ImageSharp/SixLabors.Fonts/Unicode/SpanWordEnumerator.cs

// Copyright (c) Six Labors.
// Licensed under the Six Labors Split License.

using System;

namespace SixLabors.Fonts.Unicode {
  /// <summary>
  /// An enumerator for retrieving word-boundary segments from a <see cref="ReadOnlySpan{Char}"/>.
  /// <br/>
  /// Implements the Unicode Word Boundary Algorithm. UAX #29
  /// <see href="https://www.unicode.org/reports/tr29/"/>
  /// <br/>
  /// Methods are pattern-matched by compiler to allow using foreach pattern.
  /// </summary>
  public ref struct SpanWordEnumerator
  {
      private ReadOnlySpan<char> source;
      private int sourceOffset;
      private int codePointOffset;

      /// <summary>
      /// Initializes a new instance of the <see cref="SpanWordEnumerator"/> struct.
      /// </summary>
      /// <param name="source">The buffer to read from.</param>
      public SpanWordEnumerator(ReadOnlySpan<char> source)
      {
          this.source = source;
          this.sourceOffset = 0;
          this.codePointOffset = 0;
          this.Current = default;
      }

      /// <summary>
      /// Gets the element in the collection at the current position of the enumerator.
      /// </summary>
      public WordSegment Current { get; private set; }

      /// <summary>
      /// Returns an enumerator that iterates through the collection.
      /// </summary>
      /// <returns>An enumerator that iterates through the collection.</returns>
      public readonly SpanWordEnumerator GetEnumerator() => this;

      /// <summary>
      /// Advances the enumerator to the next element of the collection.
      /// </summary>
      /// <returns>
      /// <see langword="true"/> if the enumerator was successfully advanced to the next element;
      /// <see langword="false"/> if the enumerator has passed the end of the collection.
      /// </returns>
      public bool MoveNext()
      {
          if (this.sourceOffset >= this.source.Length)
          {
              return false;
          }

          int segmentStart = this.sourceOffset;
          int segmentCodePointStart = this.codePointOffset;
          WordBreakCodePoint current = this.ReadForward(this.sourceOffset);
          int currentEnd = current.Utf16End;
          int boundaryCodePoint = this.codePointOffset + 1;

          while (currentEnd < this.source.Length)
          {
              WordBreakCodePoint next = this.ReadForward(currentEnd);
              if (this.IsBoundary(current, next))
              {
                  break;
              }

              current = next;
              currentEnd = current.Utf16End;
              boundaryCodePoint++;
          }

          this.Current = new WordSegment(
              this.source[segmentStart..currentEnd],
              segmentStart,
              segmentCodePointStart,
              boundaryCodePoint - segmentCodePointStart);

          this.sourceOffset = currentEnd;
          this.codePointOffset = boundaryCodePoint;
          return true;
      }

      private readonly bool IsBoundary(in WordBreakCodePoint current, in WordBreakCodePoint next)
      {
          // WB3, WB3a, WB3b, and WB3c are evaluated before the ignore rule, so they
          // use the adjacent code points exactly as they appear in the source.
          if (current.Is(WordBreakClass.CarriageReturn) && next.Is(WordBreakClass.LineFeed))
          {
              return false;
          }

          if (IsNewline(current) || IsNewline(next))
          {
              return true;
          }

          if (current.Is(WordBreakClass.ZeroWidthJoiner)
              && CodePoint.GetGraphemeClusterClass(next.CodePoint) == GraphemeClusterClass.ExtendedPictographic)
          {
              return false;
          }

          if (current.Is(WordBreakClass.WSegSpace) && next.Is(WordBreakClass.WSegSpace))
          {
              return false;
          }

          if (IsIgnored(next))
          {
              return false;
          }

          WordBreakCodePoint left = this.GetEffectivePrevious(current);
          WordBreakClass right = next.Class;

          if (IsAHLetter(left.Class) && IsAHLetter(right))
          {
              return false;
          }

          if (IsAHLetter(left.Class)
              && IsMidLetterMidNumLetQ(right)
              && this.TryGetNextSignificant(next.Utf16End, out WordBreakCodePoint after)
              && IsAHLetter(after.Class))
          {
              return false;
          }

          if (IsAHLetter(right)
              && IsMidLetterMidNumLetQ(left.Class)
              && this.TryGetPreviousSignificant(left.Utf16Start, out WordBreakCodePoint before)
              && IsAHLetter(before.Class))
          {
              return false;
          }

          if (left.Is(WordBreakClass.HebrewLetter) && right == WordBreakClass.SingleQuote)
          {
              return false;
          }

          if (left.Is(WordBreakClass.HebrewLetter)
              && right == WordBreakClass.DoubleQuote
              && this.TryGetNextSignificant(next.Utf16End, out after)
              && after.Is(WordBreakClass.HebrewLetter))
          {
              return false;
          }

          if (right == WordBreakClass.HebrewLetter
              && left.Is(WordBreakClass.DoubleQuote)
              && this.TryGetPreviousSignificant(left.Utf16Start, out before)
              && before.Is(WordBreakClass.HebrewLetter))
          {
              return false;
          }

          if (left.Is(WordBreakClass.Numeric) && right == WordBreakClass.Numeric)
          {
              return false;
          }

          if (IsAHLetter(left.Class) && right == WordBreakClass.Numeric)
          {
              return false;
          }

          if (left.Is(WordBreakClass.Numeric) && IsAHLetter(right))
          {
              return false;
          }

          if (right == WordBreakClass.Numeric
              && IsMidNumMidNumLetQ(left.Class)
              && this.TryGetPreviousSignificant(left.Utf16Start, out before)
              && before.Is(WordBreakClass.Numeric))
          {
              return false;
          }

          if (left.Is(WordBreakClass.Numeric)
              && IsMidNumMidNumLetQ(right)
              && this.TryGetNextSignificant(next.Utf16End, out after)
              && after.Is(WordBreakClass.Numeric))
          {
              return false;
          }

          if (left.Is(WordBreakClass.Katakana) && right == WordBreakClass.Katakana)
          {
              return false;
          }

          if (IsAHLetterNumericKatakanaExtendNumLet(left.Class) && right == WordBreakClass.ExtendNumLet)
          {
              return false;
          }

          if (left.Is(WordBreakClass.ExtendNumLet) && IsAHLetterNumericKatakana(right))
          {
              return false;
          }

          if (left.Is(WordBreakClass.RegionalIndicator)
              && right == WordBreakClass.RegionalIndicator
              && (this.CountRegionalIndicatorsBefore(next.Utf16Start) & 1) == 1)
          {
              return false;
          }

          return true;
      }

      private readonly WordBreakCodePoint GetEffectivePrevious(in WordBreakCodePoint current)
      {
          if (!IsIgnored(current))
          {
              return current;
          }

          int scanEnd = current.Utf16Start;
          while (this.TryReadBackward(scanEnd, out WordBreakCodePoint previous))
          {
              if (!IsIgnored(previous))
              {
                  // WB4 deliberately stops ignoring after sot and hard line breaks.
                  return IsNewline(previous) ? current : previous;
              }

              scanEnd = previous.Utf16Start;
          }

          return current;
      }

      private readonly int CountRegionalIndicatorsBefore(int utf16End)
      {
          int count = 0;
          int scanEnd = utf16End;
          while (this.TryGetPreviousSignificant(scanEnd, out WordBreakCodePoint previous))
          {
              if (!previous.Is(WordBreakClass.RegionalIndicator))
              {
                  break;
              }

              count++;
              scanEnd = previous.Utf16Start;
          }

          return count;
      }

      private readonly bool TryGetPreviousSignificant(int utf16End, out WordBreakCodePoint codePoint)
      {
          int scanEnd = utf16End;
          while (this.TryReadBackward(scanEnd, out codePoint))
          {
              if (!IsIgnored(codePoint))
              {
                  return true;
              }

              scanEnd = codePoint.Utf16Start;
          }

          codePoint = default;
          return false;
      }

      private readonly bool TryGetNextSignificant(int utf16Start, out WordBreakCodePoint codePoint)
      {
          int scanStart = utf16Start;
          while (this.TryReadForward(scanStart, out codePoint))
          {
              if (!IsIgnored(codePoint))
              {
                  return true;
              }

              scanStart = codePoint.Utf16End;
          }

          codePoint = default;
          return false;
      }

      private readonly WordBreakCodePoint ReadForward(int utf16Start)
      {
          CodePoint codePoint = CodePoint.DecodeFromUtf16At(this.source, utf16Start, out int charsConsumed);
          return new WordBreakCodePoint(codePoint, CodePoint.GetWordBreakClass(codePoint), utf16Start, utf16Start + charsConsumed);
      }

      private readonly bool TryReadForward(int utf16Start, out WordBreakCodePoint codePoint)
      {
          if (utf16Start >= this.source.Length)
          {
              codePoint = default;
              return false;
          }

          codePoint = this.ReadForward(utf16Start);
          return true;
      }

      private readonly bool TryReadBackward(int utf16End, out WordBreakCodePoint codePoint)
      {
          if (utf16End <= 0)
          {
              codePoint = default;
              return false;
          }

          int utf16Start = utf16End - 1;
          if (utf16Start > 0
              && char.IsLowSurrogate(this.source[utf16Start])
              && char.IsHighSurrogate(this.source[utf16Start - 1]))
          {
              utf16Start--;
          }

          codePoint = this.ReadForward(utf16Start);
          return true;
      }

      private static bool IsAHLetter(WordBreakClass cls)
          => cls is WordBreakClass.ALetter or WordBreakClass.HebrewLetter;

      private static bool IsAHLetterNumericKatakana(WordBreakClass cls)
          => IsAHLetter(cls) || cls is WordBreakClass.Numeric or WordBreakClass.Katakana;

      private static bool IsAHLetterNumericKatakanaExtendNumLet(WordBreakClass cls)
          => IsAHLetterNumericKatakana(cls) || cls == WordBreakClass.ExtendNumLet;

      private static bool IsIgnored(in WordBreakCodePoint codePoint) => IsIgnored(codePoint.Class);

      private static bool IsIgnored(WordBreakClass cls)
          => cls is WordBreakClass.Extend or WordBreakClass.Format or WordBreakClass.ZeroWidthJoiner;

      private static bool IsMidLetterMidNumLetQ(WordBreakClass cls)
          => cls is WordBreakClass.MidLetter or WordBreakClass.MidNumLet or WordBreakClass.SingleQuote;

      private static bool IsMidNumMidNumLetQ(WordBreakClass cls)
          => cls is WordBreakClass.MidNum or WordBreakClass.MidNumLet or WordBreakClass.SingleQuote;

      private static bool IsNewline(in WordBreakCodePoint codePoint)
          => codePoint.Class is WordBreakClass.CarriageReturn or WordBreakClass.LineFeed or WordBreakClass.Newline;

      private readonly struct WordBreakCodePoint
      {
          public WordBreakCodePoint(CodePoint codePoint, WordBreakClass cls, int utf16Start, int utf16End)
          {
              this.CodePoint = codePoint;
              this.Class = cls;
              this.Utf16Start = utf16Start;
              this.Utf16End = utf16End;
          }

          public CodePoint CodePoint { get; }

          public WordBreakClass Class { get; }

          public int Utf16Start { get; }

          public int Utf16End { get; }

          public bool Is(WordBreakClass cls) => this.Class == cls;
      }
  }
}