Font

Glyph names

Use GlyphName::toUnicode() when a glyph name is available but a Unicode character mapping is not. It returns a UTF-8 string without loading a font.

<?php

use Alto\Font\Glyph\GlyphName;

GlyphName::toUnicode('eacute');       // "é"
GlyphName::toUnicode('A.swash');      // "A"
GlyphName::toUnicode('f_f_i');        // "ffi"
GlyphName::toUnicode('uni00410301');  // "A\u{0301}"
GlyphName::toUnicode('u1F600');       // "\u{1F600}"
GlyphName::toUnicode('.notdef');      // ""

The conversion follows the Adobe Glyph List specification, version 2.9:

  • Remove the first period and everything after it, then split at underscores.
  • Resolve each component using the complete Adobe Glyph List.
  • Interpret uni followed by groups of four uppercase hexadecimal digits as BMP scalars, or u followed by four to six uppercase hexadecimal digits as one Unicode scalar.
  • Omit unknown components, malformed hexadecimal names, surrogate values, and values above U+10FFFF. Concatenate the remaining components.

For example, A_unknown_B returns AB. A surrogate anywhere in one uni component discards that entire component: uni0041D800_B returns B. Lowercase hexadecimal digits are not accepted. A UTF-16 surrogate pair in a uni name is not a supplementary character; use a single u name instead.

The result is not normalized: fi maps to U+FB01, while f_i maps to two characters. Private-use scalars and U+0000 are retained. Treat the result as text data; escape it appropriately when displaying it in a terminal or HTML.

Zapf Dingbats

Enable the additional ITC Zapf Dingbats mapping only for the font whose PostScript name is ZapfDingbats:

<?php

GlyphName::toUnicode('a1', zapfDingbats: true); // "\u{2701}"
GlyphName::toUnicode('a1');                    // ""

In this mode the Dingbats list takes precedence, followed by the standard Adobe mapping and the same Unicode-name rules.

Scope and cost

This API interprets names you provide. To look up a character in a loaded font, use Font::glyphIdForCodepoint(); PDF encoding dictionaries and /ToUnicode maps belong to the PDF reader.

The two bundled mappings load once, on demand. Caller-provided names are never cached. Processing is iterative and linear in the name length, with memory proportional to the input and output; there is no arbitrary name length cutoff. Callers processing untrusted files should apply their own input-size limits.

The bundled data provenance records the pinned Adobe revision, license, checksums, and offline regeneration command.