5912240727

Committed 19 Aug 2023 03:39PM UTC coverage: 22.685% (+0.3%) from 22.422%

Build # 5912240727

Build Type

push

github

Committed by

neon-sunset

Commit Message

feat: enumerators improvements

Run Details

122 of 776 branches covered (15.72%)

Branch coverage included in aggregate %.

65 of 65 new or added lines in 4 files covered. (100.0%)

439 of 1697 relevant lines covered (25.87%)

30109.06 hits per line

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

52.29

/src/U8String.Enumeration.cs

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

using Rune = System.Text.Rune;

namespace U8Primitives;

#pragma warning disable IDE0032, IDE0057 // Use auto property and index operator. Why: Perf, struct layout, accuracy and codegen.
public readonly partial struct U8String
{
    /// <summary>
    /// Returns a collection of <see cref="char"/>s over the provided string.
    /// </summary>
    /// <remarks>
    /// This is a lazily-evaluated allocation-free collection.
    /// </remarks>
    public U8Chars Chars
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get => new(this);
    }

    /// <summary>
    /// Returns a collection of <see cref="Rune"/>s over the provided string.
    /// </summary>
    /// <remarks>
    /// This is a lazily-evaluated allocation-free collection.
    /// </remarks>
    public U8Runes Runes
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get => new(this);
    }

    /// <summary>
    /// Returns a collection of lines over the provided string.
    /// </summary>
    /// <remarks>
    /// This is a lazily-evaluated allocation-free collection.
    /// </remarks>
    public U8Lines Lines
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get => new(this);
    }

    // Bad codegen still :(
    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public Enumerator GetEnumerator() => new(this);

    IEnumerator<byte> IEnumerable<byte>.GetEnumerator() => GetEnumerator();
    IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();

    public struct Enumerator : IEnumerator<byte>
    {
        readonly byte[]? _value;
        readonly int _offset;
        readonly int _length;
        int _index;

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public Enumerator(U8String value)
        {
            _value = value._value;
            _offset = value.Offset;
            _length = value.Length;
            _index = -1;
        }

        // Still cheaper than MemoryMarshal clever variants
        public readonly byte Current => _value![_offset + _index];

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext() => (uint)(++_index) < (uint)_length;
        // {
        //     var index = _index;
        //     if (++index < _length)
        //     {
        //         // Current = Unsafe.Add(
        //         //     ref MemoryMarshal.GetArrayDataReference(_value!),
        //         //     (nint)(uint)(_offset + index));
        //         Current = _value![_offset + index];
        //         _index = index;
        //         return true;
        //     }

        //     return false;
        // }

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public void Reset() => _index = -1;

        readonly object IEnumerator.Current => Current;
        readonly void IDisposable.Dispose() { }
    }
}

/// <summary>
/// A collection of chars in a provided <see cref="U8String"/>.
/// </summary>
public struct U8Chars : ICollection<char>, IEnumerable<char, U8Chars.Enumerator>
{
    readonly U8String _value;

    int _count;

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public U8Chars(U8String value)
    {
        _value = value;
        _count = value.IsEmpty ? 0 : -1;
    }

    /// <summary>
    /// The number of chars in the current <see cref="U8String"/>.
    /// </summary>
    public int Count
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get
        {
            // Somehow the codegen here is underwhelming
            var count = _count;
            if (count >= 0)
            {
                return count;
            }
            return _count = Count(_value.UnsafeSpan);

            static int Count(ReadOnlySpan<byte> value)
            {
                Debug.Assert(!value.IsEmpty);

                // TODO: Is this enough?
                return Encoding.UTF8.GetCharCount(value);
            }
        }
    }

    // TODO: Wow, this seems to be terribly broken on surrogate chars and 
    // there is no easy way to fix it without sacrificing performance.
    // Perhaps it is worth just do the transcoding iteration here and warn the users
    // instead of straight up producing UB or throwing exceptions???
    public readonly bool Contains(char item) => _value.Contains(item);

    public readonly void CopyTo(char[] destination, int index)
    {
        var value = _value;
        if (!value.IsEmpty)
        {
            Encoding.UTF8.GetChars(value.UnsafeSpan, destination.AsSpan()[index..]);
        }
    }

    public readonly void Deconstruct(out char first, out char second)
    {
        this.Deconstruct<U8Chars, Enumerator, char>(out first, out second);
    }

    public readonly void Deconstruct(out char first, out char second, out char third)
    {
        this.Deconstruct<U8Chars, Enumerator, char>(out first, out second, out third);
    }

    public char[] ToArray()
    {
        var value = _value;
        if (!value.IsEmpty)
        {
            var chars = new char[Count];
            Encoding.UTF8.GetChars(value.UnsafeSpan, chars);
            return chars;
        }

        return Array.Empty<char>();
    }

    public List<char> ToList()
    {
        var value = _value;
        if (!value.IsEmpty)
        {
            var count = Count;
            var chars = new List<char>(count);
            CollectionsMarshal.SetCount(chars, count);
            var span = CollectionsMarshal.AsSpan(chars);

            Encoding.UTF8.GetChars(value.UnsafeSpan, span);
            return chars;
        }

        return new List<char>();
    }

    public readonly Enumerator GetEnumerator() => new(_value);

    readonly IEnumerator<char> IEnumerable<char>.GetEnumerator() => new Enumerator(_value);
    readonly IEnumerator IEnumerable.GetEnumerator() => new Enumerator(_value);

    public struct Enumerator : IEnumerator<char>
    {
        // TODO: refactor layout
        readonly byte[]? _value;
        readonly int _offset;
        readonly int _length;
        int _nextByteIdx;
        uint _currentCharPair;

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public Enumerator(U8String value)
        {
            _value = value._value;
            _offset = value.Offset;
            _length = value.Length;
            _nextByteIdx = 0;
        }

        // TODO
        public readonly char Current => (char)_currentCharPair;

        // TODO: This looks terrible, there must be a better way
        // to convert UTF-8 to UTF-16 with an enumerator.
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext()
        {
            var (offset, length, nextByteIdx, currentCharPair) =
                (_offset, _length, _nextByteIdx, _currentCharPair);

            if (currentCharPair < char.MaxValue)
            {
                if ((uint)nextByteIdx < (uint)length)
                {
                    var span = _value!.SliceUnsafe(offset + nextByteIdx, length - nextByteIdx);
                    var firstByte = MemoryMarshal.GetReference(span);
                    if (U8Info.IsAsciiByte(firstByte))
                    {
                        // Fast path because Rune.DecodeFromUtf8 won't inline
                        // making UTF-8 push us more and more towards anglocentrism.
                        _nextByteIdx = nextByteIdx + 1;
                        _currentCharPair = firstByte;
                        return true;
                    }

                    var status = Rune.DecodeFromUtf8(span, out var rune, out var bytesConsumed);
                    Debug.Assert(status is OperationStatus.Done);

                    _nextByteIdx = nextByteIdx + bytesConsumed;

                    if (rune.IsBmp)
                    {
                        _currentCharPair = (uint)rune.Value;
                        return true;
                    }

                    // I wonder if this just explodes on BigEndian
                    var runeValue = (uint)rune.Value;
                    var highSurrogate = (char)((runeValue + ((0xD800u - 0x40u) << 10)) >> 10);
                    var lowSurrogate = (char)((runeValue & 0x3FFu) + 0xDC00u);
                    _currentCharPair = highSurrogate + ((uint)lowSurrogate << 16);
                    return true;
                }

                return false;
            }

            _currentCharPair = currentCharPair >> 16;
            return true;
        }

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public void Reset() => _nextByteIdx = 0;

        readonly object IEnumerator.Current => Current;
        readonly void IDisposable.Dispose() { }
    }

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

/// <summary>
/// A collection of Runes (unicode scalar values) in a provided <see cref="U8String"/>.
/// </summary>
public struct U8Runes : ICollection<Rune>, IEnumerable<Rune, U8Runes.Enumerator>
{
    readonly U8String _value;

    // If we bring up non-ascii counting to ascii level, we might not need this
    // similar to LineCollection.
    int _count;

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public U8Runes(U8String value)
    {
        _value = value;
        _count = value.IsEmpty ? 0 : -1;
    }

    /// <summary>
    /// The number of Runes (unicode scalar values) in the current <see cref="U8String"/>.
    /// </summary>
    public int Count
    {
        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        get
        {
            // Somehow the codegen here is underwhelming
            var count = _count;
            if (count >= 0)
            {
                return count;
            }

            return _count = Count(_value.UnsafeSpan);

            static int Count(ReadOnlySpan<byte> value)
            {
                Debug.Assert(!value.IsEmpty);

                // TODO: SIMD non-continuation byte counting
                var runeCount = (int)(nint)Polyfills.Text.Ascii.GetIndexOfFirstNonAsciiByte(value);
                value = value.SliceUnsafe(runeCount);

                for (var i = 0; (uint)i < (uint)value.Length; i += U8Info.RuneLength(value.AsRef(i)))
                {
                    runeCount++;
                }

                return runeCount;
            }
        }
    }

    public readonly bool Contains(Rune item) => _value.Contains(item);

    public readonly void CopyTo(Rune[] destination, int index)
    {
        // TODO: Simple SIMD widen ASCII to UTF-32 (ideally widen+validate in place instead of double traversal)
        // TODO: Consistency and correctness? Implement single-pass vectorized conversion?
        foreach (var rune in this)
        {
            destination[index++] = rune;
        }
    }

    public readonly void Deconstruct(out Rune first, out Rune second)
    {
        this.Deconstruct<U8Runes, Enumerator, Rune>(out first, out second);
    }

    public readonly void Deconstruct(out Rune first, out Rune second, out Rune third)
    {
        this.Deconstruct<U8Runes, Enumerator, Rune>(out first, out second, out third);
    }

    public Rune[] ToArray() => this.ToArray<U8Runes, Enumerator, Rune>();

    public List<Rune> ToList() => this.ToList<U8Runes, Enumerator, Rune>();

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly Enumerator GetEnumerator() => new(_value);

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

    public struct Enumerator : IEnumerator<Rune>
    {
        readonly byte[]? _value;
        readonly int _offset;
        readonly int _length;
        int _index;

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public Enumerator(U8String value)
        {
            _value = value._value;
            _offset = value.Offset;
            _length = value.Length;
        }

        public Rune Current { get; private set; }

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext()
        {
            var index = _index;
            if (index < _length)
            {
                ref var ptr = ref MemoryMarshal
                    .GetArrayDataReference(_value!)
                    .Add(_offset + index);

                Current = U8Conversions.CodepointToRune(ref ptr, out var size);
                _index = index + size;
                return true;
            }

            return false;
        }

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public void Reset() => _index = -1;

        readonly object IEnumerator.Current => Current;
        readonly void IDisposable.Dispose() { }
    }

    readonly bool ICollection<Rune>.IsReadOnly => true;
    readonly void ICollection<Rune>.Add(Rune item) => throw new NotImplementedException();
    readonly void ICollection<Rune>.Clear() => throw new NotImplementedException();
    readonly bool ICollection<Rune>.Remove(Rune item) => throw new NotImplementedException();
}

/// <summary>
/// A collection of lines in a provided <see cref="U8String"/>.
/// </summary>
public struct U8Lines : ICollection<U8String>, IU8Enumerable<U8Lines.Enumerator>
{
    readonly U8String _value;

    // We might not need this. Although counting is O(n), the absolute performance
    // is very good, and on AVX2/512 - it's basically instantenous.
    int _count;

    /// <summary>
    /// Creates a new line enumeration over the provided string.
    /// </summary>
    /// <param name="value">The string to enumerate over.</param>
    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public U8Lines(U8String value)
    {
        _value = value;
        _count = value.IsEmpty ? 0 : -1;
    }

    /// <summary>
    /// The number of lines in the current <see cref="U8String"/>.
    /// </summary>
    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?
            // (likely no - an empty line is still a line)
            return _count = _value.UnsafeSpan.Count((byte)'\n') + 1;
        }
    }

    public readonly bool Contains(U8String item)
    {
        return !item.Contains((byte)'\n') && _value.Contains(item);
    }

    public void CopyTo(U8String[] destination, int index)
    {
        this.CopyTo<U8Lines, Enumerator, U8String>(destination.AsSpan()[index..]);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly void Deconstruct(out U8String first, out U8String second)
    {
        this.Deconstruct<U8Lines, Enumerator, U8String>(out first, out second);
    }

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
    {
        this.Deconstruct<U8Lines, Enumerator, U8String>(out first, out second, out third);
    }

    public U8String[] ToArray() => this.ToArray<U8Lines, Enumerator, U8String>();
    public List<U8String> ToList() => this.ToList<U8Lines, Enumerator, U8String>();

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

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

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

    /// <summary>
    /// A struct that enumerates lines over a string.
    /// </summary>
    public struct Enumerator : IU8Enumerator
    {
        // TODO 1: Ensure this is aligned with Rust's .lines() implementation, or not?
        // private static readonly SearchValues<byte> NewLine = SearchValues.Create("\r\n"u8);
        // TODO 2: Consider using 'InnerOffsets'
        private readonly byte[]? _value;
        private U8Range _remaining;
        private U8Range _current;

        /// <summary>
        /// Creates a new line enumerator over the provided string.
        /// </summary>
        /// <param name="value">The string to enumerate over.</param>
        public Enumerator(U8String value)
        {
            _value = value._value;
            _remaining = value._inner;
        }

        /// <summary>
        /// Returns the current line.
        /// </summary>
        public readonly U8String Current => new(_value, _current.Offset, _current.Length);

        /// <summary>
        /// Advances the enumerator to the next line.
        /// </summary>
        [MethodImpl(MethodImplOptions.AggressiveInlining)] // Surprisingly smaller codegen than when not inlined
        public bool MoveNext()
        {
            var remaining = _remaining;
            if (remaining.Length > 0)
            {
                var span = _value!.SliceUnsafe(remaining.Offset, remaining.Length);
                var idx = span.IndexOf((byte)'\n');

                if ((uint)idx < (uint)span.Length)
                {
                    var cutoff = idx;
                    if (idx > 0 && span.AsRef().Add(idx - 1) is (byte)'\r')
                    {
                        cutoff--;
                    }

                    _current = new(remaining.Offset, cutoff);
                    _remaining = new(remaining.Offset + idx + 1, remaining.Length - idx - 1);
                }
                else
                {
                    // We've reached EOF, but we still need to return 'true' for this final
                    // iteration so that the caller can query the Current property once more.
                    _current = new(remaining.Offset, remaining.Length);
                    _remaining = default;
                }

                return true;
            }

            return false;
        }

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

1	using System.Buffers;
2	using System.Collections;
3	using System.Diagnostics;
4	using System.Runtime.InteropServices;
5	using System.Text;
6	using U8Primitives.Abstractions;
7
8	using Rune = System.Text.Rune;
9
10	namespace U8Primitives;
11
12	#pragma warning disable IDE0032, IDE0057 // Use auto property and index operator. Why: Perf, struct layout, accuracy and codegen.
13	public readonly partial struct U8String
14	{
15	/// <summary>
16	/// Returns a collection of <see cref="char"/>s over the provided string.
17	/// </summary>
18	/// <remarks>
19	/// This is a lazily-evaluated allocation-free collection.
20	/// </remarks>
21	public U8Chars Chars
22	{
23	[MethodImpl(MethodImplOptions.AggressiveInlining)]
24	get => new(this);	24✔
25	}
26
27	/// <summary>
28	/// Returns a collection of <see cref="Rune"/>s over the provided string.
29	/// </summary>
30	/// <remarks>
31	/// This is a lazily-evaluated allocation-free collection.
32	/// </remarks>
33	public U8Runes Runes
34	{
35	[MethodImpl(MethodImplOptions.AggressiveInlining)]
36	get => new(this);	30✔
37	}
38
39	/// <summary>
40	/// Returns a collection of lines over the provided string.
41	/// </summary>
42	/// <remarks>
43	/// This is a lazily-evaluated allocation-free collection.
44	/// </remarks>
45	public U8Lines Lines
46	{
47	[MethodImpl(MethodImplOptions.AggressiveInlining)]
48	get => new(this);	×
49	}
50
51	// Bad codegen still :(
52	[MethodImpl(MethodImplOptions.AggressiveInlining)]
53	public Enumerator GetEnumerator() => new(this);	12✔
54
55	IEnumerator<byte> IEnumerable<byte>.GetEnumerator() => GetEnumerator();	×
56	IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	6✔
57
58	public struct Enumerator : IEnumerator<byte>
59	{
60	readonly byte[]? _value;
61	readonly int _offset;
62	readonly int _length;
63	int _index;
64
65	[MethodImpl(MethodImplOptions.AggressiveInlining)]
66	public Enumerator(U8String value)
67	{
68	_value = value._value;	12✔
69	_offset = value.Offset;	12✔
70	_length = value.Length;	12✔
71	_index = -1;	12✔
72	}	12✔
73
74	// Still cheaper than MemoryMarshal clever variants
75	public readonly byte Current => _value![_offset + _index];	3,138✔
76
77	[MethodImpl(MethodImplOptions.AggressiveInlining)]
78	public bool MoveNext() => (uint)(++_index) < (uint)_length;	3,156✔
79	// {
80	// var index = _index;
81	// if (++index < _length)
82	// {
83	// // Current = Unsafe.Add(
84	// // ref MemoryMarshal.GetArrayDataReference(_value!),
85	// // (nint)(uint)(_offset + index));
86	// Current = _value![_offset + index];
87	// _index = index;
88	// return true;
89	// }
90
91	// return false;
92	// }
93
94	[MethodImpl(MethodImplOptions.AggressiveInlining)]
95	public void Reset() => _index = -1;	×
96
97	readonly object IEnumerator.Current => Current;	1,569✔
98	readonly void IDisposable.Dispose() { }	6✔
99	}
100	}
101
102	/// <summary>
103	/// A collection of chars in a provided <see cref="U8String"/>.
104	/// </summary>
105	public struct U8Chars : ICollection<char>, IEnumerable<char, U8Chars.Enumerator>
106	{
107	readonly U8String _value;
108
109	int _count;
110
111	[MethodImpl(MethodImplOptions.AggressiveInlining)]
112	public U8Chars(U8String value)
113	{
114	_value = value;	24✔
115	_count = value.IsEmpty ? 0 : -1;	24✔
116	}	24✔
117
118	/// <summary>
119	/// The number of chars in the current <see cref="U8String"/>.
120	/// </summary>
121	public int Count
122	{
123	[MethodImpl(MethodImplOptions.AggressiveInlining)]
124	get
125	{
126	// Somehow the codegen here is underwhelming
127	var count = _count;	23✔
128	if (count >= 0)	23✔
129	{
130	return count;	8✔
131	}
132	return _count = Count(_value.UnsafeSpan);	15✔
133
134	static int Count(ReadOnlySpan<byte> value)
135	{
136	Debug.Assert(!value.IsEmpty);
137
138	// TODO: Is this enough?
139	return Encoding.UTF8.GetCharCount(value);	15✔
140	}
141	}
142	}
143
144	// TODO: Wow, this seems to be terribly broken on surrogate chars and
145	// there is no easy way to fix it without sacrificing performance.
146	// Perhaps it is worth just do the transcoding iteration here and warn the users
147	// instead of straight up producing UB or throwing exceptions???
148	public readonly bool Contains(char item) => _value.Contains(item);	×
149
150	public readonly void CopyTo(char[] destination, int index)
151	{
152	var value = _value;	×
153	if (!value.IsEmpty)	×
154	{
155	Encoding.UTF8.GetChars(value.UnsafeSpan, destination.AsSpan()[index..]);	×
156	}
157	}	×
158
159	public readonly void Deconstruct(out char first, out char second)
160	{
161	this.Deconstruct<U8Chars, Enumerator, char>(out first, out second);	×
162	}	×
163
164	public readonly void Deconstruct(out char first, out char second, out char third)
165	{
166	this.Deconstruct<U8Chars, Enumerator, char>(out first, out second, out third);	×
167	}	×
168
169	public char[] ToArray()
170	{
171	var value = _value;	6✔
172	if (!value.IsEmpty)	6✔
173	{
174	var chars = new char[Count];	5✔
175	Encoding.UTF8.GetChars(value.UnsafeSpan, chars);	5✔
176	return chars;	5✔
177	}
178
179	return Array.Empty<char>();	1✔
180	}
181
182	public List<char> ToList()
183	{
184	var value = _value;	×
185	if (!value.IsEmpty)	×
186	{
187	var count = Count;	×
188	var chars = new List<char>(count);	×
189	CollectionsMarshal.SetCount(chars, count);	×
190	var span = CollectionsMarshal.AsSpan(chars);	×
191
192	Encoding.UTF8.GetChars(value.UnsafeSpan, span);	×
193	return chars;	×
194	}
195
196	return new List<char>();	×
197	}
198
199	public readonly Enumerator GetEnumerator() => new(_value);	6✔
200
201	readonly IEnumerator<char> IEnumerable<char>.GetEnumerator() => new Enumerator(_value);	×
202	readonly IEnumerator IEnumerable.GetEnumerator() => new Enumerator(_value);	6✔
203
204	public struct Enumerator : IEnumerator<char>
205	{
206	// TODO: refactor layout
207	readonly byte[]? _value;
208	readonly int _offset;
209	readonly int _length;
210	int _nextByteIdx;
211	uint _currentCharPair;
212
213	[MethodImpl(MethodImplOptions.AggressiveInlining)]
214	public Enumerator(U8String value)
215	{
216	_value = value._value;	12✔
217	_offset = value.Offset;	12✔
218	_length = value.Length;	12✔
219	_nextByteIdx = 0;	12✔
220	}	12✔
221
222	// TODO
223	public readonly char Current => (char)_currentCharPair;	1,402✔
224
225	// TODO: This looks terrible, there must be a better way
226	// to convert UTF-8 to UTF-16 with an enumerator.
227	[MethodImpl(MethodImplOptions.AggressiveInlining)]
228	public bool MoveNext()
229	{
230	var (offset, length, nextByteIdx, currentCharPair) =	1,420✔
231	(_offset, _length, _nextByteIdx, _currentCharPair);	1,420✔
232
233	if (currentCharPair < char.MaxValue)	1,420✔
234	{
235	if ((uint)nextByteIdx < (uint)length)	1,302✔
236	{
237	var span = _value!.SliceUnsafe(offset + nextByteIdx, length - nextByteIdx);	1,284✔
238	var firstByte = MemoryMarshal.GetReference(span);	1,284✔
239	if (U8Info.IsAsciiByte(firstByte))	1,284✔
240	{
241	// Fast path because Rune.DecodeFromUtf8 won't inline
242	// making UTF-8 push us more and more towards anglocentrism.
243	_nextByteIdx = nextByteIdx + 1;	314✔
244	_currentCharPair = firstByte;	314✔
245	return true;	314✔
246	}
247
248	var status = Rune.DecodeFromUtf8(span, out var rune, out var bytesConsumed);	970✔
249	Debug.Assert(status is OperationStatus.Done);
250
251	_nextByteIdx = nextByteIdx + bytesConsumed;	970✔
252
253	if (rune.IsBmp)	970✔
254	{
255	_currentCharPair = (uint)rune.Value;	852✔
256	return true;	852✔
257	}
258
259	// I wonder if this just explodes on BigEndian
260	var runeValue = (uint)rune.Value;	118✔
261	var highSurrogate = (char)((runeValue + ((0xD800u - 0x40u) << 10)) >> 10);	118✔
262	var lowSurrogate = (char)((runeValue & 0x3FFu) + 0xDC00u);	118✔
263	_currentCharPair = highSurrogate + ((uint)lowSurrogate << 16);	118✔
264	return true;	118✔
265	}
266
267	return false;	18✔
268	}
269
270	_currentCharPair = currentCharPair >> 16;	118✔
271	return true;	118✔
272	}
273
274	[MethodImpl(MethodImplOptions.AggressiveInlining)]
275	public void Reset() => _nextByteIdx = 0;	×
276
277	readonly object IEnumerator.Current => Current;	701✔
278	readonly void IDisposable.Dispose() { }	6✔
279	}
280
281	readonly bool ICollection<char>.IsReadOnly => true;	×
282	readonly void ICollection<char>.Add(char item) => throw new NotSupportedException();	×
283	readonly void ICollection<char>.Clear() => throw new NotSupportedException();	×
284	readonly bool ICollection<char>.Remove(char item) => throw new NotSupportedException();	×
285	}
286
287	/// <summary>
288	/// A collection of Runes (unicode scalar values) in a provided <see cref="U8String"/>.
289	/// </summary>
290	public struct U8Runes : ICollection<Rune>, IEnumerable<Rune, U8Runes.Enumerator>
291	{
292	readonly U8String _value;
293
294	// If we bring up non-ascii counting to ascii level, we might not need this
295	// similar to LineCollection.
296	int _count;
297
298	[MethodImpl(MethodImplOptions.AggressiveInlining)]
299	public U8Runes(U8String value)
300	{
301	_value = value;	30✔
302	_count = value.IsEmpty ? 0 : -1;	30✔
303	}	30✔
304
305	/// <summary>
306	/// The number of Runes (unicode scalar values) in the current <see cref="U8String"/>.
307	/// </summary>
308	public int Count
309	{
310	[MethodImpl(MethodImplOptions.AggressiveInlining)]
311	get
312	{
313	// Somehow the codegen here is underwhelming
314	var count = _count;	24✔
315	if (count >= 0)	24✔
316	{
317	return count;	9✔
318	}
319
320	return _count = Count(_value.UnsafeSpan);	15✔
321
322	static int Count(ReadOnlySpan<byte> value)
323	{
324	Debug.Assert(!value.IsEmpty);
325
326	// TODO: SIMD non-continuation byte counting
327	var runeCount = (int)(nint)Polyfills.Text.Ascii.GetIndexOfFirstNonAsciiByte(value);	15✔
328	value = value.SliceUnsafe(runeCount);	15✔
329
330	for (var i = 0; (uint)i < (uint)value.Length; i += U8Info.RuneLength(value.AsRef(i)))	2,940✔
331	{
332	runeCount++;	1,455✔
333	}
334
335	return runeCount;	15✔
336	}
337	}
338	}
339
340	public readonly bool Contains(Rune item) => _value.Contains(item);	642✔
341
342	public readonly void CopyTo(Rune[] destination, int index)
343	{
344	// TODO: Simple SIMD widen ASCII to UTF-32 (ideally widen+validate in place instead of double traversal)
345	// TODO: Consistency and correctness? Implement single-pass vectorized conversion?
346	foreach (var rune in this)	×
347	{
348	destination[index++] = rune;	×
349	}
350	}	×
351
352	public readonly void Deconstruct(out Rune first, out Rune second)
353	{
354	this.Deconstruct<U8Runes, Enumerator, Rune>(out first, out second);	×
355	}	×
356
357	public readonly void Deconstruct(out Rune first, out Rune second, out Rune third)
358	{
359	this.Deconstruct<U8Runes, Enumerator, Rune>(out first, out second, out third);	×
360	}	×
361
362	public Rune[] ToArray() => this.ToArray<U8Runes, Enumerator, Rune>();	6✔
363
364	public List<Rune> ToList() => this.ToList<U8Runes, Enumerator, Rune>();	×
365
366	[MethodImpl(MethodImplOptions.AggressiveInlining)]
367	public readonly Enumerator GetEnumerator() => new(_value);	659✔
368
369	readonly IEnumerator<Rune> IEnumerable<Rune>.GetEnumerator() => GetEnumerator();	642✔
370	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	6✔
371
372	public struct Enumerator : IEnumerator<Rune>
373	{
374	readonly byte[]? _value;
375	readonly int _offset;
376	readonly int _length;
377	int _index;
378
379	[MethodImpl(MethodImplOptions.AggressiveInlining)]
380	public Enumerator(U8String value)
381	{
382	_value = value._value;	659✔
383	_offset = value.Offset;	659✔
384	_length = value.Length;	659✔
385	}	659✔
386
387	public Rune Current { get; private set; }
388
389	[MethodImpl(MethodImplOptions.AggressiveInlining)]
390	public bool MoveNext()
391	{
392	var index = _index;	55,597✔
393	if (index < _length)	55,597✔
394	{
395	ref var ptr = ref MemoryMarshal	55,574✔
396	.GetArrayDataReference(_value!)	55,574✔
397	.Add(_offset + index);	55,574✔
398
399	Current = U8Conversions.CodepointToRune(ref ptr, out var size);	55,574✔
400	_index = index + size;	55,574✔
401	return true;	55,574✔
402	}
403
404	return false;	23✔
405	}
406
407	[MethodImpl(MethodImplOptions.AggressiveInlining)]
408	public void Reset() => _index = -1;	×
409
410	readonly object IEnumerator.Current => Current;	642✔
411	readonly void IDisposable.Dispose() { }	653✔
412	}
413
414	readonly bool ICollection<Rune>.IsReadOnly => true;	×
415	readonly void ICollection<Rune>.Add(Rune item) => throw new NotImplementedException();	×
416	readonly void ICollection<Rune>.Clear() => throw new NotImplementedException();	×
417	readonly bool ICollection<Rune>.Remove(Rune item) => throw new NotImplementedException();	×
418	}
419
420	/// <summary>
421	/// A collection of lines in a provided <see cref="U8String"/>.
422	/// </summary>
423	public struct U8Lines : ICollection<U8String>, IU8Enumerable<U8Lines.Enumerator>
424	{
425	readonly U8String _value;
426
427	// We might not need this. Although counting is O(n), the absolute performance
428	// is very good, and on AVX2/512 - it's basically instantenous.
429	int _count;
430
431	/// <summary>
432	/// Creates a new line enumeration over the provided string.
433	/// </summary>
434	/// <param name="value">The string to enumerate over.</param>
435	[MethodImpl(MethodImplOptions.AggressiveInlining)]
436	public U8Lines(U8String value)
437	{
438	_value = value;	×
439	_count = value.IsEmpty ? 0 : -1;	×
440	}	×
441
442	/// <summary>
443	/// The number of lines in the current <see cref="U8String"/>.
444	/// </summary>
445	public int Count
446	{
447	[MethodImpl(MethodImplOptions.AggressiveInlining)]
448	get
449	{
450	var count = _count;	×
451	if (count >= 0)	×
452	{
453	return count;	×
454	}
455
456	// Matches the behavior of string.Split('\n').Length for "hello\n"
457	// TODO: Should we break consistency and not count the very last segment if it is empty?
458	// (likely no - an empty line is still a line)
459	return _count = _value.UnsafeSpan.Count((byte)'\n') + 1;	×
460	}
461	}
462
463	public readonly bool Contains(U8String item)
464	{
465	return !item.Contains((byte)'\n') && _value.Contains(item);	×
466	}
467
468	public void CopyTo(U8String[] destination, int index)
469	{
470	this.CopyTo<U8Lines, Enumerator, U8String>(destination.AsSpan()[index..]);	×
471	}	×
472
473	[MethodImpl(MethodImplOptions.AggressiveInlining)]
474	public readonly void Deconstruct(out U8String first, out U8String second)
475	{
476	this.Deconstruct<U8Lines, Enumerator, U8String>(out first, out second);	×
477	}	×
478
479	[MethodImpl(MethodImplOptions.AggressiveInlining)]
480	public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
481	{
482	this.Deconstruct<U8Lines, Enumerator, U8String>(out first, out second, out third);	×
483	}	×
484
485	public U8String[] ToArray() => this.ToArray<U8Lines, Enumerator, U8String>();	×
486	public List<U8String> ToList() => this.ToList<U8Lines, Enumerator, U8String>();	×
487
488	/// <summary>
489	/// Returns a <see cref="Enumerator"/> over the provided string.
490	/// </summary>
491	[MethodImpl(MethodImplOptions.AggressiveInlining)]
492	public readonly Enumerator GetEnumerator() => new(_value);	×
493
494	readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();	×
495	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	×
496
497	readonly bool ICollection<U8String>.IsReadOnly => true;	×
498	readonly void ICollection<U8String>.Add(U8String item) => throw new NotSupportedException();	×
499	readonly void ICollection<U8String>.Clear() => throw new NotSupportedException();	×
500	readonly bool ICollection<U8String>.Remove(U8String item) => throw new NotSupportedException();	×
501
502	/// <summary>
503	/// A struct that enumerates lines over a string.
504	/// </summary>
505	public struct Enumerator : IU8Enumerator
506	{
507	// TODO 1: Ensure this is aligned with Rust's .lines() implementation, or not?
508	// private static readonly SearchValues<byte> NewLine = SearchValues.Create("\r\n"u8);
509	// TODO 2: Consider using 'InnerOffsets'
510	private readonly byte[]? _value;
511	private U8Range _remaining;
512	private U8Range _current;
513
514	/// <summary>
515	/// Creates a new line enumerator over the provided string.
516	/// </summary>
517	/// <param name="value">The string to enumerate over.</param>
518	public Enumerator(U8String value)
519	{
520	_value = value._value;	×
521	_remaining = value._inner;	×
522	}	×
523
524	/// <summary>
525	/// Returns the current line.
526	/// </summary>
527	public readonly U8String Current => new(_value, _current.Offset, _current.Length);	×
528
529	/// <summary>
530	/// Advances the enumerator to the next line.
531	/// </summary>
532	[MethodImpl(MethodImplOptions.AggressiveInlining)] // Surprisingly smaller codegen than when not inlined
533	public bool MoveNext()
534	{
535	var remaining = _remaining;	×
536	if (remaining.Length > 0)	×
537	{
538	var span = _value!.SliceUnsafe(remaining.Offset, remaining.Length);	×
539	var idx = span.IndexOf((byte)'\n');	×
540
541	if ((uint)idx < (uint)span.Length)	×
542	{
543	var cutoff = idx;	×
544	if (idx > 0 && span.AsRef().Add(idx - 1) is (byte)'\r')	×
545	{
546	cutoff--;	×
547	}
548
549	_current = new(remaining.Offset, cutoff);	×
550	_remaining = new(remaining.Offset + idx + 1, remaining.Length - idx - 1);	×
551	}
552	else
553	{
554	// We've reached EOF, but we still need to return 'true' for this final
555	// iteration so that the caller can query the Current property once more.
556	_current = new(remaining.Offset, remaining.Length);	×
557	_remaining = default;	×
558	}
559
560	return true;	×
561	}
562
563	return false;	×
564	}
565
566	readonly object IEnumerator.Current => Current;	×
567	readonly void IEnumerator.Reset() => throw new NotSupportedException();	×
568	readonly void IDisposable.Dispose() { }	×
569	}
570	}

neon-sunset / U8String / 5912240727

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