From f48c4821f7f41f9547c53a4004f3c53f213147c4 Mon Sep 17 00:00:00 2001
From: Sergey Lyubka <valenok@gmail.com>
Date: Wed, 31 Jul 2013 08:56:42 +0100
Subject: [PATCH] Added handling of "*" quantifier and non-greedy match

---
 slre.c      | 14 +++++++++-----
 unit_test.c | 10 ++++++++++
 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/slre.c b/slre.c
index ac2fb5f..f51ec05 100644
--- a/slre.c
+++ b/slre.c
@@ -74,8 +74,7 @@ struct regex_info {
 };
 
 static int get_op_len(const char *re) {
-  return re[0] == '\\' ? 2 :
-    (re[0] == '*' || re[0] == '+') && re[1] == '?' ? 2 : 1;
+  return re[0] == '\\' ? 2 : 1;
 }
 
 static int is_quantifier(const char *re) {
@@ -123,11 +122,15 @@ static int m1(const char *re, int re_len, const char *s, int s_len,
         j += m1(re + i, step, s + j, s_len - j, caps, info);
         i++;
         continue;
-      } else if (re[i + step] == '+') {
-        int j2 = j, nj = 0, n1, n2, ni, next_step;
+      } else if (re[i + step] == '+' || re[i + step] == '*') {
+        int j2 = j, nj = 0, n1, n2, ni, next_step, non_greedy = 0;
 
         /* Points to the regexp code after the quantifier */
         next_step = get_op_len(re + i + step);
+        if (i + step + 1 < re_len && re[i + step + 1] == '?') {
+          non_greedy = 1;
+          next_step++;
+        }
         ni = i + step + next_step;
 
         while ((n1 = m1(re + i, step, s + j2, s_len - j2, caps, info)) > 0) {
@@ -138,9 +141,10 @@ static int m1(const char *re, int re_len, const char *s, int s_len,
                               s_len - (j2 + n1), caps, info)) > 0) {
             nj = j2 + n1 + n2;
           }
+          if (nj > 0 && non_greedy) break;
           j2 += n1;
         }
-        FAIL_IF(nj == 0, static_error_no_match);
+        FAIL_IF(re[i + step] == '+' && nj == 0, static_error_no_match);
         return nj;
       }
     }
diff --git a/unit_test.c b/unit_test.c
index 373a127..ff91f46 100644
--- a/unit_test.c
+++ b/unit_test.c
@@ -68,6 +68,7 @@ int main(void) {
   ASSERT(slre_match("()+", "fooklmn", 7, NULL, &msg) == 0);
   ASSERT(strcmp(msg, static_error_no_match) == 0);
 
+  /* Balancing brackets */
   ASSERT(slre_match("(x))", "fooklmn", 7, NULL, &msg) == 0);
   ASSERT(strcmp(msg, static_error_unbalanced_brackets) == 0);
   ASSERT(slre_match("(", "fooklmn", 7, NULL, &msg) == 0);
@@ -76,9 +77,18 @@ int main(void) {
   ASSERT(slre_match("klz?mn", "fooklmn", 7, NULL, &msg) == 7);
   ASSERT(slre_match("fa?b", "fooklmn", 7, NULL, &msg) == 0);
 
+  /* Brackets & capturing */
   ASSERT(slre_match("^(te)", "tenacity subdues all", 20, NULL, &msg) == 2);
   ASSERT(slre_match("(bc)", "abcdef", 6, NULL, &msg) == 3);
   ASSERT(slre_match(".(d.)", "abcdef", 6, NULL, &msg) == 5);
+  ASSERT(slre_match(".(d.)\\)?", "abcdef", 6, NULL, &msg) == 5);
+
+  /* Greedy vs non-greedy */
+  ASSERT(slre_match(".+c", "abcabc", 6, NULL, &msg) == 6);
+  ASSERT(slre_match(".+?c", "abcabc", 6, NULL, &msg) == 3);
+  ASSERT(slre_match(".*?c", "abcabc", 6, NULL, &msg) == 3);
+  ASSERT(slre_match(".*c", "abcabc", 6, NULL, &msg) == 6);
+  ASSERT(slre_match("bc.d?k?b+", "abcabc", 6, NULL, &msg) == 5);
 
   printf("Unit test %s (total test: %d, failed tests: %d)\n",
          static_failed_tests > 0 ? "FAILED" : "PASSED",