# Git issue 498: Conditional negative lookahead inside positive lookahead fails to match

Conditional node needed an additional member that points to the true branch.

Author: Matthew Barnett

changelog.txt

@@ -1,3 +1,8 @@
+Version: 2023.6.3
+
+    # Git issue 498: Conditional negative lookahead inside positive lookahead fails to match
+    Conditional node needed an additional member that points to the true branch.
+
 Version: 2023.5.5
 
     Removed semicolon after 'else' in 'munge_name'.
@@ -86,11 +91,11 @@ Version: 2022.3.2
 
     Git issue 453: Document last supported python2 version
 
-        Added a brief reference to the last version to support Python 2 in README.rst.
+    Added a brief reference to the last version to support Python 2 in README.rst.
 
     Git issue 456: RegexFlag exists in re, but not regex
 
-        Updated the flags to use enum now that regex supports only Python 3.6+.
+    Updated the flags to use enum now that regex supports only Python 3.6+.
 
 Version: 2022.1.21
 
@@ -1225,13 +1230,13 @@ Version: 2013.1.25
 Version: 2013.1.24
 
     Hg issue 86: Enhance API of captures() to enable retrieval of ALL groups at once, as a dictionary
-            Added capturesdict() method to match object.
+        Added capturesdict() method to match object.
 
     Hg issue 87: Allow duplicate names of groups
-            Now allowed.
+        Now allowed.
 
     Hg issue 88: regex.match() hangs
-            Fixed.
+        Fixed.
 
 Version: 2013.1.20
 

regex_3/_regex.c

@@ -291,6 +291,7 @@ typedef struct RE_Node {
     union {
         struct {
             RE_NextNode next_2;
+            struct RE_Node* true_node; /* Used by a CONDITIONAL node. */
         } nonstring;
         struct {
             /* Used only if (node->status & RE_STATUS_STRING) is true. */
@@ -15272,7 +15273,7 @@ Py_LOCAL_INLINE(int) basic_match(RE_State* state, BOOL search) {
                 /* It's a negative lookaround that's failed. Go to the 'true'
                  * branch.
                  */
-                node = conditional->next_1.node;
+                node = conditional->nonstring.true_node;
 
             goto advance;
         }
@@ -16990,7 +16991,6 @@ Py_LOCAL_INLINE(int) basic_match(RE_State* state, BOOL search) {
 
             if (!lookaround->match) {
                 /* It's a negative lookaround that's failed. */
-
                 node = lookaround->nonstring.next_2.node;
                 goto advance;
             }
@@ -22986,6 +22986,14 @@ Py_LOCAL_INLINE(void) skip_one_way_branches(PatternObject* pattern) {
                 node->nonstring.next_2.node = next->next_1.node;
                 modified = TRUE;
             }
+
+            /* Check the true branch for CONDITIONAL. */
+            next = node->nonstring.true_node;
+            if (next && next->op == RE_OP_BRANCH &&
+              !next->nonstring.true_node) {
+                node->nonstring.true_node = next->next_1.node;
+                modified = TRUE;
+            }
         }
     } while (modified);
 
@@ -24443,6 +24451,7 @@ Py_LOCAL_INLINE(int) build_CONDITIONAL(RE_CompileArgs* args) {
     /* end test node -> true branch -> end node */
     add_node(end_test_node, subargs.start);
     add_node(subargs.end, end_node);
+    test_node->nonstring.true_node = subargs.start;
 
     if (args->code[0] == RE_OP_NEXT) {
         /* There's a false branch. */

regex_3/regex.py

@@ -241,7 +241,7 @@
   "VERSION1", "X", "VERBOSE", "W", "WORD", "error", "Regex", "__version__",
   "__doc__", "RegexFlag"]
 
-__version__ = "2.5.128"
+__version__ = "2.5.129"
 
 # --------------------------------------------------------------------
 # Public interface.

regex_3/test_regex.py

@@ -4332,6 +4332,12 @@ def test_hg_bugs(self):
         # Git issue 494: Backtracking failure matching regex ^a?(a?)b?c\1$ against string abca
         self.assertEqual(regex.search(r"^a?(a?)b?c\1$", "abca").span(), (0, 4))
 
+        # Git issue 498: Conditional negative lookahead inside positive lookahead fails to match
+        self.assertEqual(regex.match(r"(?(?=a).|..)", "ab").span(), (0, 1))
+        self.assertEqual(regex.match(r"(?(?=b).|..)", "ab").span(), (0, 2))
+        self.assertEqual(regex.match(r"(?(?!a).|..)", "ab").span(), (0, 2))
+        self.assertEqual(regex.match(r"(?(?!b).|..)", "ab").span(), (0, 1))
+
     def test_fuzzy_ext(self):
         self.assertEqual(bool(regex.fullmatch(r'(?r)(?:a){e<=1:[a-z]}', 'e')),
           True)

setup.py

@@ -7,7 +7,7 @@
 
 setup(
     name='regex',
-    version='2023.5.5',
+    version='2023.6.3',
     description='Alternative regular expression module, to replace re.',
     long_description=long_description,
     long_description_content_type='text/x-rst',