5868414274

pending completion

Build # 5868414274

Build Type

push

github

Committed by

neon-sunset

Commit Message

feat: consolidate CopyTo/ToArray/List logic, naming and Rune methods

Run Details

105 of 704 branches covered (14.91%)

Branch coverage included in aggregate %.

46 of 46 new or added lines in 6 files covered. (100.0%)

372 of 1547 relevant lines covered (24.05%)

460.03 hits per line

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

61.88

/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>
{
    readonly U8String _value;

    int _count;

    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    public U8Chars(U8String value)
    {
        if (!value.IsEmpty)
        {
            _value = value;
            _count = -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 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)
        {
            if (!value.IsEmpty)
            {
                _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>
{
    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)
    {
        if (!value.IsEmpty)
        {
            _value = value;
            _count = -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.CharLength(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;
        }
    }

    [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)
        {
            if (!value.IsEmpty)
            {
                _value = value._value;
                _offset = value.Offset;
                _length = value.Length;
            }
        }

        public Rune Current { get; private set; }

        [MethodImpl(MethodImplOptions.AggressiveInlining)]
        public bool MoveNext()
        {
            // TODO: Optimize for codegen, this one isn't great
            var index = _index;
            if (index < _length)
            {
                Rune.DecodeFromUtf8(
                    _value!.SliceUnsafe(_offset + index, _length - index),
                    out var rune,
                    out var consumed);

                Current = rune;
                _index = index + consumed;
                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)
    {
        if (!value.IsEmpty)
        {
            _value = value;
            _count = -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>(destination.AsSpan()[index..]);
    }

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

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

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

    /// <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)
        {
            if (!value.IsEmpty)
            {
                _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>
106	{
107	readonly U8String _value;
108
109	int _count;
110
111	[MethodImpl(MethodImplOptions.AggressiveInlining)]
112	public U8Chars(U8String value)
113	{
114	if (!value.IsEmpty)	24✔
115	{
116	_value = value;	20✔
117	_count = -1;	20✔
118	}
119	}	24✔
120
121	/// <summary>
122	/// The number of chars in the current <see cref="U8String"/>.
123	/// </summary>
124	public int Count
125	{
126	[MethodImpl(MethodImplOptions.AggressiveInlining)]
127	get
128	{
129	// Somehow the codegen here is underwhelming
130	var count = _count;	24✔
131	if (count >= 0)	24✔
132	{
133	return count;	9✔
134	}
135	return _count = Count(_value.UnsafeSpan);	15✔
136
137	static int Count(ReadOnlySpan<byte> value)
138	{
139	Debug.Assert(!value.IsEmpty);
140
141	// TODO: Is this enough?
142	return Encoding.UTF8.GetCharCount(value);	15✔
143	}
144	}
145	}
146
147	// TODO: Wow, this seems to be terribly broken on surrogate chars and
148	// there is no easy way to fix it without sacrificing performance.
149	// Perhaps it is worth just do the transcoding iteration here and warn the users
150	// instead of straight up producing UB or throwing exceptions???
151	public readonly bool Contains(char item) => _value.Contains(item);	×
152
153	public readonly void CopyTo(char[] destination, int index)
154	{
155	var value = _value;	5✔
156	if (!value.IsEmpty)	5✔
157	{
158	Encoding.UTF8.GetChars(value.UnsafeSpan, destination.AsSpan(index));	5✔
159	}
160	}	5✔
161
162	public readonly Enumerator GetEnumerator() => new(_value);	6✔
163
164	readonly IEnumerator<char> IEnumerable<char>.GetEnumerator() => new Enumerator(_value);	×
165	readonly IEnumerator IEnumerable.GetEnumerator() => new Enumerator(_value);	6✔
166
167	public struct Enumerator : IEnumerator<char>
168	{
169	// TODO: refactor layout
170	readonly byte[]? _value;
171	readonly int _offset;
172	readonly int _length;
173	int _nextByteIdx;
174	uint _currentCharPair;
175
176	[MethodImpl(MethodImplOptions.AggressiveInlining)]
177	public Enumerator(U8String value)
178	{
179	if (!value.IsEmpty)	12✔
180	{
181	_value = value._value;	10✔
182	_offset = value.Offset;	10✔
183	_length = value.Length;	10✔
184	_nextByteIdx = 0;	10✔
185	}
186	}	12✔
187
188	// TODO
189	public readonly char Current => (char)_currentCharPair;	1,402✔
190
191	// TODO: This looks terrible, there must be a better way
192	// to convert UTF-8 to UTF-16 with an enumerator.
193	[MethodImpl(MethodImplOptions.AggressiveInlining)]
194	public bool MoveNext()
195	{
196	var (offset, length, nextByteIdx, currentCharPair) =	1,420✔
197	(_offset, _length, _nextByteIdx, _currentCharPair);	1,420✔
198
199	if (currentCharPair < char.MaxValue)	1,420✔
200	{
201	if ((uint)nextByteIdx < (uint)length)	1,302✔
202	{
203	var span = _value!.SliceUnsafe(offset + nextByteIdx, length - nextByteIdx);	1,284✔
204	var firstByte = MemoryMarshal.GetReference(span);	1,284✔
205	if (U8Info.IsAsciiByte(firstByte))	1,284✔
206	{
207	// Fast path because Rune.DecodeFromUtf8 won't inline
208	// making UTF-8 push us more and more towards anglocentrism.
209	_nextByteIdx = nextByteIdx + 1;	314✔
210	_currentCharPair = firstByte;	314✔
211	return true;	314✔
212	}
213
214	var status = Rune.DecodeFromUtf8(span, out var rune, out var bytesConsumed);	970✔
215	Debug.Assert(status is OperationStatus.Done);
216
217	_nextByteIdx = nextByteIdx + bytesConsumed;	970✔
218
219	if (rune.IsBmp)	970✔
220	{
221	_currentCharPair = (uint)rune.Value;	852✔
222	return true;	852✔
223	}
224
225	// I wonder if this just explodes on BigEndian
226	var runeValue = (uint)rune.Value;	118✔
227	var highSurrogate = (char)((runeValue + ((0xD800u - 0x40u) << 10)) >> 10);	118✔
228	var lowSurrogate = (char)((runeValue & 0x3FFu) + 0xDC00u);	118✔
229	_currentCharPair = highSurrogate + ((uint)lowSurrogate << 16);	118✔
230	return true;	118✔
231	}
232
233	return false;	18✔
234	}
235
236	_currentCharPair = currentCharPair >> 16;	118✔
237	return true;	118✔
238	}
239
240	[MethodImpl(MethodImplOptions.AggressiveInlining)]
241	public void Reset() => _nextByteIdx = 0;	×
242
243	readonly object IEnumerator.Current => Current;	701✔
244	readonly void IDisposable.Dispose() { }	6✔
245	}
246
247	readonly bool ICollection<char>.IsReadOnly => true;	×
248	readonly void ICollection<char>.Add(char item) => throw new NotSupportedException();	×
249	readonly void ICollection<char>.Clear() => throw new NotSupportedException();	×
250	readonly bool ICollection<char>.Remove(char item) => throw new NotSupportedException();	×
251	}
252
253	/// <summary>
254	/// A collection of Runes (unicode scalar values) in a provided <see cref="U8String"/>.
255	/// </summary>
256	public struct U8Runes : ICollection<Rune>
257	{
258	readonly U8String _value;
259
260	// If we bring up non-ascii counting to ascii level, we might not need this
261	// similar to LineCollection.
262	int _count;
263
264	[MethodImpl(MethodImplOptions.AggressiveInlining)]
265	public U8Runes(U8String value)
266	{
267	if (!value.IsEmpty)	30✔
268	{
269	_value = value;	25✔
270	_count = -1;	25✔
271	}
272	}	30✔
273
274	/// <summary>
275	/// The number of Runes (unicode scalar values) in the current <see cref="U8String"/>.
276	/// </summary>
277	public int Count
278	{
279	[MethodImpl(MethodImplOptions.AggressiveInlining)]
280	get
281	{
282	// Somehow the codegen here is underwhelming
283	var count = _count;	24✔
284	if (count >= 0)	24✔
285	{
286	return count;	9✔
287	}
288
289	return _count = Count(_value.UnsafeSpan);	15✔
290
291	static int Count(ReadOnlySpan<byte> value)
292	{
293	Debug.Assert(!value.IsEmpty);
294
295	// TODO: SIMD non-continuation byte counting
296	var runeCount = (int)(nint)Polyfills.Text.Ascii.GetIndexOfFirstNonAsciiByte(value);	15✔
297	value = value.SliceUnsafe(runeCount);	15✔
298
299	for (var i = 0; (uint)i < (uint)value.Length; i += U8Info.CharLength(value.AsRef(i)))	2,940✔
300	{
301	runeCount++;	1,455✔
302	}
303
304	return runeCount;	15✔
305	}
306	}
307	}
308
309	public readonly bool Contains(Rune item) => _value.Contains(item);	642✔
310
311	public readonly void CopyTo(Rune[] destination, int index)
312	{
313	// TODO: Simple SIMD widen ASCII to UTF-32 (ideally widen+validate in place instead of double traversal)
314	// TODO: Consistency and correctness? Implement single-pass vectorized conversion?
315	foreach (var rune in this)	1,294✔
316	{
317	destination[index++] = rune;	642✔
318	}
319	}	5✔
320
321	[MethodImpl(MethodImplOptions.AggressiveInlining)]
322	public readonly Enumerator GetEnumerator() => new(_value);	659✔
323
324	readonly IEnumerator<Rune> IEnumerable<Rune>.GetEnumerator() => GetEnumerator();	642✔
325	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	6✔
326
327	public struct Enumerator : IEnumerator<Rune>
328	{
329	readonly byte[]? _value;
330	readonly int _offset;
331	readonly int _length;
332	int _index;
333
334	[MethodImpl(MethodImplOptions.AggressiveInlining)]
335	public Enumerator(U8String value)
336	{
337	if (!value.IsEmpty)	659✔
338	{
339	_value = value._value;	657✔
340	_offset = value.Offset;	657✔
341	_length = value.Length;	657✔
342	}
343	}	659✔
344
345	public Rune Current { get; private set; }
346
347	[MethodImpl(MethodImplOptions.AggressiveInlining)]
348	public bool MoveNext()
349	{
350	// TODO: Optimize for codegen, this one isn't great
351	var index = _index;	55,597✔
352	if (index < _length)	55,597✔
353	{
354	Rune.DecodeFromUtf8(	55,574✔
355	_value!.SliceUnsafe(_offset + index, _length - index),	55,574✔
356	out var rune,	55,574✔
357	out var consumed);	55,574✔
358
359	Current = rune;	55,574✔
360	_index = index + consumed;	55,574✔
361	return true;	55,574✔
362	}
363
364	return false;	23✔
365	}
366
367	[MethodImpl(MethodImplOptions.AggressiveInlining)]
368	public void Reset() => _index = -1;	×
369
370	readonly object IEnumerator.Current => Current;	642✔
371	readonly void IDisposable.Dispose() { }	653✔
372	}
373
374	readonly bool ICollection<Rune>.IsReadOnly => true;	×
375	readonly void ICollection<Rune>.Add(Rune item) => throw new NotImplementedException();	×
376	readonly void ICollection<Rune>.Clear() => throw new NotImplementedException();	×
377	readonly bool ICollection<Rune>.Remove(Rune item) => throw new NotImplementedException();	×
378	}
379
380	/// <summary>
381	/// A collection of lines in a provided <see cref="U8String"/>.
382	/// </summary>
383	public struct U8Lines : ICollection<U8String>, IU8Enumerable<U8Lines.Enumerator>
384	{
385	readonly U8String _value;
386
387	// We might not need this. Although counting is O(n), the absolute performance
388	// is very good, and on AVX2/512 - it's basically instantenous.
389	int _count;
390
391	/// <summary>
392	/// Creates a new line enumeration over the provided string.
393	/// </summary>
394	/// <param name="value">The string to enumerate over.</param>
395	[MethodImpl(MethodImplOptions.AggressiveInlining)]
396	public U8Lines(U8String value)
397	{
398	if (!value.IsEmpty)	×
399	{
400	_value = value;	×
401	_count = -1;	×
402	}
403	}	×
404
405	/// <summary>
406	/// The number of lines in the current <see cref="U8String"/>.
407	/// </summary>
408	public int Count
409	{
410	[MethodImpl(MethodImplOptions.AggressiveInlining)]
411	get
412	{
413	var count = _count;	×
414	if (count >= 0)	×
415	{
416	return count;	×
417	}
418
419	// Matches the behavior of string.Split('\n').Length for "hello\n"
420	// TODO: Should we break consistency and not count the very last segment if it is empty?
421	// (likely no - an empty line is still a line)
422	return _count = _value.UnsafeSpan.Count((byte)'\n') + 1;	×
423	}
424	}
425
426	public readonly bool Contains(U8String item)
427	{
428	return !item.Contains((byte)'\n') && _value.Contains(item);	×
429	}
430
431	public void CopyTo(U8String[] destination, int index)
432	{
433	this.CopyTo<U8Lines, Enumerator>(destination.AsSpan()[index..]);	×
434	}	×
435
436	[MethodImpl(MethodImplOptions.AggressiveInlining)]
437	public readonly void Deconstruct(out U8String first, out U8String second)
438	{
439	this.Deconstruct<U8Lines, Enumerator>(out first, out second);	×
440	}	×
441
442	[MethodImpl(MethodImplOptions.AggressiveInlining)]
443	public readonly void Deconstruct(out U8String first, out U8String second, out U8String third)
444	{
445	this.Deconstruct<U8Lines, Enumerator>(out first, out second, out third);	×
446	}	×
447
448	public U8String[] ToArray() => this.ToArray<U8Lines, Enumerator>();	×
449	public List<U8String> ToList() => this.ToList<U8Lines, Enumerator>();	×
450
451	/// <summary>
452	/// Returns a <see cref="Enumerator"/> over the provided string.
453	/// </summary>
454	[MethodImpl(MethodImplOptions.AggressiveInlining)]
455	public readonly Enumerator GetEnumerator() => new(_value);	×
456
457	readonly IEnumerator<U8String> IEnumerable<U8String>.GetEnumerator() => GetEnumerator();	×
458	readonly IEnumerator IEnumerable.GetEnumerator() => GetEnumerator();	×
459
460	readonly bool ICollection<U8String>.IsReadOnly => true;	×
461	readonly void ICollection<U8String>.Add(U8String item) => throw new NotSupportedException();	×
462	readonly void ICollection<U8String>.Clear() => throw new NotSupportedException();	×
463	readonly bool ICollection<U8String>.Remove(U8String item) => throw new NotSupportedException();	×
464
465	/// <summary>
466	/// A struct that enumerates lines over a string.
467	/// </summary>
468	public struct Enumerator : IU8Enumerator
469	{
470	// TODO 1: Ensure this is aligned with Rust's .lines() implementation, or not?
471	// private static readonly SearchValues<byte> NewLine = SearchValues.Create("\r\n"u8);
472	// TODO 2: Consider using 'InnerOffsets'
473	private readonly byte[]? _value;
474	private U8Range _remaining;
475	private U8Range _current;
476
477	/// <summary>
478	/// Creates a new line enumerator over the provided string.
479	/// </summary>
480	/// <param name="value">The string to enumerate over.</param>
481	public Enumerator(U8String value)
482	{
483	if (!value.IsEmpty)	×
484	{
485	_value = value._value;	×
486	_remaining = value._inner;	×
487	}
488	}	×
489
490	/// <summary>
491	/// Returns the current line.
492	/// </summary>
493	public readonly U8String Current => new(_value, _current.Offset, _current.Length);	×
494
495	/// <summary>
496	/// Advances the enumerator to the next line.
497	/// </summary>
498	[MethodImpl(MethodImplOptions.AggressiveInlining)] // Surprisingly smaller codegen than when not inlined
499	public bool MoveNext()
500	{
501	var remaining = _remaining;	×
502	if (remaining.Length > 0)	×
503	{
504	var span = _value!.SliceUnsafe(remaining.Offset, remaining.Length);	×
505	var idx = span.IndexOf((byte)'\n');	×
506
507	if ((uint)idx < (uint)span.Length)	×
508	{
509	var cutoff = idx;	×
510	if (idx > 0 && span.AsRef().Add(idx - 1) is (byte)'\r')	×
511	{
512	cutoff--;	×
513	}
514
515	_current = new(remaining.Offset, cutoff);	×
516	_remaining = new(remaining.Offset + idx + 1, remaining.Length - idx - 1);	×
517	}
518	else
519	{
520	// We've reached EOF, but we still need to return 'true' for this final
521	// iteration so that the caller can query the Current property once more.
522	_current = new(remaining.Offset, remaining.Length);	×
523	_remaining = default;	×
524	}
525
526	return true;	×
527	}
528
529	return false;	×
530	}
531
532	readonly object IEnumerator.Current => Current;	×
533	readonly void IEnumerator.Reset() => throw new NotSupportedException();	×
534	readonly void IDisposable.Dispose() { }	×
535	}
536	}

neon-sunset / U8String / 5868414274

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