fix(producer): stop copying the whole document per character when scanning fonts

Studio preview re-runs font injection on every save; scanning a multi-MB
bundled document built a per-character array twice, about 0.3s of pure
allocation per request. Iterate the strings directly; the requested text
subset is unchanged.
This commit is contained in:
Miguel Angel Simon Sierra
2026-09-19 04:51:54 -04:00
parent 607775bb3a
commit 93b24ca6a7
2 changed files with 27 additions and 1 deletions
@@ -230,4 +230,26 @@ describe("Google Fonts text subsetting", () => {
expect(text).toContain("A");
});
it("scans a multi-MB document without building a per-character array", async () => {
const html = `<!doctype html><html><head><style>
p { font-family: "Noto Performance Test", sans-serif; }
</style></head><body><p>旅</p><i data-payload="${"xyz".repeat(400_000)}"></i></body></html>`;
const realFrom = Array.from;
let wholeDocumentCopies = 0;
Array.from = ((source: unknown, ...rest: unknown[]) => {
if (typeof source === "string" && source.length > 100_000) wholeDocumentCopies += 1;
return (realFrom as (...args: unknown[]) => unknown[])(source, ...rest);
}) as typeof Array.from;
let text: string;
try {
text = await subsetTextFor(html);
} finally {
Array.from = realFrom;
}
expect(wholeDocumentCopies).toBe(0);
expect(text.startsWith("<!")).toBe(true);
for (const character of "xyz旅") expect(text).toContain(character);
});
});
@@ -1410,8 +1410,12 @@ function extractGoogleFontsText(html: string): string | undefined {
// Intentional over-approximation: raw html includes base64, scripts, and
// class names, but they collapse in the Set and the budget gate catches bloat.
// Iterate strings directly: Array.from(html) allocates one element per code point (~0.3s per 5 MB).
const sourceCharacters = new Set<string>();
for (const character of html) sourceCharacters.add(character);
for (const character of decodedBodyText) sourceCharacters.add(character);
const uniqueCharacters = new Set<string>();
for (const character of new Set([...Array.from(html), ...Array.from(decodedBodyText)])) {
for (const character of sourceCharacters) {
addCaseClosure(uniqueCharacters, character, locales);
}