5872750434

pending completion

Build # 5872750434

Build Type

push

github

Committed by

neon-sunset

Commit Message

feat: U8SplitOptions.Trim impl. on Split

Run Details

105 of 742 branches covered (14.15%)

Branch coverage included in aggregate %.

4 of 4 new or added lines in 2 files covered. (100.0%)

372 of 1600 relevant lines covered (23.25%)

444.8 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/src/U8String.Splitting.cs

using System.Collections;
using System.Runtime.InteropServices;
using System.Text;
using U8Primitives.Abstractions;
using U8Primitives.InteropServices;

#pragma warning disable RCS1085, RCS1085FadeOut, IDE0032 // Use auto-implemented property. Why: readable fields.
namespace U8Primitives;

public readonly partial struct U8String
{
    public U8SplitPair SplitFirst(byte separator)
    {
        if (!U8Info.IsAsciiByte(separator))
        {
            // TODO: EH UX
            ThrowHelpers.ArgumentOutOfRange();
        }

        var source = this;
        if (!source.IsEmpty)
        {
            var span = source.UnsafeSpan;
            var index = span.IndexOf(separator);
            if (index >= 0)
            {
                return new(source, index, 1);
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    public U8SplitPair SplitFirst(char separator) => char.IsAscii(separator)
        ? SplitFirst((byte)separator)
        : SplitFirstUnchecked(separator.NonAsciiToUtf8(out _));

    public U8SplitPair SplitFirst(Rune separator) => separator.IsAscii
        ? SplitFirst((byte)separator.Value)
        : SplitFirstUnchecked(separator.NonAsciiToUtf8(out _));

    public U8SplitPair SplitFirst(U8String separator)
    {
        var source = this;
        if (!source.IsEmpty)
        {
            if (!separator.IsEmpty)
            {
                var span = source.UnsafeSpan;
                var index = span.IndexOf(separator.UnsafeSpan);
                if (index >= 0)
                {
                    return new(source, index, separator.Length);
                }
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    // It would be *really nice* to aggressively inline this method
    // but the way validation is currently implemented does not significantly
    // benefit from splitting on UTF-8 literals while possibly risking
    // running out of inlining budget significantly regressing performance everywhere else.
    public U8SplitPair SplitFirst(ReadOnlySpan<byte> separator)
    {
        var source = this;
        if (!source.IsEmpty)
        {
            if (!separator.IsEmpty)
            {
                var span = source.UnsafeSpan;
                var index = span.IndexOf(separator);
                if (index >= 0)
                {
                    // Same as with Slice(int, int), this might dereference past the end of the string.
                    // TODO: Do something about it if it's ever an issue.
                    if (U8Info.IsContinuationByte(source.UnsafeRefAdd(index)) ||
                        U8Info.IsContinuationByte(source.UnsafeRefAdd(index + separator.Length)))
                    {
                        ThrowHelpers.InvalidSplit();
                    }

                    return new(source, index, separator.Length);
                }
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public U8SplitPair SplitFirstUnchecked(ReadOnlySpan<byte> separator)
    {
        var source = this;
        if (!source.IsEmpty)
        {
            if (!separator.IsEmpty)
            {
                var span = source.UnsafeSpan;
                var index = span.IndexOf(separator);
                if (index >= 0)
                {
                    return new(source, index, separator.Length);
                }
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    public U8SplitPair SplitLast(byte separator)
    {
        if (!U8Info.IsAsciiByte(separator))
        {
            // TODO: EH UX
            ThrowHelpers.ArgumentOutOfRange();
        }

        var source = this;
        if (!source.IsEmpty)
        {
            var span = source.UnsafeSpan;
            var index = span.LastIndexOf(separator);
            if (index >= 0)
            {
                return new(source, index, 1);
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    public U8SplitPair SplitLast(char separator) => char.IsAscii(separator)
        ? SplitLast((byte)separator)
        : SplitLastUnchecked(separator.NonAsciiToUtf8(out _));

    public U8SplitPair SplitLast(Rune separator) => separator.IsAscii
        ? SplitLast((byte)separator.Value)
        : SplitLastUnchecked(separator.NonAsciiToUtf8(out _));

    public U8SplitPair SplitLast(U8String separator)
    {
        var source = this;
        if (!source.IsEmpty)
        {
            if (!separator.IsEmpty)
            {
                var span = source.UnsafeSpan;
                var index = span.LastIndexOf(separator.UnsafeSpan);
                if (index >= 0)
                {
                    return new(source, index, separator.Length);
                }
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    public U8SplitPair SplitLast(ReadOnlySpan<byte> separator)
    {
        var source = this;
        if (!source.IsEmpty)
        {
            if (!separator.IsEmpty)
            {
                var span = source.UnsafeSpan;
                var index = span.LastIndexOf(separator);
                if (index >= 0)
                {
                    if (U8Info.IsContinuationByte(source.UnsafeRefAdd(index)) ||
                        U8Info.IsContinuationByte(source.UnsafeRefAdd(index + separator.Length)))
                    {
                        ThrowHelpers.InvalidSplit();
                    }

                    return new(source, index, separator.Length);
                }
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public U8SplitPair SplitLastUnchecked(ReadOnlySpan<byte> separator)
    {
        var source = this;
        if (!source.IsEmpty)
        {
            if (!separator.IsEmpty)
            {
                var span = source.UnsafeSpan;
                var index = span.LastIndexOf(separator);
                if (index >= 0)
                {
                    return new(source, index, separator.Length);
                }
            }

            return U8SplitPair.NotFound(source);
        }

        return default;
    }

    public U8Split<byte> Split(byte separator)
    {
        if (!U8Info.IsAsciiByte(separator))
        {
            ThrowHelpers.ArgumentOutOfRange();
        }

        return new(this, separator);
    }

    public ConfiguredU8Split<byte> Split(byte separator, U8SplitOptions options)
    {
        if (!U8Info.IsAsciiByte(separator))
        {
            ThrowHelpers.ArgumentOutOfRange();
        }

        return new(this, separator, options);
    }

    public U8Split<char> Split(char separator)
    {
        if (char.IsSurrogate(separator))
        {
            ThrowHelpers.ArgumentOutOfRange();
        }

        return new(this, separator);
    }

    public ConfiguredU8Split<char> Split(char separator, U8SplitOptions options)
    {
        if (char.IsSurrogate(separator))
        {
            ThrowHelpers.ArgumentOutOfRange();
        }

        return new(this, separator, options);
    }

    public U8Split<Rune> Split(Rune separator) => new(this, separator);

    public ConfiguredU8Split<Rune> Split(Rune separator, U8SplitOptions options) => new(this, separator, options);

    public U8Split Split(U8String separator)
    {
        return !separator.IsEmpty ? new(this, separator) : default;
    }

    public ConfiguredU8Split<U8String> Split(U8String separator, U8SplitOptions options)
    {
        return !separator.IsEmpty ? new(this, separator, options) : default;
    }

    public U8Split<byte[]> Split(byte[] separator)
    {
        if (!IsValid(separator))
        {
            // TODO: EH UX
            ThrowHelpers.InvalidSplit();
        }

        var source = this;
        return (!source.IsEmpty && separator != null) ? new(source, separator) : default;
    }

    public ConfiguredU8Split<byte[]> Split(byte[] separator, U8SplitOptions options)
    {
        if (!IsValid(separator))
        {
            // TODO: EH UX
            ThrowHelpers.InvalidSplit();
        }

        var source = this;
        return (!source.IsEmpty && separator != null) ? new(source, separator, options) : default;
    }
}

public readonly record struct U8SplitPair
{
    readonly U8String _value;
    readonly int _offset;
    readonly int _stride;

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    internal U8SplitPair(U8String value, int offset, int stride)
    {
        _value = value;
        _offset = offset;
        _stride = stride;
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public static U8SplitPair NotFound(U8String value)
    {
        return new(value, value.Length, 0);
    }

    public U8String Segment
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get => U8Marshal.Slice(_value, 0, _offset);
    }

    public U8String Remainder
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get => U8Marshal.Slice(_value, _offset + _stride);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public void Deconstruct(out U8String segment, out U8String remainder)
    {
        segment = Segment;
        remainder = Remainder;
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public static implicit operator (U8String, U8String)(U8SplitPair value)
    {
        return (value.Segment, value.Remainder);
    }
}

public struct U8Split : ICollection<U8String>, IU8Enumerable<U8Split.Enumerator>
{
    readonly U8String _value;
    readonly U8String _separator;
    int _count;

    internal U8Split(U8String value, U8String separator)
    {
        if (!value.IsEmpty)
        {
            _value = value;
            _separator = separator;
            _count = -1;
        }
    }

    public int Count
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get
        {
            var count = _count;
            if (count >= 0)
            {
                return count;
            }

            // Matches the behavior of string.Split('\n').Length for "hello\n"
            // TODO: Should we break consistency and not count the very last segment if it is empty?
            return _count = Count(_value.UnsafeSpan, _separator) + 1;

            static int Count(ReadOnlySpan<byte> value, ReadOnlySpan<byte> separator)
            {
                return U8Searching.Count(value, separator);
            }
        }
    }

    public readonly bool Contains(U8String item)
    {
        var separator = _separator;
        var overlaps = item.Contains(separator);

        return !overlaps && _value.Contains(item);
    }

    public void CopyTo(U8String[] array, int index)
    {
        this.CopyTo<U8Split, Enumerator>(array.AsSpan()[index..]);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly void Deconstruct(out U8String first, out U8String second)
    {
        this.Deconstruct<U8Split, Enumerator>(out first, out second);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
    {
        this.Deconstruct<U8Split, Enumerator>(out first, out second, out third);
    }

    public U8String[] ToArray() => this.ToArray<U8Split, Enumerator>();
    public List<U8String> ToList() => this.ToList<U8Split, Enumerator>();

    /// <summary>
    /// Returns a <see cref="Enumerator"/> over the provided string.
    /// </summary>
    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly Enumerator GetEnumerator() => new(_value, _separator);

    readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();
    readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();
    readonly bool ICollection<U8String>.IsReadOnly => true;

    [StructLayout(LayoutKind.Auto)]
    public struct Enumerator : IU8Enumerator
    {
        readonly byte[]? _value;
        readonly U8String _separator;
        U8Range _current;
        U8Range _remaining;

        internal Enumerator(U8String value, U8String separator)
        {
            _value = value._value;
            _separator = separator;
            _remaining = value._inner;
        }

        public readonly U8String Current => new(_value, _current.Offset, _current.Length);

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext()
        {
            var remaining = _remaining;
            if (remaining.Length > 0)
            {
                var value = _value!.SliceUnsafe(remaining.Offset, remaining.Length);
                var separator = _separator;
                var index = value.IndexOf(separator.UnsafeSpan);
                if (index >= 0)
                {
                    _current = new(remaining.Offset, index);
                    _remaining = new(
                        remaining.Offset + index + separator.Length,
                        remaining.Length - index - separator.Length);
                }
                else
                {
                    _current = remaining;
                    _remaining = default;
                }

                return true;
            }

            return false;
        }

        readonly object IEnumerator.Current => Current;
        readonly void IEnumerator.Reset() => throw new NotSupportedException();
        readonly void IDisposable.Dispose() { }
    }

    readonly void ICollection<U8String>.Add(U8String item) => throw new NotSupportedException();
    readonly void ICollection<U8String>.Clear() => throw new NotSupportedException();
    readonly bool ICollection<U8String>.Remove(U8String item) => throw new NotSupportedException();
}

// TODO: I don't think this is a successful design. JIT/ILC can't optimize away
// ASCII char literal conversion causing the size of MoveNext() to explode. This needs an overhaul,
// possibly disambiguating on struct vs separator types or maybe even 3 options:
// primitive, span (ref split aka SplitRef? SplitSpan?) and U8String / byte[]
[StructLayout(LayoutKind.Auto)]
public struct U8Split<TSeparator> :
    ICollection<U8String>,
    IU8Enumerable<U8Split<TSeparator>.Enumerator>
{
    readonly U8String _value;
    readonly TSeparator? _separator;
    int _count;

    // TODO: Move value.IsEmpty -> count = 0 check here
    internal U8Split(U8String value, TSeparator? separator)
    {
        if (!value.IsEmpty)
        {
            _value = value;
            _separator = separator;
            _count = -1;
        }
    }

    public int Count
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get
        {
            var count = _count;
            if (count >= 0)
            {
                return count;
            }

            // Matches the behavior of string.Split('\n').Length for "hello\n"
            // TODO: Should we break consistency and not count the very last segment if it is empty?
            return _count = Count(_value, _separator) + 1;

            static int Count(ReadOnlySpan<byte> value, TSeparator? separator)
            {
                return U8Searching.Count(value, separator);
            }
        }
    }

    public readonly bool Contains(U8String item)
    {
        var separator = _separator;
        var overlaps = U8Searching.Contains(item, separator);

        return !overlaps && _value.Contains(item);
    }

    public void CopyTo(U8String[] array, int index)
    {
        this.CopyTo<U8Split<TSeparator>, Enumerator>(array.AsSpan()[index..]);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly void Deconstruct(out U8String first, out U8String second)
    {
        this.Deconstruct<U8Split<TSeparator>, Enumerator>(out first, out second);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
    {
        this.Deconstruct<U8Split<TSeparator>, Enumerator>(out first, out second, out third);
    }

    public U8String[] ToArray() => this.ToArray<U8Split<TSeparator>, Enumerator>();
    public List<U8String> ToList() => this.ToList<U8Split<TSeparator>, Enumerator>();

    /// <summary>
    /// Returns a <see cref="Enumerator"/> over the provided string.
    /// </summary>
    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly Enumerator GetEnumerator() => new(_value, _separator);

    readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();
    readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();
    readonly bool ICollection<U8String>.IsReadOnly => true;

    [StructLayout(LayoutKind.Auto)]
    public struct Enumerator : IU8Enumerator
    {
        readonly byte[]? _value;
        readonly TSeparator? _separator;
        readonly U8Size _separatorSize;
        U8Range _current;
        U8Range _remaining;

        internal Enumerator(U8String value, TSeparator? separator)
        {
            _value = value._value;
            _separator = separator;
            _separatorSize = U8Info.GetSize(separator);
            _remaining = value._inner;
        }

        public readonly U8String Current => new(_value, _current.Offset, _current.Length);

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext()
        {
            var remaining = _remaining;
            if (remaining.Length > 0)
            {
                var size = _separatorSize;
                var value = _value!.SliceUnsafe(remaining.Offset, remaining.Length);
                var index = U8Searching.IndexOf(value, _separator, size);
                if (index >= 0)
                {
                    _current = new(remaining.Offset, index);
                    _remaining = new(
                        remaining.Offset + index + (int)size,
                        remaining.Length - index - (int)size);
                }
                else
                {
                    _current = remaining;
                    _remaining = default;
                }

                return true;
            }

            return false;
        }

        readonly object IEnumerator.Current => Current;
        readonly void IEnumerator.Reset() => throw new NotSupportedException();
        readonly void IDisposable.Dispose() { }
    }

    readonly void ICollection<U8String>.Add(U8String item) => throw new NotSupportedException();
    readonly void ICollection<U8String>.Clear() => throw new NotSupportedException();
    readonly bool ICollection<U8String>.Remove(U8String item) => throw new NotSupportedException();
}

public readonly struct ConfiguredU8Split<TSeparator> :
    IU8Enumerable<ConfiguredU8Split<TSeparator>.Enumerator>
{
    readonly U8String _value;
    readonly TSeparator? _separator;
    readonly U8SplitOptions _options;

    internal ConfiguredU8Split(U8String value, TSeparator? separator, U8SplitOptions options)
    {
        _value = value;
        _separator = separator;
        _options = options;
    }

    public readonly Enumerator GetEnumerator() => new(_value, _separator, _options);

    readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();
    readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();

    public struct Enumerator : IU8Enumerator
    {
        readonly byte[]? _value;
        readonly TSeparator? _separator;
        readonly U8Size _separatorSize;
        readonly U8SplitOptions _options;
        U8Range _current;
        U8Range _remaining;

        internal Enumerator(U8String value, TSeparator? separator, U8SplitOptions options)
        {
            _value = value._value;
            _separator = separator;
            _separatorSize = U8Info.GetSize(separator);
            _options = options;
            _remaining = value._inner;
        }

        public readonly U8String Current => new(_value, _current.Offset, _current.Length);

        // TODO: Not most efficient but it works for now
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext()
        {
        Next:
            var remaining = _remaining;
            if (remaining.Length > 0)
            {
                var size = _separatorSize;
                var value = _value!.SliceUnsafe(remaining.Offset, remaining.Length);
                var index = U8Searching.IndexOf(value, _separator, size);
                if (index >= 0)
                {
                    _current = (_options & U8SplitOptions.Trim) != U8SplitOptions.Trim
                        ? new(remaining.Offset, index)
                        : TrimEntry(_value!, new(remaining.Offset, index));
                    _remaining = new(
                        remaining.Offset + index + (int)size,
                        remaining.Length - index - (int)size);
                }
                else
                {
                    _current = (_options & U8SplitOptions.Trim) != U8SplitOptions.Trim
                        ? remaining
                        : TrimEntry(_value!, remaining);
                    _remaining = default;
                }

                if ((_options & U8SplitOptions.RemoveEmpty) is U8SplitOptions.RemoveEmpty
                    && _current.Length is 0)
                {
                    goto Next;
                }

                return true;
            }

            return false;
        }

        private static U8Range TrimEntry(byte[] value, U8Range range)
        {
            // This could have been done better but works for now.
            return new U8String(value, range).Trim()._inner;
        }

        readonly object IEnumerator.Current => Current;
        readonly void IEnumerator.Reset() => throw new NotSupportedException();
        readonly void IDisposable.Dispose() { }
    }
}

// // TODO:
// public ref struct U8RefSplit
// {
//     readonly U8String _value;
//     readonly ReadOnlySpan<byte> _separator;
//     // Can't force JIT/ILC to emit proper layout here which is sad, and explicit layout ruins codegen
//     // int _count;
// }

1	using System.Collections;
2	using System.Runtime.InteropServices;
3	using System.Text;
4	using U8Primitives.Abstractions;
5	using U8Primitives.InteropServices;
6
7	#pragma warning disable RCS1085, RCS1085FadeOut, IDE0032 // Use auto-implemented property. Why: readable fields.
8	namespace U8Primitives;
9
10	public readonly partial struct U8String
11	{
12	public U8SplitPair SplitFirst(byte separator)
13	{
14	if (!U8Info.IsAsciiByte(separator))	×
15	{
16	// TODO: EH UX
17	ThrowHelpers.ArgumentOutOfRange();	×
18	}
19
20	var source = this;	×
21	if (!source.IsEmpty)	×
22	{
23	var span = source.UnsafeSpan;	×
24	var index = span.IndexOf(separator);	×
25	if (index >= 0)	×
26	{
27	return new(source, index, 1);	×
28	}
29
30	return U8SplitPair.NotFound(source);	×
31	}
32
33	return default;	×
34	}
35
36	public U8SplitPair SplitFirst(char separator) => char.IsAscii(separator)	×
37	? SplitFirst((byte)separator)	×
38	: SplitFirstUnchecked(separator.NonAsciiToUtf8(out _));	×
39
40	public U8SplitPair SplitFirst(Rune separator) => separator.IsAscii	×
41	? SplitFirst((byte)separator.Value)	×
42	: SplitFirstUnchecked(separator.NonAsciiToUtf8(out _));	×
43
44	public U8SplitPair SplitFirst(U8String separator)
45	{
46	var source = this;	×
47	if (!source.IsEmpty)	×
48	{
49	if (!separator.IsEmpty)	×
50	{
51	var span = source.UnsafeSpan;	×
52	var index = span.IndexOf(separator.UnsafeSpan);	×
53	if (index >= 0)	×
54	{
55	return new(source, index, separator.Length);	×
56	}
57	}
58
59	return U8SplitPair.NotFound(source);	×
60	}
61
62	return default;	×
63	}
64
65	// It would be really nice to aggressively inline this method
66	// but the way validation is currently implemented does not significantly
67	// benefit from splitting on UTF-8 literals while possibly risking
68	// running out of inlining budget significantly regressing performance everywhere else.
69	public U8SplitPair SplitFirst(ReadOnlySpan<byte> separator)
70	{
71	var source = this;	×
72	if (!source.IsEmpty)	×
73	{
74	if (!separator.IsEmpty)	×
75	{
76	var span = source.UnsafeSpan;	×
77	var index = span.IndexOf(separator);	×
78	if (index >= 0)	×
79	{
80	// Same as with Slice(int, int), this might dereference past the end of the string.
81	// TODO: Do something about it if it's ever an issue.
82	if (U8Info.IsContinuationByte(source.UnsafeRefAdd(index)) \|\|	×
83	U8Info.IsContinuationByte(source.UnsafeRefAdd(index + separator.Length)))	×
84	{
85	ThrowHelpers.InvalidSplit();	×
86	}
87
88	return new(source, index, separator.Length);	×
89	}
90	}
91
92	return U8SplitPair.NotFound(source);	×
93	}
94
95	return default;	×
96	}
97
98	[MethodImpl(MethodImplOptions.AggressiveInlining)]
99	public U8SplitPair SplitFirstUnchecked(ReadOnlySpan<byte> separator)
100	{
101	var source = this;	×
102	if (!source.IsEmpty)	×
103	{
104	if (!separator.IsEmpty)	×
105	{
106	var span = source.UnsafeSpan;	×
107	var index = span.IndexOf(separator);	×
108	if (index >= 0)	×
109	{
110	return new(source, index, separator.Length);	×
111	}
112	}
113
114	return U8SplitPair.NotFound(source);	×
115	}
116
117	return default;	×
118	}
119
120	public U8SplitPair SplitLast(byte separator)
121	{
122	if (!U8Info.IsAsciiByte(separator))	×
123	{
124	// TODO: EH UX
125	ThrowHelpers.ArgumentOutOfRange();	×
126	}
127
128	var source = this;	×
129	if (!source.IsEmpty)	×
130	{
131	var span = source.UnsafeSpan;	×
132	var index = span.LastIndexOf(separator);	×
133	if (index >= 0)	×
134	{
135	return new(source, index, 1);	×
136	}
137
138	return U8SplitPair.NotFound(source);	×
139	}
140
141	return default;	×
142	}
143
144	public U8SplitPair SplitLast(char separator) => char.IsAscii(separator)	×
145	? SplitLast((byte)separator)	×
146	: SplitLastUnchecked(separator.NonAsciiToUtf8(out _));	×
147
148	public U8SplitPair SplitLast(Rune separator) => separator.IsAscii	×
149	? SplitLast((byte)separator.Value)	×
150	: SplitLastUnchecked(separator.NonAsciiToUtf8(out _));	×
151
152	public U8SplitPair SplitLast(U8String separator)
153	{
154	var source = this;	×
155	if (!source.IsEmpty)	×
156	{
157	if (!separator.IsEmpty)	×
158	{
159	var span = source.UnsafeSpan;	×
160	var index = span.LastIndexOf(separator.UnsafeSpan);	×
161	if (index >= 0)	×
162	{
163	return new(source, index, separator.Length);	×
164	}
165	}
166
167	return U8SplitPair.NotFound(source);	×
168	}
169
170	return default;	×
171	}
172
173	public U8SplitPair SplitLast(ReadOnlySpan<byte> separator)
174	{
175	var source = this;	×
176	if (!source.IsEmpty)	×
177	{
178	if (!separator.IsEmpty)	×
179	{
180	var span = source.UnsafeSpan;	×
181	var index = span.LastIndexOf(separator);	×
182	if (index >= 0)	×
183	{
184	if (U8Info.IsContinuationByte(source.UnsafeRefAdd(index)) \|\|	×
185	U8Info.IsContinuationByte(source.UnsafeRefAdd(index + separator.Length)))	×
186	{
187	ThrowHelpers.InvalidSplit();	×
188	}
189
190	return new(source, index, separator.Length);	×
191	}
192	}
193
194	return U8SplitPair.NotFound(source);	×
195	}
196
197	return default;	×
198	}
199
200	[MethodImpl(MethodImplOptions.AggressiveInlining)]
201	public U8SplitPair SplitLastUnchecked(ReadOnlySpan<byte> separator)
202	{
203	var source = this;	×
204	if (!source.IsEmpty)	×
205	{
206	if (!separator.IsEmpty)	×
207	{
208	var span = source.UnsafeSpan;	×
209	var index = span.LastIndexOf(separator);	×
210	if (index >= 0)	×
211	{
212	return new(source, index, separator.Length);	×
213	}
214	}
215
216	return U8SplitPair.NotFound(source);	×
217	}
218
219	return default;	×
220	}
221
222	public U8Split<byte> Split(byte separator)
223	{
224	if (!U8Info.IsAsciiByte(separator))	×
225	{
226	ThrowHelpers.ArgumentOutOfRange();	×
227	}
228
229	return new(this, separator);	×
230	}
231
232	public ConfiguredU8Split<byte> Split(byte separator, U8SplitOptions options)
233	{
234	if (!U8Info.IsAsciiByte(separator))	×
235	{
236	ThrowHelpers.ArgumentOutOfRange();	×
237	}
238
239	return new(this, separator, options);	×
240	}
241
242	public U8Split<char> Split(char separator)
243	{
244	if (char.IsSurrogate(separator))	×
245	{
246	ThrowHelpers.ArgumentOutOfRange();	×
247	}
248
249	return new(this, separator);	×
250	}
251
252	public ConfiguredU8Split<char> Split(char separator, U8SplitOptions options)
253	{
254	if (char.IsSurrogate(separator))	×
255	{
256	ThrowHelpers.ArgumentOutOfRange();	×
257	}
258
259	return new(this, separator, options);	×
260	}
261
262	public U8Split<Rune> Split(Rune separator) => new(this, separator);	×
263
264	public ConfiguredU8Split<Rune> Split(Rune separator, U8SplitOptions options) => new(this, separator, options);	×
265
266	public U8Split Split(U8String separator)
267	{
268	return !separator.IsEmpty ? new(this, separator) : default;	×
269	}
270
271	public ConfiguredU8Split<U8String> Split(U8String separator, U8SplitOptions options)
272	{
273	return !separator.IsEmpty ? new(this, separator, options) : default;	×
274	}
275
276	public U8Split<byte[]> Split(byte[] separator)
277	{
278	if (!IsValid(separator))	×
279	{
280	// TODO: EH UX
281	ThrowHelpers.InvalidSplit();	×
282	}
283
284	var source = this;	×
285	return (!source.IsEmpty && separator != null) ? new(source, separator) : default;	×
286	}
287
288	public ConfiguredU8Split<byte[]> Split(byte[] separator, U8SplitOptions options)
289	{
290	if (!IsValid(separator))	×
291	{
292	// TODO: EH UX
293	ThrowHelpers.InvalidSplit();	×
294	}
295
296	var source = this;	×
297	return (!source.IsEmpty && separator != null) ? new(source, separator, options) : default;	×
298	}
299	}
300
301	public readonly record struct U8SplitPair
302	{
303	readonly U8String _value;
304	readonly int _offset;
305	readonly int _stride;
306
307	[MethodImpl(MethodImplOptions.AggressiveInlining)]
308	internal U8SplitPair(U8String value, int offset, int stride)
309	{
310	_value = value;	×
311	_offset = offset;	×
312	_stride = stride;	×
313	}	×
314
315	[MethodImpl(MethodImplOptions.AggressiveInlining)]
316	public static U8SplitPair NotFound(U8String value)
317	{
318	return new(value, value.Length, 0);	×
319	}
320
321	public U8String Segment
322	{
323	[MethodImpl(MethodImplOptions.AggressiveInlining)]
324	get => U8Marshal.Slice(_value, 0, _offset);	×
325	}
326
327	public U8String Remainder
328	{
329	[MethodImpl(MethodImplOptions.AggressiveInlining)]
330	get => U8Marshal.Slice(_value, _offset + _stride);	×
331	}
332
333	[MethodImpl(MethodImplOptions.AggressiveInlining)]
334	public void Deconstruct(out U8String segment, out U8String remainder)
335	{
336	segment = Segment;	×
337	remainder = Remainder;	×
338	}	×
339
340	[MethodImpl(MethodImplOptions.AggressiveInlining)]
341	public static implicit operator (U8String, U8String)(U8SplitPair value)
342	{
343	return (value.Segment, value.Remainder);	×
344	}
345	}
346
347	public struct U8Split : ICollection<U8String>, IU8Enumerable<U8Split.Enumerator>
348	{
349	readonly U8String _value;
350	readonly U8String _separator;
351	int _count;
352
353	internal U8Split(U8String value, U8String separator)
354	{
355	if (!value.IsEmpty)	×
356	{
357	_value = value;	×
358	_separator = separator;	×
359	_count = -1;	×
360	}
361	}	×
362
363	public int Count
364	{
365	[MethodImpl(MethodImplOptions.AggressiveInlining)]
366	get
367	{
368	var count = _count;	×
369	if (count >= 0)	×
370	{
371	return count;	×
372	}
373
374	// Matches the behavior of string.Split('\n').Length for "hello\n"
375	// TODO: Should we break consistency and not count the very last segment if it is empty?
376	return _count = Count(_value.UnsafeSpan, _separator) + 1;	×
377
378	static int Count(ReadOnlySpan<byte> value, ReadOnlySpan<byte> separator)
379	{
380	return U8Searching.Count(value, separator);	×
381	}
382	}
383	}
384
385	public readonly bool Contains(U8String item)
386	{
387	var separator = _separator;	×
388	var overlaps = item.Contains(separator);	×
389
390	return !overlaps && _value.Contains(item);	×
391	}
392
393	public void CopyTo(U8String[] array, int index)
394	{
395	this.CopyTo<U8Split, Enumerator>(array.AsSpan()[index..]);	×
396	}	×
397
398	[MethodImpl(MethodImplOptions.AggressiveInlining)]
399	public readonly void Deconstruct(out U8String first, out U8String second)
400	{
401	this.Deconstruct<U8Split, Enumerator>(out first, out second);	×
402	}	×
403
404	[MethodImpl(MethodImplOptions.AggressiveInlining)]
405	public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
406	{
407	this.Deconstruct<U8Split, Enumerator>(out first, out second, out third);	×
408	}	×
409
410	public U8String[] ToArray() => this.ToArray<U8Split, Enumerator>();	×
411	public List<U8String> ToList() => this.ToList<U8Split, Enumerator>();	×
412
413	/// <summary>
414	/// Returns a <see cref="Enumerator"/> over the provided string.
415	/// </summary>
416	[MethodImpl(MethodImplOptions.AggressiveInlining)]
417	public readonly Enumerator GetEnumerator() => new(_value, _separator);	×
418
419	readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();	×
420	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	×
421	readonly bool ICollection<U8String>.IsReadOnly => true;	×
422
423	[StructLayout(LayoutKind.Auto)]
424	public struct Enumerator : IU8Enumerator
425	{
426	readonly byte[]? _value;
427	readonly U8String _separator;
428	U8Range _current;
429	U8Range _remaining;
430
431	internal Enumerator(U8String value, U8String separator)
432	{
433	_value = value._value;	×
434	_separator = separator;	×
435	_remaining = value._inner;	×
436	}	×
437
438	public readonly U8String Current => new(_value, _current.Offset, _current.Length);	×
439
440	[MethodImpl(MethodImplOptions.AggressiveInlining)]
441	public bool MoveNext()
442	{
443	var remaining = _remaining;	×
444	if (remaining.Length > 0)	×
445	{
446	var value = _value!.SliceUnsafe(remaining.Offset, remaining.Length);	×
447	var separator = _separator;	×
448	var index = value.IndexOf(separator.UnsafeSpan);	×
449	if (index >= 0)	×
450	{
451	_current = new(remaining.Offset, index);	×
452	_remaining = new(	×
453	remaining.Offset + index + separator.Length,	×
454	remaining.Length - index - separator.Length);	×
455	}
456	else
457	{
458	_current = remaining;	×
459	_remaining = default;	×
460	}
461
462	return true;	×
463	}
464
465	return false;	×
466	}
467
468	readonly object IEnumerator.Current => Current;	×
469	readonly void IEnumerator.Reset() => throw new NotSupportedException();	×
470	readonly void IDisposable.Dispose() { }	×
471	}
472
473	readonly void ICollection<U8String>.Add(U8String item) => throw new NotSupportedException();	×
474	readonly void ICollection<U8String>.Clear() => throw new NotSupportedException();	×
475	readonly bool ICollection<U8String>.Remove(U8String item) => throw new NotSupportedException();	×
476	}
477
478	// TODO: I don't think this is a successful design. JIT/ILC can't optimize away
479	// ASCII char literal conversion causing the size of MoveNext() to explode. This needs an overhaul,
480	// possibly disambiguating on struct vs separator types or maybe even 3 options:
481	// primitive, span (ref split aka SplitRef? SplitSpan?) and U8String / byte[]
482	[StructLayout(LayoutKind.Auto)]
483	public struct U8Split<TSeparator> :
484	ICollection<U8String>,
485	IU8Enumerable<U8Split<TSeparator>.Enumerator>
486	{
487	readonly U8String _value;
488	readonly TSeparator? _separator;
489	int _count;
490
491	// TODO: Move value.IsEmpty -> count = 0 check here
492	internal U8Split(U8String value, TSeparator? separator)
493	{
494	if (!value.IsEmpty)	×
495	{
496	_value = value;	×
497	_separator = separator;	×
498	_count = -1;	×
499	}
500	}	×
501
502	public int Count
503	{
504	[MethodImpl(MethodImplOptions.AggressiveInlining)]
505	get
506	{
507	var count = _count;	×
508	if (count >= 0)	×
509	{
510	return count;	×
511	}
512
513	// Matches the behavior of string.Split('\n').Length for "hello\n"
514	// TODO: Should we break consistency and not count the very last segment if it is empty?
515	return _count = Count(_value, _separator) + 1;	×
516
517	static int Count(ReadOnlySpan<byte> value, TSeparator? separator)
518	{
519	return U8Searching.Count(value, separator);	×
520	}
521	}
522	}
523
524	public readonly bool Contains(U8String item)
525	{
526	var separator = _separator;	×
527	var overlaps = U8Searching.Contains(item, separator);	×
528
529	return !overlaps && _value.Contains(item);	×
530	}
531
532	public void CopyTo(U8String[] array, int index)
533	{
534	this.CopyTo<U8Split<TSeparator>, Enumerator>(array.AsSpan()[index..]);	×
535	}	×
536
537	[MethodImpl(MethodImplOptions.AggressiveInlining)]
538	public readonly void Deconstruct(out U8String first, out U8String second)
539	{
540	this.Deconstruct<U8Split<TSeparator>, Enumerator>(out first, out second);	×
541	}	×
542
543	[MethodImpl(MethodImplOptions.AggressiveInlining)]
544	public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
545	{
546	this.Deconstruct<U8Split<TSeparator>, Enumerator>(out first, out second, out third);	×
547	}	×
548
549	public U8String[] ToArray() => this.ToArray<U8Split<TSeparator>, Enumerator>();	×
550	public List<U8String> ToList() => this.ToList<U8Split<TSeparator>, Enumerator>();	×
551
552	/// <summary>
553	/// Returns a <see cref="Enumerator"/> over the provided string.
554	/// </summary>
555	[MethodImpl(MethodImplOptions.AggressiveInlining)]
556	public readonly Enumerator GetEnumerator() => new(_value, _separator);	×
557
558	readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();	×
559	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	×
560	readonly bool ICollection<U8String>.IsReadOnly => true;	×
561
562	[StructLayout(LayoutKind.Auto)]
563	public struct Enumerator : IU8Enumerator
564	{
565	readonly byte[]? _value;
566	readonly TSeparator? _separator;
567	readonly U8Size _separatorSize;
568	U8Range _current;
569	U8Range _remaining;
570
571	internal Enumerator(U8String value, TSeparator? separator)
572	{
573	_value = value._value;	×
574	_separator = separator;	×
575	_separatorSize = U8Info.GetSize(separator);	×
576	_remaining = value._inner;	×
577	}	×
578
579	public readonly U8String Current => new(_value, _current.Offset, _current.Length);	×
580
581	[MethodImpl(MethodImplOptions.AggressiveInlining)]
582	public bool MoveNext()
583	{
584	var remaining = _remaining;	×
585	if (remaining.Length > 0)	×
586	{
587	var size = _separatorSize;	×
588	var value = _value!.SliceUnsafe(remaining.Offset, remaining.Length);	×
589	var index = U8Searching.IndexOf(value, _separator, size);	×
590	if (index >= 0)	×
591	{
592	_current = new(remaining.Offset, index);	×
593	_remaining = new(	×
594	remaining.Offset + index + (int)size,	×
595	remaining.Length - index - (int)size);	×
596	}
597	else
598	{
599	_current = remaining;	×
600	_remaining = default;	×
601	}
602
603	return true;	×
604	}
605
606	return false;	×
607	}
608
609	readonly object IEnumerator.Current => Current;	×
610	readonly void IEnumerator.Reset() => throw new NotSupportedException();	×
611	readonly void IDisposable.Dispose() { }	×
612	}
613
614	readonly void ICollection<U8String>.Add(U8String item) => throw new NotSupportedException();	×
615	readonly void ICollection<U8String>.Clear() => throw new NotSupportedException();	×
616	readonly bool ICollection<U8String>.Remove(U8String item) => throw new NotSupportedException();	×
617	}
618
619	public readonly struct ConfiguredU8Split<TSeparator> :
620	IU8Enumerable<ConfiguredU8Split<TSeparator>.Enumerator>
621	{
622	readonly U8String _value;
623	readonly TSeparator? _separator;
624	readonly U8SplitOptions _options;
625
626	internal ConfiguredU8Split(U8String value, TSeparator? separator, U8SplitOptions options)
627	{
628	_value = value;	×
629	_separator = separator;	×
630	_options = options;	×
631	}	×
632
633	public readonly Enumerator GetEnumerator() => new(_value, _separator, _options);	×
634
635	readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();	×
636	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	×
637
638	public struct Enumerator : IU8Enumerator
639	{
640	readonly byte[]? _value;
641	readonly TSeparator? _separator;
642	readonly U8Size _separatorSize;
643	readonly U8SplitOptions _options;
644	U8Range _current;
645	U8Range _remaining;
646
647	internal Enumerator(U8String value, TSeparator? separator, U8SplitOptions options)
648	{
649	_value = value._value;	×
650	_separator = separator;	×
651	_separatorSize = U8Info.GetSize(separator);	×
652	_options = options;	×
653	_remaining = value._inner;	×
654	}	×
655
656	public readonly U8String Current => new(_value, _current.Offset, _current.Length);	×
657
658	// TODO: Not most efficient but it works for now
659	[MethodImpl(MethodImplOptions.AggressiveInlining)]
660	public bool MoveNext()
661	{
662	Next:
663	var remaining = _remaining;	×
664	if (remaining.Length > 0)	×
665	{
666	var size = _separatorSize;	×
667	var value = _value!.SliceUnsafe(remaining.Offset, remaining.Length);	×
668	var index = U8Searching.IndexOf(value, _separator, size);	×
669	if (index >= 0)	×
670	{
671	_current = (_options & U8SplitOptions.Trim) != U8SplitOptions.Trim	×
672	? new(remaining.Offset, index)	×
673	: TrimEntry(_value!, new(remaining.Offset, index));	×
674	_remaining = new(	×
675	remaining.Offset + index + (int)size,	×
676	remaining.Length - index - (int)size);	×
677	}
678	else
679	{
680	_current = (_options & U8SplitOptions.Trim) != U8SplitOptions.Trim	×
681	? remaining	×
682	: TrimEntry(_value!, remaining);	×
683	_remaining = default;	×
684	}
685
686	if ((_options & U8SplitOptions.RemoveEmpty) is U8SplitOptions.RemoveEmpty	×
687	&& _current.Length is 0)	×
688	{
689	goto Next;
690	}
691
692	return true;	×
693	}
694
695	return false;	×
696	}
697
698	private static U8Range TrimEntry(byte[] value, U8Range range)
699	{
700	// This could have been done better but works for now.
701	return new U8String(value, range).Trim()._inner;	×
702	}
703
704	readonly object IEnumerator.Current => Current;	×
705	readonly void IEnumerator.Reset() => throw new NotSupportedException();	×
706	readonly void IDisposable.Dispose() { }	×
707	}
708	}
709
710	// // TODO:
711	// public ref struct U8RefSplit
712	// {
713	// readonly U8String _value;
714	// readonly ReadOnlySpan<byte> _separator;
715	// // Can't force JIT/ILC to emit proper layout here which is sad, and explicit layout ruins codegen
716	// // int _count;
717	// }

neon-sunset / U8String / 5872750434

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous