blume 1.7.2 → 2.0.0

package/src/deploy/headers.ts

@@ -1,3 +1,4 @@
+import { crossOriginDiscoveryPaths } from "../ai/ai-catalog.ts";
 import {
   API_CATALOG_PATH,
   API_CATALOG_TYPE,
@@ -21,8 +22,9 @@ import type { ResolvedConfig } from "../core/schema.ts";
  * escape this because they carry `<meta charset>`; the raw endpoints have only
  * the HTTP header. Pinning `charset=utf-8` here matches the Content-Type these
  * same routes already send from the dev/server runtime (see
- * `astro/templates.ts`). Hosts that don't read `_headers` (Vercel, S3) ignore
- * the file harmlessly.
+ * `astro/templates.ts`). Hosts that don't read `_headers` ignore the file
+ * harmlessly; Vercel gets the same rules in `vercel.json` (see
+ * {@link buildVercelHeaders}).
  */
 
 /**
@@ -55,10 +57,30 @@ const HEADER_RULES: readonly {
   },
 ];
 
+/** Where the generated endpoint serves content-source images from. */
+const CONTENT_ASSETS_ROOT = "/blume-assets";
+
+/**
+ * The policy a content-source SVG is served with: an opaque origin with
+ * scripts off, so an uploaded SVG that carries a `<script>` stays inert when
+ * opened directly. The endpoint sends it from the server and dev runtimes;
+ * {@link headerRules} carries it to static hosts.
+ */
+export const SVG_ASSET_POLICY = "sandbox";
+
 /**
- * `_headers` contents: a `/*.<ext>` glob per rule with an indented
- * `Content-Type` line, in the two-space format Netlify and Cloudflare read. The
- * glob carries the served prefix (`{deployment.base}{basePath}` for the
+ * One static-host header rule: a served path (a `*` glob spanning path
+ * segments, or an exact path) and the header it sets there.
+ */
+export interface HeaderRule {
+  name: string;
+  path: string;
+  value: string;
+}
+
+/**
+ * The header rules every static host should apply, in `_headers` glob syntax.
+ * The glob carries the served prefix (`{deployment.base}{basePath}` for the
  * Markdown mirrors, `{deployment.base}` for the root `.txt` files) so the rules
  * still match once the site is mounted under a subpath (`/docs/*.md`); the
  * wildcard spans path segments, so a nested route like `/docs/ja/intro.md`
@@ -68,41 +90,126 @@ const HEADER_RULES: readonly {
  * exact-path rule for the root page advertises the agent-discovery resources —
  * the static-host counterpart of the Vercel routing-config injection.
  */
-export const buildNetlifyHeaders = (
+export const headerRules = (
   config: ResolvedConfig,
   homeLinkHeader?: string | null
-): string => {
-  const deployBase = normalizeBasePath(config.deployment.base);
-  const rules = HEADER_RULES.map((rule) => {
+): HeaderRule[] => {
+  const deployBase = normalizeBasePath(config.deployment.options.base);
+  const rules: HeaderRule[] = HEADER_RULES.map((rule) => {
     const prefix = rule.underBasePath
       ? `${deployBase}${config.basePath}`
       : deployBase;
-    return `${prefix}/*.${rule.ext}\n  Content-Type: ${rule.contentType}`;
+    return {
+      name: "Content-Type",
+      path: `${prefix}/*.${rule.ext}`,
+      value: rule.contentType,
+    };
   });
   if (homeLinkHeader) {
-    rules.push(`${deployBase}/\n  Link: ${homeLinkHeader}`);
+    rules.push({ name: "Link", path: `${deployBase}/`, value: homeLinkHeader });
   }
   // The well-known discovery files are extensionless, so without an explicit
   // rule a static host serves their registered media types as octet-stream or
   // text/plain.
   if (hasApiCatalog(config)) {
-    rules.push(
-      `${deployBase}${API_CATALOG_PATH}\n  Content-Type: ${API_CATALOG_TYPE}`
-    );
+    rules.push({
+      name: "Content-Type",
+      path: `${deployBase}${API_CATALOG_PATH}`,
+      value: API_CATALOG_TYPE,
+    });
   }
-  if (config.ai.webBotAuth.keys.length > 0) {
-    rules.push(
-      `${deployBase}${SIGNATURES_DIRECTORY_PATH}\n  Content-Type: ${SIGNATURES_DIRECTORY_TYPE}`
-    );
+  if (config.agents.webBotAuth.keys.length > 0) {
+    rules.push({
+      name: "Content-Type",
+      path: `${deployBase}${SIGNATURES_DIRECTORY_PATH}`,
+      value: SIGNATURES_DIRECTORY_TYPE,
+    });
+  }
+  // Agent registries fetch the discovery documents cross-origin; a static
+  // host sends no CORS header unless told to.
+  for (const path of crossOriginDiscoveryPaths(config)) {
+    rules.push({
+      name: "Access-Control-Allow-Origin",
+      path: `${deployBase}${path}`,
+      value: "*",
+    });
   }
   // Published skills live at the deployment base, outside `basePath` — the
   // `.md` charset rule above misses them whenever a basePath is set, and the
   // RFC wants archives served as application/gzip explicitly.
-  if (config.ai.skills) {
+  if (config.agents.skills) {
     rules.push(
-      `${deployBase}/.well-known/agent-skills/*.md\n  Content-Type: text/markdown; charset=utf-8`,
-      `${deployBase}/.well-known/agent-skills/*.tar.gz\n  Content-Type: application/gzip`
+      {
+        name: "Content-Type",
+        path: `${deployBase}/.well-known/agent-skills/*.md`,
+        value: "text/markdown; charset=utf-8",
+      },
+      {
+        name: "Content-Type",
+        path: `${deployBase}/.well-known/agent-skills/*.tar.gz`,
+        value: "application/gzip",
+      }
     );
   }
-  return `${rules.join("\n")}\n`;
+  // Images a content source downloads (a CMS upload, a remote page's figure)
+  // are served from the docs origin, and an SVG opened directly is a document
+  // that can run script. A sandbox policy keeps it inert; an `<img>` that
+  // embeds it is unaffected.
+  rules.push({
+    name: "Content-Security-Policy",
+    path: `${deployBase}${CONTENT_ASSETS_ROOT}/*.svg`,
+    value: SVG_ASSET_POLICY,
+  });
+  return rules;
 };
+
+/**
+ * `_headers` contents: each {@link headerRules} entry as its path with an
+ * indented header line, in the two-space format Netlify and Cloudflare read.
+ */
+export const buildNetlifyHeaders = (
+  config: ResolvedConfig,
+  homeLinkHeader?: string | null
+): string =>
+  `${headerRules(config, homeLinkHeader)
+    .map((rule) => `${rule.path}\n  ${rule.name}: ${rule.value}`)
+    .join("\n")}\n`;
+
+/** One entry of `vercel.json`'s `headers` array. */
+export interface VercelHeader {
+  headers: { key: string; value: string }[];
+  source: string;
+}
+
+/** The characters `path-to-regexp` would read as syntax in a literal path. */
+const SOURCE_SYNTAX = /[\\(){}?+:]/gu;
+
+/**
+ * A `_headers` path as a `vercel.json` `source`, which `path-to-regexp` reads:
+ * the literal parts are escaped, and a `*` glob becomes the `(.*)` group that
+ * spans path segments as the glob does. An exact directory path (the homepage
+ * rule's `/docs/`) drops its trailing slash, the URL Vercel serves it at.
+ */
+const vercelSource = (path: string): string => {
+  const trimmed =
+    path.length > 1 && path.endsWith("/") ? path.slice(0, -1) : path;
+  return trimmed
+    .split("*")
+    .map((part) => part.replaceAll(SOURCE_SYNTAX, String.raw`\$&`))
+    .join("(.*)");
+};
+
+/**
+ * The {@link headerRules} as `vercel.json` `headers` entries. Vercel reads no
+ * `_headers` file, so a static deploy of `dist/` carries the same rules here —
+ * the charset on the raw Markdown and text files, the homepage `Link` header,
+ * and the media types and CORS header of the discovery files.
+ */
+export const buildVercelHeaders = (
+  config: ResolvedConfig,
+  homeLinkHeader?: string | null
+): VercelHeader[] =>
+  headerRules(config, homeLinkHeader).map((rule) => ({
+    headers: [{ key: rule.name, value: rule.value }],
+    source: vercelSource(rule.path),
+  }));

package/src/deploy/node-headers.ts

@@ -0,0 +1,198 @@
+import { existsSync } from "node:fs";
+import { readFile, rename, writeFile } from "node:fs/promises";
+
+import { join } from "pathe";
+
+import { crossOriginDiscoveryPaths } from "../ai/ai-catalog.ts";
+import {
+  API_CATALOG_PATH,
+  API_CATALOG_TYPE,
+  hasApiCatalog,
+} from "../ai/api-catalog.ts";
+import {
+  SIGNATURES_DIRECTORY_PATH,
+  SIGNATURES_DIRECTORY_TYPE,
+} from "../ai/web-bot-auth.ts";
+import { normalizeBasePath } from "../core/base-path.ts";
+import type { BlumeProject } from "../core/project-graph.ts";
+import type { ResolvedConfig } from "../core/schema.ts";
+import { SVG_ASSET_POLICY } from "./headers.ts";
+import { distDir } from "./platforms/paths.ts";
+import type { BuildLog } from "./platforms/types.ts";
+
+/**
+ * The response headers the Node standalone server can't set itself. Its static
+ * handler (`send`) types a file by its extension alone and sends no CORS
+ * header, so the extensionless well-known files go out as
+ * `application/octet-stream` and a registry reading the discovery documents
+ * from another origin is blocked — the half of the `_headers` rules
+ * (`deploy/headers.ts`) that a Node server has no file to read from. The Markdown
+ * and text charsets need no rule here: `send` already adds `charset=UTF-8` to
+ * `text/*` types.
+ */
+
+/** The headers one exact served path (deployment base included) gets. */
+export interface NodeHeaderRule {
+  headers: Record<string, string>;
+  path: string;
+}
+
+/** The server entry `@astrojs/node` writes, which the wrapper takes over. */
+export const NODE_ENTRY_FILE = "entry.mjs";
+/** Where Astro's own entry moves, beside the wrapper that imports it. */
+export const NODE_ASTRO_ENTRY_FILE = "astro-entry.mjs";
+/** First line of the wrapper, so a second pass over the same build is a no-op. */
+const WRAPPER_MARKER = "// blume:node-headers";
+
+/** The header rules for this config, merged per path. */
+export const nodeHeaderRules = (config: ResolvedConfig): NodeHeaderRule[] => {
+  const deployBase = normalizeBasePath(config.deployment.options.base);
+  const byPath = new Map<string, Record<string, string>>();
+  const add = (path: string, name: string, value: string): void => {
+    byPath.set(path, { ...byPath.get(path), [name]: value });
+  };
+  if (hasApiCatalog(config)) {
+    add(API_CATALOG_PATH, "Content-Type", API_CATALOG_TYPE);
+  }
+  if (config.agents.webBotAuth.keys.length > 0) {
+    add(SIGNATURES_DIRECTORY_PATH, "Content-Type", SIGNATURES_DIRECTORY_TYPE);
+  }
+  for (const path of crossOriginDiscoveryPaths(config)) {
+    add(path, "Access-Control-Allow-Origin", "*");
+  }
+  return [...byPath].map(([path, headers]) => ({
+    headers,
+    path: `${deployBase}${path}`,
+  }));
+};
+
+/** Exact-path rules, as `[path, headers]` pairs for the wrapper's lookup. */
+const exactRules = (
+  rules: readonly NodeHeaderRule[]
+): [string, Record<string, string>][] =>
+  rules
+    .filter((rule) => !rule.path.includes("*"))
+    .map((rule) => [rule.path, rule.headers]);
+
+/**
+ * Glob rules (`/blume-assets/*.svg`), as `[prefix, suffix, headers]` — the
+ * text before and after the one `*`, which spans path segments.
+ */
+const patternRules = (
+  rules: readonly NodeHeaderRule[]
+): [string, string, Record<string, string>][] =>
+  rules
+    .filter((rule) => rule.path.includes("*"))
+    .map((rule) => {
+      const star = rule.path.indexOf("*");
+      return [
+        rule.path.slice(0, star),
+        rule.path.slice(star + 1),
+        rule.headers,
+      ];
+    });
+
+/**
+ * The entry that replaces Astro's: it stamps the rules' headers on the
+ * response, then hands the request to Astro's handler. `send` leaves a
+ * Content-Type that is already set alone, so the registered media type wins.
+ *
+ * It keeps Astro's contract — the `handler`, `options`, and `startServer`
+ * exports `astro preview` and a middleware-mode host import — and its
+ * autostart: Astro's entry is imported with autostart off (the env var is read
+ * when that module evaluates, so the import has to be dynamic), and the
+ * wrapper starts the server itself with its listener in front of Astro's.
+ */
+export const nodeEntryWrapper = (rules: readonly NodeHeaderRule[]): string =>
+  `${WRAPPER_MARKER}
+// Written by \`blume build\`: Astro's server entry is ./${NODE_ASTRO_ENTRY_FILE}.
+// This wrapper sets the media types and CORS headers of the .well-known
+// discovery files, which the standalone server's static handler can't, then
+// hands every request to Astro.
+const RULES = new Map(${JSON.stringify(exactRules(rules))});
+const PATTERNS = ${JSON.stringify(patternRules(rules))};
+const applyHeaders = (req, res) => {
+  const path = (req.url ?? "").split("?")[0];
+  const headers =
+    RULES.get(path) ??
+    PATTERNS.find(([prefix, suffix]) => path.startsWith(prefix) && path.endsWith(suffix))?.[2];
+  if (headers) {
+    for (const [name, value] of Object.entries(headers)) {
+      res.setHeader(name, value);
+    }
+  }
+};
+const previous = process.env.ASTRO_NODE_AUTOSTART;
+process.env.ASTRO_NODE_AUTOSTART = "disabled";
+const astro = await import("./${NODE_ASTRO_ENTRY_FILE}");
+if (previous === undefined) {
+  delete process.env.ASTRO_NODE_AUTOSTART;
+} else {
+  process.env.ASTRO_NODE_AUTOSTART = previous;
+}
+export const { options } = astro;
+export const handler = (req, res, ...rest) => {
+  applyHeaders(req, res);
+  return astro.handler(req, res, ...rest);
+};
+export const startServer = () => {
+  const started = astro.startServer();
+  // Stamp the headers only where the adapter exposes its HTTP server as
+  // expected; a different shape serves the files without them rather than
+  // failing to start.
+  const httpServer = started?.server?.server;
+  if (typeof httpServer?.prependListener === "function") {
+    httpServer.prependListener("request", applyHeaders);
+  }
+  return started;
+};
+if (options.mode === "standalone" && previous !== "disabled") {
+  startServer();
+}
+`;
+
+/**
+ * Put the header wrapper in front of a Node server build's entry. A build
+ * with no rule (no API catalog, AI catalog, MCP server, signatures directory,
+ * or downloaded content assets) leaves Astro's entry alone.
+ */
+export const wrapNodeEntry = async (
+  project: BlumeProject,
+  log: BuildLog
+): Promise<void> => {
+  // Content assets a CMS source downloaded ship as static files, which the
+  // standalone server serves without the SVG sandbox static hosts get from
+  // `_headers`; only a build that has them needs the rule.
+  const assetsDir = join(distDir(project.context), "client", "blume-assets");
+  const rules = [
+    ...nodeHeaderRules(project.config),
+    ...(existsSync(assetsDir)
+      ? [
+          {
+            headers: { "Content-Security-Policy": SVG_ASSET_POLICY },
+            path: `${normalizeBasePath(project.config.deployment.options.base)}/blume-assets/*.svg`,
+          },
+        ]
+      : []),
+  ];
+  if (rules.length === 0) {
+    return;
+  }
+  const serverDir = join(distDir(project.context), "server");
+  const entry = join(serverDir, NODE_ENTRY_FILE);
+  if (!existsSync(entry)) {
+    log.warn(
+      `Could not find the Node server entry at ${entry}, so the .well-known discovery files are served without their media types and CORS headers.`
+    );
+    return;
+  }
+  const source = await readFile(entry, "utf-8");
+  if (source.startsWith(WRAPPER_MARKER)) {
+    return;
+  }
+  await rename(entry, join(serverDir, NODE_ASTRO_ENTRY_FILE));
+  await writeFile(entry, nodeEntryWrapper(rules), "utf-8");
+  log.success(
+    "Wired the .well-known discovery headers into the Node server entry"
+  );
+};