ICU-3337 better detect incorrect usage of Regex APIs.

X-SVN-Rev: 16595

icu4c/source/i18n/rematch.cpp

@@ -721,6 +721,11 @@ RegexMatcher &RegexMatcher::reset(const UnicodeString &input) {
     return *this;
 }
 
+RegexMatcher &RegexMatcher::reset(const UChar *) {
+    fDeferredStatus = U_INTERNAL_PROGRAM_ERROR;
+    return *this;
+}
+
 
 RegexMatcher &RegexMatcher::reset(int32_t position, UErrorCode &status) {
     if (U_FAILURE(status)) {

icu4c/source/i18n/repattrn.cpp

@@ -324,6 +324,15 @@ RegexMatcher *RegexPattern::matcher(const UnicodeString &input,
     return retMatcher;
 };
 
+RegexMatcher *RegexPattern::matcher(const UChar *input,
+                                    UErrorCode          &status)  const 
+{
+    RegexMatcher *retMatcher = NULL;
+    if (U_SUCCESS(status)) {
+        status = U_UNSUPPORTED_ERROR;
+    }
+    return retMatcher;
+}
 
 
 //---------------------------------------------------------------------

icu4c/source/i18n/unicode/regex.h

@@ -247,6 +247,22 @@ public:
     virtual RegexMatcher *matcher(const UnicodeString &input,
         UErrorCode          &status) const;
 
+private:
+    /**
+     * Cause a compilation error if an application accidently attempts to
+     *   create a matcher with a (UChar *) string as input rather than
+     *   a UnicodeString.  Avoids a dangling reference to a temporary string.
+     * <p>
+     * To efficiently work with UChar *strings, wrap the data in a UnicodeString
+     * using one of the aliasing constructors, such as
+     * <code>UnicodeString(UBool isTerminated, const UChar *text, int32_t textLength);</code>
+     *
+     * @internal
+     */
+    virtual RegexMatcher *matcher(const UChar *input,
+        UErrorCode          &status) const;
+public:
+
 
    /**
     * Creates a RegexMatcher that will match against this pattern.  The
@@ -446,6 +462,22 @@ public:
     RegexMatcher(const UnicodeString &regexp, const UnicodeString &input,
         uint32_t flags, UErrorCode &status);
 
+private:
+    /**
+     * Cause a compilation error if an application accidently attempts to
+     *   create a matcher with a (UChar *) string as input rather than
+     *   a UnicodeString.    Avoids a dangling reference to a temporary string.
+     * <p>
+     * To efficiently work with UChar *strings, wrap the data in a UnicodeString
+     * using one of the aliasing constructors, such as
+     * <code>UnicodeString(UBool isTerminated, const UChar *text, int32_t textLength);</code>
+     *
+     * @internal
+     */
+    RegexMatcher(const UnicodeString &regexp, const UChar *input,
+        uint32_t flags, UErrorCode &status);
+public:
+
 
    /**
     *   Destructor.
@@ -654,6 +686,20 @@ public:
     */
     virtual RegexMatcher &reset(const UnicodeString &input);
 
+private:
+    /**
+     * Cause a compilation error if an application accidently attempts to
+     *   reset a matcher with a (UChar *) string as input rather than
+     *   a UnicodeString.    Avoids a dangling reference to a temporary string.
+     * <p>
+     * To efficiently work with UChar *strings, wrap the data in a UnicodeString
+     * using one of the aliasing constructors, such as
+     * <code>UnicodeString(UBool isTerminated, const UChar *text, int32_t textLength);</code>
+     *
+     * @internal
+     */
+    virtual RegexMatcher &reset(const UChar *input);
+public:
 
    /**
     *   Returns the input string being matched.  The returned string is not a copy,
@@ -821,7 +867,6 @@ private:
     friend class RegexPattern;
     friend class RegexCImpl;
 
-
     //
     //  MatchAt   This is the internal interface to the match engine itself.
     //            Match status comes back in matcher member variables.

icu4c/source/test/intltest/regextst.cpp

@@ -940,6 +940,26 @@ void RegexTest::API_Match() {
         delete p;
     }
 
+    //
+    // Compilation error on reset with UChar *
+    //   These were a hazard that people were stumbling over with runtime errors.
+    //   Changed them to compiler errors by adding private methods that more closely
+    //   matched the incorrect use of the functions.
+    //
+#if 0
+    {
+        UErrorCode status = U_ZERO_ERROR;
+        UChar ucharString[20];
+        RegexMatcher m(".", 0, status);
+        m.reset(ucharString);  // should not compile.
+
+        RegexPattern *p = RegexPattern::compile(".", 0, status);
+        RegexMatcher *m2 = p->matcher(ucharString, status);    //  should not compile.
+
+        RegexMatcher m3(".", ucharString, 0, status);  //  Should not compile
+    }
+#endif
+
 }
 
 
@@ -1362,7 +1382,8 @@ void RegexTest::API_Pattern() {
     REGEX_CHECK_STATUS;
     REGEX_ASSERT(pat1->getDynamicClassID() == RegexPattern::getStaticClassID());
     REGEX_ASSERT(pat1->getDynamicClassID() != NULL);
-    RegexMatcher *m = pat1->matcher("Hello, World", status);
+    UnicodeString Hello("Hello, world.");
+    RegexMatcher *m = pat1->matcher(Hello, status);
     REGEX_ASSERT(pat1->getDynamicClassID() != m->getDynamicClassID());
     REGEX_ASSERT(m->getDynamicClassID() == RegexMatcher::getStaticClassID());
     REGEX_ASSERT(m->getDynamicClassID() != NULL);
@@ -1766,7 +1787,7 @@ void RegexTest::PerlTests() {
     //   Coming out, capture group 2 is the pattern, capture group 3 is the flags.
     //
     RegexPattern *flagPat = RegexPattern::compile("('?)(.*)\\1(.*)", 0, pe, status);
-    RegexMatcher* flagMat = flagPat->matcher("", status);
+    RegexMatcher* flagMat = flagPat->matcher(status);
 
     //
     // The Perl tests reference several perl-isms, which are evaluated/substituted
@@ -1783,11 +1804,11 @@ void RegexTest::PerlTests() {
 
     //  regexp for $-[0], $+[2], etc.
     RegexPattern *groupsPat = RegexPattern::compile("\\$([+\\-])\\[(\\d+)\\]", 0, pe, status);
-    RegexMatcher *groupsMat = groupsPat->matcher("", status);
+    RegexMatcher *groupsMat = groupsPat->matcher(status);
     
     //  regexp for $0, $1, $2, etc.
     RegexPattern *cgPat = RegexPattern::compile("\\$(\\d+)", 0, pe, status);
-    RegexMatcher *cgMat = cgPat->matcher("", status);
+    RegexMatcher *cgMat = cgPat->matcher(status);
 
 
     //