ref(wcwidth): add datahash

Author: natemoo-re

src/wcwidth.c

@@ -1,5 +1,6 @@
 /* wcwidth.c - Unicode character width lookup
- * Unicode 16.0 - generated by tasks/gen-wcwidth.ts on 2026-05-28
+ * Unicode 16.0 - generated by tasks/gen-wcwidth.ts
+ * Data hash: 2397fe3ee973c18a
  *
  * Only zero-width (combining marks + default ignorable) and double-width
  * (wide/fullwidth) codepoints are stored. Everything else defaults to

tasks/gen-wcwidth.ts

@@ -181,6 +181,14 @@ function formatUint8Array(values: number[], indent = "  "): string {
   return lines.join("\n");
 }
 
+async function sha256Hex(input: string): Promise<string> {
+  let bytes = new TextEncoder().encode(input);
+  let digest = await crypto.subtle.digest("SHA-256", bytes);
+  return Array.from(new Uint8Array(digest))
+    .map((byte) => byte.toString(16).padStart(2, "0"))
+    .join("");
+}
+
 let [derivedCategoryText, derivedCorePropsText, eastAsianWidthText] =
   await Promise.all([
     fetchText("extracted/DerivedGeneralCategory.txt"),
@@ -307,11 +315,24 @@ console.error(`special_large_ranges: ${largeRanges.length} entries`);
 console.error(`BMP dirty blocks:     ${dirtyBlocks} / 512`);
 console.error(`Table data:           ${tableBytes} bytes`);
 
-let date = new Date().toISOString().slice(0, 10);
+// Fingerprint the generated tables (not the source text or comments) so the
+// header changes only when the emitted data would actually differ. Re-running
+// against identical Unicode data is byte-for-byte reproducible.
+let dataHash = (await sha256Hex(
+  JSON.stringify({
+    unicode: UNICODE_BASE,
+    bmpFilter: Array.from(bmpFilter),
+    smallPacked,
+    largeStarts,
+    largeCounts,
+    largeWidths,
+  }),
+)).slice(0, 16);
 
 let output = `\
 /* wcwidth.c - Unicode character width lookup
- * Unicode 16.0 - generated by tasks/gen-wcwidth.ts on ${date}
+ * Unicode 16.0 - generated by tasks/gen-wcwidth.ts
+ * Data hash: ${dataHash}
  *
  * Only zero-width (combining marks + default ignorable) and double-width
  * (wide/fullwidth) codepoints are stored. Everything else defaults to
@@ -423,3 +444,5 @@ if (Deno.args.includes("--check")) {
   await Deno.writeTextFile("src/wcwidth.c", output);
   console.error("Wrote src/wcwidth.c");
 }
+
+export {};