I like your "only idea" of just doing a static character width map! It actually works well for my purposes. Sometimes, for performance reasons or because you don't have easy access to a DOM, you may just want a quick hacky standalone calculator calibrated to a single font. So here's one calibrated to Helvetica; pass a string and a font size:
const widths = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2796875,0.2765625,0.3546875,0.5546875,0.5546875,0.8890625,0.665625,0.190625,0.3328125,0.3328125,0.3890625,0.5828125,0.2765625,0.3328125,0.2765625,0.3015625,0.5546875,0.5546875,0.5546875,0.5546875,0.5546875,0.5546875,0.5546875,0.5546875,0.5546875,0.5546875,0.2765625,0.2765625,0.584375,0.5828125,0.584375,0.5546875,1.0140625,0.665625,0.665625,0.721875,0.721875,0.665625,0.609375,0.7765625,0.721875,0.2765625,0.5,0.665625,0.5546875,0.8328125,0.721875,0.7765625,0.665625,0.7765625,0.721875,0.665625,0.609375,0.721875,0.665625,0.94375,0.665625,0.665625,0.609375,0.2765625,0.3546875,0.2765625,0.4765625,0.5546875,0.3328125,0.5546875,0.5546875,0.5,0.5546875,0.5546875,0.2765625,0.5546875,0.5546875,0.221875,0.240625,0.5,0.221875,0.8328125,0.5546875,0.5546875,0.5546875,0.5546875,0.3328125,0.5,0.2765625,0.5546875,0.5,0.721875,0.5,0.5,0.5,0.3546875,0.259375,0.353125,0.5890625]
const avg = 0.5279276315789471
function measureText(str, fontSize) {
return Array.from(str).reduce(
(acc, cur) => acc + (widths[cur.charCodeAt(0)] ?? avg), 0
) * fontSize
}
That giant ugly array is ASCII character widths indexed by character code. So this just supports ASCII (otherwise it assumes an average character width). Fortunately, width basically scales linearly with font size, so it works pretty well at any font size. It's noticeably lacking any awareness of kerning or ligatures or whatever.
To "calibrate" I just rendered every character up to charCode 126 (the mighty tilde) on an svg and got the bounding box and saved it to this array; more code and explanation and demo here.