From df7d633f59c90dba89b11701e6ca8504935e9a86 Mon Sep 17 00:00:00 2001
From: Peter Edberg <pedberg@unicode.org>
Date: Wed, 28 Apr 2010 03:55:58 +0000
Subject: [PATCH] ICU-7651 Add test case (segfaults without the fix in r27992)

X-SVN-Rev: 27994
---
 icu4c/source/test/intltest/regextst.cpp | 39 +++++++++++++++++++++++++
 icu4c/source/test/intltest/regextst.h   |  1 +
 2 files changed, 40 insertions(+)

diff --git a/icu4c/source/test/intltest/regextst.cpp b/icu4c/source/test/intltest/regextst.cpp
index 185e52caa6f..98748658744 100644
--- a/icu4c/source/test/intltest/regextst.cpp
+++ b/icu4c/source/test/intltest/regextst.cpp
@@ -98,6 +98,9 @@ void RegexTest::runIndexedTest( int32_t index, UBool exec, const char* &name, ch
         case 14: name = "PreAllocatedUTextCAPI";
           if (exec) PreAllocatedUTextCAPI();
           break;
+        case 15: name = "Bug 7651";
+          if (exec) Bug7651();
+          break;
 
         default: name = "";
             break; //needed to end loop
@@ -4608,5 +4611,41 @@ void RegexTest::PreAllocatedUTextCAPI () {
     utext_close(&patternText);
 }
 
+//--------------------------------------------------------------
+//
+//  Bug7651   Regex pattern that exceeds default operator stack depth in matcher.
+//
+//---------------------------------------------------------------
+void RegexTest::Bug7651() {
+    UnicodeString pattern1("((?<![A-Za-z0-9])[#\\uff03][A-Za-z0-9_][A-Za-z0-9_\\u00c0-\\u00d6\\u00c8-\\u00f6\\u00f8-\\u00ff]*|(?<![A-Za-z0-9_])[@\\uff20][A-Za-z0-9_]+(?:\\/[\\w-]+)?|(https?\\:\\/\\/|www\\.)\\S+(?<![\\!\\),\\.:;\\]\\u0080-\\uFFFF])|\\$[A-Za-z]+)");
+    //  The following should exceed the default operator stack depth in the matcher, i.e. force the matcher to malloc instead of using fSmallData.
+    UnicodeString pattern2("((https?\\:\\/\\/|www\\.)\\S+(?<![\\!\\),\\.:;\\]\\u0080-\\uFFFF])|(?<![A-Za-z0-9_])[\\@\\uff20][A-Za-z0-9_]+(?:\\/[\\w\\-]+)?|(?<![A-Za-z0-9])[\\#\\uff03][A-Za-z0-9_][A-Za-z0-9_\\u00c0-\\u00d6\\u00c8-\\u00f6\\u00f8-\\u00ff]*|\\$[A-Za-z]+)");
+    UnicodeString s("#ff @abcd This is test");
+    RegexPattern  *REPattern = NULL;
+    RegexMatcher  *REMatcher = NULL;
+    UErrorCode status = U_ZERO_ERROR;
+    UParseError pe;
+
+    REPattern = RegexPattern::compile(pattern1, 0, pe, status);
+    REGEX_CHECK_STATUS;
+    REMatcher = REPattern->matcher(s, status);
+    REGEX_CHECK_STATUS;
+    REGEX_ASSERT(REMatcher->find());
+    REGEX_ASSERT(REMatcher->start(status) == 0);
+    delete REPattern;
+    delete REMatcher;
+    status = U_ZERO_ERROR;
+
+    REPattern = RegexPattern::compile(pattern2, 0, pe, status);
+    REGEX_CHECK_STATUS;
+    REMatcher = REPattern->matcher(s, status);
+    REGEX_CHECK_STATUS;
+    REGEX_ASSERT(REMatcher->find());
+    REGEX_ASSERT(REMatcher->start(status) == 0);
+    delete REPattern;
+    delete REMatcher;
+    status = U_ZERO_ERROR;
+ }
+
 #endif  /* !UCONFIG_NO_REGULAR_EXPRESSIONS  */
 
diff --git a/icu4c/source/test/intltest/regextst.h b/icu4c/source/test/intltest/regextst.h
index 3bf592210d3..627ad5b249f 100644
--- a/icu4c/source/test/intltest/regextst.h
+++ b/icu4c/source/test/intltest/regextst.h
@@ -40,6 +40,7 @@ public:
     virtual void API_Replace_UTF8();
     virtual void PerlTestsUTF8();
     virtual void PreAllocatedUTextCAPI();
+    virtual void Bug7651();
 
     // The following functions are internal to the regexp tests.
     virtual void assertUText(const char *expected, UText *actual, const char *file, int line);