Meziantou.Framework

Bidirectional text

edit

Arabic and Hebrew are written from right to left, and their numbers and the Latin words inside them from left to right. A text is stored in the order it is read; displaying it needs the Unicode Bidirectional Algorithm (UAX #9), which UnicodeBidi implements in the two steps the annex defines: the levels of a paragraph, then the display order of each line.

Order a line for display

var paragraph = "car \u05D0\u05D1\u05D2!";
Span<byte> levels = stackalloc byte[paragraph.Length];
var paragraphLevel = UnicodeBidi.ResolveLevels(paragraph, UnicodeBidiDirection.Auto, levels);

// For each line, once the paragraph is broken into lines:
UnicodeBidi.ResetLineLevels(paragraph, paragraphLevel, levels);
Span<int> visualToLogical = stackalloc int[paragraph.Length];
UnicodeBidi.GetVisualOrder(paragraph, levels, visualToLogical);

Assert.Equal<int>([0, 1, 2, 3, 6, 5, 4, 7], visualToLogical);
  1. ResolveLevels gives the embedding level of each UTF-16 code unit of a paragraph (the two code units of a surrogate pair have the same level), and returns the level of the paragraph. A character whose level is odd is displayed from right to left. UnicodeBidiDirection.Auto takes the direction of the paragraph from its first strong character, left to right when it has none; LeftToRight and RightToLeft set it.
  2. Break the paragraph into lines (see Line breaking), then call ResetLineLevels for each line: it resets the levels of the separators and of the whitespace at the end of the line (rule L1).
  3. GetVisualOrder gives, for each position on the screen from the left, the index of the character to display there.

Details

  • UnicodeBidi.MayRequireBidi is a fast check: when it returns false, every level is 0 and the display order is the order of the text. Skip the algorithm for such text.
  • GetParagraphLevel gives the level of a paragraph without resolving the levels of its characters.
  • ResolveLevels has an overload that takes its working storage from you (GetWorkspaceLength gives its length) and does not allocate.
  • The algorithm orders characters; it does not mirror them. A right-to-left character such as a parenthesis is displayed mirrored: UnicodeProperties.TryGetBidiMirroringGlyph gives its mirrored character (rule L4).
  • A text renderer also needs shaping: the letters of Arabic change their form with their neighbors. To draw text on images, the text layout of Meziantou.Framework.Imaging does all of it.