@anthusai/papyrus 2.0.0-next.7 → 2.0.0-next.70

package/amplify/plugins/reader-steering/schema.ts

@@ -0,0 +1,206 @@
+import { a } from "@aws-amplify/backend";
+import { readerActions } from "./function/resource";
+
+/**
+ * Staff and IAM only: no guest, API key or reader rule on any model. The reader-actions function writes the
+ * tables directly with its own role, and editors may list them; admins may delete. `id` is the lookup key:
+ * the siteSubject for a profile, the lowercase handle for a claim, `subjectKey#readerId#dimension` for a vote (`subjectKey#readerId#kind` for a flag; the key is the bare id for items and submissions and `type:id` for the rest) and
+ * `sub-<sha256 of the canonical URL>` for a submission (so a canonical URL is one submission).
+ */
+const staffOnly = (allow: any) => [allow.group("editor").to(["read"]), allow.group("admin").to(["read", "delete"])];
+
+export function readerSteeringSchemaDefinitions() {
+  return {
+    ReaderProfile: a
+      .model({
+        id: a.id().required(),
+        siteSubject: a.string().required(),
+        readerId: a.string().required(),
+        handle: a.string(),
+        state: a.string().required(),
+        karma: a.integer(),
+      })
+      .disableOperations(["subscriptions"])
+      .secondaryIndexes((index) => [index("readerId").name("profilesByReader").queryField("listReaderProfilesByReader")])
+      .authorization(staffOnly),
+    ReaderHandle: a
+      .model({
+        id: a.id().required(),
+        readerId: a.string().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .authorization(staffOnly),
+    ReaderVote: a
+      .model({
+        id: a.id().required(),
+        subjectKey: a.string().required(),
+        subjectType: a.string(),
+        readerId: a.string().required(),
+        dimension: a.string().required(),
+        value: a.integer().required(),
+        voterRole: a.string().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .authorization(staffOnly),
+    ReaderFlag: a
+      .model({
+        id: a.id().required(),
+        subjectKey: a.string().required(),
+        subjectType: a.string().required(),
+        readerId: a.string().required(),
+        kind: a.string().required(),
+        reason: a.string().required(),
+        targetKey: a.string(),
+        suggestedName: a.string(),
+        voterRole: a.string().required(),
+        createdAt: a.datetime().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .authorization(staffOnly),
+    ReaderComment: a
+      .model({
+        id: a.id().required(),
+        subjectKey: a.string().required(),
+        readerId: a.string().required(),
+        body: a.string().required(),
+        parentId: a.string(),
+        ratingContext: a.json(),
+        state: a.string().required(),
+        signals: a.json(),
+        moderation: a.string().required(),
+        reportCount: a.integer(),
+        reviewedAt: a.datetime(),
+        editedAt: a.datetime(),
+        deletedAt: a.datetime(),
+        createdAt: a.datetime().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .secondaryIndexes((index) => [index("moderation").sortKeys(["createdAt"]).name("commentsByModeration").queryField("listReaderCommentsByModerationAndCreatedAt")])
+      .authorization(staffOnly),
+    ReaderSubmission: a
+      .model({
+        id: a.id().required(),
+        canonicalUrl: a.string().required(),
+        domain: a.string().required(),
+        title: a.string().required(),
+        kind: a.string().required(),
+        note: a.string(),
+        readerId: a.string().required(),
+        state: a.string().required(),
+        similar: a.string().array(),
+        signals: a.json(),
+        extractionState: a.string().required(),
+        itemId: a.string(),
+        moderation: a.string().required(),
+        reportCount: a.integer(),
+        reviewedAt: a.datetime(),
+        createdAt: a.datetime().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .secondaryIndexes((index) => [
+        index("domain").sortKeys(["createdAt"]).name("submissionsByDomain").queryField("listReaderSubmissionsByDomainAndCreatedAt"),
+        index("moderation").sortKeys(["createdAt"]).name("submissionsByModeration").queryField("listReaderSubmissionsByModerationAndCreatedAt"),
+      ])
+      .authorization(staffOnly),
+    ReaderReport: a
+      .model({
+        id: a.id().required(),
+        targetType: a.string().required(),
+        target: a.string().required(),
+        reporterReaderId: a.string().required(),
+        reason: a.string().required(),
+        state: a.string().required(),
+        closedAt: a.datetime(),
+        createdAt: a.datetime().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .secondaryIndexes((index) => [index("target").sortKeys(["createdAt"]).name("reportsByTarget").queryField("listReaderReportsByTargetAndCreatedAt")])
+      .authorization(staffOnly),
+    ModerationAction: a
+      .model({
+        id: a.id().required(),
+        by: a.string().required(),
+        actorSub: a.string().required(),
+        action: a.string().required(),
+        targetType: a.string().required(),
+        target: a.string().required(),
+        reason: a.string(),
+        fromState: a.string().required(),
+        toState: a.string().required(),
+        fromModeration: a.string(),
+        toModeration: a.string(),
+        undoOf: a.string(),
+        undoneBy: a.string(),
+        at: a.datetime().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .secondaryIndexes((index) => [index("target").sortKeys(["at"]).name("actionsByTarget").queryField("listModerationActionsByTargetAndAt")])
+      .authorization(staffOnly),
+    Proposal: a
+      .model({
+        id: a.id().required(),
+        kind: a.string().required(),
+        title: a.string().required(),
+        summary: a.string(),
+        body: a.string().required(),
+        targets: a.json().required(),
+        evidence: a.json(),
+        proposer: a.string().required(),
+        state: a.string().required(),
+        history: a.json().required(),
+        communityResult: a.json(),
+        createdAt: a.datetime().required(),
+      })
+      .disableOperations(["subscriptions"])
+      .secondaryIndexes((index) => [index("state").sortKeys(["createdAt"]).name("proposalsByState").queryField("listProposalsByStateAndCreatedAt")])
+      .authorization(staffOnly),
+    ProposalReceipt: a.customType({
+      status: a.string().required(),
+      id: a.string(),
+      error: a.string(),
+      message: a.string(),
+    }),
+    fileProposal: a
+      .mutation()
+      .arguments({
+        kind: a.string().required(),
+        title: a.string().required(),
+        body: a.string().required(),
+        summary: a.string(),
+        targets: a.json(),
+        evidence: a.json(),
+      })
+      .returns(a.ref("ProposalReceipt"))
+      .authorization((allow) => [allow.group("editor"), allow.group("admin")])
+      .handler(a.handler.function(readerActions)),
+    transitionProposal: a
+      .mutation()
+      .arguments({
+        proposalId: a.string().required(),
+        state: a.string().required(),
+        reason: a.string(),
+        result: a.string(),
+      })
+      .returns(a.ref("ProposalReceipt"))
+      .authorization((allow) => [allow.group("editor"), allow.group("admin")])
+      .handler(a.handler.function(readerActions)),
+    ModerationReceipt: a.customType({
+      status: a.string().required(),
+      id: a.string(),
+      error: a.string(),
+      message: a.string(),
+    }),
+    moderateReader: a
+      .mutation()
+      .arguments({
+        action: a.string().required(),
+        targetType: a.string().required(),
+        target: a.string().required(),
+        reason: a.string(),
+        undoOf: a.string(),
+      })
+      .returns(a.ref("ModerationReceipt"))
+      .authorization((allow) => [allow.group("editor"), allow.group("admin")])
+      .handler(a.handler.function(readerActions)),
+  };
+}

package/amplify/plugins/registry.js

@@ -1,8 +1,10 @@
-import { editorialChecksPlugin } from "./editorial-checks/papyrus.plugin.js";
+import { newsletterPlugin } from "./newsletter/papyrus.plugin.js";
+import { readerSteeringPlugin } from "./reader-steering/papyrus.plugin.js";
 import { signupPlugin } from "./signup/papyrus.plugin.js";
 /** Plugins are packaged with Papyrus and pinned by its version: there is no registry to fetch from. */
 export const KNOWN_PLUGINS = {
-    "editorial-checks": editorialChecksPlugin,
+    newsletter: newsletterPlugin,
+    "reader-steering": readerSteeringPlugin,
     signup: signupPlugin,
 };
 export function resolvePlugins(requests) {

package/amplify/plugins/registry.ts

@@ -1,10 +1,12 @@
 import type { PapyrusPlugin, PluginRequest } from "./types";
-import { editorialChecksPlugin } from "./editorial-checks/papyrus.plugin";
+import { newsletterPlugin } from "./newsletter/papyrus.plugin";
+import { readerSteeringPlugin } from "./reader-steering/papyrus.plugin";
 import { signupPlugin } from "./signup/papyrus.plugin";
 
 /** Plugins are packaged with Papyrus and pinned by its version: there is no registry to fetch from. */
 export const KNOWN_PLUGINS: Record<string, PapyrusPlugin<any, any>> = {
-  "editorial-checks": editorialChecksPlugin,
+  newsletter: newsletterPlugin,
+  "reader-steering": readerSteeringPlugin,
   signup: signupPlugin,
 };
 

package/amplify/plugins/types.ts

@@ -1,3 +1,4 @@
+import type { Function as LambdaFunction } from "aws-cdk-lib/aws-lambda";
 import type { Construct } from "constructs";
 
 /** How a site asks for a plugin in `papyrus.config.ts`: `plugins: ["signup"]` or `plugins: [{ name: "signup", forms: ["news"] }]`. */
@@ -9,6 +10,8 @@ export type PluginManifest = {
   /** Public contract version. Bump it when a mutation input, a model field, an element name or an attribute changes. No compatibility shims. */
   contract: number;
   models: string[];
+  /** Custom queries it adds (operations that read). */
+  queries?: string[];
   mutations: string[];
   /** What an anonymous visitor may call. Everything else is closed to guests. */
   guest: { mutations: string[] };
@@ -22,6 +25,8 @@ export type PluginWiringContext = {
   createStack: (name: string) => Construct;
   graphqlApiId: string;
   graphqlEndpoint: string;
+  /** The site's `content-actions` function: it runs the projection tasks a plugin may add (for example `reader-snapshot`). */
+  contentActionsLambda: LambdaFunction;
 };
 
 export type PapyrusPlugin<Config, Functions = unknown> = {

package/amplify/plugins/wire.js

@@ -12,6 +12,7 @@ export function wirePlugins(backend, plugins, contentActionsLambda) {
         createStack: (name) => backend.createStack(name),
         graphqlApiId: graphqlApi.apiId,
         graphqlEndpoint: backend.data.resources.cfnResources.cfnGraphqlApi.attrGraphQlUrl,
+        contentActionsLambda,
     };
     for (const { plugin, config } of plugins)
         plugin.wire(backend, config, context);

package/amplify/plugins/wire.ts

@@ -15,6 +15,7 @@ export function wirePlugins(backend: any, plugins: ResolvedPlugin[], contentActi
     createStack: (name: string) => backend.createStack(name),
     graphqlApiId: graphqlApi.apiId as string,
     graphqlEndpoint: backend.data.resources.cfnResources.cfnGraphqlApi.attrGraphQlUrl as string,
+    contentActionsLambda,
   };
   for (const { plugin, config } of plugins) plugin.wire(backend, config, context);
   contentActionsLambda.addEnvironment(PLUGINS_ENVIRONMENT_VARIABLE, plugins.map(({ plugin }) => plugin.manifest.name).join(","));

package/amplify/publication/projection-lease.js

@@ -0,0 +1,30 @@
+import { Stack } from "aws-cdk-lib";
+import { Effect, PolicyStatement } from "aws-cdk-lib/aws-iam";
+/** The shared lock table's name (created by the cloud knowledge run stack). The Lambda finds it through this SSM parameter. */
+export const PROJECTION_LOCK_TABLE_PARAMETER = "/papyrus/projection-lock/table-name";
+export const PROJECTION_LOCK_TABLE_NAME = "papyrus-projection-locks";
+export const PROJECTION_LOCK_TABLE_ACTIONS = ["dynamodb:PutItem", "dynamodb:DeleteItem", "dynamodb:UpdateItem", "dynamodb:GetItem"];
+export function projectionLeaseParameterArn(region, account) {
+    return `arn:aws:ssm:${region}:${account}:parameter${PROJECTION_LOCK_TABLE_PARAMETER}`;
+}
+export function projectionLeaseTableArn(region, account) {
+    return `arn:aws:dynamodb:${region}:${account}:table/${PROJECTION_LOCK_TABLE_NAME}*`;
+}
+/**
+ * Let a projecting Lambda take and release the per-publication projection lease: read the table-name
+ * parameter and write items in the lock table, nothing broader. The table itself is not created here;
+ * a deployment without it keeps projecting without a lease.
+ */
+export function addProjectionLeaseAccess(lambda) {
+    const stack = Stack.of(lambda);
+    lambda.addToRolePolicy(new PolicyStatement({
+        effect: Effect.ALLOW,
+        actions: ["ssm:GetParameter"],
+        resources: [projectionLeaseParameterArn(stack.region, stack.account)],
+    }));
+    lambda.addToRolePolicy(new PolicyStatement({
+        effect: Effect.ALLOW,
+        actions: PROJECTION_LOCK_TABLE_ACTIONS,
+        resources: [projectionLeaseTableArn(stack.region, stack.account)],
+    }));
+}

package/amplify/publication/projection-lease.ts

@@ -0,0 +1,39 @@
+import { Stack } from "aws-cdk-lib";
+import { Effect, PolicyStatement } from "aws-cdk-lib/aws-iam";
+import type { IFunction } from "aws-cdk-lib/aws-lambda";
+
+/** The shared lock table's name (created by the cloud knowledge run stack). The Lambda finds it through this SSM parameter. */
+export const PROJECTION_LOCK_TABLE_PARAMETER = "/papyrus/projection-lock/table-name";
+export const PROJECTION_LOCK_TABLE_NAME = "papyrus-projection-locks";
+export const PROJECTION_LOCK_TABLE_ACTIONS = ["dynamodb:PutItem", "dynamodb:DeleteItem", "dynamodb:UpdateItem", "dynamodb:GetItem"];
+
+export function projectionLeaseParameterArn(region: string, account: string): string {
+  return `arn:aws:ssm:${region}:${account}:parameter${PROJECTION_LOCK_TABLE_PARAMETER}`;
+}
+
+export function projectionLeaseTableArn(region: string, account: string): string {
+  return `arn:aws:dynamodb:${region}:${account}:table/${PROJECTION_LOCK_TABLE_NAME}*`;
+}
+
+/**
+ * Let a projecting Lambda take and release the per-publication projection lease: read the table-name
+ * parameter and write items in the lock table, nothing broader. The table itself is not created here;
+ * a deployment without it keeps projecting without a lease.
+ */
+export function addProjectionLeaseAccess(lambda: IFunction): void {
+  const stack = Stack.of(lambda);
+  lambda.addToRolePolicy(
+    new PolicyStatement({
+      effect: Effect.ALLOW,
+      actions: ["ssm:GetParameter"],
+      resources: [projectionLeaseParameterArn(stack.region, stack.account)],
+    }),
+  );
+  lambda.addToRolePolicy(
+    new PolicyStatement({
+      effect: Effect.ALLOW,
+      actions: PROJECTION_LOCK_TABLE_ACTIONS,
+      resources: [projectionLeaseTableArn(stack.region, stack.account)],
+    }),
+  );
+}

package/amplify/publication/redirect-and-index.js

@@ -12,6 +12,9 @@ async function lookUpRedirectTarget(uri) {
 }
 
 function rewriteToIndexDocument(uri) {
+  if (uri === "/404") {
+    return "/404.html";
+  }
   if (uri.endsWith("/")) {
     return uri + "index.html";
   }
@@ -22,6 +25,88 @@ function rewriteToIndexDocument(uri) {
   return uri;
 }
 
+// Pattern redirects: data from the site's projection.json, spliced between the markers at deploy
+// time by redirect-rules.ts. Each rule is {p: regex, t: target with $1..$9, s: status, x: excluded paths}.
+const patternRules = /*PAPYRUS_PATTERN_RULES*/[]/*END_PAPYRUS_PATTERN_RULES*/;
+
+function lookUpPatternRedirect(uri) {
+  for (let index = 0; index < patternRules.length; index += 1) {
+    const rule = patternRules[index];
+    if (rule.x.indexOf(uri) !== -1) {
+      continue;
+    }
+    const match = new RegExp(rule.p).exec(uri);
+    if (match !== null) {
+      return {
+        status: rule.s,
+        target: rule.t.replace(/\$([0-9])/g, function (whole, group) {
+          return match[Number(group)] || "";
+        }),
+      };
+    }
+  }
+  return null;
+}
+
+// Content negotiation for agents: when the site publishes Markdown twins (standard front end, agents.markdown), a request
+// whose Accept header prefers text/markdown over text/html is rewritten to the .md object before the cache lookup. The
+// flag is spliced in by redirect-rules.ts at deploy time.
+const negotiateMarkdown = /*PAPYRUS_NEGOTIATE_MARKDOWN*/false/*END_PAPYRUS_NEGOTIATE_MARKDOWN*/;
+
+function qualityOf(accept, mediaType) {
+  let quality = 0;
+  const parts = accept.toLowerCase().split(",");
+  for (let index = 0; index < parts.length; index += 1) {
+    const fields = parts[index].split(";");
+    if (fields[0].trim() !== mediaType) {
+      continue;
+    }
+    quality = 1;
+    for (let field = 1; field < fields.length; field += 1) {
+      const pair = fields[field].trim();
+      if (pair.indexOf("q=") === 0) {
+        quality = Number(pair.substring(2));
+        if (isNaN(quality)) {
+          quality = 0;
+        }
+      }
+    }
+  }
+  return quality;
+}
+
+function prefersMarkdown(request) {
+  const header = request.headers && request.headers.accept;
+  if (!header || !header.value) {
+    return false;
+  }
+  const markdown = qualityOf(header.value, "text/markdown");
+  return markdown > 0 && markdown > qualityOf(header.value, "text/html");
+}
+
+function markdownTwin(uri) {
+  if (uri === "/404.html" || uri.length < 6 || uri.substring(uri.length - 5) !== ".html") {
+    return null;
+  }
+  return uri.substring(0, uri.length - 5) + ".md";
+}
+
+// Directory-style sites keep a page's twin at /x/index.md, so a bare /x.md is rewritten there. The edge cannot test for
+// existence, hence the rule: a last segment <name>.md with a name other than index. A path prefix (a base path) is just
+// part of <path>. The flag is spliced in by redirect-rules.ts at deploy time from urls.style.
+const directoryTwins = /*PAPYRUS_DIRECTORY_TWINS*/false/*END_PAPYRUS_DIRECTORY_TWINS*/;
+
+function directoryTwin(uri) {
+  if (uri.length < 5 || uri.substring(uri.length - 3) !== ".md") {
+    return null;
+  }
+  const lastSegment = uri.substring(uri.lastIndexOf("/") + 1);
+  if (lastSegment === "index.md" || lastSegment === ".md") {
+    return null;
+  }
+  return uri.substring(0, uri.length - 3) + "/index.md";
+}
+
 async function handler(event) {
   const request = event.request;
   const redirectTarget = await lookUpRedirectTarget(request.uri);
@@ -32,6 +117,26 @@ async function handler(event) {
       headers: { location: { value: redirectTarget } },
     };
   }
+  const patternRedirect = lookUpPatternRedirect(request.uri);
+  if (patternRedirect !== null) {
+    return {
+      statusCode: patternRedirect.status,
+      statusDescription: "Moved",
+      headers: { location: { value: patternRedirect.target } },
+    };
+  }
   request.uri = rewriteToIndexDocument(request.uri);
+  if (directoryTwins) {
+    const bareTwin = directoryTwin(request.uri);
+    if (bareTwin !== null) {
+      request.uri = bareTwin;
+    }
+  }
+  if (negotiateMarkdown && prefersMarkdown(request)) {
+    const twin = markdownTwin(request.uri);
+    if (twin !== null) {
+      request.uri = twin;
+    }
+  }
   return request;
 }

package/amplify/publication/redirect-rules.js

@@ -0,0 +1,73 @@
+import { existsSync, readFileSync } from "node:fs";
+import { resolve } from "node:path";
+/** Mirrors DIRECTORY_HTML_PATTERN and DIRECTORY_SLASH_PATTERN in standard/config.py. */
+const DIRECTORY_HTML_PATTERN = "^/(?!(?:.*/)?index\\.html$)(.+)\\.html$";
+const DIRECTORY_SLASH_PATTERN = "^/((?:[^/]+/)*[^/.]+)$";
+const START_MARKER = "/*PAPYRUS_PATTERN_RULES*/";
+const END_MARKER = "/*END_PAPYRUS_PATTERN_RULES*/";
+/**
+ * The pattern rules a site declares in projection.json: its `redirects.patterns`, plus, for
+ * `urls.style: "directory"`, `/x.html` to `/x/`. Mirrors `StandardSettings.edge_rules` in Python.
+ */
+export function edgeRulesFromSettings(settings) {
+    const rules = (settings.redirects?.patterns ?? []).map((rule) => ({
+        p: rule.pattern,
+        t: rule.to,
+        s: rule.status ?? 301,
+        x: rule.except ?? [],
+    }));
+    if (settings.urls?.style === "directory") {
+        // /x.html to /x/ (a directory index file such as /x/index.html is served as it is), then /x to /x/.
+        rules.push({ p: DIRECTORY_HTML_PATTERN, t: "/$1/", s: 301, x: ["/404.html"] });
+        rules.push({ p: DIRECTORY_SLASH_PATTERN, t: "/$1/", s: 301, x: ["/404"] });
+    }
+    return rules;
+}
+/** A standard-front-end site publishes Markdown twins unless it turns `agents.markdown` off. */
+export function negotiatesMarkdown(settings) {
+    return settings.site !== undefined && settings.agents?.markdown !== false;
+}
+/** A directory-style site keeps a page's Markdown twin at `/x/index.md`; the edge then serves a bare `/x.md` from there. */
+export function usesDirectoryTwins(settings) {
+    return negotiatesMarkdown(settings) && settings.urls?.style === "directory";
+}
+export function readEdgeConfig(staticDir, baseDirectory = process.cwd()) {
+    if (staticDir === undefined)
+        return { rules: [], negotiateMarkdown: false, directoryTwins: false };
+    const settingsPath = resolve(baseDirectory, staticDir, "projection.json");
+    if (!existsSync(settingsPath))
+        return { rules: [], negotiateMarkdown: false, directoryTwins: false };
+    const settings = JSON.parse(readFileSync(settingsPath, "utf8"));
+    return { rules: edgeRulesFromSettings(settings), negotiateMarkdown: negotiatesMarkdown(settings), directoryTwins: usesDirectoryTwins(settings) };
+}
+export function readEdgeRules(staticDir, baseDirectory = process.cwd()) {
+    return readEdgeConfig(staticDir, baseDirectory).rules;
+}
+const NEGOTIATE_START = "/*PAPYRUS_NEGOTIATE_MARKDOWN*/";
+const NEGOTIATE_END = "/*END_PAPYRUS_NEGOTIATE_MARKDOWN*/";
+const TWINS_START = "/*PAPYRUS_DIRECTORY_TWINS*/";
+const TWINS_END = "/*END_PAPYRUS_DIRECTORY_TWINS*/";
+function spliceFlag(source, start, end, label) {
+    const from = source.indexOf(start);
+    const to = source.indexOf(end);
+    if (from === -1 || to === -1 || to < from) {
+        throw new Error(`redirect function source has no ${label} markers`);
+    }
+    return source.slice(0, from + start.length) + "true" + source.slice(to);
+}
+/** The CloudFront Function source with the site's pattern rules and Markdown negotiation flag spliced in. */
+export function compileRedirectFunctionSource(source, rules, options = {}) {
+    const start = source.indexOf(START_MARKER);
+    const end = source.indexOf(END_MARKER);
+    if (start === -1 || end === -1 || end < start) {
+        throw new Error("redirect function source has no pattern-rule markers");
+    }
+    let compiled = rules.length === 0 ? source : source.slice(0, start + START_MARKER.length) + JSON.stringify(rules) + source.slice(end);
+    if (options.negotiateMarkdown === true) {
+        compiled = spliceFlag(compiled, NEGOTIATE_START, NEGOTIATE_END, "markdown negotiation");
+    }
+    if (options.directoryTwins === true) {
+        compiled = spliceFlag(compiled, TWINS_START, TWINS_END, "directory twin");
+    }
+    return compiled;
+}

package/amplify/publication/redirect-rules.ts

@@ -0,0 +1,93 @@
+import { existsSync, readFileSync } from "node:fs";
+import { resolve } from "node:path";
+
+/** One pattern redirect as the edge function reads it: regex, target with $1..$9, status, excluded paths. */
+export type EdgeRedirectRule = { p: string; t: string; s: number; x: string[] };
+
+/** Mirrors DIRECTORY_HTML_PATTERN and DIRECTORY_SLASH_PATTERN in standard/config.py. */
+const DIRECTORY_HTML_PATTERN = "^/(?!(?:.*/)?index\\.html$)(.+)\\.html$";
+const DIRECTORY_SLASH_PATTERN = "^/((?:[^/]+/)*[^/.]+)$";
+
+const START_MARKER = "/*PAPYRUS_PATTERN_RULES*/";
+const END_MARKER = "/*END_PAPYRUS_PATTERN_RULES*/";
+
+type ProjectionRedirects = { patterns?: { pattern: string; to: string; status?: number; except?: string[] }[] };
+type ProjectionSettings = { site?: unknown; urls?: { style?: string }; redirects?: ProjectionRedirects; agents?: { markdown?: boolean } };
+
+/**
+ * The pattern rules a site declares in projection.json: its `redirects.patterns`, plus, for
+ * `urls.style: "directory"`, `/x.html` to `/x/`. Mirrors `StandardSettings.edge_rules` in Python.
+ */
+export function edgeRulesFromSettings(settings: ProjectionSettings): EdgeRedirectRule[] {
+  const rules = (settings.redirects?.patterns ?? []).map((rule) => ({
+    p: rule.pattern,
+    t: rule.to,
+    s: rule.status ?? 301,
+    x: rule.except ?? [],
+  }));
+  if (settings.urls?.style === "directory") {
+    // /x.html to /x/ (a directory index file such as /x/index.html is served as it is), then /x to /x/.
+    rules.push({ p: DIRECTORY_HTML_PATTERN, t: "/$1/", s: 301, x: ["/404.html"] });
+    rules.push({ p: DIRECTORY_SLASH_PATTERN, t: "/$1/", s: 301, x: ["/404"] });
+  }
+  return rules;
+}
+
+/** A standard-front-end site publishes Markdown twins unless it turns `agents.markdown` off. */
+export function negotiatesMarkdown(settings: ProjectionSettings): boolean {
+  return settings.site !== undefined && settings.agents?.markdown !== false;
+}
+
+/** A directory-style site keeps a page's Markdown twin at `/x/index.md`; the edge then serves a bare `/x.md` from there. */
+export function usesDirectoryTwins(settings: ProjectionSettings): boolean {
+  return negotiatesMarkdown(settings) && settings.urls?.style === "directory";
+}
+
+export type EdgeConfig = { rules: EdgeRedirectRule[]; negotiateMarkdown: boolean; directoryTwins: boolean };
+
+export function readEdgeConfig(staticDir: string | undefined, baseDirectory: string = process.cwd()): EdgeConfig {
+  if (staticDir === undefined) return { rules: [], negotiateMarkdown: false, directoryTwins: false };
+  const settingsPath = resolve(baseDirectory, staticDir, "projection.json");
+  if (!existsSync(settingsPath)) return { rules: [], negotiateMarkdown: false, directoryTwins: false };
+  const settings = JSON.parse(readFileSync(settingsPath, "utf8")) as ProjectionSettings;
+  return { rules: edgeRulesFromSettings(settings), negotiateMarkdown: negotiatesMarkdown(settings), directoryTwins: usesDirectoryTwins(settings) };
+}
+
+export function readEdgeRules(staticDir: string | undefined, baseDirectory: string = process.cwd()): EdgeRedirectRule[] {
+  return readEdgeConfig(staticDir, baseDirectory).rules;
+}
+
+const NEGOTIATE_START = "/*PAPYRUS_NEGOTIATE_MARKDOWN*/";
+const NEGOTIATE_END = "/*END_PAPYRUS_NEGOTIATE_MARKDOWN*/";
+const TWINS_START = "/*PAPYRUS_DIRECTORY_TWINS*/";
+const TWINS_END = "/*END_PAPYRUS_DIRECTORY_TWINS*/";
+
+function spliceFlag(source: string, start: string, end: string, label: string): string {
+  const from = source.indexOf(start);
+  const to = source.indexOf(end);
+  if (from === -1 || to === -1 || to < from) {
+    throw new Error(`redirect function source has no ${label} markers`);
+  }
+  return source.slice(0, from + start.length) + "true" + source.slice(to);
+}
+
+/** The CloudFront Function source with the site's pattern rules and Markdown negotiation flag spliced in. */
+export function compileRedirectFunctionSource(
+  source: string,
+  rules: EdgeRedirectRule[],
+  options: { negotiateMarkdown?: boolean; directoryTwins?: boolean } = {},
+): string {
+  const start = source.indexOf(START_MARKER);
+  const end = source.indexOf(END_MARKER);
+  if (start === -1 || end === -1 || end < start) {
+    throw new Error("redirect function source has no pattern-rule markers");
+  }
+  let compiled = rules.length === 0 ? source : source.slice(0, start + START_MARKER.length) + JSON.stringify(rules) + source.slice(end);
+  if (options.negotiateMarkdown === true) {
+    compiled = spliceFlag(compiled, NEGOTIATE_START, NEGOTIATE_END, "markdown negotiation");
+  }
+  if (options.directoryTwins === true) {
+    compiled = spliceFlag(compiled, TWINS_START, TWINS_END, "directory twin");
+  }
+  return compiled;
+}

package/amplify/publication/s3-projection.js

@@ -7,11 +7,13 @@ import { SqsDestination } from "aws-cdk-lib/aws-lambda-destinations";
 import * as route53 from "aws-cdk-lib/aws-route53";
 import * as s3 from "aws-cdk-lib/aws-s3";
 import * as sqs from "aws-cdk-lib/aws-sqs";
+import { readFileSync } from "node:fs";
 import { dirname, resolve } from "node:path";
 import { fileURLToPath } from "node:url";
 import { assertValidPublicationConfig } from "../site-backend-config.js";
 import { addRenditionsFunction } from "./renditions-function.js";
 import { RENDITIONS_FUNCTION_NAME_ENV } from "./renditions-settings.js";
+import { compileRedirectFunctionSource, readEdgeConfig } from "./redirect-rules.js";
 import { addInvalidationPathAlarm } from "./invalidation-alarm.js";
 import { distributionIdFromArn, s3ProjectionEnvironment, s3ProjectionPolicyStatements } from "./s3-projection-settings.js";
 const redirectFunctionSourcePath = resolve(dirname(fileURLToPath(import.meta.url)), "redirect-and-index.js");
@@ -68,9 +70,13 @@ export function addS3ProjectionPublication(scope, props, media) {
     assertValidPublicationConfig({ kind: "s3-projection", ...props });
     const bucket = createSiteBucket(scope, props.retainBucketOnDelete !== false);
     const keyValueStore = new cloudfront.KeyValueStore(scope, "RedirectsKeyValueStore");
+    const edgeConfig = readEdgeConfig(props.staticDir);
     const redirectFunction = new cloudfront.Function(scope, "RedirectAndIndexFunction", {
         runtime: cloudfront.FunctionRuntime.JS_2_0,
-        code: cloudfront.FunctionCode.fromFile({ filePath: redirectFunctionSourcePath }),
+        code: cloudfront.FunctionCode.fromInline(compileRedirectFunctionSource(readFileSync(redirectFunctionSourcePath, "utf8"), edgeConfig.rules, {
+            negotiateMarkdown: edgeConfig.negotiateMarkdown,
+            directoryTwins: edgeConfig.directoryTwins,
+        })),
         keyValueStore,
     });
     const deadLetterQueue = new sqs.Queue(scope, "ProjectionDeadLetterQueue", {