@react-x11/components 0.9.0 → 0.11.0

package/src/html/css/decode.ts

@@ -0,0 +1,130 @@
+// A stylesheet's bytes as text.
+//
+// A host that fetches a stylesheet has bytes and, often, the charset its
+// protocol named; a host that decodes them itself hands over text, and this
+// is not asked. For the first, CSS says which encoding the bytes are in
+// (CSS 2.1 4.4, CSS Syntax 3 3.2), in this order:
+//
+//   1. a byte order mark, which the Encoding standard lets override the rest;
+//   2. the charset the protocol gave, HTTP's `Content-Type` parameter;
+//   3. an `@charset "…";` rule at the very start, byte for byte — naming
+//      UTF-16 there means UTF-8, since a sheet that could spell the rule in
+//      ASCII bytes is not in UTF-16;
+//   4. what the referrer says: a `<link charset>`, then the encoding of the
+//      document, or of the stylesheet that imported this one;
+//   5. UTF-8.
+//
+// A name that names no encoding is passed over, as a step that says nothing.
+
+/** The `TextDecoder` slice this uses, through `globalThis`: `src/` compiles
+ *  with `types: []`, and a runtime without one gets UTF-8 alone. */
+type DecoderConstructor = new (label: string) => {
+  readonly encoding: string;
+  decode(input: Uint8Array): string;
+};
+
+function decoderFor(label: string): InstanceType<DecoderConstructor> | null {
+  const Decoder = (globalThis as { TextDecoder?: DecoderConstructor })
+    .TextDecoder;
+  if (!Decoder) return null;
+  try {
+    return new Decoder(label.trim());
+  } catch {
+    // an unknown label throws a RangeError, which here means "no answer"
+    return null;
+  }
+}
+
+/** `@charset "` in ASCII bytes, which the rule has to be spelt as exactly. */
+const CHARSET_RULE = [
+  0x40, 0x63, 0x68, 0x61, 0x72, 0x73, 0x65, 0x74, 0x20, 0x22,
+];
+
+/** The label an `@charset` rule at the very start names, or null. It counts
+ *  only as exactly `@charset "label";` within the first 1024 bytes. */
+function charsetRule(bytes: Uint8Array): string | null {
+  if (bytes.length < CHARSET_RULE.length + 2) return null;
+  for (let i = 0; i < CHARSET_RULE.length; i += 1) {
+    if (bytes[i] !== CHARSET_RULE[i]) return null;
+  }
+  const end = Math.min(bytes.length - 1, 1024);
+  let label = '';
+  for (let i = CHARSET_RULE.length; i < end; i += 1) {
+    if (bytes[i] === 0x22) return bytes[i + 1] === 0x3b ? label : null;
+    label += String.fromCharCode(bytes[i]);
+  }
+  return null;
+}
+
+export interface DecodedStylesheet {
+  text: string;
+  /** The encoding it was decoded with, canonical — which a stylesheet it
+   *  imports falls back to. */
+  encoding: string;
+}
+
+/**
+ * Decode a stylesheet's bytes. `charset` is what the protocol said, and
+ * `fallbacks` what the referrer does, most specific first.
+ */
+export function decodeStylesheet(
+  bytes: Uint8Array,
+  charset: string | undefined,
+  fallbacks: readonly (string | undefined)[],
+): DecodedStylesheet {
+  // a byte order mark wins, and is not part of the text
+  const bom =
+    bytes[0] === 0xef && bytes[1] === 0xbb && bytes[2] === 0xbf
+      ? 'utf-8'
+      : bytes[0] === 0xfe && bytes[1] === 0xff
+        ? 'utf-16be'
+        : bytes[0] === 0xff && bytes[1] === 0xfe
+          ? 'utf-16le'
+          : null;
+  const labels: (string | null | undefined)[] = [bom, charset];
+  const rule = charsetRule(bytes);
+  if (rule !== null) {
+    const named = decoderFor(rule)?.encoding;
+    labels.push(named === 'utf-16be' || named === 'utf-16le' ? 'utf-8' : rule);
+  }
+  labels.push(...fallbacks, 'utf-8');
+  for (const label of labels) {
+    if (!label) continue;
+    const decoder = decoderFor(label);
+    if (decoder)
+      return { text: decoder.decode(bytes), encoding: decoder.encoding };
+  }
+  return { text: utf8(bytes), encoding: 'utf-8' };
+}
+
+/** UTF-8 by hand, for a runtime with no `TextDecoder`. */
+function utf8(bytes: Uint8Array): string {
+  let out = '';
+  let i = bytes[0] === 0xef && bytes[1] === 0xbb && bytes[2] === 0xbf ? 3 : 0;
+  while (i < bytes.length) {
+    const b = bytes[i];
+    let code = 0xfffd;
+    let size = 1;
+    if (b < 0x80) code = b;
+    else if (b >= 0xc2 && b < 0xe0 && i + 1 < bytes.length) {
+      code = ((b & 0x1f) << 6) | (bytes[i + 1] & 0x3f);
+      size = 2;
+    } else if (b >= 0xe0 && b < 0xf0 && i + 2 < bytes.length) {
+      code =
+        ((b & 0x0f) << 12) |
+        ((bytes[i + 1] & 0x3f) << 6) |
+        (bytes[i + 2] & 0x3f);
+      size = 3;
+    } else if (b >= 0xf0 && b < 0xf5 && i + 3 < bytes.length) {
+      code =
+        ((b & 0x07) << 18) |
+        ((bytes[i + 1] & 0x3f) << 12) |
+        ((bytes[i + 2] & 0x3f) << 6) |
+        (bytes[i + 3] & 0x3f);
+      size = 4;
+    }
+    out += String.fromCodePoint(code);
+    i += size;
+  }
+  return out;
+}