From 118efa19c68be597b01c42aaa5ba921e2c77dde2 Mon Sep 17 00:00:00 2001
From: Andy Heninger <andy.heninger@gmail.com>
Date: Thu, 21 Apr 2011 00:07:50 +0000
Subject: [PATCH] ICU-8486 UText not checking for Bogus UnicodeStrings, causing
 crash in regular expressions.

X-SVN-Rev: 29840
---
 icu4c/source/common/utext.cpp           | 22 +++++++++-------------
 icu4c/source/test/intltest/regextst.cpp | 20 ++++++++++++++++++++
 icu4c/source/test/intltest/regextst.h   |  3 ++-
 3 files changed, 31 insertions(+), 14 deletions(-)

diff --git a/icu4c/source/common/utext.cpp b/icu4c/source/common/utext.cpp
index e0cc5b76c69..10ebc3296b3 100644
--- a/icu4c/source/common/utext.cpp
+++ b/icu4c/source/common/utext.cpp
@@ -2620,20 +2620,9 @@ U_CDECL_END
 
 U_CAPI UText * U_EXPORT2
 utext_openUnicodeString(UText *ut, UnicodeString *s, UErrorCode *status) {
-    // TODO:  use openConstUnicodeString, then add in the differences.
-    //
-    ut = utext_setup(ut, 0, status);
+    ut = utext_openConstUnicodeString(ut, s, status);
     if (U_SUCCESS(*status)) {
-        ut->pFuncs              = &unistrFuncs;
-        ut->context             = s;
-        ut->providerProperties  = I32_FLAG(UTEXT_PROVIDER_STABLE_CHUNKS)|
-                                  I32_FLAG(UTEXT_PROVIDER_WRITABLE);
-
-        ut->chunkContents       = s->getBuffer();
-        ut->chunkLength         = s->length();
-        ut->chunkNativeStart    = 0;
-        ut->chunkNativeLimit    = ut->chunkLength;
-        ut->nativeIndexingLimit = ut->chunkLength;
+        ut->providerProperties |= I32_FLAG(UTEXT_PROVIDER_WRITABLE);
     }
     return ut;
 }
@@ -2642,6 +2631,13 @@ utext_openUnicodeString(UText *ut, UnicodeString *s, UErrorCode *status) {
 
 U_CAPI UText * U_EXPORT2
 utext_openConstUnicodeString(UText *ut, const UnicodeString *s, UErrorCode *status) {
+    if (U_SUCCESS(*status) && s->isBogus()) {
+        // The UnicodeString is bogus, but we still need to detach the UText
+        //   from whatever it was hooked to before, if anything.
+        utext_openUChars(ut, NULL, 0, status);
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        return ut;
+    }
     ut = utext_setup(ut, 0, status);
     //    note:  use the standard (writable) function table for UnicodeString.
     //           The flag settings disable writing, so having the functions in
diff --git a/icu4c/source/test/intltest/regextst.cpp b/icu4c/source/test/intltest/regextst.cpp
index 34fda5c93aa..20c8e235888 100644
--- a/icu4c/source/test/intltest/regextst.cpp
+++ b/icu4c/source/test/intltest/regextst.cpp
@@ -108,6 +108,9 @@ void RegexTest::runIndexedTest( int32_t index, UBool exec, const char* &name, ch
         case 17: name = "Bug 7740";
             if (exec) Bug7740();
             break;
+        case 18: name = "Bug 8479";
+            if (exec) Bug8479();
+            break;
 
         default: name = "";
             break; //needed to end loop
@@ -5046,7 +5049,24 @@ void RegexTest::Bug7740() {
     delete m;
 }
 
+// Bug 8479:  was crashing whith a Bogus UnicodeString as input.
 
+void RegexTest::Bug8479() {
+    UErrorCode status = U_ZERO_ERROR;
+
+    RegexMatcher* const pMatcher = new RegexMatcher("\\Aboo\\z", UREGEX_DOTALL|UREGEX_CASE_INSENSITIVE, status);
+    REGEX_CHECK_STATUS;
+    if (U_SUCCESS(status))
+    {
+        UnicodeString str;
+        str.setToBogus();
+        pMatcher->reset(str);
+        status = U_ZERO_ERROR;
+        pMatcher->matches(status);
+        REGEX_ASSERT(status == U_ILLEGAL_ARGUMENT_ERROR);
+        delete pMatcher;
+    }
+}
      
 
 #endif  /* !UCONFIG_NO_REGULAR_EXPRESSIONS  */
diff --git a/icu4c/source/test/intltest/regextst.h b/icu4c/source/test/intltest/regextst.h
index cbae0e4a152..10ebb269e6c 100644
--- a/icu4c/source/test/intltest/regextst.h
+++ b/icu4c/source/test/intltest/regextst.h
@@ -1,6 +1,6 @@
 /********************************************************************
  * COPYRIGHT:
- * Copyright (c) 2002-2010, International Business Machines Corporation and
+ * Copyright (c) 2002-2011, International Business Machines Corporation and
  * others. All Rights Reserved.
  ********************************************************************/
 
@@ -43,6 +43,7 @@ public:
     virtual void PreAllocatedUTextCAPI();
     virtual void Bug7651();
     virtual void Bug7740();
+    virtual void Bug8479();
 
     // The following functions are internal to the regexp tests.
     virtual void assertUText(const char *expected, UText *actual, const char *file, int line);