Enhance date parsing in WHOIS and add tests for registrar expiration dates

Updated the `toISO` function to handle various timezone formats in WHOIS date strings. Added tests to verify parsing of registrar registration expiration dates and EDUCAUSE format dates in the normalization process. This improves the accuracy of date handling in WHOIS responses.

src/lib/dates.test.ts

@@ -18,4 +18,11 @@ test("toISO parses ISO and common whois formats", () => {
 
   const mdy = toISO("Jan 02 2023");
   expect(mdy).toBe("2023-01-02T00:00:00Z");
+  // Registrar style timezone offsets
+  const plus0000 = toISO("2025-03-23T10:53:03+0000");
+  expect(plus0000).toBe("2025-03-23T10:53:03Z");
+  const plus0000Space = toISO("2025-03-23 10:53:03+0000");
+  expect(plus0000Space).toBe("2025-03-23T10:53:03Z");
+  const plus0530 = toISO("2025-03-23T10:53:03+05:30");
+  expect(plus0530).toBe("2025-03-23T05:23:03Z");
 });

src/lib/dates.ts

@@ -11,9 +11,9 @@ export function toISO(
   // Try several structured formats seen in WHOIS outputs (treat as UTC when no TZ provided)
   const tryFormats = [
     // 2023-01-02 03:04:05Z or without Z
-    /^(\d{4})-(\d{2})-(\d{2})[ T](\d{2}):(\d{2}):(\d{2})(?:Z)?$/,
+    /^(\d{4})-(\d{2})-(\d{2})[ T](\d{2}):(\d{2}):(\d{2})(?:Z|([+-]\d{2})(?::?(\d{2}))?)?$/,
     // 2023/01/02 03:04:05
-    /^(\d{4})\/(\d{2})\/(\d{2})[ T](\d{2}):(\d{2}):(\d{2})$/,
+    /^(\d{4})\/(\d{2})\/(\d{2})[ T](\d{2}):(\d{2}):(\d{2})(?:Z|([+-]\d{2})(?::?(\d{2}))?)?$/,
     // 02-Jan-2023
     /^(\d{2})-([A-Za-z]{3})-(\d{4})$/,
     // Jan 02 2023
@@ -69,17 +69,26 @@ function parseWithRegex(m: RegExpMatchArray, _re: RegExp): Date | undefined {
   try {
     // If the matched string contains time components, parse as Y-M-D H:M:S
     if (m[0].includes(":")) {
-      const [_, y, mo, d, hh, mm, ss] = m;
-      return new Date(
-        Date.UTC(
-          Number(y),
-          Number(mo) - 1,
-          Number(d),
-          Number(hh),
-          Number(mm),
-          Number(ss),
-        ),
+      const [_, y, mo, d, hh, mm, ss, offH, offM] = m;
+      // Base time as UTC
+      let dt = Date.UTC(
+        Number(y),
+        Number(mo) - 1,
+        Number(d),
+        Number(hh),
+        Number(mm),
+        Number(ss),
       );
+      // Apply timezone offset if present (e.g., +0000, -0500, +05:30)
+      if (offH) {
+        const sign = offH.startsWith("-") ? -1 : 1;
+        const hours = Math.abs(Number(offH));
+        const minutes = offM ? Number(offM) : 0;
+        const offsetMs = sign * (hours * 60 + minutes) * 60 * 1000;
+        // The captured time is local with an explicit offset; convert to UTC
+        dt -= offsetMs;
+      }
+      return new Date(dt);
     }
     // If the matched string contains hyphens, treat as DD-MMM-YYYY
     if (m[0].includes("-")) {

src/whois/normalize.test.ts

@@ -96,6 +96,49 @@ DNSSEC: unsigned
   expect(rec.nameservers && rec.nameservers.length === 2).toBe(true);
 });
 
+test("WHOIS registrar response with Registrar Registration Expiration Date", () => {
+  const text = `
+Domain Name: EXAMPLE.US
+Registrar WHOIS Server: whois.registrar.test
+Registrar URL: http://www.registrar.test
+Updated Date: 2025-03-23T10:53:03+0000
+Creation Date: 2020-04-24T15:03:39+0000
+Registrar Registration Expiration Date: 2027-04-23T00:00:00+0000
+Registrar: Registrar LLC
+`;
+  const rec = normalizeWhois(
+    "example.us",
+    "us",
+    text,
+    "whois.registrar.test",
+    "2025-01-01T00:00:00Z",
+  );
+  expect(Boolean(rec.expirationDate)).toBe(true);
+  expect(rec.expirationDate).toBe("2027-04-23T00:00:00Z");
+});
+
+test("WHOIS .edu EDUCAUSE format", () => {
+  const text = `
+This Registry database contains ONLY .EDU domains.
+
+Domain Name: TUFTS.EDU
+
+Domain record activated:    22-Jun-1987
+Domain record last updated: 02-Jul-2025
+Domain expires:             31-Jul-2026
+`;
+  const rec = normalizeWhois(
+    "tufts.edu",
+    "edu",
+    text,
+    "whois.educause.edu",
+    "2025-01-01T00:00:00Z",
+  );
+  expect(rec.creationDate).toBe("1987-06-22T00:00:00Z");
+  expect(rec.updatedDate).toBe("2025-07-02T00:00:00Z");
+  expect(rec.expirationDate).toBe("2026-07-31T00:00:00Z");
+});
+
 test("Privacy redacted WHOIS normalizes without contacts", () => {
   const text = `
 Domain Name: EXAMPLE.COM

src/whois/normalize.ts

@@ -31,17 +31,29 @@ export function normalizeWhois(
     "domain create date",
     "created",
     "registered",
+    // EDUCAUSE (.edu)
+    "domain record activated",
   ]);
   const updatedDate = anyValue(map, [
     "updated date",
     "last updated",
     "last modified",
     "modified",
+    // EDUCAUSE (.edu)
+    "domain record last updated",
   ]);
   const expirationDate = anyValue(map, [
     "registry expiry date",
+    "registry expiration date",
     "expiry date",
     "expiration date",
+    // Registrar-side synonyms commonly seen after referrals
+    "registrar registration expiration date",
+    "registrar registration expiry date",
+    "registrar expiration date",
+    "registrar expiry date",
+    // EDUCAUSE (.edu)
+    "domain expires",
     "paid-till",
     "expires on",
     "renewal date",