ICU 56 update step 1

source/test/intltest/usettest.cpp

 /*
 ********************************************************************************
-*   Copyright (C) 1999-2014 International Business Machines Corporation and
+*   Copyright (C) 1999-2015 International Business Machines Corporation and
 *   others. All Rights Reserved.
 ********************************************************************************
 *   Date        Name        Description
 *   10/20/99    alan        Creation.
 *   03/22/2000  Madhu       Added additional tests
 ********************************************************************************
 */
 
 #include <stdio.h>
 
@@ skipping 69 matching lines @@
         CASE(14,TestCloseOver);
         CASE(15,TestEscapePattern);
         CASE(16,TestInvalidCodePoint);
         CASE(17,TestSymbolTable);
         CASE(18,TestSurrogate);
         CASE(19,TestPosixClasses);
         CASE(20,TestIteration);
         CASE(21,TestFreezable);
         CASE(22,TestSpan);
         CASE(23,TestStringSpan);
+        CASE(24,TestUCAUnsafeBackwards);
         default: name = ""; break;
     }
 }
 
 static const char NOT[] = "%%%%";
 
 /** 
  * UVector was improperly copying contents
  * This code will crash this is still true
  */
@@ skipping 1604 matching lines @@
             errln("FAIL: UnicodeSet constructor");
             continue;
         }
         expectContainment(set,
                           CharsToUnicodeString("abc\\U00010000"),
                           CharsToUnicodeString("\\uD800;\\uDC00")); // split apart surrogate-pair
         if (set.size() != 4) {
             errln((UnicodeString)"FAIL: " + UnicodeString(DATA[i], -1, US_INV) + ".size() == " + 
                   set.size() + ", expected 4");
         }
+
+        {
+          UErrorCode subErr = U_ZERO_ERROR;
+          checkRoundTrip(set);
+          checkSerializeRoundTrip(set, subErr);
+        }
     }
 }
 
 void UnicodeSetTest::TestExhaustive() {
     // exhaustive tests. Simulate UnicodeSets with integers.
     // That gives us very solid tests (except for large memory tests).
 
     int32_t limit = 128;
 
     UnicodeSet x, y, z, aa;
 
     for (int32_t i = 0; i < limit; ++i) {
         bitsToSet(i, x);
         logln((UnicodeString)"Testing " + i + ", " + x);
         _testComplement(i, x, y);
 
+        UnicodeSet &toTest = bitsToSet(i, aa);
+
         // AS LONG AS WE ARE HERE, check roundtrip
-        checkRoundTrip(bitsToSet(i, aa));
+        checkRoundTrip(toTest);
+        UErrorCode ec = U_ZERO_ERROR;
+        checkSerializeRoundTrip(toTest, ec);
 
         for (int32_t j = 0; j < limit; ++j) {
             _testAdd(i,j,  x,y,z);
             _testXor(i,j,  x,y,z);
             _testRetain(i,j,  x,y,z);
             _testRemove(i,j,  x,y,z);
         }
     }
 }
 
@@ skipping 138 matching lines @@
     }
     return pairs;
 }
 
 /**
  * Basic consistency check for a few items.
  * That the iterator works, and that we can create a pattern and
  * get the same thing back
  */
 void UnicodeSetTest::checkRoundTrip(const UnicodeSet& s) {
-    UErrorCode ec = U_ZERO_ERROR;
+    {
+        UnicodeSet t(s);
+        checkEqual(s, t, "copy ct");
+    }
 
-    UnicodeSet t(s);
-    checkEqual(s, t, "copy ct");
+    {
+        UnicodeSet t(0xabcd, 0xdef0);  // dummy contents should be overwritten
+        t = s;
+        checkEqual(s, t, "operator=");
+    }
 
-    t = s;
-    checkEqual(s, t, "operator=");
+    {
+        UnicodeSet t;
+        copyWithIterator(t, s, FALSE);
+        checkEqual(s, t, "iterator roundtrip");
+    }
 
-    copyWithIterator(t, s, FALSE);
-    checkEqual(s, t, "iterator roundtrip");
+    {
+        UnicodeSet t;
+        copyWithIterator(t, s, TRUE); // try range
+        checkEqual(s, t, "iterator roundtrip");
+    }
 
-    copyWithIterator(t, s, TRUE); // try range
-    checkEqual(s, t, "iterator roundtrip");
-        
-    UnicodeString pat; s.toPattern(pat, FALSE);
-    t.applyPattern(pat, ec);
-    if (U_FAILURE(ec)) {
-        errln("FAIL: applyPattern");
-        return;
-    } else {
-        checkEqual(s, t, "toPattern(false)");
+    {
+        UnicodeSet t;
+        UnicodeString pat;
+        UErrorCode ec = U_ZERO_ERROR;
+        s.toPattern(pat, FALSE);
+        t.applyPattern(pat, ec);
+        if (U_FAILURE(ec)) {
+            errln("FAIL: toPattern(escapeUnprintable=FALSE), applyPattern - %s", u_errorName(ec));
+            return;
+        } else {
+            checkEqual(s, t, "toPattern(false)");
+        }
     }
-        
-    s.toPattern(pat, TRUE);
-    t.applyPattern(pat, ec);
-    if (U_FAILURE(ec)) {
-        errln("FAIL: applyPattern");
-        return;
-    } else {
-        checkEqual(s, t, "toPattern(true)");
+
+    {
+        UnicodeSet t;
+        UnicodeString pat;
+        UErrorCode ec = U_ZERO_ERROR;
+        s.toPattern(pat, TRUE);
+        t.applyPattern(pat, ec);
+        if (U_FAILURE(ec)) {
+            errln("FAIL: toPattern(escapeUnprintable=TRUE), applyPattern - %s", u_errorName(ec));
+            return;
+        } else {
+            checkEqual(s, t, "toPattern(true)");
+        }
     }
 }
-    
+
+void UnicodeSetTest::checkSerializeRoundTrip(const UnicodeSet& t, UErrorCode &status) {
+  if(U_FAILURE(status)) return;
+  int32_t len = t.serialize(serializeBuffer.getAlias(), serializeBuffer.getCapacity(), status);
+  if(status == U_BUFFER_OVERFLOW_ERROR) {
+    status = U_ZERO_ERROR;
+    serializeBuffer.resize(len);
+    len = t.serialize(serializeBuffer.getAlias(), serializeBuffer.getCapacity(), status);
+    // let 2nd error stand
+  }
+  if(U_FAILURE(status)) {
+    errln("checkSerializeRoundTrip: error %s serializing buffer\n", u_errorName(status));
+    return;
+  }
+  UnicodeSet deserialized(serializeBuffer.getAlias(), len, UnicodeSet::kSerialized, status);
+  if(U_FAILURE(status)) {
+    errln("checkSerializeRoundTrip: error %s deserializing buffer: buf %p len %d, original %d\n", u_errorName(status), serializeBuffer.getAlias(), len, t.getRangeCount());
+    return;
+  }
+
+  checkEqual(t, deserialized, "Set was unequal when deserialized");
+}
+
 void UnicodeSetTest::copyWithIterator(UnicodeSet& t, const UnicodeSet& s, UBool withRange) {
     t.clear();
     UnicodeSetIterator it(s);
     if (withRange) {
         while (it.nextRange()) {
             if (it.isString()) {
                 t.add(it.getString());
             } else {
                 t.add(it.getCodepoint(), it.getCodepointEnd());
             }
         }
     } else {
         while (it.next()) {
             if (it.isString()) {
                 t.add(it.getString());
             } else {
                 t.add(it.getCodepoint());
             }
         }
     }
 }
     
 UBool UnicodeSetTest::checkEqual(const UnicodeSet& s, const UnicodeSet& t, const char* message) {
+  assertEquals(UnicodeString("RangeCount: ","") + message, s.getRangeCount(), t.getRangeCount());
+  assertEquals(UnicodeString("size: ","") + message, s.size(), t.size());
     UnicodeString source; s.toPattern(source, TRUE);
     UnicodeString result; t.toPattern(result, TRUE);
     if (s != t) {
         errln((UnicodeString)"FAIL: " + message
               + "; source = " + source
               + "; result = " + result
               );
         return FALSE;
     } else {
         logln((UnicodeString)"Ok: " + message
@@ skipping 1846 matching lines @@
     string16=UNICODE_STRING_SIMPLE("abbcdabcdabd");
     s16=string16.getBuffer();
     length16=string16.length();
     if( set.spanBack(s16, 12, USET_SPAN_CONTAINED)!=0 ||
         set.spanBack(s16, 12, USET_SPAN_SIMPLE)!=6 ||
         set.spanBack(s16, 5, USET_SPAN_SIMPLE)!=0
     ) {
         errln("FAIL: UnicodeSet(%s).spanBack(while longest match) returns the wrong value", pattern);
     }
 }
+
+/**
+ * Including collationroot.h fails here with
+1>c:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\include\driverspecs.h(142): error C2008: '$' : unexpected in macro definition
+ *  .. so, we skip this test on Windows.
+ * 
+ * the cause is that  intltest builds with /Za which disables language extensions - which means
+ *  windows header files can't be used.
+ */
+#if !UCONFIG_NO_COLLATION && !U_PLATFORM_HAS_WIN32_API
+#include "collationroot.h"
+#include "collationtailoring.h"
+#endif
+
+void UnicodeSetTest::TestUCAUnsafeBackwards() {
+#if U_PLATFORM_HAS_WIN32_API
+    infoln("Skipping TestUCAUnsafeBackwards() - can't include collationroot.h on Windows without language extensions!");
+#elif !UCONFIG_NO_COLLATION
+    UErrorCode errorCode = U_ZERO_ERROR;
+
+    // Get the unsafeBackwardsSet
+    const CollationCacheEntry *rootEntry = CollationRoot::getRootCacheEntry(errorCode);
+    if(U_FAILURE(errorCode)) {
+      dataerrln("FAIL: %s getting root cache entry", u_errorName(errorCode));
+      return;
+    }
+    //const UVersionInfo &version = rootEntry->tailoring->version;
+    const UnicodeSet *unsafeBackwardSet = rootEntry->tailoring->unsafeBackwardSet;
+
+    checkSerializeRoundTrip(*unsafeBackwardSet, errorCode);
+
+    if(!logKnownIssue("11891","UnicodeSet fails to round trip on CollationRoot...unsafeBackwards set")) {
+        // simple test case
+        // TODO(ticket #11891): Simplify this test function to this simple case. Rename it appropriately.
+        // TODO(ticket #11891): Port test to Java. Is this a bug there, too?
+        UnicodeSet surrogates;
+        surrogates.add(0xd83a);  // a lead surrogate
+        surrogates.add(0xdc00, 0xdfff);  // a range of trail surrogates
+        UnicodeString pat;
+        surrogates.toPattern(pat, FALSE);  // bad: [ 0xd83a, 0xdc00, 0x2d, 0xdfff ]
+        // TODO: Probably fix either UnicodeSet::_generatePattern() or _appendToPat()
+        // so that at least one type of surrogate code points are escaped,
+        // or (minimally) so that adjacent lead+trail surrogate code points are escaped.
+        errorCode = U_ZERO_ERROR;
+        UnicodeSet s2;
+        s2.applyPattern(pat, errorCode);  // looks like invalid range [ 0x1e800, 0x2d, 0xdfff ]
+        if(U_FAILURE(errorCode)) {
+            errln("FAIL: surrogates to/from pattern - %s", u_errorName(errorCode));
+        } else {
+            checkEqual(surrogates, s2, "surrogates to/from pattern");
+        }
+        // This occurs in the UCA unsafe-backwards set.
+        checkRoundTrip(*unsafeBackwardSet);
+    }
+#endif
+}