@rangojs/router 0.0.0-experimental.eb0645d3 → 0.0.0-experimental.f1468e3c

package/src/router/middleware.ts

@@ -1,16 +1,10 @@
 /// <reference types="vite/types/importMeta.d.ts" />
-/**
- * Middleware Execution
- *
- * True middleware that wraps the entire RSC handler.
- * - `await next()` returns actual Response
- * - Can modify response headers
- * - Can catch errors from RSC rendering
- * - Forgiving API: if middleware doesn't return, original response is used
- */
 
 import { contextGet, contextSet } from "../context-var.js";
+import { escapeRegExp } from "../regex-escape.js";
+import { parsePattern as parseRoutePattern } from "./pattern-matching.js";
 import { safeDecodeURIComponent } from "./url-params.js";
+import { fireAndForgetWaitUntil } from "../types/request-scope.js";
 import type {
   CollectedMiddleware,
   MiddlewareCollectableEntry,
@@ -20,28 +14,29 @@ import type {
   ResponseHolder,
 } from "./middleware-types.js";
 import { _getRequestContext } from "../server/request-context.js";
+import {
+  EXTERNAL_REDIRECT_MARKER,
+  isExternalRedirect,
+  markExternalRedirect,
+} from "../redirect-origin.js";
 import { isAutoGeneratedRouteName } from "../route-name.js";
 import { appendMetric, createMetricsStore } from "./metrics.js";
+import { observePhase, PHASES } from "./instrument.js";
 import { stripInternalParams } from "./handler-context.js";
 import { isWebSocketUpgradeResponse } from "../response-utils.js";
 
-// Re-export types and cookie utilities for backward compatibility
+// Re-export types consumed through this module's path.
 export type {
   CookieOptions,
-  CollectedMiddleware,
-  MiddlewareCollectableEntry,
   MiddlewareContext,
   MiddlewareEntry,
   MiddlewareFn,
-  ResponseHolder,
 } from "./middleware-types.js";
-export { parseCookies, serializeCookie } from "./middleware-cookies.js";
 
 const MIDDLEWARE_METRIC_DEPTH = 1;
-/** Ignore post-next() durations below this threshold (measurement noise). */
 const POST_METRIC_MIN_DURATION_MS = 0.01;
 
-function getMiddlewareMetricBase<TEnv>(
+function getMiddlewareMetricLabel<TEnv>(
   entry: MiddlewareEntry<TEnv>,
   ordinal: number,
 ): string {
@@ -49,24 +44,49 @@ function getMiddlewareMetricBase<TEnv>(
   const scope = entry.pattern ?? "*";
 
   if (handlerName) {
-    return `${handlerName}@${scope}`;
+    return `middleware:${handlerName}@${scope}`;
   }
 
-  return `${scope}#${ordinal + 1}`;
-}
-
-function getMiddlewareMetricLabel<TEnv>(
-  entry: MiddlewareEntry<TEnv>,
-  ordinal: number,
-): string {
-  return `middleware:${getMiddlewareMetricBase(entry, ordinal)}`;
+  return `middleware:${scope}#${ordinal + 1}`;
 }
 
 /**
- * Parse a route pattern into regex and param names
- * Supports: *, /path, /path/*, /path/:param, /path/:param/*
+ * Compile a middleware scope pattern to a regex + param names.
+ *
+ * Middleware scopes reuse the route pattern parser (`parsePattern` from
+ * pattern-matching.ts), so they support the same param forms as routes —
+ * optional (`:x?`), constrained (`:x(en|gb)`), and suffix (`:x.html`) — in
+ * addition to the trailing-`*` wildcard middleware relies on. Before this
+ * unification the middleware-side parser handled only static, bare `:param`,
+ * and trailing `*`, so e.g. `router.use("/:locale(en|gb)/*", mw)` silently
+ * named the param "locale(en|gb)" and never enforced the constraint.
+ *
+ * Middleware matching semantics deliberately differ from route matching, so we
+ * emit the regex here rather than route through `compilePattern`:
+ * - `*` alone matches every path (`/^.*$/`).
+ * - A trailing `*` segment is an OPTIONAL subtree match (`(?:/.*)?`): `/admin/*`
+ *   matches `/admin`, `/admin/`, and `/admin/users`. It contributes no param
+ *   name (unlike route wildcards, which capture `*`).
+ * - A NON-trailing `*` is also OPTIONAL (`(?:/.*)?`), matching zero-or-more
+ *   intermediate segments: `/a/<star>/b` matches both `/a/b` and `/a/x/b`. This
+ *   mirrors the pre-unification parser, which compiled every `*` part as
+ *   optional regardless of position.
+ * - A pattern without a trailing `*` tolerates a trailing slash (`/?$`).
+ * - Constraints are baked into the regex as an alternation so `matchMiddleware`
+ *   (a bare `regex.test`) enforces them without extra validation. Constraint
+ *   values are matched against the raw (still URL-encoded) path segment, which
+ *   matches the pre-unification middleware behavior (it never decoded for
+ *   matching); the constraint string is regex-escaped so values like `en.gb`
+ *   are treated literally.
+ *
+ * The route segment parser only recognizes `/`-prefixed segments, but the
+ * pre-unification middleware parser split on `/` and dropped empty parts, so a
+ * leading slash was irrelevant: `use("admin/*")` and `use("/admin/*")` scoped
+ * identically. Normalize a non-`*` pattern to have a leading slash before
+ * parsing so that behavior is preserved (without it, `parseRoutePattern("admin/*")`
+ * drops the static `admin` and the scope explodes to every path).
  */
-export function parsePattern(pattern: string): {
+export function compileMiddlewarePattern(pattern: string): {
   regex: RegExp;
   paramNames: string[];
 } {
@@ -74,45 +94,47 @@ export function parsePattern(pattern: string): {
     return { regex: /^.*$/, paramNames: [] };
   }
 
+  const normalizedPattern = pattern.startsWith("/") ? pattern : `/${pattern}`;
+  const segments = parseRoutePattern(normalizedPattern);
   const paramNames: string[] = [];
   let regexStr = "^";
+  let hasTrailingWildcard = false;
 
-  const parts = pattern.split("/").filter(Boolean);
-
-  for (let i = 0; i < parts.length; i++) {
-    const part = parts[i];
+  for (let i = 0; i < segments.length; i++) {
+    const segment = segments[i];
 
-    if (part === "*") {
-      // Wildcard - match rest of path
+    if (segment.type === "wildcard") {
+      // Optional subtree match (parity with the original middleware parser,
+      // which compiled every `*` as `(?:/.*)?`). A trailing `*` matches the
+      // subtree; a non-trailing `*` matches zero-or-more intermediate segments,
+      // so `/a/<star>/b` still matches `/a/b`.
       regexStr += "(?:/.*)?";
-    } else if (part.startsWith(":")) {
-      // Param
-      const paramName = part.slice(1);
-      paramNames.push(paramName);
-      regexStr += "/([^/]+)";
+      if (i === segments.length - 1) {
+        hasTrailingWildcard = true;
+      }
+    } else if (segment.type === "param") {
+      paramNames.push(segment.value);
+      const suffixPattern = segment.suffix ? escapeRegExp(segment.suffix) : "";
+      const valuePattern = segment.constraint
+        ? `(${segment.constraint.map(escapeRegExp).join("|")})`
+        : "([^/]+)";
+      if (segment.optional) {
+        regexStr += `(?:/${valuePattern}${suffixPattern})?`;
+      } else {
+        regexStr += `/${valuePattern}${suffixPattern}`;
+      }
     } else {
-      // Literal
-      regexStr += "/" + escapeRegex(part);
+      // Static literal
+      regexStr += "/" + escapeRegExp(segment.value);
     }
   }
 
-  // If pattern doesn't end with *, match exact or with trailing segments
-  if (!pattern.endsWith("*")) {
-    regexStr += "/?$";
-  } else {
-    regexStr += "$";
-  }
+  // Without a trailing `*`, match exactly with an optional trailing slash.
+  regexStr += hasTrailingWildcard ? "$" : "/?$";
 
   return { regex: new RegExp(regexStr), paramNames };
 }
 
-/**
- * Escape special regex characters
- */
-function escapeRegex(str: string): string {
-  return str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
-}
-
 /**
  * Extract params from a pathname using a pattern's regex and param names.
  *
@@ -186,14 +208,22 @@ export function createMiddlewareContext<TEnv>(
     return responseHolder.response;
   };
 
+  // Capture reqCtx once: the request-scoped platform fields
+  // (originalUrl, executionContext, waitUntil) are immutable per request,
+  // so snapshotting beats re-reading ALS on every access. The lazy getters
+  // below (routeName, theme, setTheme) stay lazy because those can change
+  // during `await next()`.
+  const reqCtx = _getRequestContext();
   return {
     request,
     url,
-    originalUrl: new URL(request.url),
+    originalUrl: reqCtx?.originalUrl ?? new URL(request.url),
     pathname: url.pathname,
     searchParams: url.searchParams,
     env: env as MiddlewareContext<TEnv>["env"],
     params,
+    executionContext: reqCtx?.executionContext,
+    waitUntil: reqCtx ? reqCtx.waitUntil.bind(reqCtx) : fireAndForgetWaitUntil,
     // Getter: re-derives from request context on each access so that global
     // middleware sees the matched route name after await next().
     get routeName(): MiddlewareContext<TEnv>["routeName"] {
@@ -252,9 +282,13 @@ export function createMiddlewareContext<TEnv>(
 
     reverse:
       reverse ??
-      ((name: string) => {
+      ((
+        name: string,
+        _params?: Record<string, string>,
+        _search?: Record<string, unknown>,
+      ) => {
         throw new Error(
-          `ctx.reverse() is not available - route map was not provided to middleware context`,
+          `ctx.reverse(${JSON.stringify(name)}) is not available: no route map is bound to this middleware context.`,
         );
       }),
 
@@ -288,9 +322,16 @@ export function matchMiddleware<TEnv>(
       continue;
     }
 
-    // Check if pathname matches
-    if (entry.regex.test(pathname)) {
-      const params = extractParams(pathname, entry.regex, entry.paramNames);
+    // Run the scope regex ONCE per entry. The old code ran test() then, on a
+    // hit, extractParams' match() — a second full pass over the same string for
+    // every matching entry on the per-request hot path. The regexes carry no
+    // `g` flag, so there is no lastIndex statefulness across this single match.
+    const m = pathname.match(entry.regex);
+    if (m) {
+      const params: Record<string, string> = {};
+      for (let i = 0; i < entry.paramNames.length; i++) {
+        params[entry.paramNames[i]] = safeDecodeURIComponent(m[i + 1] || "");
+      }
       matches.push({ entry, params });
     }
   }
@@ -298,6 +339,88 @@ export function matchMiddleware<TEnv>(
   return matches;
 }
 
+// Set-Cookie is appended; for other headers stubOverridesNonCookie=true
+// overwrites (chain ran to completion), false fills only missing slots (an
+// explicit short-circuit Response's own headers win).
+function mergeStubHeaders(
+  target: Headers,
+  stub: Headers,
+  stubOverridesNonCookie: boolean,
+): void {
+  stub.forEach((value, name) => {
+    // The reserved external-redirect marker is internal and never a trust
+    // signal; never copy a stub value (e.g. a stray ctx.header() call) onto a
+    // browser-facing response. The opt-in is the out-of-band brand.
+    if (name.toLowerCase() === EXTERNAL_REDIRECT_MARKER) return;
+    if (name.toLowerCase() === "set-cookie") {
+      target.append(name, value);
+    } else if (stubOverridesNonCookie || !target.has(name)) {
+      target.set(name, value);
+    }
+  });
+}
+
+// Set-Cookie is deduped so a nested inner executeMiddleware that already merged
+// the same reqCtx cookies does not duplicate them; other headers fill if missing.
+function mergeReqCtxStub(
+  target: Headers,
+  reqCtx: ReturnType<typeof _getRequestContext>,
+): void {
+  if (!reqCtx) return;
+  const stubCookies = reqCtx.res.headers.getSetCookie();
+  if (stubCookies.length > 0) {
+    const existing = new Set(target.getSetCookie());
+    for (const cookie of stubCookies) {
+      if (!existing.has(cookie)) {
+        target.append("set-cookie", cookie);
+      }
+    }
+  }
+  reqCtx.res.headers.forEach((value, name) => {
+    // Never propagate the reserved external-redirect marker (see mergeStubHeaders).
+    if (name.toLowerCase() === EXTERNAL_REDIRECT_MARKER) return;
+    if (name !== "set-cookie" && !target.has(name)) {
+      target.set(name, value);
+    }
+  });
+}
+
+// Clone `base` with stub headers merged into a fresh Headers (the clone keeps
+// the body mutable for post-next() modifications). Set-Cookie is always
+// appended; other headers obey stubOverridesNonCookie (see mergeStubHeaders).
+// mergeReqCtx folds in RequestContext stub cookies/headers; the intercept
+// short-circuit path passes false (its reqCtx headers are not merged here),
+// which is the one deliberate divergence between the call sites.
+function mergeResponse(
+  base: Response,
+  stub: Headers,
+  opts: { stubOverridesNonCookie: boolean; mergeReqCtx: boolean },
+): Response {
+  const mergedHeaders = new Headers(base.headers);
+  // The reserved external-redirect marker is never a trust signal and must never
+  // reach the browser. The guard strips it on 3xx redirects; strip it here too so
+  // a forged value cannot ride a non-3xx middleware response (which the 3xx-only
+  // guard would not touch) to the client. The opt-in is the out-of-band brand.
+  mergedHeaders.delete(EXTERNAL_REDIRECT_MARKER);
+  mergeStubHeaders(mergedHeaders, stub, opts.stubOverridesNonCookie);
+  if (opts.mergeReqCtx) {
+    mergeReqCtxStub(mergedHeaders, _getRequestContext());
+  }
+  const merged = new Response(base.body, {
+    status: base.status,
+    statusText: base.statusText,
+    headers: mergedHeaders,
+  });
+  // Transfer the out-of-band external-redirect brand across this rebuild: a
+  // middleware short-circuit `return redirect(url, { external: true })` reaches
+  // the open-redirect guard only after this merge, and the brand lives on the
+  // Response object, not in its headers.
+  if (isExternalRedirect(base)) {
+    markExternalRedirect(merged);
+  }
+  return merged;
+}
+
 /**
  * Execute middleware chain
  *
@@ -306,7 +429,7 @@ export function matchMiddleware<TEnv>(
  * - `ctx.headers` available before and after `await next()`
  * - `ctx.header()` shorthand for setting a single header
  * - Forgiving: if middleware doesn't return, uses the downstream response
- * - Short-circuit: return Response to stop chain
+ * - Short-circuit: return OR throw a Response to stop chain
  * - Error catching: try/catch around `next()` works
  */
 export async function executeMiddleware<TEnv>(
@@ -336,47 +459,16 @@ export async function executeMiddleware<TEnv>(
       // End of chain - call actual RSC handler
       const response = await finalHandler();
 
-      // Merge headers set on stub into the real response.
-      // Use append for Set-Cookie to preserve multiple cookies.
-      const mergedHeaders = new Headers(response.headers);
-      stubResponse.headers.forEach((value, name) => {
-        if (name.toLowerCase() === "set-cookie") {
-          mergedHeaders.append(name, value);
-        } else {
-          mergedHeaders.set(name, value);
-        }
-      });
-      // Also merge shared RequestContext stub (cookies written via cookies().set()).
-      // Dedup Set-Cookie: an inner executeMiddleware (route-level middleware)
-      // may have already merged the same reqCtx cookies into the response.
-      const reqCtx = _getRequestContext();
-      if (reqCtx) {
-        const stubCookies = reqCtx.res.headers.getSetCookie();
-        if (stubCookies.length > 0) {
-          const existing = new Set(mergedHeaders.getSetCookie());
-          for (const cookie of stubCookies) {
-            if (!existing.has(cookie)) {
-              mergedHeaders.append("set-cookie", cookie);
-            }
-          }
-        }
-        reqCtx.res.headers.forEach((value, name) => {
-          if (name !== "set-cookie" && !mergedHeaders.has(name)) {
-            mergedHeaders.set(name, value);
-          }
-        });
-      }
-
       if (isWebSocketUpgradeResponse(response)) {
         responseHolder.response = response;
         return response;
       }
 
-      // Clone response with merged headers (mutable for post-next() modifications)
-      responseHolder.response = new Response(response.body, {
-        status: response.status,
-        statusText: response.statusText,
-        headers: mergedHeaders,
+      // Chain ran to completion: stub headers overwrite (stubOverridesNonCookie)
+      // and reqCtx stub headers are merged in.
+      responseHolder.response = mergeResponse(response, stubResponse.headers, {
+        stubOverridesNonCookie: true,
+        mergeReqCtx: true,
       });
 
       return responseHolder.response;
@@ -434,12 +526,28 @@ export async function executeMiddleware<TEnv>(
       return nextPromise;
     };
 
+    // Wrap the middleware (including its downstream next() chain) in its span
+    // via the unified phase API. metric:false — the middleware's perf metric is
+    // its exclusive pre/post own-time, recorded directly above and below, finer
+    // than a single wrap. Spans nest by async context, so this onions
+    // middleware-over-middleware and the core handler underneath. Pass-through
+    // when neither surface is active.
     let result: Response | void;
     try {
-      result = await entry.handler(ctx, wrappedNext);
+      result = await observePhase(PHASES.middleware(metricLabel), () =>
+        entry.handler(ctx, wrappedNext),
+      );
     } catch (error) {
-      finishMiddleware();
-      throw error;
+      // Thrown Response is short-circuit control flow, not an error.
+      // Fall through to the `if (result instanceof Response)` branch below
+      // so stub headers and request-context cookies merge as they do for
+      // an explicit `return new Response(...)`. Real errors propagate.
+      if (error instanceof Response) {
+        result = error;
+      } else {
+        finishMiddleware();
+        throw error;
+      }
     }
     finishMiddleware();
 
@@ -460,45 +568,18 @@ export async function executeMiddleware<TEnv>(
 
     // Explicit return takes precedence (middleware short-circuit).
     // Merge stub headers (from ctx.header before this point) and
-    // RequestContext stub headers (from ctx.setCookie) into the
+    // RequestContext stub headers (from cookies().set()) into the
     // returned Response so they are not lost.
     if (result instanceof Response) {
       if (isWebSocketUpgradeResponse(result)) {
         responseHolder.response = result;
         return result;
       }
-      const mergedHeaders = new Headers(result.headers);
-      stubResponse.headers.forEach((value, name) => {
-        if (name.toLowerCase() === "set-cookie") {
-          mergedHeaders.append(name, value);
-        } else if (!mergedHeaders.has(name)) {
-          mergedHeaders.set(name, value);
-        }
-      });
-      // Also merge shared RequestContext stub (cookies written via setCookie).
-      // Dedup Set-Cookie: an inner executeMiddleware (route-level middleware)
-      // may have already merged the same reqCtx cookies into the response.
-      const reqCtx = _getRequestContext();
-      if (reqCtx) {
-        const stubCookies = reqCtx.res.headers.getSetCookie();
-        if (stubCookies.length > 0) {
-          const existing = new Set(mergedHeaders.getSetCookie());
-          for (const cookie of stubCookies) {
-            if (!existing.has(cookie)) {
-              mergedHeaders.append("set-cookie", cookie);
-            }
-          }
-        }
-        reqCtx.res.headers.forEach((value, name) => {
-          if (name !== "set-cookie" && !mergedHeaders.has(name)) {
-            mergedHeaders.set(name, value);
-          }
-        });
-      }
-      const merged = new Response(result.body, {
-        status: result.status,
-        statusText: result.statusText,
-        headers: mergedHeaders,
+      // Explicit short-circuit: the returned Response's own headers win
+      // (stubOverridesNonCookie=false); reqCtx stub headers still merge in.
+      const merged = mergeResponse(result, stubResponse.headers, {
+        stubOverridesNonCookie: false,
+        mergeReqCtx: true,
       });
       responseHolder.response = merged;
       return merged;
@@ -548,21 +629,7 @@ export async function executeMiddleware<TEnv>(
   // set-cookie on an upgrade is not meaningful.
   const reqCtx = _getRequestContext();
   if (reqCtx && !isWebSocketUpgradeResponse(finalResponse)) {
-    const stubCookies = reqCtx.res.headers.getSetCookie();
-    if (stubCookies.length > 0) {
-      const existingCookies = new Set(finalResponse.headers.getSetCookie());
-      for (const cookie of stubCookies) {
-        if (!existingCookies.has(cookie)) {
-          finalResponse.headers.append("set-cookie", cookie);
-        }
-      }
-    }
-    // Fill in non-cookie headers that aren't already on the response
-    reqCtx.res.headers.forEach((value, name) => {
-      if (name !== "set-cookie" && !finalResponse.headers.has(name)) {
-        finalResponse.headers.set(name, value);
-      }
-    });
+    mergeReqCtxStub(finalResponse.headers, reqCtx);
   }
 
   return finalResponse;
@@ -573,7 +640,7 @@ export async function executeMiddleware<TEnv>(
  *
  * Intercepts use a shared stubResponse from the request context. This function:
  * - Runs middleware in sequence with a simple next() chain
- * - Returns Response if any middleware short-circuits (returns Response or redirects BEFORE next())
+ * - Returns Response if any middleware short-circuits (returns OR throws a Response, or redirects, BEFORE next())
  * - Returns null if all middleware calls next() - headers set after next() remain on stubResponse
  *
  * @param middlewares - Array of middleware functions
@@ -611,6 +678,7 @@ export async function executeInterceptMiddleware<TEnv>(
       return stubResponse;
     }
 
+    const ordinal = index;
     const middleware = middlewares[index++];
     const ctx = createMiddlewareContext(
       request,
@@ -632,7 +700,29 @@ export async function executeInterceptMiddleware<TEnv>(
       return next();
     };
 
-    const result = await middleware(ctx, guardedNext);
+    // Span-wrap each intercept middleware as rango.middleware (metric:false —
+    // intercept runs inside the render phase already metered by render:total, so
+    // it contributes a span but no separate perf metric). Bare MiddlewareFns
+    // have no pattern, so the label is scoped to "*" like a pattern-less entry.
+    const label = getMiddlewareMetricLabel(
+      { handler: middleware, pattern: null } as MiddlewareEntry<TEnv>,
+      ordinal,
+    );
+
+    let result: Response | void;
+    try {
+      result = await observePhase(PHASES.middleware(label), () =>
+        middleware(ctx, guardedNext),
+      );
+    } catch (error) {
+      // Thrown Response is short-circuit control flow, parity with the
+      // explicit-return path below. Real errors propagate.
+      if (error instanceof Response) {
+        result = error;
+      } else {
+        throw error;
+      }
+    }
 
     if (result instanceof Response) {
       earlyResponse = result;
@@ -656,21 +746,13 @@ export async function executeInterceptMiddleware<TEnv>(
     });
 
     if (hasStubHeaders) {
-      // Clone and merge headers from stub into early response.
-      // Only fill in missing headers — the returned Response's explicit
-      // headers take precedence, matching executeMiddleware behavior.
-      const mergedHeaders = new Headers(response.headers);
-      stubResponse.headers.forEach((value, name) => {
-        if (name.toLowerCase() === "set-cookie") {
-          mergedHeaders.append(name, value);
-        } else if (!mergedHeaders.has(name)) {
-          mergedHeaders.set(name, value);
-        }
-      });
-      return new Response(response.body, {
-        status: response.status,
-        statusText: response.statusText,
-        headers: mergedHeaders,
+      // Only fill in missing headers — the returned Response's explicit headers
+      // take precedence (stubOverridesNonCookie=false), matching executeMiddleware.
+      // mergeReqCtx=false: the intercept path deliberately does NOT merge reqCtx
+      // stub headers here (pinned by intercept-middleware-headers.test.ts).
+      return mergeResponse(response, stubResponse.headers, {
+        stubOverridesNonCookie: false,
+        mergeReqCtx: false,
       });
     }
     return response;
@@ -711,7 +793,6 @@ export async function executeLoaderMiddleware<TEnv>(
       regex: null,
       paramNames: [],
       handler,
-      mountPrefix: null,
     } as MiddlewareEntry<TEnv>,
     params,
   }));