Merged in feature/BOOK-401-handle-new-redirects (pull request #2942)

feat(BOOK-401): redirect-service with tests and typings

* Switch to typescript

* add new dependencies for tests and typescript

* Switch to typescript

* add new dependencies for tests and typescript

* rewrite to typescript and include tests

* rewrite to typescript and include tests

* refactor: update README and scripts for CSV handling; rename update script to generate

* include csv-data folder


Approved-by: Linus Flood

apps/scandic-redirect/scripts/generateRedirectFile/checkPrerequisites.ts

@@ -0,0 +1,18 @@
+import fs from "node:fs"
+import { langs } from "./lang"
+import { csvFilePath } from "./paths"
+
+export function checkPrerequisites() {
+  const missingLangs = langs.reduce((acc, lang) => {
+    const filepath = csvFilePath(lang)
+    if (!fs.existsSync(filepath)) {
+      return [...acc, filepath]
+    }
+
+    return acc
+  }, [] as string[])
+
+  if (missingLangs.length > 0) {
+    throw new Error(`Missing CSV file:\n${missingLangs.join("\n")}`)
+  }
+}

apps/scandic-redirect/scripts/generateRedirectFile/createRedirectEntry.test.ts

@@ -0,0 +1,59 @@
+import { describe, it, expect } from "vitest"
+import { createRedirectEntry } from "./createRedirectEntry"
+describe("createRedirectEntry", () => {
+  describe("top level domain from old site to new", () => {
+    it(".com", () => {
+      expect(
+        createRedirectEntry({
+          oldUrl: "https://scandichotels.com/old-path",
+          newUrl: "https://www.scandichotels.com/en/new-path",
+        })
+      ).toEqual({ from: "/en/old-path", to: "/en/new-path" })
+    })
+
+    it(".se", () => {
+      expect(
+        createRedirectEntry({
+          oldUrl: "https://scandichotels.se/old-path",
+          newUrl: "https://www.scandichotels.com/sv/new-path",
+        })
+      ).toEqual({ from: "/sv/old-path", to: "/sv/new-path" })
+    })
+
+    it(".dk", () => {
+      expect(
+        createRedirectEntry({
+          oldUrl: "https://scandichotels.dk/old-path",
+          newUrl: "https://www.scandichotels.com/da/new-path",
+        })
+      ).toEqual({ from: "/da/old-path", to: "/da/new-path" })
+    })
+
+    it(".de", () => {
+      expect(
+        createRedirectEntry({
+          oldUrl: "https://scandichotels.de/old-path",
+          newUrl: "https://www.scandichotels.com/de/new-path",
+        })
+      ).toEqual({ from: "/de/old-path", to: "/de/new-path" })
+    })
+
+    it(".fi", () => {
+      expect(
+        createRedirectEntry({
+          oldUrl: "https://scandichotels.fi/old-path",
+          newUrl: "https://www.scandichotels.com/fi/new-path",
+        })
+      ).toEqual({ from: "/fi/old-path", to: "/fi/new-path" })
+    })
+
+    it(".no", () => {
+      expect(
+        createRedirectEntry({
+          oldUrl: "https://scandichotels.no/old-path",
+          newUrl: "https://www.scandichotels.com/no/new-path",
+        })
+      ).toEqual({ from: "/no/old-path", to: "/no/new-path" })
+    })
+  })
+})

apps/scandic-redirect/scripts/generateRedirectFile/createRedirectEntry.ts

@@ -0,0 +1,17 @@
+import { akamaiRedirect, removeDomain } from "./utils"
+
+export function createRedirectEntry({
+  oldUrl,
+  newUrl,
+}: {
+  oldUrl: string
+  newUrl: string
+}) {
+  const from = removeDomain(akamaiRedirect(oldUrl))
+  const to = removeDomain(newUrl)
+
+  return {
+    from,
+    to,
+  }
+}

apps/scandic-redirect/scripts/generateRedirectFile/index.ts

@@ -0,0 +1,68 @@
+import fs from "node:fs"
+
+import csvToJson from "convert-csv-to-json"
+import { Lang, langs } from "./lang"
+import { csvFilePath, outputFilepath } from "./paths"
+import { checkPrerequisites } from "./checkPrerequisites"
+import { createRedirectEntry } from "./createRedirectEntry"
+
+const csvHeaders = {
+  current: "Old URL",
+  redirect: "Redirect URL",
+} as const
+
+type Entry = {
+  [key in (typeof csvHeaders)[keyof typeof csvHeaders]]: string
+}
+
+async function convertCsvToJson<T extends Record<string, unknown>>(
+  lang: Lang
+): Promise<T[]> {
+  const json = csvToJson
+    .utf8Encoding()
+    .fieldDelimiter(",")
+    .getJsonFromCsv(csvFilePath(lang))
+
+  return json
+}
+
+async function createRedirectFile(data: Entry[], lang: Lang) {
+  const finalUrls = data.reduce(
+    (acc, url) => {
+      const { from, to } = createRedirectEntry({
+        oldUrl: url[csvHeaders.current],
+        newUrl: url[csvHeaders.redirect],
+      })
+
+      return {
+        ...acc,
+        [from]: to,
+      }
+    },
+    {} as Record<string, string>
+  )
+
+  try {
+    fs.writeFileSync(outputFilepath(lang), JSON.stringify(finalUrls), {
+      encoding: "utf-8",
+    })
+  } catch (e) {
+    console.error(`Unable to create redirect file for lang: ${lang}`, e)
+  }
+}
+
+checkPrerequisites()
+console.log("All prerequisites met. Generating redirect files...")
+let totalCount = 0
+for (const lang of langs.toSorted()) {
+  const perf = performance.now()
+  const data = await convertCsvToJson<Entry>(lang)
+  await createRedirectFile(data, lang)
+  const duration = (performance.now() - perf).toFixed(0)
+  console.log(
+    `Created redirect file for ${lang}: Processed ${data.length} entries in ${duration}ms`
+  )
+  totalCount += data.length
+}
+
+console.log(`\n🥳 All done!\nProcessed a total of ${totalCount} entries.`)

apps/scandic-redirect/scripts/generateRedirectFile/lang.ts

@@ -0,0 +1,2 @@
+export const langs = ["da", "de", "en", "fi", "no", "sv"] as const
+export type Lang = (typeof langs)[number]

apps/scandic-redirect/scripts/generateRedirectFile/paths.ts

@@ -0,0 +1,17 @@
+import { Lang } from "./lang"
+import path from "node:path"
+
+export function csvFilePath(lang: Lang) {
+  return `${import.meta.dirname}/../data/csv/${lang}.csv`
+}
+
+export function jsonFilePath(lang: Lang) {
+  return `${import.meta.dirname}/../data/json/${lang}.json`
+}
+
+export function outputFilepath(lang: Lang) {
+  return path.resolve(
+    import.meta.dirname,
+    `../../netlify/functions/data/${lang}.json`
+  )
+}

apps/scandic-redirect/scripts/generateRedirectFile/utils.test.ts

@@ -0,0 +1,114 @@
+import { describe, it, expect } from "vitest"
+import { akamaiRedirect, removeDomain } from "./utils"
+
+describe("removeDomain", () => {
+  it("https scandichotels.com with path", () => {
+    expect(removeDomain("https://scandichotels.com/foo")).toBe("/foo")
+  })
+
+  it("http www scandichotels.com no path", () => {
+    expect(removeDomain("http://www.scandichotels.com")).toBe("")
+  })
+
+  it("stage subdomain with de TLD, query and hash", () => {
+    expect(
+      removeDomain("https://stage.scandichotels.de/path?query=1#frag")
+    ).toBe("/path?query=1#frag")
+  })
+
+  it("prod subdomain with fi TLD and trailing slash", () => {
+    expect(removeDomain("http://prod.scandichotels.fi/")).toBe("/")
+  })
+
+  it("test subdomain with dk TLD and deeper path", () => {
+    expect(removeDomain("https://test.scandichotels.dk/some/page")).toBe(
+      "/some/page"
+    )
+  })
+
+  it("no TLD with unicode path", () => {
+    expect(removeDomain("https://scandichotels.no/öä")).toBe("/öä")
+  })
+
+  it("se TLD root", () => {
+    expect(removeDomain("https://scandichotels.se")).toBe("")
+  })
+
+  it("input without protocol should be unchanged", () => {
+    expect(removeDomain("scandichotels.com/foo")).toBe("scandichotels.com/foo")
+  })
+
+  it("different domain containing scandichotels should be unchanged", () => {
+    expect(removeDomain("https://example.com/scandichotels.com/foo")).toBe(
+      "https://example.com/scandichotels.com/foo"
+    )
+  })
+
+  it("uppercase scheme/domain should be unchanged (regex is case-sensitive)", () => {
+    expect(removeDomain("HTTPS://WWW.SCANDICHOTELS.COM/Path")).toBe(
+      "HTTPS://WWW.SCANDICHOTELS.COM/Path"
+    )
+  })
+
+  it("only removes the leading matching domain when it appears at the start", () => {
+    const input = "https://scandichotels.com/https://scandichotels.com/foo"
+    expect(removeDomain(input)).toBe("/https://scandichotels.com/foo")
+  })
+})
+
+describe("akamaiRedirect", () => {
+  it("https scandichotels.com with path", () => {
+    expect(akamaiRedirect("https://scandichotels.com/foo")).toBe("/en/foo")
+  })
+
+  it("http www scandichotels.com no path", () => {
+    expect(akamaiRedirect("http://www.scandichotels.com")).toBe("/en")
+  })
+
+  it("stage subdomain with de TLD, query and hash", () => {
+    expect(
+      akamaiRedirect("https://stage.scandichotels.de/path?query=1#frag")
+    ).toBe("/de/path?query=1#frag")
+  })
+
+  it("prod subdomain with fi TLD and trailing slash", () => {
+    expect(akamaiRedirect("http://prod.scandichotels.fi/")).toBe("/fi/")
+  })
+
+  it("test subdomain with dk TLD and deeper path", () => {
+    expect(akamaiRedirect("https://test.scandichotels.dk/some/page")).toBe(
+      "/da/some/page"
+    )
+  })
+
+  it("no TLD with unicode path", () => {
+    expect(akamaiRedirect("https://scandichotels.no/öä")).toBe("/no/öä")
+  })
+
+  it("se TLD root", () => {
+    expect(akamaiRedirect("https://scandichotels.se")).toBe("/sv")
+  })
+
+  it("input without protocol should be unchanged", () => {
+    expect(akamaiRedirect("scandichotels.com/foo")).toBe(
+      "scandichotels.com/foo"
+    )
+  })
+
+  it("different domain containing scandichotels should be unchanged", () => {
+    expect(akamaiRedirect("https://example.com/scandichotels.com/foo")).toBe(
+      "https://example.com/scandichotels.com/foo"
+    )
+  })
+
+  it("uppercase scheme/domain should be unchanged (regex is case-sensitive)", () => {
+    expect(akamaiRedirect("HTTPS://WWW.SCANDICHOTELS.COM/Path")).toBe(
+      "HTTPS://WWW.SCANDICHOTELS.COM/Path"
+    )
+  })
+
+  it("only replaces the leading matching domain when it appears at the start", () => {
+    const input = "https://scandichotels.com/https://scandichotels.com/foo"
+    expect(akamaiRedirect(input)).toBe("/en/https://scandichotels.com/foo")
+  })
+})

apps/scandic-redirect/scripts/generateRedirectFile/utils.ts

@@ -0,0 +1,68 @@
+/**
+ * Removes the scheme and Scandic domain prefix from a URL string.
+ *
+ * Strips leading "http://" or "https://" and an optional "www|test|stage|prod" subdomain
+ * for hosts matching scandichotels.(com|de|dk|fi|no|se).
+ *
+ * @param str - Input string or URL.
+ * @returns The input with the Scandic domain portion removed.
+ */
+export function removeDomain(str: string) {
+  return str.replace(
+    /^https?:\/\/((www|test|stage|prod)\.)?scandichotels.(com|de|dk|fi|no|se)/,
+    ""
+  )
+}
+
+/**
+ * Converts an absolute Scandic Hotels URL to a localized path prefix based on the top-level domain.
+ *
+ * Replaces the leading "http(s)://[www|test|stage|prod.]scandichotels.{tld}" portion with a locale prefix:
+ * - .com -> /en
+ * - .de  -> /de
+ * - .dk  -> /da
+ * - .fi  -> /fi
+ * - .no  -> /no
+ * - .se  -> /sv
+ *
+ * The function recognizes http and https schemes and optional subdomains (www, test, stage, prod).
+ * If the input does not match the expected host pattern, the original string is returned unchanged.
+ *
+ * Examples:
+ * - "https://scandichotels.com/hotels"  -> "/en/hotels"
+ * - "http://www.scandichotels.de"       -> "/de"
+ * - "https://stage.scandichotels.se/?"  -> "/sv/?"
+ *
+ * @param str - The URL string to convert.
+ * @returns The input string with the Scandic Hotels host replaced by the locale path prefix,
+ *          or the original string if no matching Scandic host is found.
+ *
+ * @remarks
+ * - This is a simple regex-based replacement and does not perform full URL parsing or validation.
+ * - The remainder of the path, query string, and fragment (if present) are preserved after the locale prefix.
+ * - If the original URL ends exactly at the TLD (no trailing slash), the result will be the locale prefix without a trailing slash.
+ */
+export function akamaiRedirect(str: string) {
+  return str.replace(
+    /^https?:\/\/((www|test|stage|prod)\.)?scandichotels.(com|de|dk|fi|no|se)/,
+    (...match) => {
+      if (match[3]) {
+        switch (match[3]) {
+          case "com":
+            return "/en"
+          case "de":
+            return "/de"
+          case "dk":
+            return "/da"
+          case "fi":
+            return "/fi"
+          case "no":
+            return "/no"
+          case "se":
+            return "/sv"
+        }
+      }
+      return ""
+    }
+  )
+}

apps/scandic-redirect/scripts/update.mjs

@@ -1,138 +0,0 @@
-import fs from 'node:fs';
-import path from 'node:path';
-
-import csvToJson from 'convert-csv-to-json';
-
-const langs = ['da', 'de', 'en', 'fi', 'no', 'sv'];
-const csvHeaders = {
-  current: 'Current URL',
-  redirect: 'Redirect URL',
-};
-
-function csvFilePath(lang) {
-  return `${import.meta.dirname}/csv/${lang}.csv`;
-}
-
-function jsonFilePath(lang) {
-  return `${import.meta.dirname}/json/${lang}.json`;
-}
-
-function outputFilepath(lang) {
-  return path.resolve(
-    import.meta.dirname,
-    `../netlify/functions/data/${lang}.json`
-  );
-}
-
-function removeDomain(str) {
-  return str.replace(
-    /^https?:\/\/((www|test|stage|prod)\.)?scandichotels.(com|de|dk|fi|no|se)/,
-    ''
-  );
-}
-
-function akamaiRedirect(str) {
-  return str.replace(
-    /^https?:\/\/((www|test|stage|prod)\.)?scandichotels.(com|de|dk|fi|no|se)/,
-    (...match) => {
-      if (match[3]) {
-        switch (match[3]) {
-          case 'com':
-            return '/en';
-          case 'de':
-            return '/de';
-          case 'dk':
-            return '/da';
-          case 'fi':
-            return '/fi';
-          case 'no':
-            return '/no';
-          case 'se':
-            return '/sv';
-        }
-      }
-      return '';
-    }
-  );
-}
-
-function checkPrerequisites() {
-  const missingLangs = langs.reduce((acc, lang) => {
-    const filepath = csvFilePath(lang);
-    if (!fs.existsSync(filepath)) {
-      acc.push(filepath);
-    }
-    return acc;
-  }, []);
-
-  if (missingLangs.length > 0) {
-    console.error(`Missing CSV file:\n${missingLangs.join('\n')}`);
-    process.exit(1);
-  }
-}
-
-// convert-csv-to-json writes async without callback support
-// so we workaround it be overriding console.log which it uses when it is done
-async function convertCsvToJson() {
-  return new Promise((resolve, reject) => {
-    const _consoleLog = console.log;
-    let resolved = 0;
-    console.log = function (str) {
-      if (str.indexOf('File saved:') >= 0) {
-        resolved++;
-      }
-
-      if (resolved === langs.length) {
-        console.log = _consoleLog;
-        resolve();
-      }
-    };
-
-    for (const lang of langs) {
-      csvToJson
-        .utf8Encoding()
-        .fieldDelimiter(',')
-        .generateJsonFileFromCsv(csvFilePath(lang), jsonFilePath(lang));
-    }
-
-    setTimeout(() => {
-      reject('timeout');
-    }, 5000);
-  });
-}
-
-async function makeOutput() {
-  for (const lang of langs) {
-    try {
-      const json = JSON.parse(
-        fs.readFileSync(jsonFilePath(lang), {
-          encoding: 'utf-8',
-        })
-      );
-
-      if (Array.isArray(json)) {
-        const finalUrls = json.reduce((acc, url) => {
-          const from = removeDomain(akamaiRedirect(url[csvHeaders.current]));
-          const to = removeDomain(url[csvHeaders.redirect]);
-
-          return {
-            ...acc,
-            [from]: to,
-          };
-        }, {});
-
-        fs.writeFileSync(outputFilepath(lang), JSON.stringify(finalUrls), {
-          encoding: 'utf-8',
-        });
-      } else {
-        throw new Error(`JSON was not an array: ${jsonFilePath(lang)}`);
-      }
-    } catch (e) {
-      console.error(e);
-    }
-  }
-}
-
-checkPrerequisites();
-await convertCsvToJson();
-await makeOutput();