Workaround for insufficiently sized dest buffer

dotnet · Jun 29, 2023 · 42e3269 · 42e3269
1 parent 3e34f7b
commit 42e3269
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 44 deletions.
diff --git a/docs/design/features/globalization-hybrid-mode.md b/docs/design/features/globalization-hybrid-mode.md
@@ -428,14 +428,3 @@ Behavioural changes compared to ICU
    - Final sigma behavior correction:
 
      ICU-based case change does not respect final-sigma rule, but hybrid does, so "ΒΌΛΟΣ" -> "βόλος", not "βόλοσ".
-
-   - Below cases will throw exception because of insufficiently sized destination buffer
-
-      - Capitalizing the German letter ß (sharp S) gives SS when using Apple native functions.
-
-      - Capitalizing ligatures gives different result on Apple platforms, eg. "\uFB00" (ﬀ) uppercase (FF)
-
-      - Capitalizing "\u0149" (ŉ) on Apple platforms returns combination of  "\u02BC" (ʼ) and N -> (ʼN)
-
-
-
diff --git a/src/libraries/System.Globalization/tests/System/Globalization/TextInfoTests.cs b/src/libraries/System.Globalization/tests/System/Globalization/TextInfoTests.cs
@@ -396,29 +396,24 @@ public static IEnumerable<object[]> ToUpper_TestData()
                 // RAINBOW (outside the BMP and does not case)
                 yield return new object[] { cultureName, "\U0001F308", "\U0001F308" };
 
-                if (!PlatformDetection.IsHybridGlobalizationOnOSX)
-                {
-                    // Unicode defines some codepoints which expand into multiple codepoints
-                    // when cased (see SpecialCasing.txt from UNIDATA for some examples). We have never done
-                    // these sorts of expansions, since it would cause string lengths to change when cased,
-                    // which is non-intuitive. In addition, there are some context sensitive mappings which
-                    // we also don't preform.
-                    // es-zed does not case to SS when uppercased.
-                    // on OSX, capitalizing the German letter ß (sharp S) gives SS
-                    yield return new object[] { cultureName, "\u00DF", "\u00DF" };
-                    yield return new object[] { cultureName, "stra\u00DFe", "STRA\u00DFE" };
-                    if (!PlatformDetection.IsNlsGlobalization)
-                        yield return new object[] { cultureName, "st\uD801\uDC37ra\u00DFe", "ST\uD801\uDC0FRA\u00DFE" };
-
-                    // Ligatures do not expand when cased.
-                    // on OSX, this is uppercase to "FF"
-                    yield return new object[] { cultureName, "\uFB00", "\uFB00" };
-
-                    // Precomposed character with no uppercase variant, we don't want to "decompose" this
-                    // as part of casing.
-                    // on OSX, this is uppercased to "ʼN"
-                    yield return new object[] { cultureName, "\u0149", "\u0149" };
-                }
+
+                // Unicode defines some codepoints which expand into multiple codepoints
+                // when cased (see SpecialCasing.txt from UNIDATA for some examples). We have never done
+                // these sorts of expansions, since it would cause string lengths to change when cased,
+                // which is non-intuitive. In addition, there are some context sensitive mappings which
+                // we also don't preform.
+                // es-zed does not case to SS when uppercased.
+                yield return new object[] { cultureName, "\u00DF", "\u00DF" };
+                yield return new object[] { cultureName, "stra\u00DFe", "STRA\u00DFE" };
+                if (!PlatformDetection.IsNlsGlobalization && !PlatformDetection.IsHybridGlobalizationOnOSX)
+                    yield return new object[] { cultureName, "st\uD801\uDC37ra\u00DFe", "ST\uD801\uDC0FRA\u00DFE" };
+
+                // Ligatures do not expand when cased.
+                yield return new object[] { cultureName, "\uFB00", "\uFB00" };
+
+                // Precomposed character with no uppercase variant, we don't want to "decompose" this
+                // as part of casing.
+                yield return new object[] { cultureName, "\u0149", "\u0149" };
             }
 
             // Turkish i

diff --git a/src/native/libs/System.Globalization.Native/pal_casing.m b/src/native/libs/System.Globalization.Native/pal_casing.m
@@ -66,12 +66,28 @@ int32_t GlobalizationNative_ChangeCaseNative(const uint16_t* localeName, int32_t
 
     int32_t srcIdx = 0, dstIdx = 0, isError = 0;
     uint16_t dstCodepoint;
-    while (srcIdx < result.length)
+    if (result.length <= cwDstLength)
     {
-        dstCodepoint = [result characterAtIndex:srcIdx++];
-        Append(lpDst, dstIdx, cwDstLength, dstCodepoint, isError);
-        if (isError)
-            return isError;
+        while (srcIdx < result.length)
+        {
+            dstCodepoint = [result characterAtIndex:srcIdx++];
+            Append(lpDst, dstIdx, cwDstLength, dstCodepoint, isError);
+            if (isError)
+                return isError;
+        }
+    }
+    else
+    {
+        while (srcIdx < cwSrcLength)
+        {           
+            NSString *src = [NSString stringWithCharacters: lpSrc + srcIdx length: 1];
+            srcIdx++;
+            NSString *dst = bToUpper ? [src uppercaseStringWithLocale:currentLocale] : [src lowercaseStringWithLocale:currentLocale];
+            dstCodepoint = dst.length > 1 ? [src characterAtIndex: 0] : [dst characterAtIndex: 0];
+            Append(lpDst, dstIdx, cwDstLength, dstCodepoint, isError);
+            if (isError)
+                return isError;
+        }
     }
     return Success;
 }
@@ -90,12 +106,28 @@ int32_t GlobalizationNative_ChangeCaseInvariantNative(const uint16_t* lpSrc, int
 
     int32_t srcIdx = 0, dstIdx = 0, isError = 0;
     uint16_t dstCodepoint;
-    while (srcIdx < result.length)
+    if (result.length <= cwDstLength)
+    {
+        while (srcIdx < result.length)
+        {
+            dstCodepoint = [result characterAtIndex:srcIdx++];
+            Append(lpDst, dstIdx, cwDstLength, dstCodepoint, isError);
+            if (isError)
+                return isError;
+        }
+    }
+    else
     {
-        dstCodepoint = [result characterAtIndex:srcIdx++];
-        Append(lpDst, dstIdx, cwDstLength, dstCodepoint, isError);
-        if (isError)
-            return isError;
+        while (srcIdx < cwSrcLength)
+        {           
+            NSString *src = [NSString stringWithCharacters: lpSrc + srcIdx length: 1];
+            srcIdx++;
+            NSString *dst = bToUpper ? src.uppercaseString : src.lowercaseString;
+            dstCodepoint = dst.length > 1 ? [src characterAtIndex: 0] : [dst characterAtIndex: 0];
+            Append(lpDst, dstIdx, cwDstLength, dstCodepoint, isError);
+            if (isError)
+                return isError;
+        }
     }
     return Success;
 }