From 964b93a09deb7e166041c584ace83b2989f41671 Mon Sep 17 00:00:00 2001
From: Jonathan Marler <johnnymarler@gmail.com>
Date: Sat, 6 Mar 2021 07:19:02 -0700
Subject: [PATCH] fix mishandling of ^ inside an expression

---
 Makefile      |  1 +
 re.c          |  1 +
 tests/test1.c | 20 ++++++++++++++++++--
 3 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/Makefile b/Makefile
index 857d2ee..05fe7ec 100644
--- a/Makefile
+++ b/Makefile
@@ -101,6 +101,7 @@ test: all
 	@$(PYTHON) ./scripts/regex_test_neg.py [012345-9]             $(NRAND_TESTS)
 	@$(PYTHON) ./scripts/regex_test_neg.py [0-56789]              $(NRAND_TESTS)
 	@$(PYTHON) ./scripts/regex_test_neg.py .*123faerdig           $(NRAND_TESTS)
+	@$(PYTHON) ./scripts/regex_test_neg.py a^                     $(NRAND_TESTS)
 	@echo
 	@echo
 	@./tests/test2
diff --git a/re.c b/re.c
index 20d1474..34831af 100644
--- a/re.c
+++ b/re.c
@@ -393,6 +393,7 @@ static int matchone(regex_t p, char c)
     case NOT_ALPHA:      return !matchalphanum(c);
     case WHITESPACE:     return  matchwhitespace(c);
     case NOT_WHITESPACE: return !matchwhitespace(c);
+    case BEGIN:          return 0;
     default:             return  (p.u.ch == c);
   }
 }
diff --git a/tests/test1.c b/tests/test1.c
index 5fdfe74..666d18b 100644
--- a/tests/test1.c
+++ b/tests/test1.c
@@ -101,11 +101,12 @@ int main()
     int should_fail;
     int length;
     int correctlen;
-    size_t ntests = sizeof(test_vector) / sizeof(*test_vector);
+    size_t nvector_tests = sizeof(test_vector) / sizeof(*test_vector);
+    size_t ntests = nvector_tests + 1;
     size_t nfailed = 0;
     size_t i;
 
-    for (i = 0; i < ntests; ++i)
+    for (i = 0; i < nvector_tests; ++i)
     {
         pattern = test_vector[i][1];
         text = test_vector[i][2];
@@ -141,6 +142,21 @@ int main()
         }
     }
 
+    // regression test for unhandled BEGIN in the middle of an expression
+    // we need to test text strings with all possible values for the second
+    // byte because re.c was matching it against an uninitalized value, so
+    // it could be anything
+    pattern = "a^";
+    for (i = 0; i < 255; i++) {
+      char text_buf[] = { 'a', i, '\0' };
+      int m = re_match(pattern, text_buf, &length);
+      if (m != -1) {
+        fprintf(stderr, "[%lu/%lu]: pattern '%s' matched '%s' unexpectedly", ntests, ntests, pattern, text_buf);
+        nfailed += 1;
+        break;
+      }
+    }
+
     // printf("\n");
     printf("%lu/%lu tests succeeded.\n", ntests - nfailed, ntests);
     printf("\n");