ICU-57131.0.1.tar.gz

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

diff --git a/icuSources/common/bmpset.cpp b/icuSources/common/bmpset.cpp
index 83cc9064f7e62da4484c51868264046202f3e027..7cd32eb99c9e33f6ec31aca6fbf779d1fb2f5694 100644 (file)
--- a/icuSources/common/bmpset.cpp
+++ b/icuSources/common/bmpset.cpp
@@ -1,7 +1,7 @@
 /*
 ******************************************************************************
 *
 /*
 ******************************************************************************
 *

-*   Copyright (C) 2007-2011, International Business Machines
+*   Copyright (C) 2007-2012, International Business Machines

 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************
 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************


@@ -107,7 +107,7 @@ static void set32x64Bits(uint32_t table[64], int32_t start, int32_t limit) {
         // limit<=0x800. If limit==0x800 then limitLead=32 and limitTrail=0.
         // In that case, bits=1<<limitLead is undefined but the bits value
         // is not used because trail<limitTrail is already false.
         // limit<=0x800. If limit==0x800 then limitLead=32 and limitTrail=0.
         // In that case, bits=1<<limitLead is undefined but the bits value
         // is not used because trail<limitTrail is already false.

-        bits=1<<limitLead;
+        bits=(uint32_t)1<<((limitLead == 0x20) ? (limitLead - 1) : limitLead);

         for(trail=0; trail<limitTrail; ++trail) {
             table[trail]|=bits;
         }
         for(trail=0; trail<limitTrail; ++trail) {
             table[trail]|=bits;
         }


@@ -690,16 +690,9 @@ BMPSet::spanBackUTF8(const uint8_t *s, int32_t length, USetSpanCondition spanCon
 
         int32_t prev=length;
         UChar32 c;
 
         int32_t prev=length;
         UChar32 c;

-        if(b<0xc0) {
-            // trail byte: collect a multi-byte character
-            c=utf8_prevCharSafeBody(s, 0, &length, b, -1);
-            if(c<0) {
-                c=0xfffd;
-            }
-        } else {
-            // lead byte in last-trail position
-            c=0xfffd;
-        }
+        // trail byte: collect a multi-byte character
+        // (or  lead byte in last-trail position)
+        c=utf8_prevCharSafeBody(s, 0, &length, b, -3);

         // c is a valid code point, not ASCII, not a surrogate
         if(c<=0x7ff) {
             if((USetSpanCondition)((table7FF[c&0x3f]&((uint32_t)1<<(c>>6)))!=0) != spanCondition) {
         // c is a valid code point, not ASCII, not a surrogate
         if(c<=0x7ff) {
             if((USetSpanCondition)((table7FF[c&0x3f]&((uint32_t)1<<(c>>6)))!=0) != spanCondition) {