ICU-62123.0.1.tar.gz

[apple/icu.git] / icuSources / common / normalizer2.cpp

diff --git a/icuSources/common/normalizer2.cpp b/icuSources/common/normalizer2.cpp
index eacc57168bb1632bd1425a7fa8ccd01aa66ac93c..0f12dfcff42fbdab0dcb7edc82b4d8691f3c3725 100644 (file)
--- a/icuSources/common/normalizer2.cpp
+++ b/icuSources/common/normalizer2.cpp
@@ -1,12 +1,14 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html

 /*
 *******************************************************************************
 *
 /*
 *******************************************************************************
 *

-*   Copyright (C) 2009-2014, International Business Machines
+*   Copyright (C) 2009-2016, International Business Machines

 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
 *   file name:  normalizer2.cpp
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
 *   file name:  normalizer2.cpp

-*   encoding:   US-ASCII
+*   encoding:   UTF-8

 *   tab size:   8 (not used)
 *   indentation:4
 *
 *   tab size:   8 (not used)
 *   indentation:4
 *


@@ -18,7 +20,9 @@
 
 #if !UCONFIG_NO_NORMALIZATION
 
 
 #if !UCONFIG_NO_NORMALIZATION
 

+#include "unicode/edits.h"

 #include "unicode/normalizer2.h"
 #include "unicode/normalizer2.h"

+#include "unicode/stringoptions.h"

 #include "unicode/unistr.h"
 #include "unicode/unorm.h"
 #include "cstring.h"
 #include "unicode/unistr.h"
 #include "unicode/unorm.h"
 #include "cstring.h"


@@ -31,6 +35,7 @@
 using icu::Normalizer2Impl;
 
 // NFC/NFD data machine-generated by gennorm2 --csource
 using icu::Normalizer2Impl;
 
 // NFC/NFD data machine-generated by gennorm2 --csource

+#define INCLUDED_FROM_NORMALIZER2_CPP

 #include "norm2_nfc_data.h"
 
 U_NAMESPACE_BEGIN
 #include "norm2_nfc_data.h"
 
 U_NAMESPACE_BEGIN


@@ -39,6 +44,20 @@ U_NAMESPACE_BEGIN
 
 Normalizer2::~Normalizer2() {}
 
 
 Normalizer2::~Normalizer2() {}
 

+void
+Normalizer2::normalizeUTF8(uint32_t /*options*/, StringPiece src, ByteSink &sink,
+                           Edits *edits, UErrorCode &errorCode) const {
+    if (U_FAILURE(errorCode)) {
+        return;
+    }
+    if (edits != nullptr) {
+        errorCode = U_UNSUPPORTED_ERROR;
+        return;
+    }
+    UnicodeString src16 = UnicodeString::fromUTF8(src);
+    normalize(src16, errorCode).toUTF8(sink);
+}
+

 UBool
 Normalizer2::getRawDecomposition(UChar32, UnicodeString &) const {
     return FALSE;
 UBool
 Normalizer2::getRawDecomposition(UChar32, UnicodeString &) const {
     return FALSE;


@@ -54,6 +73,11 @@ Normalizer2::getCombiningClass(UChar32 /*c*/) const {
     return 0;
 }
 
     return 0;
 }
 

+UBool
+Normalizer2::isNormalizedUTF8(StringPiece s, UErrorCode &errorCode) const {
+    return U_SUCCESS(errorCode) && isNormalized(UnicodeString::fromUTF8(s), errorCode);
+}
+

 // Normalizer2 implementation for the old UNORM_NONE.
 class NoopNormalizer2 : public Normalizer2 {
     virtual ~NoopNormalizer2();
 // Normalizer2 implementation for the old UNORM_NONE.
 class NoopNormalizer2 : public Normalizer2 {
     virtual ~NoopNormalizer2();


@@ -61,7 +85,7 @@ class NoopNormalizer2 : public Normalizer2 {
     virtual UnicodeString &
     normalize(const UnicodeString &src,
               UnicodeString &dest,
     virtual UnicodeString &
     normalize(const UnicodeString &src,
               UnicodeString &dest,

-              UErrorCode &errorCode) const {
+              UErrorCode &errorCode) const U_OVERRIDE {

         if(U_SUCCESS(errorCode)) {
             if(&dest!=&src) {
                 dest=src;
         if(U_SUCCESS(errorCode)) {
             if(&dest!=&src) {
                 dest=src;


@@ -71,10 +95,27 @@ class NoopNormalizer2 : public Normalizer2 {
         }
         return dest;
     }
         }
         return dest;
     }

+    virtual void
+    normalizeUTF8(uint32_t options, StringPiece src, ByteSink &sink,
+                  Edits *edits, UErrorCode &errorCode) const U_OVERRIDE {
+        if(U_SUCCESS(errorCode)) {
+            if (edits != nullptr) {
+                if ((options & U_EDITS_NO_RESET) == 0) {
+                    edits->reset();
+                }
+                edits->addUnchanged(src.length());
+            }
+            if ((options & U_OMIT_UNCHANGED_TEXT) == 0) {
+                sink.Append(src.data(), src.length());
+            }
+            sink.Flush();
+        }
+    }
+

     virtual UnicodeString &
     normalizeSecondAndAppend(UnicodeString &first,
                              const UnicodeString &second,
     virtual UnicodeString &
     normalizeSecondAndAppend(UnicodeString &first,
                              const UnicodeString &second,

-                             UErrorCode &errorCode) const {
+                             UErrorCode &errorCode) const U_OVERRIDE {

         if(U_SUCCESS(errorCode)) {
             if(&first!=&second) {
                 first.append(second);
         if(U_SUCCESS(errorCode)) {
             if(&first!=&second) {
                 first.append(second);


@@ -87,7 +128,7 @@ class NoopNormalizer2 : public Normalizer2 {
     virtual UnicodeString &
     append(UnicodeString &first,
            const UnicodeString &second,
     virtual UnicodeString &
     append(UnicodeString &first,
            const UnicodeString &second,

-           UErrorCode &errorCode) const {
+           UErrorCode &errorCode) const U_OVERRIDE {

         if(U_SUCCESS(errorCode)) {
             if(&first!=&second) {
                 first.append(second);
         if(U_SUCCESS(errorCode)) {
             if(&first!=&second) {
                 first.append(second);


@@ -98,25 +139,29 @@ class NoopNormalizer2 : public Normalizer2 {
         return first;
     }
     virtual UBool
         return first;
     }
     virtual UBool

-    getDecomposition(UChar32, UnicodeString &) const {
+    getDecomposition(UChar32, UnicodeString &) const U_OVERRIDE {

         return FALSE;
     }
         return FALSE;
     }

-    // No need to override the default getRawDecomposition().
+    // No need to U_OVERRIDE the default getRawDecomposition().
+    virtual UBool
+    isNormalized(const UnicodeString &, UErrorCode &errorCode) const U_OVERRIDE {
+        return U_SUCCESS(errorCode);
+    }

     virtual UBool
     virtual UBool

-    isNormalized(const UnicodeString &, UErrorCode &) const {
-        return TRUE;
+    isNormalizedUTF8(StringPiece, UErrorCode &errorCode) const U_OVERRIDE {
+        return U_SUCCESS(errorCode);

     }
     virtual UNormalizationCheckResult
     }
     virtual UNormalizationCheckResult

-    quickCheck(const UnicodeString &, UErrorCode &) const {
+    quickCheck(const UnicodeString &, UErrorCode &) const U_OVERRIDE {

         return UNORM_YES;
     }
     virtual int32_t
         return UNORM_YES;
     }
     virtual int32_t

-    spanQuickCheckYes(const UnicodeString &s, UErrorCode &) const {
+    spanQuickCheckYes(const UnicodeString &s, UErrorCode &) const U_OVERRIDE {

         return s.length();
     }
         return s.length();
     }

-    virtual UBool hasBoundaryBefore(UChar32) const { return TRUE; }
-    virtual UBool hasBoundaryAfter(UChar32) const { return TRUE; }
-    virtual UBool isInert(UChar32) const { return TRUE; }
+    virtual UBool hasBoundaryBefore(UChar32) const U_OVERRIDE { return TRUE; }
+    virtual UBool hasBoundaryAfter(UChar32) const U_OVERRIDE { return TRUE; }
+    virtual UBool isInert(UChar32) const U_OVERRIDE { return TRUE; }

 };
 
 NoopNormalizer2::~NoopNormalizer2() {}
 };
 
 NoopNormalizer2::~NoopNormalizer2() {}