[iOS] Implement IgnoreSymbols for CompreInfo in managed code (#120462)

Implementation of `CompareOptions.IgnoreSymbols` for `IndexOf`, `IsPrefix`, `IsSuffix`, `Compare` on iOS. Because ObjectiveC string APIs (https://developer.apple.com/documentation/foundation/nsstring/compareoptions?language=objc) don't provide a direct alternative to `IgnoreSymbols` option, we filter the symbols from the strings in the managed code and after returning from native, we re-calculate the original index and length.

The implementation works in a following:
1. Preprocess the source and search strings by removing the symbols.
2. Invoke the native API for comparison.
3. Map the range from preprocessed source string back to the original string to get the index and length.

---------

Co-authored-by: Tarek Mahmoud Sayed <[email protected]>
Co-authored-by: Tarek Mahmoud Sayed <[email protected]>
Co-authored-by: Copilot <[email protected]>

Author: 3 people

docs/design/features/globalization-hybrid-mode.md

@@ -30,7 +30,7 @@ Due to these differences, the exact result of string compariso on Apple platform
 
 The number of `CompareOptions` and `NSStringCompareOptions` combinations are limited. Originally supported combinations can be found [here for CompareOptions](https://learn.microsoft.com/dotnet/api/system.globalization.compareoptions) and [here for NSStringCompareOptions](https://developer.apple.com/documentation/foundation/nsstringcompareoptions).
 
-- `IgnoreSymbols` is not supported because there is no equivalent in native api. Throws `PlatformNotSupportedException`.
+- `IgnoreSymbols` is supported by filtering ignorable symbols on the managed side before invoking the native API.
 
 - `IgnoreKanaType` is implemented using [`kCFStringTransformHiraganaKatakana`](https://developer.apple.com/documentation/corefoundation/kcfstringtransformhiraganakatakana?language=objc) then comparing strings.
 
@@ -71,10 +71,6 @@ The number of `CompareOptions` and `NSStringCompareOptions` combinations are lim
 
    `CompareOptions.IgnoreWidth` is mapped to `NSStringCompareOptions.NSWidthInsensitiveSearch | NSStringCompareOptions.NSLiteralSearch`
 
-- All combinations that contain below `CompareOptions` always throw `PlatformNotSupportedException`:
-
-   `IgnoreSymbols`
-
 ## String starts with / ends with
 
 Affected public APIs:
@@ -91,7 +87,7 @@ Apple Native API does not expose locale-sensitive endsWith/startsWith function.
 
 - `IgnoreSymbols`
 
-   As there is no IgnoreSymbols equivalent in NSStringCompareOptions all `CompareOptions` combinations that include `IgnoreSymbols` throw `PlatformNotSupportedException`
+   Supported by filtering ignorable symbols on the managed side prior to comparison using native API.
 
 ## String indexing
 
@@ -129,7 +125,7 @@ Not covered case:
 
 - `IgnoreSymbols`
 
-   As there is no IgnoreSymbols equivalent in NSStringCompareOptions all `CompareOptions` combinations that include `IgnoreSymbols` throw `PlatformNotSupportedException`
+   Supported by filtering ignorable symbols on the managed side prior to comparison using native API.
 
 - Some letters consist of more than one grapheme.
 

src/libraries/System.Private.CoreLib/src/System/Globalization/CompareInfo.Icu.cs

@@ -78,16 +78,15 @@ private unsafe int IcuIndexOfCore(ReadOnlySpan<char> source, ReadOnlySpan<char>
             }
             else
             {
+#if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
+                if (GlobalizationMode.Hybrid)
+                    return IndexOfCoreNative(target, source, options, fromBeginning, matchLengthPtr);
+#endif
                 // GetReference may return nullptr if the input span is defaulted. The native layer handles
                 // this appropriately; no workaround is needed on the managed side.
-
                 fixed (char* pSource = &MemoryMarshal.GetReference(source))
                 fixed (char* pTarget = &MemoryMarshal.GetReference(target))
                 {
-#if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
-                    if (GlobalizationMode.Hybrid)
-                        return IndexOfCoreNative(pTarget, target.Length, pSource, source.Length, options, fromBeginning, matchLengthPtr);
-#endif
                     if (fromBeginning)
                         return Interop.Globalization.IndexOf(_sortHandle, pTarget, target.Length, pSource, source.Length, options, matchLengthPtr);
                     else
@@ -207,7 +206,7 @@ private unsafe int IndexOfOrdinalIgnoreCaseHelper(ReadOnlySpan<char> source, Rea
             InteropCall:
 #if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
                 if (GlobalizationMode.Hybrid)
-                    return IndexOfCoreNative(b, target.Length, a, source.Length, options, fromBeginning, matchLengthPtr);
+                    return IndexOfCoreNative(target, source, options, fromBeginning, matchLengthPtr);
 #endif
                 if (fromBeginning)
                     return Interop.Globalization.IndexOf(_sortHandle, b, target.Length, a, source.Length, options, matchLengthPtr);
@@ -301,7 +300,7 @@ private unsafe int IndexOfOrdinalHelper(ReadOnlySpan<char> source, ReadOnlySpan<
             InteropCall:
 #if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
                 if (GlobalizationMode.Hybrid)
-                    return IndexOfCoreNative(b, target.Length, a, source.Length, options, fromBeginning, matchLengthPtr);
+                    return IndexOfCoreNative(target, source, options, fromBeginning, matchLengthPtr);
 #endif
                 if (fromBeginning)
                     return Interop.Globalization.IndexOf(_sortHandle, b, target.Length, a, source.Length, options, matchLengthPtr);
@@ -328,13 +327,13 @@ private unsafe bool IcuStartsWith(ReadOnlySpan<char> source, ReadOnlySpan<char>
             }
             else
             {
+#if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
+                if (GlobalizationMode.Hybrid)
+                    return NativeStartsWith(prefix, source, options);
+#endif
                 fixed (char* pSource = &MemoryMarshal.GetReference(source)) // could be null (or otherwise unable to be dereferenced)
                 fixed (char* pPrefix = &MemoryMarshal.GetReference(prefix))
                 {
-#if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
-                    if (GlobalizationMode.Hybrid)
-                        return NativeStartsWith(pPrefix, prefix.Length, pSource, source.Length, options);
-#endif
                     return Interop.Globalization.StartsWith(_sortHandle, pPrefix, prefix.Length, pSource, source.Length, options, matchLengthPtr);
                 }
             }
@@ -416,7 +415,7 @@ private unsafe bool StartsWithOrdinalIgnoreCaseHelper(ReadOnlySpan<char> source,
             InteropCall:
 #if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
                 if (GlobalizationMode.Hybrid)
-                    return NativeStartsWith(bp, prefix.Length, ap, source.Length, options);
+                    return NativeStartsWith(prefix, source, options);
 #endif
                 return Interop.Globalization.StartsWith(_sortHandle, bp, prefix.Length, ap, source.Length, options, matchLengthPtr);
             }
@@ -488,7 +487,7 @@ private unsafe bool StartsWithOrdinalHelper(ReadOnlySpan<char> source, ReadOnlyS
             InteropCall:
 #if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
                 if (GlobalizationMode.Hybrid)
-                    return NativeStartsWith(bp, prefix.Length, ap, source.Length, options);
+                    return NativeStartsWith(prefix, source, options);
 #endif
                 return Interop.Globalization.StartsWith(_sortHandle, bp, prefix.Length, ap, source.Length, options, matchLengthPtr);
             }
@@ -512,13 +511,13 @@ private unsafe bool IcuEndsWith(ReadOnlySpan<char> source, ReadOnlySpan<char> su
             }
             else
             {
+#if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
+                if (GlobalizationMode.Hybrid)
+                    return NativeEndsWith(suffix, source, options);
+#endif
                 fixed (char* pSource = &MemoryMarshal.GetReference(source)) // could be null (or otherwise unable to be dereferenced)
                 fixed (char* pSuffix = &MemoryMarshal.GetReference(suffix))
                 {
-#if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
-                    if (GlobalizationMode.Hybrid)
-                        return NativeEndsWith(pSuffix, suffix.Length, pSource, source.Length, options);
-#endif
                     return Interop.Globalization.EndsWith(_sortHandle, pSuffix, suffix.Length, pSource, source.Length, options, matchLengthPtr);
                 }
             }
@@ -601,7 +600,7 @@ private unsafe bool EndsWithOrdinalIgnoreCaseHelper(ReadOnlySpan<char> source, R
             InteropCall:
 #if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
                 if (GlobalizationMode.Hybrid)
-                    return NativeEndsWith(bp, suffix.Length, ap, source.Length, options);
+                    return NativeEndsWith(suffix, source, options);
 #endif
                 return Interop.Globalization.EndsWith(_sortHandle, bp, suffix.Length, ap, source.Length, options, matchLengthPtr);
             }
@@ -673,7 +672,7 @@ private unsafe bool EndsWithOrdinalHelper(ReadOnlySpan<char> source, ReadOnlySpa
             InteropCall:
 #if TARGET_MACCATALYST || TARGET_IOS || TARGET_TVOS
                 if (GlobalizationMode.Hybrid)
-                    return NativeEndsWith(bp, suffix.Length, ap, source.Length, options);
+                    return NativeEndsWith(suffix, source, options);
 #endif
                 return Interop.Globalization.EndsWith(_sortHandle, bp, suffix.Length, ap, source.Length, options, matchLengthPtr);
             }