From 49868a5a6818531eda5bce3d84b11c6d4c2c0c33 Mon Sep 17 00:00:00 2001
From: Markus Scherer <markus.icu@gmail.com>
Date: Thu, 25 Oct 2007 05:17:43 +0000
Subject: [PATCH] ICU-6001 fix ucnv_getUnicodeSet(LMBCS)

X-SVN-Rev: 22850
---
 icu4c/source/common/ucnv_lmb.c            | 21 ++++++++++-----------
 icu4c/source/test/intltest/convtest.cpp   |  2 +-
 icu4c/source/test/testdata/conversion.txt |  8 ++++++++
 3 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/icu4c/source/common/ucnv_lmb.c b/icu4c/source/common/ucnv_lmb.c
index 79606af2894..0d92c576e5a 100644
--- a/icu4c/source/common/ucnv_lmb.c
+++ b/icu4c/source/common/ucnv_lmb.c
@@ -1,6 +1,6 @@
 /*  
 **********************************************************************
-*   Copyright (C) 2000-2006, International Business Machines
+*   Copyright (C) 2000-2007, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 **********************************************************************
 *   file name:  ucnv_lmb.cpp
@@ -536,7 +536,7 @@ static const UConverterImpl _LMBCSImpl##n={\
     NULL,\
     NULL,\
     _LMBCSSafeClone,\
-    _LMBCSGetUnicodeSet\
+    ucnv_getCompleteUnicodeSet\
 };\
 static const UConverterStaticData _LMBCSStaticData##n={\
   sizeof(UConverterStaticData),\
@@ -662,15 +662,14 @@ _LMBCSSafeClone(const UConverter *cnv,
     return &newLMBCS->cnv;
 }
 
-static void
-_LMBCSGetUnicodeSet(const UConverter *cnv,
-                   const USetAdder *sa,
-                   UConverterUnicodeSet which,
-                   UErrorCode *pErrorCode) {
-    /* all but U+F6xx, see LMBCS explanation above (search for F6xx) */
-    sa->addRange(sa->set, 0, 0xf5ff);
-    sa->addRange(sa->set, 0xf700, 0x10ffff);
-}
+/*
+ * There used to be a _LMBCSGetUnicodeSet() function here (up to svn revision 20117)
+ * which added all code points except for U+F6xx
+ * because those cannot be represented in the Unicode group.
+ * However, it turns out that windows-950 has roundtrips for all of U+F6xx
+ * which means that LMBCS can convert all Unicode code points after all.
+ * We now simply use ucnv_getCompleteUnicodeSet().
+ */
 
 /* 
    Here's the basic helper function that we use when converting from
diff --git a/icu4c/source/test/intltest/convtest.cpp b/icu4c/source/test/intltest/convtest.cpp
index 189a527dc8c..c4c9e8ad5d4 100644
--- a/icu4c/source/test/intltest/convtest.cpp
+++ b/icu4c/source/test/intltest/convtest.cpp
@@ -541,7 +541,7 @@ ConversionTest::TestGetUnicodeSet2() {
         "JIS7",
         "ISO-2022-CN",
         "ISO-2022-CN-EXT",
-        // "LMBCS" TODO(markus): known bug, the fallback set is said to be missing [\uF600-\uF6FF]
+        "LMBCS"
     };
     char buffer[1024];
     int32_t i;
diff --git a/icu4c/source/test/testdata/conversion.txt b/icu4c/source/test/testdata/conversion.txt
index bdbeeb8d7f2..f7691bccf5b 100644
--- a/icu4c/source/test/testdata/conversion.txt
+++ b/icu4c/source/test/testdata/conversion.txt
@@ -1365,6 +1365,14 @@ conversion:table(nofallback) {
           :int{0}
         }
         
+        // LMBCS
+        {
+          "LMBCS",
+          "[\x00-\U0010ffff]",
+          "[]",
+          :int{0}
+        }
+
         // DBCS-only
         {
           "ibm-971",