@ascribed/astro 0.2.1-next.15 → 0.2.1-next.17

package/dist/head.js

@@ -0,0 +1,114 @@
+/**
+ * The tags of a page's head that are about the page, in order: the title and
+ * the description; the canonical link; the robots rule; the link to the
+ * Markdown version, when the build publishes one and the page is offered; the
+ * Open Graph tags a link preview reads, the card's among them when there is
+ * one; the breadcrumbs, as structured data, when there are two steps or more;
+ * and, on the page at the host's root, the site's name, as structured data.
+ * The title, the description, and the address are the record's, so the head
+ * can't disagree with the sitemap or `llms.txt`.
+ */
+export function headTags(record, options) {
+    const tags = [];
+    const meta = (attributes) => tags.push({ tag: "meta", attributes });
+    const link = (attributes) => tags.push({ tag: "link", attributes });
+    // A step's address is on the record's site when it has one.
+    const absolute = (route) => new URL(route, record.canonical ?? options.address).href;
+    const title = record.title === undefined
+        ? undefined
+        : options.title === undefined
+            ? record.title
+            : options.title(record.title, record);
+    if (title !== undefined)
+        tags.push({ tag: "title", text: title });
+    if (record.description !== undefined)
+        meta({ name: "description", content: record.description });
+    if (record.canonical !== undefined)
+        link({ rel: "canonical", href: record.canonical });
+    if (record.listing === "unlisted")
+        meta({ name: "robots", content: "noindex" });
+    if (record.listing === "unquoted")
+        meta({ name: "robots", content: "nosnippet" });
+    if (record.markdown !== undefined && record.listing !== "unlisted")
+        link({ rel: "alternate", type: "text/markdown", href: record.markdown });
+    meta({ property: "og:type", content: "website" });
+    if (options.name !== undefined)
+        meta({ property: "og:site_name", content: options.name });
+    if (title !== undefined)
+        meta({ property: "og:title", content: title });
+    if (record.description !== undefined)
+        meta({ property: "og:description", content: record.description });
+    if (record.canonical !== undefined)
+        meta({ property: "og:url", content: record.canonical });
+    const card = options.card;
+    if (card !== undefined) {
+        meta({ property: "og:image", content: new URL(card.src, options.address).href });
+        if (card.width !== undefined)
+            meta({ property: "og:image:width", content: String(card.width) });
+        if (card.height !== undefined)
+            meta({ property: "og:image:height", content: String(card.height) });
+        meta({ property: "og:image:alt", content: card.alt });
+    }
+    // Google reads a trail of two steps or more.
+    const steps = record.breadcrumbs ?? [];
+    if (steps.length >= 2) {
+        tags.push({
+            tag: "script",
+            json: {
+                "@context": "https://schema.org",
+                "@type": "BreadcrumbList",
+                itemListElement: steps.map((step, at) => ({
+                    "@type": "ListItem",
+                    position: at + 1,
+                    name: step.title,
+                    item: absolute(step.route),
+                })),
+            },
+        });
+    }
+    // The site's name, on the home page alone: the host's root.
+    if (options.name !== undefined && record.route === "/") {
+        tags.push({
+            tag: "script",
+            json: {
+                "@context": "https://schema.org",
+                "@type": "WebSite",
+                name: options.name,
+                url: absolute("/"),
+            },
+        });
+    }
+    return tags;
+}
+/**
+ * JSON for a `<script>` element: `<` written as `\u003c`, so no string in it
+ * can close the element.
+ */
+export function scriptJson(json) {
+    return JSON.stringify(json).replaceAll("<", "\\u003c");
+}
+/** The tags `Head.astro` writes that a head should have once, each with how to find it. */
+const ONCE = [
+    { name: "<title>", pattern: /<title[\s>]/gi },
+    ...["description", "robots"].map((name) => ({
+        name: `<meta name="${name}">`,
+        pattern: new RegExp(`<meta\\s[^>]*name=["']?${name}["'\\s/>]`, "gi"),
+    })),
+    { name: '<link rel="canonical">', pattern: /<link\s[^>]*rel=["']?canonical["'\s/>]/gi },
+    ...["og:type", "og:site_name", "og:title", "og:description", "og:url"].map((name) => ({
+        name: `<meta property="${name}">`,
+        pattern: new RegExp(`<meta\\s[^>]*property=["']?${name}["'\\s/>]`, "gi"),
+    })),
+];
+/**
+ * The tags that a page's `<head>` has more than once, of those `Head.astro`
+ * writes, named as `<title>` or `<meta property="og:title">` are. A layout
+ * that includes the component and writes one of them too has two.
+ */
+export function repeatedHeadTags(html) {
+    const head = /<head[\s>][\s\S]*?<\/head>/i.exec(html)?.[0];
+    if (head === undefined)
+        return [];
+    return ONCE.filter(({ pattern }) => (head.match(pattern) ?? []).length > 1).map(({ name }) => name);
+}
+//# sourceMappingURL=head.js.map

package/dist/head.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"head.js","sourceRoot":"","sources":["../src/head.ts"],"names":[],"mappings":"AAiCA;;;;;;;;;GASG;AACH,MAAM,UAAU,QAAQ,CAAC,MAAkB,EAAE,OAAoB;IAC/D,MAAM,IAAI,GAAc,EAAE,CAAC;IAC3B,MAAM,IAAI,GAAG,CAAC,UAAkC,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC,CAAC;IAC5F,MAAM,IAAI,GAAG,CAAC,UAAkC,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC,CAAC;IAC5F,4DAA4D;IAC5D,MAAM,QAAQ,GAAG,CAAC,KAAa,EAAE,EAAE,CAAC,IAAI,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,SAAS,IAAI,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC;IAE7F,MAAM,KAAK,GACT,MAAM,CAAC,KAAK,KAAK,SAAS;QACxB,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,OAAO,CAAC,KAAK,KAAK,SAAS;YAC3B,CAAC,CAAC,MAAM,CAAC,KAAK;YACd,CAAC,CAAC,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IAC5C,IAAI,KAAK,KAAK,SAAS;QAAE,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;IAClE,IAAI,MAAM,CAAC,WAAW,KAAK,SAAS;QAAE,IAAI,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,OAAO,EAAE,MAAM,CAAC,WAAW,EAAE,CAAC,CAAC;IACjG,IAAI,MAAM,CAAC,SAAS,KAAK,SAAS;QAAE,IAAI,CAAC,EAAE,GAAG,EAAE,WAAW,EAAE,IAAI,EAAE,MAAM,CAAC,SAAS,EAAE,CAAC,CAAC;IACvF,IAAI,MAAM,CAAC,OAAO,KAAK,UAAU;QAAE,IAAI,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC,CAAC;IAChF,IAAI,MAAM,CAAC,OAAO,KAAK,UAAU;QAAE,IAAI,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC,CAAC;IAClF,IAAI,MAAM,CAAC,QAAQ,KAAK,SAAS,IAAI,MAAM,CAAC,OAAO,KAAK,UAAU;QAChE,IAAI,CAAC,EAAE,GAAG,EAAE,WAAW,EAAE,IAAI,EAAE,eAAe,EAAE,IAAI,EAAE,MAAM,CAAC,QAAQ,EAAE,CAAC,CAAC;IAE3E,IAAI,CAAC,EAAE,QAAQ,EAAE,SAAS,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC,CAAC;IAClD,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS;QAAE,IAAI,CAAC,EAAE,QAAQ,EAAE,cAAc,EAAE,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;IAC1F,IAAI,KAAK,KAAK,SAAS;QAAE,IAAI,CAAC,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC,CAAC;IACxE,IAAI,MAAM,CAAC,WAAW,KAAK,SAAS;QAClC,IAAI,CAAC,EAAE,QAAQ,EAAE,gBAAgB,EAAE,OAAO,EAAE,MAAM,CAAC,WAAW,EAAE,CAAC,CAAC;IACpE,IAAI,MAAM,CAAC,SAAS,KAAK,SAAS;QAAE,IAAI,CAAC,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,CAAC,SAAS,EAAE,CAAC,CAAC;IAC5F,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC;IAC1B,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACvB,IAAI,CAAC,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;QACjF,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS;YAAE,IAAI,CAAC,EAAE,QAAQ,EAAE,gBAAgB,EAAE,OAAO,EAAE,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;QAChG,IAAI,IAAI,CAAC,MAAM,KAAK,SAAS;YAC3B,IAAI,CAAC,EAAE,QAAQ,EAAE,iBAAiB,EAAE,OAAO,EAAE,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;QACtE,IAAI,CAAC,EAAE,QAAQ,EAAE,cAAc,EAAE,OAAO,EAAE,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC;IACxD,CAAC;IAED,6CAA6C;IAC7C,MAAM,KAAK,GAAG,MAAM,CAAC,WAAW,IAAI,EAAE,CAAC;IACvC,IAAI,KAAK,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;QACtB,IAAI,CAAC,IAAI,CAAC;YACR,GAAG,EAAE,QAAQ;YACb,IAAI,EAAE;gBACJ,UAAU,EAAE,oBAAoB;gBAChC,OAAO,EAAE,gBAAgB;gBACzB,eAAe,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;oBACxC,OAAO,EAAE,UAAU;oBACnB,QAAQ,EAAE,EAAE,GAAG,CAAC;oBAChB,IAAI,EAAE,IAAI,CAAC,KAAK;oBAChB,IAAI,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC;iBAC3B,CAAC,CAAC;aACJ;SACF,CAAC,CAAC;IACL,CAAC;IACD,4DAA4D;IAC5D,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,MAAM,CAAC,KAAK,KAAK,GAAG,EAAE,CAAC;QACvD,IAAI,CAAC,IAAI,CAAC;YACR,GAAG,EAAE,QAAQ;YACb,IAAI,EAAE;gBACJ,UAAU,EAAE,oBAAoB;gBAChC,OAAO,EAAE,SAAS;gBAClB,IAAI,EAAE,OAAO,CAAC,IAAI;gBAClB,GAAG,EAAE,QAAQ,CAAC,GAAG,CAAC;aACnB;SACF,CAAC,CAAC;IACL,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,UAAU,CAAC,IAAa;IACtC,OAAO,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,GAAG,EAAE,SAAS,CAAC,CAAC;AACzD,CAAC;AAED,2FAA2F;AAC3F,MAAM,IAAI,GAAwC;IAChD,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,eAAe,EAAE;IAC7C,GAAG,CAAC,aAAa,EAAE,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QAC1C,IAAI,EAAE,eAAe,IAAI,IAAI;QAC7B,OAAO,EAAE,IAAI,MAAM,CAAC,0BAA0B,IAAI,WAAW,EAAE,IAAI,CAAC;KACrE,CAAC,CAAC;IACH,EAAE,IAAI,EAAE,wBAAwB,EAAE,OAAO,EAAE,0CAA0C,EAAE;IACvF,GAAG,CAAC,SAAS,EAAE,cAAc,EAAE,UAAU,EAAE,gBAAgB,EAAE,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QACpF,IAAI,EAAE,mBAAmB,IAAI,IAAI;QACjC,OAAO,EAAE,IAAI,MAAM,CAAC,8BAA8B,IAAI,WAAW,EAAE,IAAI,CAAC;KACzE,CAAC,CAAC;CACJ,CAAC;AAEF;;;;GAIG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC3C,MAAM,IAAI,GAAG,6BAA6B,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;IAC3D,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IAClC,OAAO,IAAI,CAAC,MAAM,CAAC,CAAC,EAAE,OAAO,EAAE,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,GAAG,CAC7E,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC,IAAI,CACnB,CAAC;AACJ,CAAC","sourcesContent":["// What a page's head says about the page, from its record: the tags\n// `Head.astro` writes, as data, so each can be tested against the record. And\n// the check `astro dev` makes that a layout doesn't write one of them again.\nimport type { PageRecord } from \"./records.js\";\n\n/** A page's social card: its image, what it shows, and its size in pixels. */\nexport interface Card {\n  /** The image's address: absolute, or from the host's root, resolved against the site's address. */\n  src: string;\n  /** What the image shows, for a reader who can't see it. */\n  alt: string;\n  width?: number;\n  height?: number;\n}\n\n/** What a site says about itself, which a record doesn't hold. */\nexport interface HeadOptions {\n  /** The whole title, from the page's title and its record. Default: the page's title. */\n  title?: ((title: string, record: PageRecord) => string) | undefined;\n  /** The site's name, for `og:site_name` and, on the page at the host's root, the site-name structured data. */\n  name?: string | undefined;\n  /** The page's social card. Without one, no image tags. */\n  card?: Card | undefined;\n  /** The site's address (Astro's `site`, or the request's), for addresses the record doesn't make absolute. */\n  address: string | URL;\n}\n\n/** One tag of the head. */\nexport type HeadTag =\n  | { tag: \"title\"; text: string }\n  | { tag: \"meta\" | \"link\"; attributes: Record<string, string> }\n  | { tag: \"script\"; json: unknown };\n\n/**\n * The tags of a page's head that are about the page, in order: the title and\n * the description; the canonical link; the robots rule; the link to the\n * Markdown version, when the build publishes one and the page is offered; the\n * Open Graph tags a link preview reads, the card's among them when there is\n * one; the breadcrumbs, as structured data, when there are two steps or more;\n * and, on the page at the host's root, the site's name, as structured data.\n * The title, the description, and the address are the record's, so the head\n * can't disagree with the sitemap or `llms.txt`.\n */\nexport function headTags(record: PageRecord, options: HeadOptions): HeadTag[] {\n  const tags: HeadTag[] = [];\n  const meta = (attributes: Record<string, string>) => tags.push({ tag: \"meta\", attributes });\n  const link = (attributes: Record<string, string>) => tags.push({ tag: \"link\", attributes });\n  // A step's address is on the record's site when it has one.\n  const absolute = (route: string) => new URL(route, record.canonical ?? options.address).href;\n\n  const title =\n    record.title === undefined\n      ? undefined\n      : options.title === undefined\n        ? record.title\n        : options.title(record.title, record);\n  if (title !== undefined) tags.push({ tag: \"title\", text: title });\n  if (record.description !== undefined) meta({ name: \"description\", content: record.description });\n  if (record.canonical !== undefined) link({ rel: \"canonical\", href: record.canonical });\n  if (record.listing === \"unlisted\") meta({ name: \"robots\", content: \"noindex\" });\n  if (record.listing === \"unquoted\") meta({ name: \"robots\", content: \"nosnippet\" });\n  if (record.markdown !== undefined && record.listing !== \"unlisted\")\n    link({ rel: \"alternate\", type: \"text/markdown\", href: record.markdown });\n\n  meta({ property: \"og:type\", content: \"website\" });\n  if (options.name !== undefined) meta({ property: \"og:site_name\", content: options.name });\n  if (title !== undefined) meta({ property: \"og:title\", content: title });\n  if (record.description !== undefined)\n    meta({ property: \"og:description\", content: record.description });\n  if (record.canonical !== undefined) meta({ property: \"og:url\", content: record.canonical });\n  const card = options.card;\n  if (card !== undefined) {\n    meta({ property: \"og:image\", content: new URL(card.src, options.address).href });\n    if (card.width !== undefined) meta({ property: \"og:image:width\", content: String(card.width) });\n    if (card.height !== undefined)\n      meta({ property: \"og:image:height\", content: String(card.height) });\n    meta({ property: \"og:image:alt\", content: card.alt });\n  }\n\n  // Google reads a trail of two steps or more.\n  const steps = record.breadcrumbs ?? [];\n  if (steps.length >= 2) {\n    tags.push({\n      tag: \"script\",\n      json: {\n        \"@context\": \"https://schema.org\",\n        \"@type\": \"BreadcrumbList\",\n        itemListElement: steps.map((step, at) => ({\n          \"@type\": \"ListItem\",\n          position: at + 1,\n          name: step.title,\n          item: absolute(step.route),\n        })),\n      },\n    });\n  }\n  // The site's name, on the home page alone: the host's root.\n  if (options.name !== undefined && record.route === \"/\") {\n    tags.push({\n      tag: \"script\",\n      json: {\n        \"@context\": \"https://schema.org\",\n        \"@type\": \"WebSite\",\n        name: options.name,\n        url: absolute(\"/\"),\n      },\n    });\n  }\n  return tags;\n}\n\n/**\n * JSON for a `<script>` element: `<` written as `\\u003c`, so no string in it\n * can close the element.\n */\nexport function scriptJson(json: unknown): string {\n  return JSON.stringify(json).replaceAll(\"<\", \"\\\\u003c\");\n}\n\n/** The tags `Head.astro` writes that a head should have once, each with how to find it. */\nconst ONCE: { name: string; pattern: RegExp }[] = [\n  { name: \"<title>\", pattern: /<title[\\s>]/gi },\n  ...[\"description\", \"robots\"].map((name) => ({\n    name: `<meta name=\"${name}\">`,\n    pattern: new RegExp(`<meta\\\\s[^>]*name=[\"']?${name}[\"'\\\\s/>]`, \"gi\"),\n  })),\n  { name: '<link rel=\"canonical\">', pattern: /<link\\s[^>]*rel=[\"']?canonical[\"'\\s/>]/gi },\n  ...[\"og:type\", \"og:site_name\", \"og:title\", \"og:description\", \"og:url\"].map((name) => ({\n    name: `<meta property=\"${name}\">`,\n    pattern: new RegExp(`<meta\\\\s[^>]*property=[\"']?${name}[\"'\\\\s/>]`, \"gi\"),\n  })),\n];\n\n/**\n * The tags that a page's `<head>` has more than once, of those `Head.astro`\n * writes, named as `<title>` or `<meta property=\"og:title\">` are. A layout\n * that includes the component and writes one of them too has two.\n */\nexport function repeatedHeadTags(html: string): string[] {\n  const head = /<head[\\s>][\\s\\S]*?<\\/head>/i.exec(html)?.[0];\n  if (head === undefined) return [];\n  return ONCE.filter(({ pattern }) => (head.match(pattern) ?? []).length > 1).map(\n    ({ name }) => name,\n  );\n}\n"]}

package/dist/headers.d.ts

@@ -0,0 +1,50 @@
+/** The headers a host sends with one path, matched exactly. */
+export interface HeaderRule {
+    /** The path, from the host's root. */
+    path: string;
+    /** Each header's value, by its name. */
+    headers: Record<string, string>;
+}
+/** The rules in a build's site output, or `undefined` when it has no list. */
+export declare function readHeaders(siteRoot: string): HeaderRule[] | undefined;
+/**
+ * The headers the rules in the build whose site output is at `siteRoot` give
+ * `pathname`, a path from the host's root: none when no rule names it. Read
+ * again when the file changes, as it does with each build in `astro dev`.
+ */
+export declare function headersFor(siteRoot: string, pathname: string): Record<string, string>;
+/**
+ * Every project's rules as one list, sorted by path: a path two projects
+ * both have keeps the first's rule, as the first entry's files win.
+ */
+export declare function mergeRules(projects: readonly (readonly HeaderRule[])[]): HeaderRule[];
+/** The rules in Netlify's `_headers` format: each path, then its headers indented below it. */
+export declare function netlifyHeaders(rules: readonly HeaderRule[]): string;
+/**
+ * The rules, each with only the headers the site's own `_headers` doesn't
+ * give its path, leaving out a rule it has whole. Empty when it has every
+ * line.
+ */
+export declare function missingRules(own: string, rules: readonly HeaderRule[]): HeaderRule[];
+/** What the build does about the host's headers. */
+export type HeadersOutcome = {
+    kind: "none";
+} | {
+    kind: "write";
+    text: string;
+} | {
+    kind: "note";
+    text: string;
+} | {
+    kind: "fail";
+    text: string;
+};
+/** What the build does about `_headers`, given every project's rules and the site. */
+export declare function headersOutcome(sources: {
+    /** The rules of every project that sets `[consumer] headers = "netlify"`. */
+    rules: readonly HeaderRule[];
+    /** Astro's `base`. */
+    base: string;
+    /** The site's own `public/_headers`, when it has one. */
+    own: string | undefined;
+}): HeadersOutcome;

package/dist/headers.js

@@ -0,0 +1,155 @@
+// The response headers a build's files need from the host:
+// `_ascribe/headers.json` in its site output (output-layout contract §1.4).
+// A Markdown version has no head, so its page's rules are headers, which only
+// the host sends. With `[consumer] headers = "netlify"`, `astro build` writes
+// them into the site's `_headers`, Netlify's format; `astro dev` sends them
+// with each Markdown version whatever the setting, so a site can be tried
+// with no host.
+//
+// A site's own `public/_headers` is never rewritten: the build checks it has
+// the lines, and fails with the lines to add when it doesn't.
+import { readFileSync, statSync } from "node:fs";
+import path from "node:path";
+import { normalizeBase } from "./project.js";
+/** The format version of `_ascribe/headers.json` this reads. */
+const VERSION = 1;
+function headersFile(siteRoot) {
+    return path.join(siteRoot, "_ascribe", "headers.json");
+}
+/** The rules in a build's site output, or `undefined` when it has no list. */
+export function readHeaders(siteRoot) {
+    const file = headersFile(siteRoot);
+    let text;
+    try {
+        text = readFileSync(file, "utf8");
+    }
+    catch {
+        return undefined;
+    }
+    const list = JSON.parse(text);
+    if (list.version !== VERSION) {
+        throw new Error(`@ascribed/astro: ${file} is version ${String(list.version)} of the headers list, and this version of the integration reads version ${VERSION}. Update @ascribed/astro and @ascribed/cli together.`);
+    }
+    return list.headers;
+}
+/** A build's rules by path, as last read, and when its file was written. */
+const byPath = new Map();
+/**
+ * The headers the rules in the build whose site output is at `siteRoot` give
+ * `pathname`, a path from the host's root: none when no rule names it. Read
+ * again when the file changes, as it does with each build in `astro dev`.
+ */
+export function headersFor(siteRoot, pathname) {
+    let written;
+    try {
+        written = statSync(headersFile(siteRoot)).mtimeMs;
+    }
+    catch {
+        return {};
+    }
+    let known = byPath.get(siteRoot);
+    if (known === undefined || known.written !== written) {
+        const rules = new Map();
+        for (const rule of readHeaders(siteRoot) ?? []) {
+            if (!rules.has(rule.path))
+                rules.set(rule.path, rule.headers);
+        }
+        known = { written, rules };
+        byPath.set(siteRoot, known);
+    }
+    return known.rules.get(pathname) ?? {};
+}
+/**
+ * Every project's rules as one list, sorted by path: a path two projects
+ * both have keeps the first's rule, as the first entry's files win.
+ */
+export function mergeRules(projects) {
+    const rules = new Map();
+    for (const list of projects) {
+        for (const rule of list)
+            if (!rules.has(rule.path))
+                rules.set(rule.path, rule);
+    }
+    return [...rules.values()].sort((a, b) => (a.path < b.path ? -1 : a.path > b.path ? 1 : 0));
+}
+/** The rules in Netlify's `_headers` format: each path, then its headers indented below it. */
+export function netlifyHeaders(rules) {
+    return rules
+        .map((rule) => `${rule.path}\n${Object.entries(rule.headers)
+        .map(([name, value]) => `  ${name}: ${value}\n`)
+        .join("")}`)
+        .join("");
+}
+/**
+ * The headers `_headers` gives each path exactly, as Netlify reads the file:
+ * a line that isn't indented is a path, an indented `Name: value` line is a
+ * header of the path above it, and a line starting with `#` is a comment.
+ * Names are compared in lower case, as Netlify compares them. A path with a
+ * wildcard or a placeholder is kept as written, and matches only itself here.
+ */
+function parseHeaders(text) {
+    const paths = new Map();
+    let current;
+    for (const line of text.split(/\r?\n/)) {
+        const trimmed = line.trim();
+        if (trimmed === "" || trimmed.startsWith("#"))
+            continue;
+        if (!/^\s/.test(line)) {
+            current = paths.get(trimmed) ?? new Set();
+            paths.set(trimmed, current);
+            continue;
+        }
+        const colon = trimmed.indexOf(":");
+        if (current === undefined || colon === -1)
+            continue;
+        current.add(headerKey(trimmed.slice(0, colon), trimmed.slice(colon + 1)));
+    }
+    return paths;
+}
+function headerKey(name, value) {
+    return `${name.trim().toLowerCase()}: ${value.trim()}`;
+}
+/**
+ * The rules, each with only the headers the site's own `_headers` doesn't
+ * give its path, leaving out a rule it has whole. Empty when it has every
+ * line.
+ */
+export function missingRules(own, rules) {
+    const has = parseHeaders(own);
+    const missing = [];
+    for (const rule of rules) {
+        const given = has.get(rule.path);
+        const headers = Object.fromEntries(Object.entries(rule.headers).filter(([name, value]) => !given?.has(headerKey(name, value))));
+        if (Object.keys(headers).length > 0)
+            missing.push({ path: rule.path, headers });
+    }
+    return missing;
+}
+/** What the build does about `_headers`, given every project's rules and the site. */
+export function headersOutcome(sources) {
+    if (sources.rules.length === 0)
+        return { kind: "none" };
+    const text = netlifyHeaders(sources.rules);
+    const base = normalizeBase(sources.base);
+    if (base !== "/") {
+        return {
+            kind: "note",
+            text: `no _headers: Astro's \`base\` is "${base}", and Netlify reads _headers only at the root of what's deployed. Put these lines in the host's own:\n${text}`,
+        };
+    }
+    if (sources.own !== undefined) {
+        const missing = missingRules(sources.own, sources.rules);
+        if (missing.length === 0) {
+            return {
+                kind: "note",
+                text: "no _headers: public/_headers is the site's own, and has Ascribe's lines.",
+            };
+        }
+        return {
+            kind: "fail",
+            text: `public/_headers is the site's own, and lacks the headers Ascribe's pages need. Add these lines to it:\n${netlifyHeaders(missing)}`,
+        };
+    }
+    return { kind: "write", text };
+}
+//# sourceMappingURL=headers.js.map

package/dist/headers.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"headers.js","sourceRoot":"","sources":["../src/headers.ts"],"names":[],"mappings":"AAAA,2DAA2D;AAC3D,4EAA4E;AAC5E,8EAA8E;AAC9E,8EAA8E;AAC9E,4EAA4E;AAC5E,0EAA0E;AAC1E,gBAAgB;AAChB,EAAE;AACF,6EAA6E;AAC7E,8DAA8D;AAC9D,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,SAAS,CAAC;AACjD,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,aAAa,EAAE,MAAM,cAAc,CAAC;AAE7C,gEAAgE;AAChE,MAAM,OAAO,GAAG,CAAC,CAAC;AAgBlB,SAAS,WAAW,CAAC,QAAgB;IACnC,OAAO,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,UAAU,EAAE,cAAc,CAAC,CAAC;AACzD,CAAC;AAED,8EAA8E;AAC9E,MAAM,UAAU,WAAW,CAAC,QAAgB;IAC1C,MAAM,IAAI,GAAG,WAAW,CAAC,QAAQ,CAAC,CAAC;IACnC,IAAI,IAAY,CAAC;IACjB,IAAI,CAAC;QACH,IAAI,GAAG,YAAY,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;IACpC,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,SAAS,CAAC;IACnB,CAAC;IACD,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAe,CAAC;IAC5C,IAAI,IAAI,CAAC,OAAO,KAAK,OAAO,EAAE,CAAC;QAC7B,MAAM,IAAI,KAAK,CACb,oBAAoB,IAAI,eAAe,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,2EAA2E,OAAO,sDAAsD,CACpM,CAAC;IACJ,CAAC;IACD,OAAO,IAAI,CAAC,OAAO,CAAC;AACtB,CAAC;AAED,4EAA4E;AAC5E,MAAM,MAAM,GAAG,IAAI,GAAG,EAA2E,CAAC;AAElG;;;;GAIG;AACH,MAAM,UAAU,UAAU,CAAC,QAAgB,EAAE,QAAgB;IAC3D,IAAI,OAAe,CAAC;IACpB,IAAI,CAAC;QACH,OAAO,GAAG,QAAQ,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC;IACpD,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,IAAI,KAAK,GAAG,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IACjC,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,OAAO,KAAK,OAAO,EAAE,CAAC;QACrD,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkC,CAAC;QACxD,KAAK,MAAM,IAAI,IAAI,WAAW,CAAC,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;YAC/C,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC;gBAAE,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QAChE,CAAC;QACD,KAAK,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC;QAC3B,MAAM,CAAC,GAAG,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;IAC9B,CAAC;IACD,OAAO,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC;AACzC,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,UAAU,CAAC,QAA4C;IACrE,MAAM,KAAK,GAAG,IAAI,GAAG,EAAsB,CAAC;IAC5C,KAAK,MAAM,IAAI,IAAI,QAAQ,EAAE,CAAC;QAC5B,KAAK,MAAM,IAAI,IAAI,IAAI;YAAE,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC;gBAAE,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IACjF,CAAC;IACD,OAAO,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AAC9F,CAAC;AAED,+FAA+F;AAC/F,MAAM,UAAU,cAAc,CAAC,KAA4B;IACzD,OAAO,KAAK;SACT,GAAG,CACF,CAAC,IAAI,EAAE,EAAE,CACP,GAAG,IAAI,CAAC,IAAI,KAAK,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC;SAC1C,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,KAAK,IAAI,KAAK,KAAK,IAAI,CAAC;SAC/C,IAAI,CAAC,EAAE,CAAC,EAAE,CAChB;SACA,IAAI,CAAC,EAAE,CAAC,CAAC;AACd,CAAC;AAED;;;;;;GAMG;AACH,SAAS,YAAY,CAAC,IAAY;IAChC,MAAM,KAAK,GAAG,IAAI,GAAG,EAAuB,CAAC;IAC7C,IAAI,OAAgC,CAAC;IACrC,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;QACvC,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5B,IAAI,OAAO,KAAK,EAAE,IAAI,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC;YAAE,SAAS;QACxD,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACtB,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,GAAG,EAAE,CAAC;YAC1C,KAAK,CAAC,GAAG,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;YAC5B,SAAS;QACX,CAAC;QACD,MAAM,KAAK,GAAG,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,OAAO,KAAK,SAAS,IAAI,KAAK,KAAK,CAAC,CAAC;YAAE,SAAS;QACpD,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,EAAE,OAAO,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAC5E,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,SAAS,CAAC,IAAY,EAAE,KAAa;IAC5C,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,KAAK,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;AACzD,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAC,GAAW,EAAE,KAA4B;IACpE,MAAM,GAAG,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;IAC9B,MAAM,OAAO,GAAiB,EAAE,CAAC;IACjC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,KAAK,GAAG,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACjC,MAAM,OAAO,GAAG,MAAM,CAAC,WAAW,CAChC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,EAAE,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC,CAC5F,CAAC;QACF,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,OAAO,EAAE,CAAC,CAAC;IAClF,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC;AASD,sFAAsF;AACtF,MAAM,UAAU,cAAc,CAAC,OAO9B;IACC,IAAI,OAAO,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC;IACxD,MAAM,IAAI,GAAG,cAAc,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;IAC3C,MAAM,IAAI,GAAG,aAAa,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IACzC,IAAI,IAAI,KAAK,GAAG,EAAE,CAAC;QACjB,OAAO;YACL,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,qCAAqC,IAAI,0GAA0G,IAAI,EAAE;SAChK,CAAC;IACJ,CAAC;IACD,IAAI,OAAO,CAAC,GAAG,KAAK,SAAS,EAAE,CAAC;QAC9B,MAAM,OAAO,GAAG,YAAY,CAAC,OAAO,CAAC,GAAG,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC;QACzD,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACzB,OAAO;gBACL,IAAI,EAAE,MAAM;gBACZ,IAAI,EAAE,0EAA0E;aACjF,CAAC;QACJ,CAAC;QACD,OAAO;YACL,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,0GAA0G,cAAc,CAAC,OAAO,CAAC,EAAE;SAC1I,CAAC;IACJ,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC;AACjC,CAAC","sourcesContent":["// The response headers a build's files need from the host:\n// `_ascribe/headers.json` in its site output (output-layout contract §1.4).\n// A Markdown version has no head, so its page's rules are headers, which only\n// the host sends. With `[consumer] headers = \"netlify\"`, `astro build` writes\n// them into the site's `_headers`, Netlify's format; `astro dev` sends them\n// with each Markdown version whatever the setting, so a site can be tried\n// with no host.\n//\n// A site's own `public/_headers` is never rewritten: the build checks it has\n// the lines, and fails with the lines to add when it doesn't.\nimport { readFileSync, statSync } from \"node:fs\";\nimport path from \"node:path\";\nimport { normalizeBase } from \"./project.js\";\n\n/** The format version of `_ascribe/headers.json` this reads. */\nconst VERSION = 1;\n\n/** The headers a host sends with one path, matched exactly. */\nexport interface HeaderRule {\n  /** The path, from the host's root. */\n  path: string;\n  /** Each header's value, by its name. */\n  headers: Record<string, string>;\n}\n\n/** `_ascribe/headers.json`. */\ninterface HeaderList {\n  version: number;\n  headers: HeaderRule[];\n}\n\nfunction headersFile(siteRoot: string): string {\n  return path.join(siteRoot, \"_ascribe\", \"headers.json\");\n}\n\n/** The rules in a build's site output, or `undefined` when it has no list. */\nexport function readHeaders(siteRoot: string): HeaderRule[] | undefined {\n  const file = headersFile(siteRoot);\n  let text: string;\n  try {\n    text = readFileSync(file, \"utf8\");\n  } catch {\n    return undefined;\n  }\n  const list = JSON.parse(text) as HeaderList;\n  if (list.version !== VERSION) {\n    throw new Error(\n      `@ascribed/astro: ${file} is version ${String(list.version)} of the headers list, and this version of the integration reads version ${VERSION}. Update @ascribed/astro and @ascribed/cli together.`,\n    );\n  }\n  return list.headers;\n}\n\n/** A build's rules by path, as last read, and when its file was written. */\nconst byPath = new Map<string, { written: number; rules: Map<string, Record<string, string>> }>();\n\n/**\n * The headers the rules in the build whose site output is at `siteRoot` give\n * `pathname`, a path from the host's root: none when no rule names it. Read\n * again when the file changes, as it does with each build in `astro dev`.\n */\nexport function headersFor(siteRoot: string, pathname: string): Record<string, string> {\n  let written: number;\n  try {\n    written = statSync(headersFile(siteRoot)).mtimeMs;\n  } catch {\n    return {};\n  }\n  let known = byPath.get(siteRoot);\n  if (known === undefined || known.written !== written) {\n    const rules = new Map<string, Record<string, string>>();\n    for (const rule of readHeaders(siteRoot) ?? []) {\n      if (!rules.has(rule.path)) rules.set(rule.path, rule.headers);\n    }\n    known = { written, rules };\n    byPath.set(siteRoot, known);\n  }\n  return known.rules.get(pathname) ?? {};\n}\n\n/**\n * Every project's rules as one list, sorted by path: a path two projects\n * both have keeps the first's rule, as the first entry's files win.\n */\nexport function mergeRules(projects: readonly (readonly HeaderRule[])[]): HeaderRule[] {\n  const rules = new Map<string, HeaderRule>();\n  for (const list of projects) {\n    for (const rule of list) if (!rules.has(rule.path)) rules.set(rule.path, rule);\n  }\n  return [...rules.values()].sort((a, b) => (a.path < b.path ? -1 : a.path > b.path ? 1 : 0));\n}\n\n/** The rules in Netlify's `_headers` format: each path, then its headers indented below it. */\nexport function netlifyHeaders(rules: readonly HeaderRule[]): string {\n  return rules\n    .map(\n      (rule) =>\n        `${rule.path}\\n${Object.entries(rule.headers)\n          .map(([name, value]) => `  ${name}: ${value}\\n`)\n          .join(\"\")}`,\n    )\n    .join(\"\");\n}\n\n/**\n * The headers `_headers` gives each path exactly, as Netlify reads the file:\n * a line that isn't indented is a path, an indented `Name: value` line is a\n * header of the path above it, and a line starting with `#` is a comment.\n * Names are compared in lower case, as Netlify compares them. A path with a\n * wildcard or a placeholder is kept as written, and matches only itself here.\n */\nfunction parseHeaders(text: string): Map<string, Set<string>> {\n  const paths = new Map<string, Set<string>>();\n  let current: Set<string> | undefined;\n  for (const line of text.split(/\\r?\\n/)) {\n    const trimmed = line.trim();\n    if (trimmed === \"\" || trimmed.startsWith(\"#\")) continue;\n    if (!/^\\s/.test(line)) {\n      current = paths.get(trimmed) ?? new Set();\n      paths.set(trimmed, current);\n      continue;\n    }\n    const colon = trimmed.indexOf(\":\");\n    if (current === undefined || colon === -1) continue;\n    current.add(headerKey(trimmed.slice(0, colon), trimmed.slice(colon + 1)));\n  }\n  return paths;\n}\n\nfunction headerKey(name: string, value: string): string {\n  return `${name.trim().toLowerCase()}: ${value.trim()}`;\n}\n\n/**\n * The rules, each with only the headers the site's own `_headers` doesn't\n * give its path, leaving out a rule it has whole. Empty when it has every\n * line.\n */\nexport function missingRules(own: string, rules: readonly HeaderRule[]): HeaderRule[] {\n  const has = parseHeaders(own);\n  const missing: HeaderRule[] = [];\n  for (const rule of rules) {\n    const given = has.get(rule.path);\n    const headers = Object.fromEntries(\n      Object.entries(rule.headers).filter(([name, value]) => !given?.has(headerKey(name, value))),\n    );\n    if (Object.keys(headers).length > 0) missing.push({ path: rule.path, headers });\n  }\n  return missing;\n}\n\n/** What the build does about the host's headers. */\nexport type HeadersOutcome =\n  | { kind: \"none\" }\n  | { kind: \"write\"; text: string }\n  | { kind: \"note\"; text: string }\n  | { kind: \"fail\"; text: string };\n\n/** What the build does about `_headers`, given every project's rules and the site. */\nexport function headersOutcome(sources: {\n  /** The rules of every project that sets `[consumer] headers = \"netlify\"`. */\n  rules: readonly HeaderRule[];\n  /** Astro's `base`. */\n  base: string;\n  /** The site's own `public/_headers`, when it has one. */\n  own: string | undefined;\n}): HeadersOutcome {\n  if (sources.rules.length === 0) return { kind: \"none\" };\n  const text = netlifyHeaders(sources.rules);\n  const base = normalizeBase(sources.base);\n  if (base !== \"/\") {\n    return {\n      kind: \"note\",\n      text: `no _headers: Astro's \\`base\\` is \"${base}\", and Netlify reads _headers only at the root of what's deployed. Put these lines in the host's own:\\n${text}`,\n    };\n  }\n  if (sources.own !== undefined) {\n    const missing = missingRules(sources.own, sources.rules);\n    if (missing.length === 0) {\n      return {\n        kind: \"note\",\n        text: \"no _headers: public/_headers is the site's own, and has Ascribe's lines.\",\n      };\n    }\n    return {\n      kind: \"fail\",\n      text: `public/_headers is the site's own, and lacks the headers Ascribe's pages need. Add these lines to it:\\n${netlifyHeaders(missing)}`,\n    };\n  }\n  return { kind: \"write\", text };\n}\n"]}

package/dist/index.d.ts

@@ -30,6 +30,15 @@ interface SharedOptions {
      * `<figure class="code-title">` with a `<figcaption>`. Default `true`.
      */
     codeTitles?: boolean;
+    /**
+     * The site's one sitemap, at `<base>sitemap.xml`, of every listed page of
+     * every project and every page of the site's own. `exclude` lists routes,
+     * from the host's root, of the site's own pages to leave out of it, such
+     * as one that carries `noindex`.
+     */
+    sitemap?: {
+        exclude?: string[];
+    };
 }
 /** Options of the integration with one project. */
 export interface OneProjectOptions extends SharedOptions {
@@ -51,6 +60,8 @@ export type AscribeOptions = OneProjectOptions | ProjectsOptions;
 /**
  * Runs `ascribe build` for each project, checks the site's routing, adds the markdown plugin, adds a
  * page for each permalink's address, and serves published files, and, with `[consumer] agents = true`,
- * `llms.txt` and each page's Markdown, each at its project's base path.
+ * `llms.txt` and each page's Markdown, each at its project's base path; and writes the site's one
+ * sitemap and `robots.txt` from every project's page records, and, with `[consumer] headers =
+ * "netlify"`, its `_headers` from their headers lists.
  */
 export default function ascribe(options: AscribeOptions): AstroIntegration;

package/dist/index.js

@@ -4,7 +4,8 @@
 //   // astro.config.mjs
 //   import ascribe from "@ascribed/astro";
 //   export default defineConfig({
-//     base: "/docs", trailingSlash: "never",     // as in ascribe.toml's [consumer]
+//     // `site`, `base`, and `trailingSlash` come from ascribe.toml's
+//     // [consumer] when they're left out here; set here, they must agree.
 //     integrations: [ascribe({ build: "site" })],
 //   });
 //
@@ -19,6 +20,11 @@
 //   // a layout
 //   import Elements from "@ascribed/astro/Elements.astro";           // <Elements /> in <head>
 //   import Availability from "@ascribed/astro/Availability.astro";   // the page's availability
+//   import Head from "@ascribed/astro/Head.astro";                   // <Head entry={entry} /> in <head>
+//
+//   // the sidebar: src/nav.ts's list, checked against the collection
+//   import { sidebar } from "@ascribed/astro/content";
+//   const groups = sidebar(nav, await getCollection("docs"));
 //
 //   // src/pages/404.astro
 //   import MissingPermalink from "@ascribed/astro/MissingPermalink.astro"; // a permalink that isn't here
@@ -33,22 +39,27 @@
 //     review: "docs",
 //   })
 //   ascribeCollection({ schema, name: "docs" })
-import { mkdirSync } from "node:fs";
+import { existsSync, mkdirSync, readFileSync } from "node:fs";
+import { writeFile } from "node:fs/promises";
 import path from "node:path";
 import { fileURLToPath } from "node:url";
 import { findBinary } from "./binary.js";
 import { codeTitles } from "./code-titles.js";
 import { SeenFiles, watchDev } from "./dev.js";
 import { agentsMiddleware, copyAgentFiles, copyPublishedFiles, filesMiddleware, folderUnder, } from "./files.js";
-import { hasPermalinks, permalinkPattern, readPermalinks } from "./permalinks.js";
-import { consumerMismatches, homeRoute, readProject } from "./project.js";
+import { WARN as HEAD_WARN } from "./head-check.js";
+import { headersFor, headersOutcome, mergeRules, readHeaders } from "./headers.js";
+import { forwardingRoutes, hasPermalinks, permalinkPattern, readPermalinks } from "./permalinks.js";
+import { consumerMismatches, homeRoute, normalizeBase, readProject, supplyRouting, } from "./project.js";
 import { basePathClashes, nameProblems, pagesUnderOtherPaths, permalinkProblems, } from "./projects.js";
+import { pageList, readRecords, routeAs, routeKey, sitemapRoute, } from "./records.js";
 import rehypeAscribeAttributes from "./rehype.js";
 import { removeDevFile, siteUrl, writeDevFile } from "./review/devfile.js";
 import { runDiff, runPrompt } from "./review/diff.js";
 import { APP_ID } from "./review/protocol.js";
 import { readRoutes } from "./review/routes.js";
 import { channelProblem, ReviewServer } from "./review/server.js";
+import { devRouted, devSitePages, rootFiles, rootFilesMiddleware, } from "./root-files.js";
 import { anchorsFor, hasWarnings, reviewFor, runBuild } from "./run.js";
 import { satteriAscribeAttributes } from "./satteri.js";
 import { copySchema } from "./schema.js";
@@ -65,7 +76,9 @@ function fail(problems, what) {
 /**
  * Runs `ascribe build` for each project, checks the site's routing, adds the markdown plugin, adds a
  * page for each permalink's address, and serves published files, and, with `[consumer] agents = true`,
- * `llms.txt` and each page's Markdown, each at its project's base path.
+ * `llms.txt` and each page's Markdown, each at its project's base path; and writes the site's one
+ * sitemap and `robots.txt` from every project's page records, and, with `[consumer] headers =
+ * "netlify"`, its `_headers` from their headers lists.
  */
 export default function ascribe(options) {
     /** With `projects`; otherwise the one entry has no name a collection gives. */
@@ -80,11 +93,47 @@ export default function ascribe(options) {
     let devCommand = false;
     let devHttps = false;
     let review;
+    /** The site's routes, as Astro resolved them, for what `astro dev` serves of the site's own. */
+    let astroRoutes = [];
+    /** The site's `public/` folder, where its own `robots.txt` would be. */
+    let publicDir = "";
+    /** What the sitemap and `robots.txt` are drawn from, given the site's pages and its routes. */
+    const rootSources = (sitePages, routed) => {
+        const base = routing?.base ?? "/";
+        const records = entries.map((entry) => readRecords(entry.siteRoot));
+        // Entries of one project share its `ascribe.toml`, and so what it says to crawlers.
+        const robots = new Map();
+        for (const [i, entry] of entries.entries()) {
+            if (!robots.has(entry.project.dir))
+                robots.set(entry.project.dir, {
+                    name: entry.name,
+                    robots: records[i]?.robots ?? { lines: [] },
+                });
+        }
+        const root = normalizeBase(base);
+        return {
+            projects: records.map(pageList),
+            routed,
+            sitePages,
+            excluded: [
+                ...forwardingRoutes(entries.map((entry) => readPermalinks(entry.siteRoot))),
+                `${root}404`,
+                `${root}500`,
+                ...(options.sitemap?.exclude ?? []),
+            ],
+            robots: [...robots.values()],
+            site: routing?.site,
+            base,
+            ownRobots: existsSync(path.join(publicDir, "robots.txt")),
+            ownSitemap: existsSync(path.join(publicDir, "sitemap.xml")),
+        };
+    };
     return {
         name: "@ascribed/astro",
         hooks: {
-            "astro:config:setup": async ({ config, command, logger, updateConfig, addDevToolbarApp, createCodegenDir, injectRoute, }) => {
+            "astro:config:setup": async ({ config, command, logger, updateConfig, addDevToolbarApp, createCodegenDir, injectRoute, addMiddleware, }) => {
                 const root = fileURLToPath(config.root);
+                publicDir = fileURLToPath(config.publicDir);
                 const list = options.projects ?? [
                     {
                         name: "",
@@ -107,8 +156,11 @@ export default function ascribe(options) {
                         schemaDir: named ? path.join(codegen, entry.name) : codegen,
                     };
                 });
+                // What astro.config leaves unset comes from [consumer]: `site` and `trailingSlash` from the
+                // first project that has them, and with one project, `base` from its base path. With
+                // `projects`, `base` stays the site's, the root every base path is under.
+                const { routing: astro, supplied } = supplyRouting(entries.map((entry) => entry.project), { base: config.base, trailingSlash: config.trailingSlash, site: config.site }, !named);
                 // `ascribe.toml`'s routing must be Astro's, or every link Ascribe writes is wrong.
-                const astro = { base: config.base, trailingSlash: config.trailingSlash, site: config.site };
                 for (const entry of entries) {
                     fail(consumerMismatches(entry.project, astro), named
                         ? `"${entry.name}"'s ascribe.toml and astro.config disagree`
@@ -210,21 +262,30 @@ export default function ascribe(options) {
                         await buildEntry(entry);
                     checkShared();
                     devCommand = command === "dev";
+                    if (devCommand) {
+                        // A tag of the head written twice, by Head.astro and the layout, is warned about.
+                        globalThis[HEAD_WARN] = (message) => logger.warn(message);
+                        addMiddleware({
+                            entrypoint: new URL("./head-check.js", import.meta.url),
+                            order: "post",
+                        });
+                    }
                     // Each permalink's address, a page that forwards to its heading. One route serves every
                     // list, at the path of the first that names anything.
                     const lists = entries.flatMap((entry) => readPermalinks(entry.siteRoot) ?? []);
                     const served = lists.find(hasPermalinks) ?? lists[0];
                     if (served !== undefined) {
                         injectRoute({
-                            pattern: permalinkPattern(served.path, config.base),
+                            pattern: permalinkPattern(served.path, astro.base),
                             entrypoint: new URL("../src/Permalink.astro", import.meta.url),
                             prerender: true,
                         });
                     }
                 }
-                const projectHome = (entry) => homeRoute(entry.project.consumer.basePath, config.trailingSlash);
+                const projectHome = (entry) => homeRoute(entry.project.consumer.basePath, astro.trailingSlash);
                 // Dev serving of `_ascribe/files/`, and of the plain output for agents; the build copies them in `astro:build:done`.
                 updateConfig({
+                    ...supplied,
                     ...(options.codeTitles === false
                         ? {}
                         : { markdown: { shikiConfig: { transformers: [codeTitles] } } }),
@@ -237,9 +298,15 @@ export default function ascribe(options) {
                                 resolveId: (id) => (id === SITE_MODULE ? `\0${SITE_MODULE}` : undefined),
                                 load: (id) => id === `\0${SITE_MODULE}`
                                     ? `export const siteRoot = ${JSON.stringify(entries[0]?.siteRoot ?? "")};\n` +
-                                        `export const home = ${JSON.stringify(homeRoute(config.base, config.trailingSlash))};\n` +
+                                        `export const astroRoot = ${JSON.stringify(root)};\n` +
+                                        `export const home = ${JSON.stringify(homeRoute(astro.base, astro.trailingSlash))};\n` +
                                         `export const collections = ${JSON.stringify(named ? Object.fromEntries(entries.map((e) => [e.name, e.siteRoot])) : {})};\n` +
-                                        `export const permalinks = ${JSON.stringify(entries.map((e) => ({ siteRoot: e.siteRoot, home: projectHome(e) })))};`
+                                        `export const permalinks = ${JSON.stringify(entries.map((e) => ({ siteRoot: e.siteRoot, home: projectHome(e) })))};\n` +
+                                        `export const routes = ${JSON.stringify(entries.map((e) => ({
+                                            siteRoot: e.siteRoot,
+                                            basePath: e.project.consumer.basePath,
+                                            trailingSlash: e.project.consumer.trailingSlash,
+                                        })))};`
                                     : undefined,
                             },
                             {
@@ -248,15 +315,22 @@ export default function ascribe(options) {
                                     for (const entry of entries) {
                                         const base = entry.project.consumer.basePath;
                                         server.middlewares.use(filesMiddleware(entry.siteRoot, base));
-                                        if (entry.agentsRoot !== undefined)
-                                            server.middlewares.use(agentsMiddleware(entry.agentsRoot, base));
+                                        if (entry.agentsRoot !== undefined) {
+                                            const siteRoot = entry.siteRoot;
+                                            server.middlewares.use(agentsMiddleware(entry.agentsRoot, base, (pathname) => headersFor(siteRoot, pathname)));
+                                        }
                                     }
+                                    // The sitemap and robots.txt, as the build would write them now.
+                                    server.middlewares.use(rootFilesMiddleware(astro.base, () => rootFiles(rootSources(devSitePages(astroRoutes, astro), devRouted(astroRoutes, astro)))));
                                 },
                             },
                         ],
                     },
                 });
             },
+            "astro:routes:resolved": ({ routes }) => {
+                astroRoutes = routes;
+            },
             "astro:server:setup": ({ server, refreshContent, logger, toolbar }) => {
                 if (!devCommand || !routing)
                     return;
@@ -331,8 +405,44 @@ export default function ascribe(options) {
                     for (const group of groups)
                         removeDevFile(group.project.dir, process.pid);
             },
-            "astro:build:done": async ({ dir, logger }) => {
+            "astro:build:done": async ({ dir, pages, logger }) => {
                 const base = routing?.base ?? "/";
+                const trailingSlash = routing?.trailingSlash ?? "ignore";
+                // Every page Astro built, by its route from the host's root. Astro lists
+                // pages only, not endpoints such as a feed, so a dot in a page's last
+                // segment, as in `/v1.2`, is still a page.
+                const built = pages.map((page) => routeAs(`${normalizeBase(base)}${page.pathname.replace(/^\//, "")}`, trailingSlash));
+                const builtKeys = new Set(built.map(routeKey));
+                const files = rootFiles(rootSources(built, (route) => builtKeys.has(routeKey(route))));
+                fail(files.problems, "the site's robots.txt");
+                for (const note of files.notes)
+                    logger[note.level](note.text);
+                const out = fileURLToPath(dir);
+                if (files.sitemap !== undefined) {
+                    await writeFile(path.join(out, "sitemap.xml"), files.sitemap);
+                    logger.info(`wrote ${sitemapRoute(base)}`);
+                }
+                if (files.robots !== undefined) {
+                    await writeFile(path.join(out, "robots.txt"), files.robots);
+                    logger.info("wrote /robots.txt");
+                }
+                // The host's headers, from the projects that name its format.
+                const ownHeaders = path.join(publicDir, "_headers");
+                const headers = headersOutcome({
+                    rules: mergeRules(entries
+                        .filter((entry) => entry.project.consumer.headers === "netlify")
+                        .map((entry) => readHeaders(entry.siteRoot) ?? [])),
+                    base,
+                    own: existsSync(ownHeaders) ? readFileSync(ownHeaders, "utf8") : undefined,
+                });
+                if (headers.kind === "fail")
+                    throw new Error(`@ascribed/astro: ${headers.text}`);
+                if (headers.kind === "note")
+                    logger.info(headers.text);
+                if (headers.kind === "write") {
+                    await writeFile(path.join(out, "_headers"), headers.text);
+                    logger.info("wrote /_headers");
+                }
                 // Entries of one project publish at one base path: the first entry's copies go last, so they win.
                 for (const entry of [...entries].reverse()) {
                     const folder = folderUnder(entry.project.consumer.basePath, base);