Harfbuzz is a text shaping engine. It solves the problem of selecting and positioning glyphs from a font given a Unicode string.
Text shaping is an integral part of preparing text for display. It is a fairly low level operation; Harfbuzz is used directly by graphic rendering libraries such as Pango, and the layout engines in Firefox, LibreOffice and Chromium. Unless you are writing one of these layout engines yourself, you will probably not need to use Harfbuzz - normally higher level libraries will turn text into glyphs for you.
However, if you are writing a layout engine or graphics library yourself, you will need to perform text shaping, and this is where Harfbuzz can help you. Here are some reasons why you need it:
OpenType fonts contain a set of glyphs, indexed by glyph ID. The glyph ID within the font does not necessarily relate to a Unicode codepoint. For instance, some fonts have the letter "a" as glyph ID 1. To pull the right glyph out of the font in order to display it, you need to consult a table within the font (the "cmap" table) which maps Unicode codepoints to glyph IDs. Text shaping turns codepoints into glyph IDs.
Many OpenType fonts contain ligatures: combinations of characters which are rendered together. For instance, it's common for the
ficombination to appear in print as the single ligature "ﬁ". Whether you should render text as
fior "ﬁ" does not depend on the input text, but on the capabilities of the font and the level of ligature application you wish to perform. Text shaping involves querying the font's ligature tables and determining what substitutions should be made.
While ligatures like "ﬁ" are typographic refinements, some languages require such substitutions to be made in order to display text correctly. In Tamil, when the letter "TTA" (ட) letter is followed by "U" (உ), the combination should appear as the single glyph "டு". The sequence of Unicode characters "டஉ" needs to be rendered as a single glyph from the font - text shaping chooses the correct glyph from the sequence of characters provided.
Similarly, each Arabic character has four different variants: within a font, there will be glyphs for the initial, medial, final, and isolated forms of each letter. Unicode only encodes one codepoint per character, and so a Unicode string will not tell you which glyph to use. Text shaping chooses the correct form of the letter and returns the correct glyph from the font that you need to render.
Other languages have marks and accents which need to be rendered in certain positions around a base character. For instance, the Moldovan language has the Cyrillic letter "zhe" (ж) with a breve accent, like so: ӂ. Some fonts will contain this character as an individual glyph, whereas other fonts will not contain a zhe-with-breve glyph but expect the rendering engine to form the character by overlaying the two glyphs ж and ˘. Where you should draw the combining breve depends on the height of the preceding glyph. Again, for Arabic, the correct positioning of vowel marks depends on the height of the character on which you are placing the mark. Text shaping tells you whether you have a precomposed glyph within your font or if you need to compose a glyph yourself out of combining marks, and if so, where to position those marks.
If this is something that you need to do, then you need a text shaping engine: you could use Uniscribe if you are using Windows; you could use CoreText on OS X; or you could use Harfbuzz. In the rest of this manual, we are going to assume that you are the implementor of a text layout engine.