Tenant naming logic adjustments

src/lib/tenant-naming.ts

@@ -0,0 +1,132 @@
+/**
+ * Deterministic tenant-name derivation for PiecedTenant CRs.
+ *
+ * Background
+ * ----------
+ * Every PiecedTenant CR's `metadata.name` becomes part of the tenant
+ * namespace, which the operator builds as `tenant-{name}` (see
+ * `pieced-operator/api/v1alpha1/piecedtenant_types.go::NamespaceName`).
+ * Kubernetes namespace names follow the RFC 1123 DNS *label* spec:
+ * lowercased alphanumeric + hyphens, must start and end with alnum,
+ * and **max 63 characters**.
+ *
+ * That gives us 63 - len("tenant-") = 56 chars to play with for the CR
+ * name itself. Anything longer is rejected at apply time, so we cap
+ * here.
+ *
+ * Format
+ * ------
+ *   kind=company  →  {slug}-{requestIdHex8}     e.g. "acme-gmbh-abc12345"
+ *   kind=personal →  p-{requestIdHex8}          e.g. "p-abc12345"
+ *
+ * The 8-hex-char suffix is taken from `tenant_requests.id` (a Postgres
+ * `gen_random_uuid()` value, set at row insert). Two motivations:
+ *
+ *   1. Uniqueness — multiple requests for the same company name no longer
+ *      collide (this is what unblocks Slice 3, multi-tenant per org).
+ *   2. Stability — the suffix is known at approval time and never changes,
+ *      so the operator and portal agree without coordination. We use the
+ *      request UUID rather than the eventual LiteLLM virtual-key UUID
+ *      because the latter doesn't exist until the operator runs.
+ *
+ * 8 hex chars = 32 bits of entropy. Collision probability with 100 active
+ * tenants per company prefix is ~1e-6; for our pilot scale that's fine.
+ *
+ * Limits
+ * ------
+ * Suffix is always 8 + 1 (hyphen) = 9 chars. Slug therefore caps at
+ * 56 - 9 = 47 chars, then we strip any trailing hyphens left by the cut.
+ *
+ * Examples
+ * --------
+ *   deriveTenantName("company",  "Acme GmbH", "abc12345-...") = "acme-gmbh-abc12345"
+ *   deriveTenantName("company",  "Müller AG", "abc12345-...") = "m-ller-ag-abc12345"   (umlaut → "-")
+ *   deriveTenantName("company",  "!!!",       "abc12345-...") = "t-abc12345"           (slug empty → "t-")
+ *   deriveTenantName("personal", "",          "abc12345-...") = "p-abc12345"
+ */
+
+export type TenantKind = "company" | "personal";
+
+const MAX_NAMESPACE_LEN = 63;
+const NAMESPACE_PREFIX = "tenant-";
+const MAX_TENANT_NAME_LEN = MAX_NAMESPACE_LEN - NAMESPACE_PREFIX.length; // 56
+const SUFFIX_HEX_LEN = 8;
+const SUFFIX_TOTAL_LEN = SUFFIX_HEX_LEN + 1; // including the joining "-"
+const MAX_SLUG_LEN = MAX_TENANT_NAME_LEN - SUFFIX_TOTAL_LEN; // 47
+
+export class InvalidRequestIdError extends Error {
+  constructor(requestId: string) {
+    super(
+      `Cannot derive tenant name: requestId "${requestId}" does not contain ${SUFFIX_HEX_LEN} hex characters`
+    );
+    this.name = "InvalidRequestIdError";
+  }
+}
+
+/**
+ * Reduce an arbitrary string to a DNS-label-safe slug. Non-alnum runs
+ * collapse to a single "-"; leading/trailing hyphens are stripped.
+ *
+ * Note this does not transliterate Unicode — "Müller" becomes "m-ller",
+ * not "mueller". That's deliberate: transliteration introduces locale
+ * dependencies (de-DE vs de-CH vs sv-SE all disagree on ä→a/ä→ae) and
+ * we'd rather have a stable, ugly slug than a pretty one that changes
+ * if we touch the locale config later. Customers see the human-readable
+ * `displayName`, not the slug.
+ */
+function slugify(input: string): string {
+  return input
+    .toLowerCase()
+    .replace(/[^a-z0-9]+/g, "-")
+    .replace(/^-+|-+$/g, "");
+}
+
+/**
+ * Extract the first 8 hex chars of a UUID string. Strips hyphens and
+ * lowercases first so callers can pass either "abc12345-..." or
+ * "ABC12345..." form. Postgres `gen_random_uuid()` already emits the
+ * canonical lowercase-hyphenated form, so this is just defense in depth
+ * against any hand-inserted IDs.
+ */
+function requestIdSuffix(requestId: string): string {
+  const hex = requestId.replace(/-/g, "").toLowerCase();
+  if (!/^[0-9a-f]{8}/.test(hex)) {
+    throw new InvalidRequestIdError(requestId);
+  }
+  return hex.slice(0, SUFFIX_HEX_LEN);
+}
+
+/**
+ * Build the PiecedTenant CR `metadata.name` for an approved tenant request.
+ *
+ * @param kind        "company" for normal customer accounts; "personal"
+ *                    for individual accounts (Slice 4 — `is_personal=true`).
+ * @param companyName Raw display name from the registration. Ignored when
+ *                    kind="personal".
+ * @param requestId   `tenant_requests.id` (Postgres UUID).
+ * @returns           A K8s-safe CR name, ≤ 56 chars, with an 8-hex suffix.
+ */
+export function deriveTenantName(
+  kind: TenantKind,
+  companyName: string,
+  requestId: string
+): string {
+  const suffix = requestIdSuffix(requestId);
+
+  if (kind === "personal") {
+    return `p-${suffix}`;
+  }
+
+  // Company branch: slug-{suffix}, with empty-slug fallback.
+  const rawSlug = slugify(companyName);
+
+  // Cap then re-trim — slicing might leave a dangling hyphen if a non-alnum
+  // run sat right at the boundary (e.g. "acme-foo-bar-..." cut to "acme-foo-").
+  const slug = rawSlug.slice(0, MAX_SLUG_LEN).replace(/-+$/, "");
+
+  if (!slug) {
+    return `t-${suffix}`;
+  }
+
+  return `${slug}-${suffix}`;
+}