From f81a01e01bc9c4430814730aaf98227e25aedfef Mon Sep 17 00:00:00 2001
From: Markus Scherer <markus.icu@gmail.com>
Date: Fri, 29 Aug 2014 23:07:42 +0000
Subject: [PATCH] ICU-10883 UTS 46 toUnicode set error for empty label
 including just xn--

X-SVN-Rev: 36297
---
 icu4c/source/common/uts46.cpp            | 34 +++++++++---------------
 icu4c/source/test/intltest/uts46test.cpp |  8 +++---
 2 files changed, 18 insertions(+), 24 deletions(-)

diff --git a/icu4c/source/common/uts46.cpp b/icu4c/source/common/uts46.cpp
index 11d5dc5d2bf..80900a019b4 100644
--- a/icu4c/source/common/uts46.cpp
+++ b/icu4c/source/common/uts46.cpp
@@ -319,9 +319,7 @@ 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);
@@ -379,12 +377,11 @@ 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;
@@ -420,9 +417,7 @@ 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;
     }
@@ -480,12 +475,11 @@ 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;
@@ -744,9 +738,7 @@ UTS46::processLabel(UnicodeString &dest,
     }
     // Validity check
     if(labelLength==0) {
-        if(toASCII) {
-            info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
-        }
+        info.labelErrors|=UIDNA_ERROR_EMPTY_LABEL;
         return replaceLabel(dest, destLabelStart, destLabelLength, *labelString, labelLength);
     }
     // labelLength>0
diff --git a/icu4c/source/test/intltest/uts46test.cpp b/icu4c/source/test/intltest/uts46test.cpp
index a1a9cbdba85..5bdf305a666 100644
--- a/icu4c/source/test/intltest/uts46test.cpp
+++ b/icu4c/source/test/intltest/uts46test.cpp
@@ -465,6 +465,7 @@ static const TestCase testCases[]={
       "1234567890123456789012345678901234567890123456789012345678901",
       UIDNA_ERROR_LABEL_TOO_LONG|UIDNA_ERROR_DOMAIN_NAME_TOO_LONG },
     // hyphen errors and empty-label errors
+    // Ticket #10883: ToUnicode also checks for empty labels.
     { ".", "B", ".", UIDNA_ERROR_EMPTY_LABEL },
     { "\\uFF0E", "B", ".", UIDNA_ERROR_EMPTY_LABEL },
     // "xn---q----jra"=="-q--a-umlaut-"
@@ -478,11 +479,13 @@ static const TestCase testCases[]={
       UIDNA_ERROR_EMPTY_LABEL|UIDNA_ERROR_LEADING_HYPHEN|UIDNA_ERROR_TRAILING_HYPHEN|
       UIDNA_ERROR_HYPHEN_3_4 },
     { "a..c", "B", "a..c", UIDNA_ERROR_EMPTY_LABEL },
+    { "a.xn--.c", "B", "a..c", UIDNA_ERROR_EMPTY_LABEL },
     { "a.-b.", "B", "a.-b.", UIDNA_ERROR_LEADING_HYPHEN },
     { "a.b-.c", "B", "a.b-.c", UIDNA_ERROR_TRAILING_HYPHEN },
     { "a.-.c", "B", "a.-.c", UIDNA_ERROR_LEADING_HYPHEN|UIDNA_ERROR_TRAILING_HYPHEN },
     { "a.bc--de.f", "B", "a.bc--de.f", UIDNA_ERROR_HYPHEN_3_4 },
     { "\\u00E4.\\u00AD.c", "B", "\\u00E4..c", UIDNA_ERROR_EMPTY_LABEL },
+    { "\\u00E4.xn--.c", "B", "\\u00E4..c", UIDNA_ERROR_EMPTY_LABEL },
     { "\\u00E4.-b.", "B", "\\u00E4.-b.", UIDNA_ERROR_LEADING_HYPHEN },
     { "\\u00E4.b-.c", "B", "\\u00E4.b-.c", UIDNA_ERROR_TRAILING_HYPHEN },
     { "\\u00E4.-.c", "B", "\\u00E4.-.c", UIDNA_ERROR_LEADING_HYPHEN|UIDNA_ERROR_TRAILING_HYPHEN },
@@ -598,10 +601,9 @@ void UTS46Test::TestSomeCases() {
         ) {
             continue;
         }
-        // ToUnicode does not set length errors.
+        // ToUnicode does not set length-overflow errors.
         uint32_t uniErrors=testCase.errors&~
-            (UIDNA_ERROR_EMPTY_LABEL|
-             UIDNA_ERROR_LABEL_TOO_LONG|
+            (UIDNA_ERROR_LABEL_TOO_LONG|
              UIDNA_ERROR_DOMAIN_NAME_TOO_LONG);
         char mode=testCase.o[0];
         if(mode=='B' || mode=='N') {
-- 
2.40.0