Open a face and map text

Add sparkles:font as a DUB dependency. Read the file yourself, by any means that keeps the bytes alive, and open a face over them:

import std.file : read;
import sparkles.font;

const bytes = cast(const(ubyte)[]) read("DejaVuSansMono.ttf");
auto opened = openFace(bytes);
if (opened.hasError)
    return; // opened.error says what is wrong, in which table and where
const face = opened.value;
assert(face.unitsPerEm == 2048);

The face borrows bytes: it holds a slice of them, and every slice it hands back points into them. Keep the buffer alive and unchanged while the face, or anything read from it, is in use. With -preview=dip1000 the compiler checks this in @safe code.

Map characters to glyphs

Build the character map once, then look codepoints up as often as you like:

const map = face.charMap.value;
const glyph = map.glyph('A'); // 0 when the face does not map it

foreach (range; map.ranges)
{
    // every codepoint in [range.first, range.last] maps to a glyph
}

charMap chooses a subtable the way HarfBuzz does and checks it; each lookup is then a binary search with no further checking.

Read names

name records decode into storage you provide, after a measuring call:

const names = face.name.value;
foreach (record; names.records)
{
    if (record.hasError || record.value.nameID != 4)
        continue;
    char[256] buffer;
    const text = record.value.decode(buffer[]);
    if (text.hasValue)
    {
        // text.value is a slice of buffer: the full font name
    }
}

A record in an encoding the library does not decode returns unsupportedCapability; its raw bytes stay available as record.value.bytes.

Name glyphs

const glyphName = face.glyphName(glyph);
if (glyphName.hasValue && glyphName.value.present)
{
    // glyphName.value.text, for example "A"
}

Names come from post, or from the CFF charset for a glyph post does not name. To name every glyph, build an index first; see the reference.