",
                              r"f'\N{GREEK CAPITAL LETTER DELTA'",
 
                              # Here are the non-f-string versions,
                              #  which should give the same errors.
                              r"'\N'",
+                             r"'\N '",
+                             r"'\N  '",
                              r"'\N{'",
                              r"'\N{GREEK CAPITAL LETTER DELTA'",
                              ])

Misc/NEWS.d/next/Core and Builtins/2022-01-24-21-24-41.bpo-46503.4UrPsE.rst

@@ -0,0 +1 @@
+Fix an assert when parsing some invalid \N escape sequences in f-strings.

Parser/pegen/parse_string.c

@@ -444,12 +444,23 @@ fstring_find_literal(Parser *p, const char **str, const char *end, int raw,
         if (!raw && ch == '\\' && s < end) {
             ch = *s++;
             if (ch == 'N') {
+                /* We need to look at and skip matching braces for "\N{name}"
+                   sequences because otherwise we'll think the opening '{'
+                   starts an expression, which is not the case with "\N".
+                   Keep looking for either a matched '{' '}' pair, or the end
+                   of the string. */
+
                 if (s < end && *s++ == '{') {
                     while (s < end && *s++ != '}') {
                     }
                     continue;
                 }
-                break;
+
+                /* This is an invalid "\N" sequence, since it's a "\N" not
+                   followed by a "{".  Just keep parsing this literal.  This
+                   error will be caught later by
+                   decode_unicode_with_escapes(). */
+                continue;
             }
             if (ch == '{' && warn_invalid_escape_sequence(p, ch, t) < 0) {
                 return -1;
@@ -493,7 +504,8 @@ fstring_find_literal(Parser *p, const char **str, const char *end, int raw,
             *literal = PyUnicode_DecodeUTF8Stateful(literal_start,
                                                     s - literal_start,
                                                     NULL, NULL);
-        } else {
+        }
+        else {
             *literal = decode_unicode_with_escapes(p, literal_start,
                                                    s - literal_start, t);
         }