@opengeni/contracts 5.5.0 → 5.5.2-canary.40656248492001

package/src/site-session-http.ts

@@ -1,4 +1,56 @@
+import { matchSiteOperation, type SiteOperation } from "./site-operations";
+
 const SITE_WORKSPACE_PREFIX = "/v1/workspaces/site-host";
+/** Site Data and Site SQL routes under any published Site: `<prefix>/published-artifacts/<id>/data|files|sql`. */
+const SITE_DATA_PATH =
+  /^\/v1\/workspaces\/site-host\/published-artifacts\/([^/]+)\/(?:data|files|sql)(?:\/|$)/u;
+/** The list filter a frame without a published Site gets: it names no Site, so lists nothing. */
+export const SITE_NO_ORIGIN_ID = "00000000-0000-0000-0000-000000000000";
+
+/**
+ * The header a host sets on every request it forwards from a Site frame. The
+ * value names the host: `1` for the console bridge, `preview` for a Codemode
+ * local preview. The API treats a request carrying it (or the pinned Site id)
+ * as Site code: only `SITE_OPERATIONS`, only the frame's own sessions.
+ */
+export const SITE_FRAME_HEADER = "x-opengeni-site-frame";
+/** SITE_FRAME_HEADER value a Codemode local preview sets. */
+export const SITE_PREVIEW_FRAME = "preview";
+/**
+ * A random id the host mints per mounted frame. A frame without a published
+ * Site (inline HTML, previews) records it on the sessions it starts and may
+ * reach only those sessions afterwards.
+ */
+export const SITE_FRAME_ID_HEADER = "x-opengeni-site-frame-id";
+/** The published Site (artifact id) a host pinned the frame to. */
+export const SITE_ID_HEADER = "x-opengeni-site-id";
+/** The exact Site version the pinned frame runs. */
+export const SITE_VERSION_HEADER = "x-opengeni-site-version";
+
+/** The Site a console Site view conversation is about, as its messages name it. */
+export type SiteConversationSite = {
+  siteId: string;
+  title: string;
+  versionId: string;
+  revision: number;
+};
+
+/**
+ * Message context for a conversation in the console's Site view, where the
+ * person edits the Site beside it: which Site and version "the Site" means,
+ * and that a change is published straight away. The API attaches it to the
+ * first message of a conversation started there; the console attaches it to
+ * each later message, with the version then current. It is ordinary,
+ * model-visible context (`modelContext`), never authority.
+ */
+export function siteConversationContext(site: SiteConversationSite): string {
+  const title = site.title.replace(/\s+/gu, " ").trim().slice(0, 200) || "Untitled Site";
+  return [
+    `The person is working on the Opengeni Site “${title}” (Site id ${site.siteId}), shown beside this chat at version ${site.revision} (version id ${site.versionId}).`,
+    "When they ask for a change to “the Site”, “this page” or a part of it, they mean this Site: read its current source, then publish the change as a new version of this Site.",
+    "A published version goes live immediately and the preview beside this chat reloads; earlier versions can be restored.",
+  ].join(" ");
+}
 
 /** A Site/Codemode SDK request named a route outside the proxied surface. */
 export class SiteSessionPathError extends Error {
@@ -10,39 +62,181 @@ export class SiteSessionPathError extends Error {
   }
 }
 
-/** Route browser/preview SDK requests to the host workspace. This is a URL
- * boundary, not an endpoint or HTTP-method allowlist. Ordinary API handlers
- * authorize every operation with the viewer or agent's existing authority. */
+/** The host serves no Site Data: it is not bound to a published Site (inline HTML, previews). */
+export class SiteDataUnsupportedError extends Error {
+  readonly code = "site_data_unsupported";
+
+  constructor() {
+    super("This Site host does not provide Site Data");
+    this.name = "SiteDataUnsupportedError";
+  }
+}
+
+function unsafePathname(pathname: string): boolean {
+  return (
+    /[%\\#\u0000-\u0020\u007f]/u.test(pathname) ||
+    pathname.split("/").some((part) => part === "." || part === "..")
+  );
+}
+
+function hostWorkspacePath(path: string, workspaceId: string): string {
+  const url = new URL(path, "http://site.invalid");
+  if (url.pathname === "/v1/config/client") {
+    const requested = url.searchParams.getAll("workspaceId");
+    if (
+      requested.length > 1 ||
+      (requested.length === 1 && requested[0] !== "site-host" && requested[0] !== workspaceId)
+    )
+      throw new SiteSessionPathError();
+    // The SDK's scoped config read uses a query, not a workspace path segment.
+    // Both the console bridge and Codemode proxy bind it to their own workspace.
+    if (workspaceId !== "site-host") url.searchParams.set("workspaceId", workspaceId);
+    return url.pathname + url.search;
+  }
+  return path.replace("/workspaces/site-host", `/workspaces/${encodeURIComponent(workspaceId)}`);
+}
+
+/** Site Data: a Site reaches only its own data, as `current`, through a host bound to a published Site. */
+function siteDataPath(path: string, rewritten: string, workspaceId: string, siteId?: string) {
+  // Inside the Site the alias stays unresolved; the host resolves it.
+  if (workspaceId === "site-host") return path;
+  // Only a host bound to a published Site serves its data.
+  if (!siteId) throw new SiteDataUnsupportedError();
+  return rewritten.replace(
+    "/published-artifacts/current/",
+    `/published-artifacts/${encodeURIComponent(siteId)}/`,
+  );
+}
+
+/**
+ * Route a request from Site code (the in-Site SDK, the console bridge, a
+ * Codemode preview) to the host workspace. This is the `SITE_OPERATIONS`
+ * allowlist: anything else is refused with SiteSessionPathError before any
+ * request is made. Ordinary API handlers still authorize every allowed
+ * operation with the viewer's own authority, and the API independently
+ * refuses frame requests outside the allowlist or to another Site's sessions.
+ *
+ * `workspaceId` `site-host` checks a path inside the Site without rewriting
+ * it. A session list must name `originSiteId=current` or `siteAccessId=current`; the host binds it to
+ * its pinned Site, or to no Site at all when the frame has none.
+ */
 export function siteSessionPath(
   path: string,
   workspaceId: string,
-  _method = "GET",
+  method = "GET",
   siteId?: string,
 ): string {
+  const pathname = path.split("?")[0]!;
+  const match = matchSiteOperation(method, pathname);
+  if (!match) throw new SiteSessionPathError();
+  const rewritten = hostWorkspacePath(path, workspaceId);
+  if (SITE_DATA_PATH.test(pathname)) return siteDataPath(path, rewritten, workspaceId, siteId);
+  if (match.operation.id !== "sessions.list") return rewritten;
+  // A Site lists only conversations available to it. Never widen to the workspace.
+  const url = new URL(rewritten, "http://site.invalid");
+  const origins = url.searchParams.getAll("originSiteId");
+  const access = url.searchParams.getAll("siteAccessId");
+  const views = url.searchParams.getAll("view");
+  if (
+    views.length > 1 ||
+    (views.length === 1 && views[0] !== "page") ||
+    origins.length + access.length !== 1 ||
+    (origins[0] ?? access[0]) !== "current"
+  )
+    throw new SiteSessionPathError();
+  if (workspaceId === "site-host") return path;
+  url.searchParams.set(
+    access.length ? "siteAccessId" : "originSiteId",
+    siteId ?? SITE_NO_ORIGIN_ID,
+  );
+  return url.pathname + url.search;
+}
+
+/**
+ * Route an agent's own SDK request through the Codemode SDK proxy to its
+ * workspace. Unlike `siteSessionPath` this is a URL prefix boundary, not an
+ * operation allowlist: the proxy signs a short-lived credential bounded by
+ * the agent's own permission ceiling and ordinary handlers authorize every
+ * operation. Site Data is not served here (an agent uses its tools).
+ * Requests a Codemode preview forwards from Site code carry the frame marker
+ * and use `siteSessionPath` instead.
+ */
+export function siteHostWorkspacePath(path: string, workspaceId: string): string {
   const pathname = path.split("?")[0]!;
   const workspace =
     pathname === SITE_WORKSPACE_PREFIX || pathname.startsWith(`${SITE_WORKSPACE_PREFIX}/`);
-  const context = pathname === "/v1/config/client";
-  if (
-    (!workspace && !context) ||
-    /[%\\#\u0000-\u0020\u007f]/u.test(pathname) ||
-    pathname.split("/").some((p) => p === "." || p === "..")
-  ) {
+  if ((!workspace && pathname !== "/v1/config/client") || unsafePathname(pathname))
     throw new SiteSessionPathError();
+  const rewritten = hostWorkspacePath(path, workspaceId);
+  const data = SITE_DATA_PATH.exec(pathname);
+  if (data) {
+    if (data[1] !== "current") throw new SiteSessionPathError();
+    return siteDataPath(path, rewritten, workspaceId);
   }
-  const rewritten = path.replace(
-    "/workspaces/site-host",
-    `/workspaces/${encodeURIComponent(workspaceId)}`,
-  );
-  // Preview has no published Site identity. Never widen a Site-only list to the workspace.
   const url = new URL(rewritten, "http://site.invalid");
   if (url.searchParams.get("originSiteId") === "current" && workspaceId !== "site-host") {
-    url.searchParams.set("originSiteId", siteId ?? "00000000-0000-0000-0000-000000000000");
+    url.searchParams.set("originSiteId", SITE_NO_ORIGIN_ID);
     return url.pathname + url.search;
   }
   return rewritten;
 }
 
+/** The Site identity a host attached to a request, from its host-set headers. */
+export type SiteFrameHeaders = {
+  /** The pinned published Site id, or null for a frame without one. */
+  siteId: string | null;
+  versionId: string | null;
+  /** The host's frame marker value (`1`, `preview`), or null when only the Site id was set. */
+  marker: string | null;
+  frameId: string | null;
+};
+
+/** Read the host-set frame headers. Null when the request did not come from a Site frame. */
+export function siteFrameFromHeaders(
+  header: (name: string) => string | null | undefined,
+): SiteFrameHeaders | null {
+  const value = (name: string) => header(name) ?? null;
+  const marker = value(SITE_FRAME_HEADER);
+  const siteId = value(SITE_ID_HEADER);
+  const versionId = value(SITE_VERSION_HEADER);
+  const frameId = value(SITE_FRAME_ID_HEADER);
+  if (marker === null && siteId === null && versionId === null && frameId === null) return null;
+  return { siteId: siteId || null, versionId: versionId || null, marker, frameId: frameId || null };
+}
+
+/**
+ * Match a request that reached a workspace route from a Site frame against
+ * `SITE_OPERATIONS`. `pathname` is the raw request path
+ * (`/v1/workspaces/<id>/...` or `/v1/config/client`). The frame's own pinned
+ * Site id maps back to `current`; any other Site id does not match.
+ */
+export function siteFrameOperation(
+  method: string,
+  pathname: string,
+  workspaceId: string,
+  frameSiteId: string | null,
+): { operation: SiteOperation; suffix: string; sessionId: string | null } | null {
+  if (unsafePathname(pathname)) return null;
+  let sitePath = pathname;
+  if (pathname !== "/v1/config/client") {
+    const prefix = `/v1/workspaces/${workspaceId}`;
+    if (pathname !== prefix && !pathname.startsWith(`${prefix}/`)) return null;
+    sitePath = `${SITE_WORKSPACE_PREFIX}${pathname.slice(prefix.length)}`;
+    const data = /^\/v1\/workspaces\/site-host\/published-artifacts\/([^/]+)\//u.exec(sitePath);
+    if (data) {
+      if (!frameSiteId || data[1] !== frameSiteId) return null;
+      sitePath = sitePath.replace(
+        `/published-artifacts/${frameSiteId}/`,
+        "/published-artifacts/current/",
+      );
+    }
+  }
+  const match = matchSiteOperation(method, sitePath);
+  if (!match) return null;
+  const session = /^\/sessions\/([^/]+)(?:\/|$)/u.exec(match.suffix);
+  return { ...match, sessionId: session ? session[1]! : null };
+}
+
 /** Preserve SDK request semantics while the host supplies identity and transport headers. */
 export function siteRequestHeaders(input: HeadersInit): Headers {
   const headers = new Headers(input);
@@ -61,3 +255,137 @@ export function siteRequestHeaders(input: HeadersInit): Headers {
   }
   return headers;
 }
+
+// ---------------------------------------------------------------------------
+// What a Site frame may put in a request body
+// ---------------------------------------------------------------------------
+
+/** Message fields a Site page may set: the text and how the model answers it. */
+const SITE_FRAME_MESSAGE_FIELDS = [
+  "text",
+  "annotations",
+  "model",
+  "reasoningEffort",
+  "latencyMode",
+  "controlEtag",
+  "expectedDraftRevision",
+  // The viewer's own IANA time zone, so the agent sees the message's local time.
+  "timeZone",
+] as const;
+
+/**
+ * Body fields a Site frame may send per operation. Anything else with a
+ * value (resources, personal resource attachments, MCP credential updates,
+ * connection accounts, model context, approval decisions, human-input
+ * answers) is the person's to give in the console, never a Site page's.
+ */
+const SITE_FRAME_BODY_FIELDS: Readonly<Record<string, readonly string[]>> = Object.freeze({
+  "sessions.send": ["type", "payload", "clientEventId"],
+  "sessions.steer": [...SITE_FRAME_MESSAGE_FIELDS, "clientEventId"],
+  "sessions.control": ["action", "reason", "clientEventId", "expectedControlEtag"],
+  "sessions.composerDraft.put": [
+    "text",
+    "annotations",
+    "resources",
+    "model",
+    "reasoningEffort",
+    "latencyMode",
+    "expectedRevision",
+  ],
+  "sessions.composerDraft.submit": [
+    ...SITE_FRAME_MESSAGE_FIELDS,
+    "resources",
+    "connectionAccounts",
+    "clientEventId",
+    "delivery",
+  ],
+  "sessions.queue.move": ["clientEventId", "expectedQueueVersion", "beforeTurnId"],
+  "sessions.queue.edit": [
+    "clientEventId",
+    "expectedTurnVersion",
+    "expectedDraftRevision",
+    "replaceDraft",
+  ],
+  "sessions.queue.steer": ["clientEventId", "expectedTurnVersion", "controlEtag"],
+  "sessions.queue.delete": ["clientEventId", "expectedTurnVersion", "reason"],
+});
+
+/** Fields allowed above only while empty: a draft carries no resources or accounts. */
+const SITE_FRAME_EMPTY_ONLY_FIELDS = new Set(["resources", "connectionAccounts"]);
+
+/**
+ * Session create fields a Site frame may send. Agent/tool choices use the
+ * ordinary viewer's defaults, never a platform-imposed Site profile.
+ * Ownership, settings, approvals and credentials remain host-owned.
+ */
+export const SITE_FRAME_SESSION_CREATE_FIELDS: readonly string[] = Object.freeze([
+  "initialMessage",
+  "requestedSessionId",
+  "idempotencyKey",
+  "clientEventId",
+  "model",
+  "reasoningEffort",
+  "latencyMode",
+  "startMode",
+  "metadata",
+  "timeZone",
+]);
+
+function bodyRecord(value: unknown): Record<string, unknown> | null {
+  return value && typeof value === "object" && !Array.isArray(value)
+    ? (value as Record<string, unknown>)
+    : null;
+}
+
+/** Absent, null, an empty list or an empty object asks for nothing. */
+function emptyBodyValue(value: unknown): boolean {
+  if (value === undefined || value === null) return true;
+  if (Array.isArray(value)) return value.length === 0;
+  if (typeof value === "object") return Object.keys(value).length === 0;
+  return false;
+}
+
+function refusedField(
+  body: Record<string, unknown>,
+  allowed: readonly string[],
+  prefix = "",
+): string | null {
+  for (const [name, value] of Object.entries(body)) {
+    if (emptyBodyValue(value)) continue;
+    if (!allowed.includes(name) || SITE_FRAME_EMPTY_ONLY_FIELDS.has(name))
+      return `${prefix}${name}`;
+  }
+  return null;
+}
+
+/**
+ * The first body field a Site frame may not send for this operation (for
+ * example `payload.resources` or `type`), or null when the body is allowed.
+ * `suffix` is the operation's site-host suffix (from `siteFrameOperation` or
+ * `matchSiteOperation`). A body that is not a JSON object is left to the
+ * route, which refuses it. Only `user.message` events are allowed:
+ * approvals and answers to questions happen in the console.
+ */
+export function siteFrameBodyRefusal(
+  operationId: string,
+  suffix: string,
+  body: unknown,
+): string | null {
+  const record = bodyRecord(body);
+  if (!record) return null;
+  if (operationId === "sessions.create")
+    return refusedField(record, SITE_FRAME_SESSION_CREATE_FIELDS);
+  const action =
+    operationId === "sessions.queue.item"
+      ? /\/queue\/[^/]+\/(move|edit|steer|delete)$/u.exec(suffix)?.[1]
+      : undefined;
+  const key = action ? `sessions.queue.${action}` : operationId;
+  const allowed = SITE_FRAME_BODY_FIELDS[key];
+  if (!allowed) return null;
+  const refused = refusedField(record, allowed);
+  if (refused) return refused;
+  if (operationId !== "sessions.send") return null;
+  if (record.type !== "user.message") return "type";
+  const payload = bodyRecord(record.payload);
+  return payload ? refusedField(payload, SITE_FRAME_MESSAGE_FIELDS, "payload.") : null;
+}