CDPT-2295 Correct paths (#63)

ministryofjustice · Jan 23, 2025 · 27e9275 · 27e9275
1 parent 32e45af
commit 27e9275
Show file tree

Hide file tree

Showing 14 changed files with 168 additions and 95 deletions.
diff --git a/conf/node/controllers/generate-indexes.js b/conf/node/controllers/generate-indexes.js
@@ -1,5 +1,6 @@
 import { intranetUrls, s3BucketName, indexCss } from "../constants.js";
 import { getAgenciesFromS3, getSnapshotsFromS3 } from "./s3.js";
+import { getAgencyPath } from "./paths.js";
 
 /**
  * Generate the root index html
@@ -16,8 +17,6 @@ export const generateRootIndex = async (bucket = s3BucketName, env) => {
     throw new Error("Env is required");
   }
 
-  const agencyPrefix = "production" === env ? '' : `${env}-`;
-
   const agencies = await getAgenciesFromS3(bucket, env);
 
   const hostname = new URL(intranetUrls[env]).hostname;
@@ -34,7 +33,10 @@ export const generateRootIndex = async (bucket = s3BucketName, env) => {
               .map(
                 (agency) => `
                 <li class="list-group-item">
-                  <a href="/${agencyPrefix}${agency}/index.html" target="_blank">${agency}</a>
+                  <a href="/${getAgencyPath(
+                    env,
+                    agency,
+                  )}/index.html" target="_blank">${agency}</a>
                 </li>`,
               )
               .join("\n")}
@@ -75,9 +77,6 @@ export const generateAgencyIndex = async (
 
   const snapshots = await getSnapshotsFromS3(bucket, env, agency);
 
-  const snapshotParent =
-    "production" === env ? `${agency}` : `${env}-${agency}`;
-
   const html = `<!doctype html><html lang="en">
     <head><title>Intranet Archive Index</title><style>${indexCss}</style></head>
     <body>
@@ -90,7 +89,9 @@ export const generateAgencyIndex = async (
               .map(
                 (snapshot) => `
                 <li class="list-group-item">
-                  <a href="/${snapshotParent}/${snapshot}/${url.hostname}/index.html" target="_blank">${snapshot}</a>
+                  <a href="/${getAgencyPath(env, agency)}/${snapshot}/${
+                  url.hostname
+                }/index.html" target="_blank">${snapshot}</a>
                 </li>`,
               )
               .join("\n")}

diff --git a/conf/node/controllers/generate-indexes.test.js b/conf/node/controllers/generate-indexes.test.js
@@ -6,7 +6,6 @@ import {
 
 import { s3BucketName } from "../constants.js";
 import { generateRootIndex, generateAgencyIndex } from "./generate-indexes.js";
-
 import { s3Options } from "./s3.js";
 
 let client;

diff --git a/conf/node/controllers/httrack.js b/conf/node/controllers/httrack.js
@@ -1,35 +1,7 @@
 import { spawn, execSync } from "node:child_process";
 import fs from "node:fs";
 
-import { intranetUrls, intranetJwts } from "../constants.js";
-
-/**
- * A helper function to get the directory for the snapshot.
- *
- * @param {Object} props
- * @param {string} props.env
- * @param {string} props.agency
- * @returns {{s3: string, fs: string}} the s3 and fs paths
- */
-
-export const getSnapshotPaths = ({ env, agency }) => {
-  // Get date in format: 2023-01-17
-  const dateString = new Date().toISOString().slice(0, 10);
-
-  if(!Object.keys(intranetJwts).includes(env)) {
-    throw new Error(`Invalid environment: ${env}`);
-  }
-
-  const s3Path =
-    env === "production"
-      ? `${agency}/${dateString}`
-      : `${env}-${agency}/${dateString}`;
-
-  const fsPath = `/tmp/snapshots/${s3Path}`;
-
-  // Return directory for the snapshot
-  return { s3: s3Path, fs: fsPath };
-};
+import { intranetJwts } from "../constants.js";
 
 /**
  * Get arguments for httrack cli.

diff --git a/conf/node/controllers/httrack.test.js b/conf/node/controllers/httrack.test.js
@@ -3,7 +3,6 @@ import http from "node:http";
 import { afterAll, it, jest } from "@jest/globals";
 
 import {
-  getSnapshotPaths,
   getHttrackArgs,
   runHttrack,
   getHttrackProgress,
@@ -14,43 +13,6 @@ import { intranetUrls, intranetJwts } from "../constants.js";
 // Skip long tests when running in watch mode.
 const skipLongTests = process.env.npm_lifecycle_event === "test:watch";
 
-describe("getSnapshotPaths", () => {
-  it("should return the s3 and fs paths - production", () => {
-    const env = "production";
-    const agency = "hq";
-
-    const paths = getSnapshotPaths({ env, agency });
-
-    expect(paths).toStrictEqual({
-      s3: `${agency}/${new Date().toISOString().slice(0, 10)}`,
-      fs: `/tmp/snapshots/${agency}/${new Date().toISOString().slice(0, 10)}`,
-    });
-  });
-
-  it("should return the s3 and fs paths - non-production", () => {
-    const env = "dev";
-    const agency = "hq";
-
-    const paths = getSnapshotPaths({ env, agency });
-
-    expect(paths).toStrictEqual({
-      s3: `dev-${agency}/${new Date().toISOString().slice(0, 10)}`,
-      fs: `/tmp/snapshots/dev-${agency}/${new Date()
-        .toISOString()
-        .slice(0, 10)}`,
-    });
-  });
-
-  it("should return the s3 and fs paths - invalid env", () => {
-    const env = "invalid";
-    const agency = "hq";
-
-    expect(() => getSnapshotPaths({ env, agency })).toThrowError(
-      `Invalid environment: ${env}`,
-    );
-  });
-});
-
 describe("getHttrackArgs", () => {
   it("should return an array of arguments", async () => {
     const env = "production";

diff --git a/conf/node/controllers/main.js b/conf/node/controllers/main.js
@@ -7,12 +7,12 @@ import {
   sensitiveFiles,
 } from "../constants.js";
 import {
-  getSnapshotPaths,
   getHttrackArgs,
   runHttrack,
   getHttrackProgress,
   waitForHttrackComplete,
 } from "./httrack.js";
+import { getAgencyPath, getSnapshotPaths } from "./paths.js";
 import { createHeartbeat, sync, writeToS3 } from "./s3.js";
 import { generateRootIndex, generateAgencyIndex } from "./generate-indexes.js";
 
@@ -69,9 +69,7 @@ export const main = async ({ env, agency, depth }) => {
   const agencyIndexHtml = await generateAgencyIndex(s3BucketName, env, agency);
   await writeToS3(
     s3BucketName,
-    "production" === env
-      ? `${agency}/index.html`
-      : `${env}-${agency}/index.html`,
+    `${getAgencyPath(env, agency)}/index.html`,
     agencyIndexHtml,
     { cacheMaxAge: 600 },
   );

diff --git a/conf/node/controllers/main.test.js b/conf/node/controllers/main.test.js
@@ -4,7 +4,7 @@ import { afterAll, beforeEach, expect, it, jest } from "@jest/globals";
 import { S3Client, ListObjectsV2Command, GetObjectCommand } from "@aws-sdk/client-s3";
 
 import { main } from "./main.js";
-import { getSnapshotPaths } from "./httrack.js";
+import { getSnapshotPaths } from "./paths.js";
 import { s3Options, s3EmptyDir } from "./s3.js";
 import { intranetUrls, s3BucketName } from "../constants.js";
 

diff --git a/conf/node/controllers/paths.js b/conf/node/controllers/paths.js
@@ -0,0 +1,45 @@
+import { intranetUrls, allowedTargetAgencies } from "../constants.js";
+
+/**
+ * A helper function to get the folder for the agency.
+ *
+ * The folder is prefixed with the environment if it is not production.
+ * Then, it is followed by the agency.
+ *
+ * @param {string} env
+ * @param {string} agency
+ * @returns {string} the folder name
+ */
+
+export const getAgencyPath = (env, agency) => {
+  if (!Object.keys(intranetUrls).includes(env)) {
+    throw new Error(`Invalid environment: ${env}`);
+  }
+
+  if (!allowedTargetAgencies.includes(agency)) {
+    throw new Error(`Invalid agency: ${agency}`);
+  }
+
+  return env === "production" ? `${agency}` : `${env}-${agency}`;
+};
+
+/**
+ * A helper function to get the directory for the snapshot.
+ *
+ * @param {Object} props
+ * @param {string} props.env
+ * @param {string} props.agency
+ * @returns {{s3: string, fs: string}} the s3 and fs paths
+ */
+
+export const getSnapshotPaths = ({ env, agency }) => {
+  // Get date in format: 2023-01-17
+  const dateString = new Date().toISOString().slice(0, 10);
+
+  const s3Path = `${getAgencyPath(env, agency)}/${dateString}`;
+
+  const fsPath = `/tmp/snapshots/${s3Path}`;
+
+  // Return directory for the snapshot
+  return { s3: s3Path, fs: fsPath };
+};
diff --git a/conf/node/controllers/paths.test.js b/conf/node/controllers/paths.test.js
@@ -0,0 +1,78 @@
+import { it } from "@jest/globals";
+
+import { getAgencyPath, getSnapshotPaths } from "./paths.js";
+
+describe("getAgencyPath", () => {
+  it("should return the folder name - production", () => {
+    const env = "production";
+    const agency = "hq";
+
+    const folder = getAgencyPath(env, agency);
+
+    expect(folder).toBe("hq");
+  });
+
+  it("should return the folder name - non-production", () => {
+    const env = "dev";
+    const agency = "hq";
+
+    const folder = getAgencyPath(env, agency);
+
+    expect(folder).toBe("dev-hq");
+  });
+
+  it("should throw an error - invalid env", () => {
+    const env = "invalid";
+    const agency = "hq";
+
+    expect(() => getAgencyPath(env, agency)).toThrowError(
+      `Invalid environment: ${env}`,
+    );
+  });
+
+  it("should throw an error - invalid agency", () => {
+    const env = "production";
+    const agency = "invalid";
+
+    expect(() => getAgencyPath(env, agency)).toThrowError(
+      `Invalid agency: ${agency}`,
+    );
+  });
+});
+
+describe("getSnapshotPaths", () => {
+  it("should return the s3 and fs paths - production", () => {
+    const env = "production";
+    const agency = "hq";
+
+    const paths = getSnapshotPaths({ env, agency });
+
+    expect(paths).toStrictEqual({
+      s3: `${agency}/${new Date().toISOString().slice(0, 10)}`,
+      fs: `/tmp/snapshots/${agency}/${new Date().toISOString().slice(0, 10)}`,
+    });
+  });
+
+  it("should return the s3 and fs paths - non-production", () => {
+    const env = "dev";
+    const agency = "hq";
+
+    const paths = getSnapshotPaths({ env, agency });
+
+    expect(paths).toStrictEqual({
+      s3: `dev-${agency}/${new Date().toISOString().slice(0, 10)}`,
+      fs: `/tmp/snapshots/dev-${agency}/${new Date()
+        .toISOString()
+        .slice(0, 10)}`,
+    });
+  });
+
+  it("should return the s3 and fs paths - invalid env", () => {
+    const env = "invalid";
+    const agency = "hq";
+
+    expect(() => getSnapshotPaths({ env, agency })).toThrowError(
+      `Invalid environment: ${env}`,
+    );
+  });
+});
diff --git a/conf/node/controllers/s3.js b/conf/node/controllers/s3.js
@@ -12,8 +12,9 @@ import {
   s3Credentials as credentials,
   s3Region,
   heartbeatEndpoint,
-  allowedTargetAgencies
+  allowedTargetAgencies,
 } from "../constants.js";
+import { getAgencyPath } from "./paths.js";
 
 /**
  * S3 client options
@@ -208,8 +209,8 @@ export const getAgenciesFromS3 = async (bucket = s3BucketName, env) => {
     folder.Prefix.replace("/", "").replace("/", ""),
   );
 
-  if(env === 'production') {
-    return folders.filter((folder) => allowedTargetAgencies.includes(folder) )
+  if (env === "production") {
+    return folders.filter((folder) => allowedTargetAgencies.includes(folder));
   }
 
   return folders
@@ -241,7 +242,7 @@ export const getSnapshotsFromS3 = async (
     throw new Error("Agency is required");
   }
 
-  const prefix = "production" === env ? `${agency}/` : `${env}-${agency}/`;
+  const prefix = `${getAgencyPath(env, agency)}/`;
 
   const command = new ListObjectsV2Command({
     Bucket: bucket,

diff --git a/conf/node/controllers/s3.test.js b/conf/node/controllers/s3.test.js
@@ -18,7 +18,7 @@ import {
   writeToS3,
   getAgenciesFromS3,
   getSnapshotsFromS3,
-} from "./s3";
+} from "./s3.js";
 
 describe("checkAccess", () => {
   let client;

diff --git a/conf/node/controllers/schedule.test.js b/conf/node/controllers/schedule.test.js
@@ -1,6 +1,6 @@
 import { expect, it, jest } from "@jest/globals";
 
-import { parseScheduleString, scheduleFunction } from "./schedule";
+import { parseScheduleString, scheduleFunction } from "./schedule.js";
 
 jest.useFakeTimers();
 

diff --git a/conf/node/middleware.js b/conf/node/middleware.js
@@ -11,7 +11,7 @@ import {
 
 /**
  * @typedef {import('express').Request & { mirror: { env: string, agency: string, depth: number } }} SpiderRequest
- * @typedef {import('express').Request & { isValid: ?boolean, agency: string, _hostname: string }} AccessRequest
+ * @typedef {import('express').Request & { isValid: ?boolean, agency: string, env: string }} AccessRequest
  */
 
 /**
@@ -106,10 +106,14 @@ export const checkSignature = (req, res, next) => {
   }
 
   // Make sure the request is for an allowed hostname
-  req._hostname = payloadObject.hostname;
-
-  if (!allowedTargetHosts.includes(req._hostname)) {
-    console.error("Error: Hostname not allowed");
+  req.env = Object.keys(intranetUrls).find(
+    (env) =>
+      intranetUrls[env] &&
+      new URL(intranetUrls[env]).hostname === payloadObject.hostname,
+  );
+
+  if (!req.env) {
+    console.error("Error: Env not found");
     res.status(403).send({ status: 403 });
     return;
   }

diff --git a/conf/node/middleware.test.js b/conf/node/middleware.test.js
@@ -47,6 +47,16 @@ describe("checkSignature middleware", () => {
     expect(res.send).toHaveBeenCalledWith({ status: 400 });
   });
 
+  it("should return 403 if hostname is not found", () => {
+    const payload = Buffer.from(
+      JSON.stringify({ expiry: Date.now() / 1000 + 10, hostname: "invalid" }),
+    ).toString("base64");
+    req.body = { sig: "signature", payload };
+    checkSignature(req, res, next);
+    expect(res.status).toHaveBeenCalledWith(403);
+    expect(res.send).toHaveBeenCalledWith({ status: 403 });
+  });
+
   it("should return 403 if request has expired", () => {
     const payload = Buffer.from(
       JSON.stringify({ expiry: Date.now() / 1000 - 10 }),