ICU-66108.tar.gz

[apple/icu.git] / icuSources / i18n / uspoof_impl.cpp

diff --git a/icuSources/i18n/uspoof_impl.cpp b/icuSources/i18n/uspoof_impl.cpp
index 745d5e897d5d579325a7f202403297aeb2568cb3..88245b7f8f9c2218fc77c87cbb3b86f9b0a9a4a1 100644 (file)
--- a/icuSources/i18n/uspoof_impl.cpp
+++ b/icuSources/i18n/uspoof_impl.cpp
@@ -52,7 +52,6 @@ SpoofImpl::SpoofImpl() {
 }
 
 void SpoofImpl::construct(UErrorCode& status) {
 }
 
 void SpoofImpl::construct(UErrorCode& status) {

-    fMagic = USPOOF_MAGIC;

     fChecks = USPOOF_ALL_CHECKS;
     fSpoofData = NULL;
     fAllowedCharsSet = NULL;
     fChecks = USPOOF_ALL_CHECKS;
     fSpoofData = NULL;
     fAllowedCharsSet = NULL;


@@ -74,17 +73,16 @@ void SpoofImpl::construct(UErrorCode& status) {
 
 // Copy Constructor, used by the user level clone() function.
 SpoofImpl::SpoofImpl(const SpoofImpl &src, UErrorCode &status)  :
 
 // Copy Constructor, used by the user level clone() function.
 SpoofImpl::SpoofImpl(const SpoofImpl &src, UErrorCode &status)  :

-        fMagic(0), fChecks(USPOOF_ALL_CHECKS), fSpoofData(NULL), fAllowedCharsSet(NULL) , 
+        fChecks(USPOOF_ALL_CHECKS), fSpoofData(NULL), fAllowedCharsSet(NULL) , 

         fAllowedLocales(NULL) {
     if (U_FAILURE(status)) {
         return;
     }
         fAllowedLocales(NULL) {
     if (U_FAILURE(status)) {
         return;
     }

-    fMagic = src.fMagic;

     fChecks = src.fChecks;
     if (src.fSpoofData != NULL) {
         fSpoofData = src.fSpoofData->addReference();
     }
     fChecks = src.fChecks;
     if (src.fSpoofData != NULL) {
         fSpoofData = src.fSpoofData->addReference();
     }

-    fAllowedCharsSet = static_cast<const UnicodeSet *>(src.fAllowedCharsSet->clone());
+    fAllowedCharsSet = src.fAllowedCharsSet->clone();

     fAllowedLocales = uprv_strdup(src.fAllowedLocales);
     if (fAllowedCharsSet == NULL || fAllowedLocales == NULL) {
         status = U_MEMORY_ALLOCATION_ERROR;
     fAllowedLocales = uprv_strdup(src.fAllowedLocales);
     if (fAllowedCharsSet == NULL || fAllowedLocales == NULL) {
         status = U_MEMORY_ALLOCATION_ERROR;


@@ -93,8 +91,6 @@ SpoofImpl::SpoofImpl(const SpoofImpl &src, UErrorCode &status)  :
 }
 
 SpoofImpl::~SpoofImpl() {
 }
 
 SpoofImpl::~SpoofImpl() {

-    fMagic = 0;                // head off application errors by preventing use of
-                               //    of deleted objects.

     if (fSpoofData != NULL) {
         fSpoofData->removeReference();   // Will delete if refCount goes to zero.
     }
     if (fSpoofData != NULL) {
         fSpoofData->removeReference();   // Will delete if refCount goes to zero.
     }


@@ -104,7 +100,7 @@ SpoofImpl::~SpoofImpl() {
 
 //  Cast this instance as a USpoofChecker for the C API.
 USpoofChecker *SpoofImpl::asUSpoofChecker() {
 
 //  Cast this instance as a USpoofChecker for the C API.
 USpoofChecker *SpoofImpl::asUSpoofChecker() {

-    return reinterpret_cast<USpoofChecker*>(this);
+    return exportForC();

 }
 
 //
 }
 
 //


@@ -112,18 +108,10 @@ USpoofChecker *SpoofImpl::asUSpoofChecker() {
 //    received from the C API.
 //
 const SpoofImpl *SpoofImpl::validateThis(const USpoofChecker *sc, UErrorCode &status) {
 //    received from the C API.
 //
 const SpoofImpl *SpoofImpl::validateThis(const USpoofChecker *sc, UErrorCode &status) {

+    auto* This = validate(sc, status);

     if (U_FAILURE(status)) {
         return NULL;
     }
     if (U_FAILURE(status)) {
         return NULL;
     }

-    if (sc == NULL) {
-        status = U_ILLEGAL_ARGUMENT_ERROR;
-        return NULL;
-    }
-    SpoofImpl *This = (SpoofImpl *)sc;
-    if (This->fMagic != USPOOF_MAGIC) {
-        status = U_INVALID_FORMAT_ERROR;
-        return NULL;
-    }

     if (This->fSpoofData != NULL && !This->fSpoofData->validateDataVersion(status)) {
         return NULL;
     }
     if (This->fSpoofData != NULL && !This->fSpoofData->validateDataVersion(status)) {
         return NULL;
     }


@@ -205,7 +193,7 @@ void SpoofImpl::setAllowedLocales(const char *localesList, UErrorCode &status) {
     }
 
     // Store the updated spoof checker state.
     }
 
     // Store the updated spoof checker state.

-    tmpSet = static_cast<UnicodeSet *>(allowedChars.clone());
+    tmpSet = allowedChars.clone();

     const char *tmpLocalesList = uprv_strdup(localesList);
     if (tmpSet == NULL || tmpLocalesList == NULL) {
         status = U_MEMORY_ALLOCATION_ERROR;
     const char *tmpLocalesList = uprv_strdup(localesList);
     if (tmpSet == NULL || tmpLocalesList == NULL) {
         status = U_MEMORY_ALLOCATION_ERROR;


@@ -377,6 +365,43 @@ URestrictionLevel SpoofImpl::getRestrictionLevel(const UnicodeString& input, UEr
     return USPOOF_MINIMALLY_RESTRICTIVE;
 }
 
     return USPOOF_MINIMALLY_RESTRICTIVE;
 }
 

+int32_t SpoofImpl::findHiddenOverlay(const UnicodeString& input, UErrorCode&) const {
+    bool sawLeadCharacter = false;
+    for (int32_t i=0; i<input.length();) {
+        UChar32 cp = input.char32At(i);
+        if (sawLeadCharacter && cp == 0x0307) {
+            return i;
+        }
+        uint8_t combiningClass = u_getCombiningClass(cp);
+        // Skip over characters except for those with combining class 0 (non-combining characters) or with
+        // combining class 230 (same class as U+0307)
+        U_ASSERT(u_getCombiningClass(0x0307) == 230);
+        if (combiningClass == 0 || combiningClass == 230) {
+            sawLeadCharacter = isIllegalCombiningDotLeadCharacter(cp);
+        }
+        i += U16_LENGTH(cp);
+    }
+    return -1;
+}
+
+static inline bool isIllegalCombiningDotLeadCharacterNoLookup(UChar32 cp) {
+    return cp == u'i' || cp == u'j' || cp == u'ı' || cp == u'ȷ' || cp == u'l' ||
+           u_hasBinaryProperty(cp, UCHAR_SOFT_DOTTED);
+}
+
+bool SpoofImpl::isIllegalCombiningDotLeadCharacter(UChar32 cp) const {
+    if (isIllegalCombiningDotLeadCharacterNoLookup(cp)) {
+        return true;
+    }
+    UnicodeString skelStr;
+    fSpoofData->confusableLookup(cp, skelStr);
+    UChar32 finalCp = skelStr.char32At(skelStr.moveIndex32(skelStr.length(), -1));
+    if (finalCp != cp && isIllegalCombiningDotLeadCharacterNoLookup(finalCp)) {
+        return true;
+    }
+    return false;
+}
+

 
 
 // Convert a text format hex number.  Utility function used by builder code.  Static.
 
 
 // Convert a text format hex number.  Utility function used by builder code.  Static.


@@ -417,12 +442,12 @@ UChar32 SpoofImpl::ScanHex(const UChar *s, int32_t start, int32_t limit, UErrorC
 //
 //-----------------------------------------
 
 //
 //-----------------------------------------
 

-CheckResult::CheckResult() : fMagic(USPOOF_CHECK_MAGIC) {
+CheckResult::CheckResult() {

     clear();
 }
 
 USpoofCheckResult* CheckResult::asUSpoofCheckResult() {
     clear();
 }
 
 USpoofCheckResult* CheckResult::asUSpoofCheckResult() {

-    return reinterpret_cast<USpoofCheckResult*>(this);
+    return exportForC();

 }
 
 //
 }
 
 //


@@ -430,22 +455,11 @@ USpoofCheckResult* CheckResult::asUSpoofCheckResult() {
 //    received from the C API.
 //
 const CheckResult* CheckResult::validateThis(const USpoofCheckResult *ptr, UErrorCode &status) {
 //    received from the C API.
 //
 const CheckResult* CheckResult::validateThis(const USpoofCheckResult *ptr, UErrorCode &status) {

-    if (U_FAILURE(status)) { return NULL; }
-    if (ptr == NULL) {
-        status = U_ILLEGAL_ARGUMENT_ERROR;
-        return NULL;
-    }
-    CheckResult *This = (CheckResult*) ptr;
-    if (This->fMagic != USPOOF_CHECK_MAGIC) {
-        status = U_INVALID_FORMAT_ERROR;
-        return NULL;
-    }
-    return This;
+    return validate(ptr, status);

 }
 
 CheckResult* CheckResult::validateThis(USpoofCheckResult *ptr, UErrorCode &status) {
 }
 
 CheckResult* CheckResult::validateThis(USpoofCheckResult *ptr, UErrorCode &status) {

-    return const_cast<CheckResult *>
-        (CheckResult::validateThis(const_cast<const USpoofCheckResult*>(ptr), status));
+    return validate(ptr, status);

 }
 
 void CheckResult::clear() {
 }
 
 void CheckResult::clear() {


@@ -532,24 +546,25 @@ uspoof_cleanupDefaultData(void) {
     if (gDefaultSpoofData) {
         // Will delete, assuming all user-level spoof checkers were closed.
         gDefaultSpoofData->removeReference();
     if (gDefaultSpoofData) {
         // Will delete, assuming all user-level spoof checkers were closed.
         gDefaultSpoofData->removeReference();

-        gDefaultSpoofData = NULL;
+        gDefaultSpoofData = nullptr;

         gSpoofInitDefaultOnce.reset();
     }
     return TRUE;
 }
 
 static void U_CALLCONV uspoof_loadDefaultData(UErrorCode& status) {
         gSpoofInitDefaultOnce.reset();
     }
     return TRUE;
 }
 
 static void U_CALLCONV uspoof_loadDefaultData(UErrorCode& status) {

-    UDataMemory *udm = udata_openChoice(NULL, "cfu", "confusables",
+    UDataMemory *udm = udata_openChoice(nullptr, "cfu", "confusables",

                                         spoofDataIsAcceptable, 
                                         spoofDataIsAcceptable, 

-                                        NULL,       // context, would receive dataVersion if supplied.
+                                        nullptr,       // context, would receive dataVersion if supplied.

                                         &status);
     if (U_FAILURE(status)) { return; }
     gDefaultSpoofData = new SpoofData(udm, status);
     if (U_FAILURE(status)) {
         delete gDefaultSpoofData;
                                         &status);
     if (U_FAILURE(status)) { return; }
     gDefaultSpoofData = new SpoofData(udm, status);
     if (U_FAILURE(status)) {
         delete gDefaultSpoofData;

+        gDefaultSpoofData = nullptr;

         return;
     }
         return;
     }

-    if (gDefaultSpoofData == NULL) {
+    if (gDefaultSpoofData == nullptr) {

         status = U_MEMORY_ALLOCATION_ERROR;
         return;
     }
         status = U_MEMORY_ALLOCATION_ERROR;
         return;
     }


@@ -590,6 +605,10 @@ SpoofData::SpoofData(const void *data, int32_t length, UErrorCode &status)
         status = U_INVALID_FORMAT_ERROR;
         return;
     }
         status = U_INVALID_FORMAT_ERROR;
         return;
     }

+    if (data == NULL) {
+        status = U_ILLEGAL_ARGUMENT_ERROR;
+        return;
+    }

     void *ncData = const_cast<void *>(data);
     fRawData = static_cast<SpoofDataHeader *>(ncData);
     if (length < fRawData->fLength) {
     void *ncData = const_cast<void *>(data);
     fRawData = static_cast<SpoofDataHeader *>(ncData);
     if (length < fRawData->fLength) {


@@ -710,9 +729,7 @@ void *SpoofData::reserveSpace(int32_t numBytes,  UErrorCode &status) {
         return NULL;
     }
     if (!fDataOwned) {
         return NULL;
     }
     if (!fDataOwned) {

-        U_ASSERT(FALSE);
-        status = U_INTERNAL_PROGRAM_ERROR;
-        return NULL;
+        UPRV_UNREACHABLE;

     }
 
     numBytes = (numBytes + 15) & ~15;   // Round up to a multiple of 16
     }
 
     numBytes = (numBytes + 15) & ~15;   // Round up to a multiple of 16