ICU-62108.0.1.tar.gz

[apple/icu.git] / icuSources / common / uts46.cpp

diff --git a/icuSources/common/uts46.cpp b/icuSources/common/uts46.cpp
index 0a9cc193eb0cecd8d4d4e90397aba9b69c40bfe0..5a23572eb64af20b0b69d2727900c950476856de 100644 (file)
--- a/icuSources/common/uts46.cpp
+++ b/icuSources/common/uts46.cpp
@@ -1,10 +1,12 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
 /*
 *******************************************************************************
-*   Copyright (C) 2010-2014, International Business Machines
+*   Copyright (C) 2010-2015, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *******************************************************************************
 *   file name:  uts46.cpp
-*   encoding:   US-ASCII
+*   encoding:   UTF-8
 *   tab size:   8 (not used)
 *   indentation:4
 *
@@ -27,8 +29,6 @@
 #include "ubidi_props.h"
 #include "ustr_imp.h"
 
-#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
-
 // Note about tests for UIDNA_ERROR_DOMAIN_NAME_TOO_LONG:
 //
 // The domain name length limit is 255 octets in an internal DNS representation
@@ -70,7 +70,7 @@ isASCIIOkBiDi(const char *s, int32_t length);
 IDNA::~IDNA() {}
 
 void
-IDNA::labelToASCII_UTF8(const StringPiece &label, ByteSink &dest,
+IDNA::labelToASCII_UTF8(StringPiece label, ByteSink &dest,
                         IDNAInfo &info, UErrorCode &errorCode) const {
     if(U_SUCCESS(errorCode)) {
         UnicodeString destString;
@@ -80,7 +80,7 @@ IDNA::labelToASCII_UTF8(const StringPiece &label, ByteSink &dest,
 }
 
 void
-IDNA::labelToUnicodeUTF8(const StringPiece &label, ByteSink &dest,
+IDNA::labelToUnicodeUTF8(StringPiece label, ByteSink &dest,
                          IDNAInfo &info, UErrorCode &errorCode) const {
     if(U_SUCCESS(errorCode)) {
         UnicodeString destString;
@@ -90,7 +90,7 @@ IDNA::labelToUnicodeUTF8(const StringPiece &label, ByteSink &dest,
 }
 
 void
-IDNA::nameToASCII_UTF8(const StringPiece &name, ByteSink &dest,
+IDNA::nameToASCII_UTF8(StringPiece name, ByteSink &dest,
                        IDNAInfo &info, UErrorCode &errorCode) const {
     if(U_SUCCESS(errorCode)) {
         UnicodeString destString;
@@ -100,7 +100,7 @@ IDNA::nameToASCII_UTF8(const StringPiece &name, ByteSink &dest,
 }
 
 void
-IDNA::nameToUnicodeUTF8(const StringPiece &name, ByteSink &dest,
+IDNA::nameToUnicodeUTF8(StringPiece name, ByteSink &dest,
                         IDNAInfo &info, UErrorCode &errorCode) const {
     if(U_SUCCESS(errorCode)) {
         UnicodeString destString;
@@ -133,19 +133,19 @@ public:
                   IDNAInfo &info, UErrorCode &errorCode) const;
 
     virtual void
-    labelToASCII_UTF8(const StringPiece &label, ByteSink &dest,
+    labelToASCII_UTF8(StringPiece label, ByteSink &dest,
                       IDNAInfo &info, UErrorCode &errorCode) const;
 
     virtual void
-    labelToUnicodeUTF8(const StringPiece &label, ByteSink &dest,
+    labelToUnicodeUTF8(StringPiece label, ByteSink &dest,
                        IDNAInfo &info, UErrorCode &errorCode) const;
 
     virtual void
-    nameToASCII_UTF8(const StringPiece &name, ByteSink &dest,
+    nameToASCII_UTF8(StringPiece name, ByteSink &dest,
                      IDNAInfo &info, UErrorCode &errorCode) const;
 
     virtual void
-    nameToUnicodeUTF8(const StringPiece &name, ByteSink &dest,
+    nameToUnicodeUTF8(StringPiece name, ByteSink &dest,
                       IDNAInfo &info, UErrorCode &errorCode) const;
 
 private:
@@ -156,7 +156,7 @@ private:
             IDNAInfo &info, UErrorCode &errorCode) const;
 
     void
-    processUTF8(const StringPiece &src,
+    processUTF8(StringPiece src,
                 UBool isLabel, UBool toASCII,
                 ByteSink &dest,
                 IDNAInfo &info, UErrorCode &errorCode) const;
@@ -182,7 +182,7 @@ private:
     int32_t
     markBadACELabel(UnicodeString &dest,
                     int32_t labelStart, int32_t labelLength,
-                    UBool toASCII, IDNAInfo &info) const;
+                    UBool toASCII, IDNAInfo &info, UErrorCode &errorCode) const;
 
     void
     checkLabelBiDi(const UChar *label, int32_t labelLength, IDNAInfo &info) const;
@@ -253,25 +253,25 @@ UTS46::nameToUnicode(const UnicodeString &name, UnicodeString &dest,
 }
 
 void
-UTS46::labelToASCII_UTF8(const StringPiece &label, ByteSink &dest,
+UTS46::labelToASCII_UTF8(StringPiece label, ByteSink &dest,
                          IDNAInfo &info, UErrorCode &errorCode) const {
     processUTF8(label, TRUE, TRUE, dest, info, errorCode);
 }
 
 void
-UTS46::labelToUnicodeUTF8(const StringPiece &label, ByteSink &dest,
+UTS46::labelToUnicodeUTF8(StringPiece label, ByteSink &dest,
                           IDNAInfo &info, UErrorCode &errorCode) const {
     processUTF8(label, TRUE, FALSE, dest, info, errorCode);
 }
 
 void
-UTS46::nameToASCII_UTF8(const StringPiece &name, ByteSink &dest,
+UTS46::nameToASCII_UTF8(StringPiece name, ByteSink &dest,
                         IDNAInfo &info, UErrorCode &errorCode) const {
     processUTF8(name, FALSE, TRUE, dest, info, errorCode);
 }
 
 void
-UTS46::nameToUnicodeUTF8(const StringPiece &name, ByteSink &dest,
+UTS46::nameToUnicodeUTF8(StringPiece name, ByteSink &dest,
                          IDNAInfo &info, UErrorCode &errorCode) const {
     processUTF8(name, FALSE, FALSE, dest, info, errorCode);
 }
@@ -321,9 +321,7 @@ UTS46::process(const UnicodeString &src,
     info.reset();
     int32_t srcLength=src.length();
     if(srcLength==0) {
-        if(toASCII) {
-            info.errors|=UIDNA_ERROR_EMPTY_LABEL;
-        }
+        info.errors|=UIDNA_ERROR_EMPTY_LABEL;
         return dest;
     }
     UChar *destArray=dest.getBuffer(srcLength);
@@ -381,12 +379,11 @@ UTS46::process(const UnicodeString &src,
                     ++i;  // '.' was copied to dest already
                     break;
                 }
-                if(toASCII) {
-                    if(i==labelStart) {
-                        info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
-                    } else if((i-labelStart)>63) {
-                        info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
-                    }
+                if(i==labelStart) {
+                    info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
+                }
+                if(toASCII && (i-labelStart)>63) {
+                    info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
                 }
                 info.errors|=info.labelErrors;
                 info.labelErrors=0;
@@ -406,7 +403,7 @@ UTS46::process(const UnicodeString &src,
 }
 
 void
-UTS46::processUTF8(const StringPiece &src,
+UTS46::processUTF8(StringPiece src,
                    UBool isLabel, UBool toASCII,
                    ByteSink &dest,
                    IDNAInfo &info, UErrorCode &errorCode) const {
@@ -422,9 +419,7 @@ UTS46::processUTF8(const StringPiece &src,
     // Arguments are fine, reset output values.
     info.reset();
     if(srcLength==0) {
-        if(toASCII) {
-            info.errors|=UIDNA_ERROR_EMPTY_LABEL;
-        }
+        info.errors|=UIDNA_ERROR_EMPTY_LABEL;
         dest.Flush();
         return;
     }
@@ -435,7 +430,7 @@ UTS46::processUTF8(const StringPiece &src,
         char stackArray[256];
         int32_t destCapacity;
         char *destArray=dest.GetAppendBuffer(srcLength, srcLength+20,
-                                             stackArray, LENGTHOF(stackArray), &destCapacity);
+                                             stackArray, UPRV_LENGTHOF(stackArray), &destCapacity);
         UBool disallowNonLDHDot=(options&UIDNA_USE_STD3_RULES)!=0;
         int32_t i;
         for(i=0;; ++i) {
@@ -482,12 +477,11 @@ UTS46::processUTF8(const StringPiece &src,
                     if(isLabel) {
                         break;  // Replacing with U+FFFD can be complicated for toASCII.
                     }
-                    if(toASCII) {
-                        if(i==labelStart) {
-                            info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
-                        } else if((i-labelStart)>63) {
-                            info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
-                        }
+                    if(i==labelStart) {
+                        info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
+                    }
+                    if(toASCII && (i-labelStart)>63) {
+                        info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
                     }
                     info.errors|=info.labelErrors;
                     info.labelErrors=0;
@@ -595,6 +589,9 @@ UTS46::processUnicode(const UnicodeString &src,
 int32_t
 UTS46::mapDevChars(UnicodeString &dest, int32_t labelStart, int32_t mappingStart,
                    UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
     int32_t length=dest.length();
     UChar *s=dest.getBuffer(dest[mappingStart]==0xdf ? length+1 : length);
     if(s==NULL) {
@@ -652,6 +649,9 @@ UTS46::mapDevChars(UnicodeString &dest, int32_t labelStart, int32_t mappingStart
         uts46Norm2.normalize(dest.tempSubString(labelStart), normalized, errorCode);
         if(U_SUCCESS(errorCode)) {
             dest.replace(labelStart, 0x7fffffff, normalized);
+            if(dest.isBogus()) {
+                errorCode=U_MEMORY_ALLOCATION_ERROR;
+            }
             return dest.length();
         }
     }
@@ -673,9 +673,16 @@ isNonASCIIDisallowedSTD3Valid(UChar32 c) {
 // Returns labelLength (= the new label length).
 static int32_t
 replaceLabel(UnicodeString &dest, int32_t destLabelStart, int32_t destLabelLength,
-             const UnicodeString &label, int32_t labelLength) {
+             const UnicodeString &label, int32_t labelLength, UErrorCode &errorCode) {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
     if(&label!=&dest) {
         dest.replace(destLabelStart, destLabelLength, label);
+        if(dest.isBogus()) {
+            errorCode=U_MEMORY_ALLOCATION_ERROR;
+            return 0;
+        }
     }
     return labelLength;
 }
@@ -685,6 +692,9 @@ UTS46::processLabel(UnicodeString &dest,
                     int32_t labelStart, int32_t labelLength,
                     UBool toASCII,
                     IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
     UnicodeString fromPunycode;
     UnicodeString *labelString;
     const UChar *label=dest.getBuffer()+labelStart;
@@ -719,7 +729,7 @@ UTS46::processLabel(UnicodeString &dest,
         fromPunycode.releaseBuffer(unicodeLength);
         if(U_FAILURE(punycodeErrorCode)) {
             info.labelErrors|=UIDNA_ERROR_PUNYCODE;
-            return markBadACELabel(dest, labelStart, labelLength, toASCII, info);
+            return markBadACELabel(dest, labelStart, labelLength, toASCII, info, errorCode);
         }
         // Check for NFC, and for characters that are not
         // valid or deviation characters according to the normalizer.
@@ -734,7 +744,7 @@ UTS46::processLabel(UnicodeString &dest,
         }
         if(!isValid) {
             info.labelErrors|=UIDNA_ERROR_INVALID_ACE_LABEL;
-            return markBadACELabel(dest, labelStart, labelLength, toASCII, info);
+            return markBadACELabel(dest, labelStart, labelLength, toASCII, info, errorCode);
         }
         labelString=&fromPunycode;
         label=fromPunycode.getBuffer();
@@ -746,10 +756,9 @@ UTS46::processLabel(UnicodeString &dest,
     }
     // Validity check
     if(labelLength==0) {
-        if(toASCII) {
-            info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
-        }
-        return replaceLabel(dest, destLabelStart, destLabelLength, *labelString, labelLength);
+        info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
+        return replaceLabel(dest, destLabelStart, destLabelLength,
+                            *labelString, labelLength, errorCode);
     }
     // labelLength>0
     if(labelLength>=4 && label[2]==0x2d && label[3]==0x2d) {
@@ -871,7 +880,7 @@ UTS46::processLabel(UnicodeString &dest,
                     info.labelErrors|=UIDNA_ERROR_LABEL_TOO_LONG;
                 }
                 return replaceLabel(dest, destLabelStart, destLabelLength,
-                                    punycode, punycodeLength);
+                                    punycode, punycodeLength, errorCode);
             } else {
                 // all-ASCII label
                 if(labelLength>63) {
@@ -884,10 +893,11 @@ UTS46::processLabel(UnicodeString &dest,
         // then leave it but make sure it does not look valid.
         if(wasPunycode) {
             info.labelErrors|=UIDNA_ERROR_INVALID_ACE_LABEL;
-            return markBadACELabel(dest, destLabelStart, destLabelLength, toASCII, info);
+            return markBadACELabel(dest, destLabelStart, destLabelLength, toASCII, info, errorCode);
         }
     }
-    return replaceLabel(dest, destLabelStart, destLabelLength, *labelString, labelLength);
+    return replaceLabel(dest, destLabelStart, destLabelLength,
+                        *labelString, labelLength, errorCode);
 }
 
 // Make sure an ACE label does not look valid.
@@ -896,7 +906,10 @@ UTS46::processLabel(UnicodeString &dest,
 int32_t
 UTS46::markBadACELabel(UnicodeString &dest,
                        int32_t labelStart, int32_t labelLength,
-                       UBool toASCII, IDNAInfo &info) const {
+                       UBool toASCII, IDNAInfo &info, UErrorCode &errorCode) const {
+    if(U_FAILURE(errorCode)) {
+        return 0;
+    }
     UBool disallowNonLDHDot=(options&UIDNA_USE_STD3_RULES)!=0;
     UBool isASCII=TRUE;
     UBool onlyLDH=TRUE;
@@ -924,6 +937,10 @@ UTS46::markBadACELabel(UnicodeString &dest,
     } while(++s<limit);
     if(onlyLDH) {
         dest.insert(labelStart+labelLength, (UChar)0xfffd);
+        if(dest.isBogus()) {
+            errorCode=U_MEMORY_ALLOCATION_ERROR;
+            return 0;
+        }
         ++labelLength;
     } else {
         if(toASCII && isASCII && labelLength>63) {
@@ -998,8 +1015,8 @@ UTS46::checkLabelBiDi(const UChar *label, int32_t labelLength, IDNAInfo &info) c
     ) {
         info.isOkBiDi=FALSE;
     }
-    // Get the directionalities of the intervening characters.
-    uint32_t mask=0;
+    // Add the directionalities of the intervening characters.
+    uint32_t mask=firstMask|lastMask;
     while(i<labelLength) {
         U16_NEXT_UNSAFE(label, i, c);
         mask|=U_MASK(u_charDirection(c));
@@ -1028,7 +1045,7 @@ UTS46::checkLabelBiDi(const UChar *label, int32_t labelLength, IDNAInfo &info) c
     // label. [...]
     // The following rule, consisting of six conditions, applies to labels
     // in BIDI domain names.
-    if(((firstMask|mask|lastMask)&R_AL_AN_MASK)!=0) {
+    if((mask&R_AL_AN_MASK)!=0) {
         info.isBiDi=TRUE;
     }
 }
@@ -1109,7 +1126,6 @@ isASCIIOkBiDi(const char *s, int32_t length) {
 
 UBool
 UTS46::isLabelOkContextJ(const UChar *label, int32_t labelLength) const {
-    const UBiDiProps *bdp=ubidi_getSingleton();
     // [IDNA2008-Tables]
     // 200C..200D  ; CONTEXTJ    # ZERO WIDTH NON-JOINER..ZERO WIDTH JOINER
     for(int32_t i=0; i<labelLength; ++i) {
@@ -1131,7 +1147,7 @@ UTS46::isLabelOkContextJ(const UChar *label, int32_t labelLength) const {
             }
             // check precontext (Joining_Type:{L,D})(Joining_Type:T)*
             for(;;) {
-                UJoiningType type=ubidi_getJoiningType(bdp, c);
+                UJoiningType type=ubidi_getJoiningType(c);
                 if(type==U_JT_TRANSPARENT) {
                     if(j==0) {
                         return FALSE;
@@ -1149,7 +1165,7 @@ UTS46::isLabelOkContextJ(const UChar *label, int32_t labelLength) const {
                     return FALSE;
                 }
                 U16_NEXT_UNSAFE(label, j, c);
-                UJoiningType type=ubidi_getJoiningType(bdp, c);
+                UJoiningType type=ubidi_getJoiningType(c);
                 if(type==U_JT_TRANSPARENT) {
                     // just skip this character
                 } else if(type==U_JT_RIGHT_JOINING || type==U_JT_DUAL_JOINING) {
@@ -1398,7 +1414,7 @@ uidna_labelToASCII_UTF8(const UIDNA *idna,
     if(!checkArgs(label, length, dest, capacity, pInfo, pErrorCode)) {
         return 0;
     }
-    StringPiece src(label, length<0 ? uprv_strlen(label) : length);
+    StringPiece src(label, length<0 ? static_cast<int32_t>(uprv_strlen(label)) : length);
     CheckedArrayByteSink sink(dest, capacity);
     IDNAInfo info;
     reinterpret_cast<const IDNA *>(idna)->labelToASCII_UTF8(src, sink, info, *pErrorCode);
@@ -1414,7 +1430,7 @@ uidna_labelToUnicodeUTF8(const UIDNA *idna,
     if(!checkArgs(label, length, dest, capacity, pInfo, pErrorCode)) {
         return 0;
     }
-    StringPiece src(label, length<0 ? uprv_strlen(label) : length);
+    StringPiece src(label, length<0 ? static_cast<int32_t>(uprv_strlen(label)) : length);
     CheckedArrayByteSink sink(dest, capacity);
     IDNAInfo info;
     reinterpret_cast<const IDNA *>(idna)->labelToUnicodeUTF8(src, sink, info, *pErrorCode);
@@ -1430,7 +1446,7 @@ uidna_nameToASCII_UTF8(const UIDNA *idna,
     if(!checkArgs(name, length, dest, capacity, pInfo, pErrorCode)) {
         return 0;
     }
-    StringPiece src(name, length<0 ? uprv_strlen(name) : length);
+    StringPiece src(name, length<0 ? static_cast<int32_t>(uprv_strlen(name)) : length);
     CheckedArrayByteSink sink(dest, capacity);
     IDNAInfo info;
     reinterpret_cast<const IDNA *>(idna)->nameToASCII_UTF8(src, sink, info, *pErrorCode);
@@ -1446,7 +1462,7 @@ uidna_nameToUnicodeUTF8(const UIDNA *idna,
     if(!checkArgs(name, length, dest, capacity, pInfo, pErrorCode)) {
         return 0;
     }
-    StringPiece src(name, length<0 ? uprv_strlen(name) : length);
+    StringPiece src(name, length<0 ? static_cast<int32_t>(uprv_strlen(name)) : length);
     CheckedArrayByteSink sink(dest, capacity);
     IDNAInfo info;
     reinterpret_cast<const IDNA *>(idna)->nameToUnicodeUTF8(src, sink, info, *pErrorCode);