refactor: pass CustomUrl to matomo tag manager pageview events

src/components/DefaultErrorMessage.tsx

@@ -50,7 +50,7 @@ export const DefaultErrorMessagePage = () => {
   return (
     <PageContainer asChild consumeCss>
       <main>
-        <PageTitle title={t("htmlTitles.errorPage")} />
+        <PageTitle title={t("htmlTitles.errorPage")} useLocationForCustomPath={true} />
         <DefaultErrorMessage applySkipToContentId={true} />
       </main>
     </PageContainer>

src/components/PageTitle.tsx

@@ -7,46 +7,91 @@
  */
 
 import { useContext, useEffect, useRef } from "react";
-import { useHref, useLocation } from "react-router";
+import { useTranslation } from "react-i18next";
+import { useLocation } from "react-router";
+import { isValidLocale } from "../i18n";
 import { log } from "../util/logger/logger";
 import { getAllDimensions } from "../util/trackingUtil";
 import { AuthContext } from "./AuthenticationContext";
+import { buildFullUrlFromPath, languagePartIndex } from "./SocialMediaMetadata";
 
-interface Props {
+interface BaseProps {
   title: string;
 }
 
+interface TrackingProps {
+  defaultUrl?: string;
+  rootId?: string;
+  context?: {
+    rootId?: string;
+    defaultUrl?: string;
+  };
+}
+
+interface WithTrackingProps extends BaseProps {
+  trackingProps: TrackingProps;
+  useLocationForCustomPath?: false;
+}
+
+interface WithNoTrackingProps extends BaseProps {
+  trackingProps?: never;
+  useLocationForCustomPath: true;
+}
+
+type Props = WithTrackingProps | WithNoTrackingProps;
+
+// NOTE: Builds the URL sent as `CustomUrl` on matomo `Pageview` events. Every
+// locale segment is stripped (unlike `getCanonicalUrl`, which only strips `nb`)
+// so that every language variant of the same page collapses onto a single URL
+// in matomo — letting us track a page's traffic as one number regardless of
+// which language the visitor used.
+export const getTrackedUrl = (pathname: string) => {
+  const parts = pathname.split("/");
+  const langIdx = languagePartIndex(parts);
+  if (isValidLocale(parts[langIdx])) {
+    parts.splice(langIdx, 1);
+  }
+
+  return buildFullUrlFromPath(parts.join("/"));
+};
+
 /**
  * A component for setting the page title and tracking a page view event.
  * @param title - The title of the page. Will update the document title tag and dispatch a page view event. The component expects this title to be stable for the lifetime of the page, meaning it should not change unless the page location changes.
+ * @param trackingProps - Optional tracking metadata for the page view event. `defaultUrl` overrides the path used for tracking instead of the current location; `rootId` is sent as the `subjectId` dimension. Either value may also be supplied nested under `context` to accept a GraphQL `TaxonomyContext` directly.
  */
-export const PageTitle = ({ title }: Props) => {
+export const PageTitle = ({ title, trackingProps }: Props) => {
   const { user, authContextLoaded } = useContext(AuthContext);
+  const { i18n } = useTranslation();
   const hasTracked = useRef(false);
 
   const location = useLocation();
-  const href = useHref(location);
+  const customPath = trackingProps?.defaultUrl ?? trackingProps?.context?.defaultUrl;
+  const subjectId = trackingProps?.rootId ?? trackingProps?.context?.rootId;
+  const trackedPath = customPath ?? location.pathname;
+  const trackedUrl = getTrackedUrl(trackedPath);
 
   useEffect(() => {
     hasTracked.current = false;
-  }, [href]);
+  }, [trackedUrl]);
 
   useEffect(() => {
     if (!authContextLoaded) return;
     if (hasTracked.current) {
       log.info("PageTitle: Page view already tracked, skipping duplicate tracking. This should not happen");
       return;
     }
-    // for debugging purposes
-    // log.info(`PageTitle: Tracking page view with title: ${title}`);
     const dimensions = getAllDimensions({ user });
     window._mtm?.push({
       page_title: title,
       event: "Pageview",
+      CustomUrl: trackedUrl,
+      languageCode: i18n.language,
+      subjectId,
       ...dimensions,
     });
     hasTracked.current = true;
-  }, [authContextLoaded, title, user]);
+  }, [authContextLoaded, title, trackedUrl, user, subjectId, i18n.language]);
 
   return <title>{title}</title>;
 };

src/components/SocialMediaMetadata.tsx

@@ -11,27 +11,40 @@ import { useLocation, useHref } from "react-router";
 import config from "../config";
 import { isValidLocale, preferredLanguages } from "../i18n";
 
+export const buildFullUrlFromPath = (path: string) => {
+  return `${config.ndlaFrontendDomain}${path}`;
+};
+
+export const languagePartIndex = (parts: string[]) => {
+  return parts.includes("article-iframe") ? 2 : 1;
+};
+
+// NOTE: Builds the `<link rel="canonical">` URL. Only the `nb` prefix is stripped
+// because bokmål is our default language and is served without a path segment;
+// other locales (`nn`, `en`, …) must remain in the URL so each language variant
+// has its own canonical and is indexed as a distinct page by search engines.
 export const getCanonicalUrl = (pathname: string) => {
-  if (!pathname.includes("article-iframe")) {
-    return `${config.ndlaFrontendDomain}${pathname}`;
+  const parts = pathname.split("/");
+  const langIdx = languagePartIndex(parts);
+  if (parts[langIdx] === config.defaultLocale) {
+    parts.splice(langIdx, 1);
   }
-  const paths = pathname.split("/");
-  if (isValidLocale(paths[2])) {
-    paths.splice(2, 1);
-  }
-  return `${config.ndlaFrontendDomain}${paths.join("/")}`;
+
+  return buildFullUrlFromPath(parts.join("/"));
 };
 
+// NOTE: Builds the `<link rel="alternate" hrefLang>` URL for a given language.
+// Any existing locale segment is removed and `alternateLanguage` is inserted in
+// its place so search engines can discover all language variants of a page from
+// any one of them.
 export const getAlternateUrl = (pathname: string, alternateLanguage: string) => {
-  if (!pathname.includes("article-iframe")) {
-    return `${config.ndlaFrontendDomain}/${alternateLanguage}${pathname}`;
-  }
-  const paths = pathname.split("/");
-  if (isValidLocale(paths[2])) {
-    paths.splice(2, 1);
+  const parts = pathname.split("/");
+  const langIdx = languagePartIndex(parts);
+  if (isValidLocale(parts[langIdx])) {
+    parts.splice(langIdx, 1);
   }
-  paths.splice(2, 0, alternateLanguage);
-  return `${config.ndlaFrontendDomain}${paths.join("/")}`;
+  parts.splice(langIdx, 0, alternateLanguage);
+  return buildFullUrlFromPath(parts.join("/"));
 };
 
 export const getAlternateLanguages = (trackableContent?: TrackableContent) => {
@@ -42,43 +55,53 @@ interface TrackableContent {
   supportedLanguages?: string[];
 }
 
-interface Props {
+interface BaseProps {
   title: string;
   description?: string;
-  path?: string;
   imageUrl?: string;
   audioUrl?: string;
   trackableContent?: TrackableContent;
   children?: ReactNode;
   type?: string;
 }
 
+interface WithCanonicalPath extends BaseProps {
+  canonicalPath: string | undefined;
+  useLocationForCanonicalPath?: false;
+}
+
+interface WithLocationForCanonicalPath extends BaseProps {
+  canonicalPath?: undefined;
+  useLocationForCanonicalPath: true;
+}
+
+type Props = WithCanonicalPath | WithLocationForCanonicalPath;
+
 export const SocialMediaMetadata = ({
   title,
   imageUrl,
   audioUrl,
   description,
-  path,
   trackableContent,
   children,
+  canonicalPath,
   type = "article",
 }: Props) => {
   const location = useLocation();
-  const href = useHref(location);
+  const hrefLocation = canonicalPath ? { pathname: canonicalPath } : location;
+  const href = useHref(hrefLocation);
+  const canonicalUrl = getCanonicalUrl(href);
+
   return (
     <>
-      <link rel="canonical" href={getCanonicalUrl(path ? path : location.pathname)} />
-      {getAlternateLanguages(trackableContent).map((alternateLanguage) => (
-        <link
-          key={alternateLanguage}
-          rel="alternate"
-          hrefLang={alternateLanguage}
-          href={getAlternateUrl(path ? path : location.pathname, alternateLanguage)}
-        />
-      ))}
+      <link rel="canonical" href={canonicalUrl} />
+      {getAlternateLanguages(trackableContent).map((alternateLanguage) => {
+        const alternateUrl = getAlternateUrl(canonicalPath ?? location.pathname, alternateLanguage);
+        return <link key={alternateLanguage} rel="alternate" hrefLang={alternateLanguage} href={alternateUrl} />;
+      })}
       {children}
       <meta property="og:type" content={type} />
-      <meta property="og:url" content={`${config.ndlaFrontendDomain}${href}`} />
+      <meta property="og:url" content={canonicalUrl} />
       {!!title && <meta property="og:title" content={`${title} - NDLA`} />}
       {!!description && <meta property="og:description" content={description} />}
       {!!description && <meta name="description" content={description} />}

src/components/__tests__/pageTitle-test.ts

@@ -0,0 +1,39 @@
+/**
+ * Copyright (c) 2019-present, NDLA.
+ *
+ * This source code is licensed under the GPLv3 license found in the
+ * LICENSE file in the root directory of this source tree.
+ *
+ */
+
+import { getTrackedUrl } from "../PageTitle";
+
+test("getTrackedUrl with article-url and empty language", () => {
+  const url = getTrackedUrl("/article/123");
+  expect(url).toMatch("https://test.ndla.no/article/123");
+});
+
+test("getTrackedUrl with article-url and non-default language", () => {
+  const url = getTrackedUrl("/nn/article/123");
+  expect(url).toMatch("https://test.ndla.no/article/123");
+});
+
+test("getTrackedUrl with article-url and default language", () => {
+  const url = getTrackedUrl("/nb/article/123");
+  expect(url).toMatch("https://test.ndla.no/article/123");
+});
+
+test("getTrackedUrl with iframe-url and nb language", () => {
+  const url = getTrackedUrl("/article-iframe/nb/urn:topic:123/1");
+  expect(url).toMatch("https://test.ndla.no/article-iframe/urn:topic:123/1");
+});
+
+test("getTrackedUrl with iframe-url and nn language", () => {
+  const url = getTrackedUrl("/article-iframe/nn/urn:topic:123/1");
+  expect(url).toMatch("https://test.ndla.no/article-iframe/urn:topic:123/1");
+});
+
+test("getTrackedUrl with iframe-url and no language", () => {
+  const url = getTrackedUrl("/article-iframe/urn:topic:123/1");
+  expect(url).toMatch("https://test.ndla.no/article-iframe/urn:topic:123/1");
+});

src/components/__tests__/socialMediaMetadata-test.ts

@@ -35,11 +35,26 @@ test("getCanonicalUrl with article-url and empty language", () => {
   expect(canonicalUrl).toMatch("https://test.ndla.no/article/123");
 });
 
+test("getCanonicalUrl with article-url and non-default language", () => {
+  const canonicalUrl = getCanonicalUrl("/nn/article/123");
+  expect(canonicalUrl).toMatch("https://test.ndla.no/nn/article/123");
+});
+
+test("getCanonicalUrl with article-url and default language", () => {
+  const canonicalUrl = getCanonicalUrl("/nb/article/123");
+  expect(canonicalUrl).toMatch("https://test.ndla.no/article/123");
+});
+
 test("getCanonicalUrl with iframe-url and nb language", () => {
   const canonicalUrl = getCanonicalUrl("/article-iframe/nb/urn:topic:123/1");
   expect(canonicalUrl).toMatch("https://test.ndla.no/article-iframe/urn:topic:123/1");
 });
 
+test("getCanonicalUrl with iframe-url and nn language", () => {
+  const canonicalUrl = getCanonicalUrl("/article-iframe/nn/urn:topic:123/1");
+  expect(canonicalUrl).toMatch("https://test.ndla.no/article-iframe/nn/urn:topic:123/1");
+});
+
 test("getCanonicalUrl with iframe-url and no language", () => {
   const canonicalUrl = getCanonicalUrl("/article-iframe/urn:topic:123/1");
   expect(canonicalUrl).toMatch("https://test.ndla.no/article-iframe/urn:topic:123/1");

src/containers/AboutPageV2/AboutPageLeaf.tsx

@@ -96,7 +96,7 @@ export const AboutPageLeaf = ({ article: _article, crumbs }: Props) => {
 
   return (
     <main>
-      <PageTitle title={getDocumentTitle(t, article.title)} />
+      <PageTitle title={getDocumentTitle(t, article.title)} useLocationForCustomPath={true} />
       <meta name="pageid" content={`${article.id}`} />
       {scripts?.map((script) => (
         <script key={script.src} src={script.src} type={script.type} async={script.async} defer={script.defer} />
@@ -108,6 +108,7 @@ export const AboutPageLeaf = ({ article: _article, crumbs }: Props) => {
         description={article.metaDescription}
         imageUrl={article.metaImage?.image.imageUrl}
         trackableContent={article}
+        useLocationForCanonicalPath={true}
       />
       <Hero variant="primary">
         <HeroBackground />

src/containers/AboutPageV2/AboutPageNode.tsx

@@ -146,7 +146,7 @@ export const AboutPageNode = ({ article, menuItems, crumbs }: Props) => {
 
   return (
     <main>
-      <PageTitle title={getDocumentTitle(t, article.title)} />
+      <PageTitle title={getDocumentTitle(t, article.title)} useLocationForCustomPath={true} />
       <meta name="pageid" content={`${article.id}`} />
       {scripts?.map((script) => (
         <script key={script.src} src={script.src} type={script.type} async={script.async} defer={script.defer} />
@@ -158,6 +158,7 @@ export const AboutPageNode = ({ article, menuItems, crumbs }: Props) => {
         description={transformedArticle.metaDescription}
         imageUrl={transformedArticle.metaImage?.image.imageUrl}
         trackableContent={transformedArticle}
+        useLocationForCanonicalPath={true}
       />
       <StyledPageContent>
         <HomeBreadcrumb items={crumbs.map((bc) => ({ ...bc, to: bc.url }))} />

src/containers/AccessDeniedPage/AccessDeniedPage.tsx

@@ -59,7 +59,7 @@ export const AccessDenied = ({ applySkipToContentId }: AccessDeniedProps) => {
   return (
     <Status code={statusCode}>
       <ErrorMessageRoot>
-        <PageTitle title={t("htmlTitles.accessDenied")} />
+        <PageTitle title={t("htmlTitles.accessDenied")} useLocationForCustomPath={true} />
         <StyledPresentationLine />
         <ErrorMessageContent>
           <ErrorMessageDescription id={applySkipToContentId ? SKIP_TO_CONTENT_ID : undefined}>

src/containers/AllSubjectsPage/AllSubjectsPage.tsx

@@ -124,7 +124,7 @@ export const AllSubjectsPage = () => {
   return (
     <StyledPageContainer asChild consumeCss>
       <main>
-        <PageTitle title={t("htmlTitles.subjectsPage")} />
+        <PageTitle title={t("htmlTitles.subjectsPage")} useLocationForCustomPath={true} />
         <HeadingWrapper>
           <Heading textStyle="heading.medium" id={SKIP_TO_CONTENT_ID}>
             {t("subjectsPage.allSubjects")}

src/containers/ArticlePage/ArticlePage.tsx

@@ -69,7 +69,7 @@ export const ArticlePage = ({ resource, skipToContentId, loading }: Props) => {
     <>
       {!!resource && !!resource.article && !!article && (
         <>
-          <PageTitle title={getDocumentTitle(t, resource, root)} />
+          <PageTitle title={getDocumentTitle(t, resource, root)} trackingProps={resource} />
           {scripts?.map((script) => (
             <script key={script.src} src={script.src} type={script.type} async={script.async} defer={script.defer} />
           ))}
@@ -89,7 +89,7 @@ export const ArticlePage = ({ resource, skipToContentId, loading }: Props) => {
             trackableContent={article}
             description={article.metaDescription}
             imageUrl={article.metaImage?.image.imageUrl}
-            path={resource.url}
+            canonicalPath={resource.url}
           />
         </>
       )}
@@ -127,13 +127,15 @@ ArticlePage.fragments = {
       nodeType
       name
       url
+      defaultUrl
       contentUri
       relevanceId
       resourceTypes {
         name
         id
       }
       context {
+        rootId
         contextId
         isArchived
         parents {

src/containers/CollectionPage/CollectionPage.tsx

@@ -135,8 +135,8 @@ const CollectionPageContent = ({ collectionLanguage, subjects, image }: Collecti
   return (
     <StyledPageContainer padding="large" asChild consumeCss>
       <main>
-        <PageTitle title={pageTitle} />
-        <SocialMediaMetadata title={metaTitle} imageUrl={image?.image.imageUrl} />
+        <PageTitle title={pageTitle} useLocationForCustomPath={true} />
+        <SocialMediaMetadata title={metaTitle} imageUrl={image?.image.imageUrl} useLocationForCanonicalPath={true} />
         <div>
           {!!image && (
             <StyledImage