Make page index retrieval work in EPUB-based books

In real EPUB-based VS books, the `index` field of `PageInfo` objects returned by
the `getCurrentPage` and `getPages` methods is undefined. Our mock VS book was
incorrect in this respect.

Correct the mock book to behave like real VS books here, and add a fallback
method to obtain a page index for EPUB-based books by looking up the page in the
list of pages returned by `MosaicBookElement.getPages`.

dev-server/static/scripts/vitalsource-mosaic-book-element.js

@@ -61,21 +61,21 @@ export class MosaicBookElement extends HTMLElement {
           absoluteURL: '/document/little-women-1',
           chapterTitle: 'Chapter One',
           cfi: '/2',
-          index: 0,
+          index: undefined,
           page: '10',
         },
         {
           absoluteURL: '/document/little-women-2',
           chapterTitle: 'Chapter Two',
           cfi: '/4',
-          index: 1,
+          index: undefined,
           page: '20',
         },
         {
           absoluteURL: '/document/little-women-3',
           chapterTitle: 'Chapter Three',
           cfi: '/6',
-          index: 2,
+          index: undefined,
           page: '30',
         },
       ];
@@ -170,6 +170,10 @@ export class MosaicBookElement extends HTMLElement {
     return this.pageData[this.pageIndex];
   }
 
+  async getPages() {
+    return { ok: true, data: [...this.pageData], status: 200 };
+  }
+
   async getTOC() {
     // In our VS test pages, the TOC entries are 1:1 with the page entries.
     // However in a real VS book there may be multiple TOC entries for different

src/annotator/integrations/test/vitalsource-test.js

@@ -236,7 +236,7 @@ describe('annotator/integrations/vitalsource', () => {
           absoluteURL: '/pages/chapter_02.xhtml',
           cfi: '/2',
           chapterTitle: 'Chapter two',
-          index: 1,
+          index: undefined,
           page: '20',
         };
       } else if (this._format === 'pbk') {
@@ -252,6 +252,12 @@ describe('annotator/integrations/vitalsource', () => {
       }
     }
 
+    async getPages() {
+      const pageData = await this.getCurrentPage();
+      const data = [pageData];
+      return { ok: true, data };
+    }
+
     async getTOC() {
       if (this._format === 'epub') {
         const toc = [
@@ -394,7 +400,7 @@ describe('annotator/integrations/vitalsource', () => {
       const pageSelector = selectors.find(s => s.type === 'PageSelector');
       assert.deepEqual(pageSelector, {
         type: 'PageSelector',
-        index: 1,
+        index: 0,
         label: '20',
       });
     });

src/annotator/integrations/vitalsource.ts

@@ -9,7 +9,11 @@ import type {
   SegmentInfo,
   SidebarLayout,
 } from '../../types/annotator';
-import type { EPUBContentSelector, Selector } from '../../types/api';
+import type {
+  EPUBContentSelector,
+  PageSelector,
+  Selector,
+} from '../../types/api';
 import type {
   ContentFrameGlobals,
   MosaicBookElement,
@@ -173,6 +177,23 @@ function makeContentFrameScrollable(frame: HTMLIFrameElement) {
   attrObserver.observe(frame, { attributes: true });
 }
 
+/**
+ * Lookup options for fetching page metadata from VitalSource.
+ *
+ * Enabling this options involves some extra work, so should be skipped if
+ * the data is not needed.
+ */
+type PageInfoOptions = {
+  /** Whether to fetch the title. */
+  includeTitle?: boolean;
+
+  /**
+   * Whether to use a fallback strategy to get the page index if not available
+   * in the {@link MosaicBookElement.getCurrentPage} response.
+   */
+  includePageIndex?: boolean;
+};
+
 /**
  * Return a copy of URL with the origin removed.
  *
@@ -338,19 +359,18 @@ export class VitalSourceContentIntegration
       page: pageLabel,
       title,
       url,
-    } = await this._getPageInfo(true /* includeTitle */);
+    } = await this._getPageInfo({ includeTitle: true, includePageIndex: true });
 
     // We generate an "EPUBContentSelector" with a CFI for all VS books,
     // although for PDF-based books the CFI is a string generated from the
     // page number.
-    const extraSelectors: Selector[] = [
-      {
-        type: 'EPUBContentSelector',
-        cfi,
-        url,
-        title,
-      },
-    ];
+    const cfiSelector: EPUBContentSelector = {
+      type: 'EPUBContentSelector',
+      cfi,
+      url,
+      title,
+    };
+    const extraSelectors: Selector[] = [cfiSelector];
 
     // Add page number if available. PDF-based books always have them.
     // Publishers are encouraged to provide page numbers for EPUB-based books,
@@ -359,11 +379,12 @@ export class VitalSourceContentIntegration
     //
     // [1] https://www.vitalsource.com/en-uk/products/vitalsource-epub3-implementation-guide-vitalsource-vvstdocsepub3implementguide?term=VST-DOCS-EPUB3IMPLEMENTGUIDE
     if (typeof pageIndex === 'number') {
-      extraSelectors.push({
+      const pageSelector: PageSelector = {
         type: 'PageSelector',
         index: pageIndex,
         label: pageLabel,
-      });
+      };
+      extraSelectors.push(pageSelector);
     }
 
     selectors.push(...extraSelectors);
@@ -451,14 +472,15 @@ export class VitalSourceContentIntegration
   /**
    * Retrieve information about the currently displayed content document or
    * page.
-   *
-   * @param includeTitle - Whether to fetch the title. This involves some extra
-   *   work so should be skipped when not required.
    */
-  async _getPageInfo(includeTitle: boolean) {
-    const [pageInfo, toc] = await Promise.all([
+  async _getPageInfo({
+    includeTitle = false,
+    includePageIndex = false,
+  }: PageInfoOptions = {}) {
+    const [pageInfo, toc, pages] = await Promise.all([
       this._bookElement.getCurrentPage(),
       includeTitle ? this._bookElement.getTOC() : undefined,
+      includePageIndex ? this._bookElement.getPages() : undefined,
     ]);
 
     // If changes in VitalSource ever mean that critical chapter/page metadata
@@ -474,7 +496,6 @@ export class VitalSourceContentIntegration
     }
 
     let title;
-
     if (toc) {
       title = pageInfo.chapterTitle;
 
@@ -490,9 +511,20 @@ export class VitalSourceContentIntegration
       }
     }
 
+    // For PDF-based books, the `pageInfo.index` property should be populated.
+    // For EPUB-based books, it may not be. In that case we try to find a
+    // page number in the complete page list instead.
+    let pageIndex = pageInfo.index;
+    if (pageIndex === undefined && pages) {
+      const index = pages.data?.findIndex(page => page.cfi === pageInfo.cfi);
+      if (index !== -1) {
+        pageIndex = index;
+      }
+    }
+
     return {
       cfi: pageInfo.cfi,
-      index: pageInfo.index,
+      index: pageIndex,
       page: pageInfo.page,
       title,
 
@@ -503,7 +535,7 @@ export class VitalSourceContentIntegration
   }
 
   async segmentInfo(): Promise<SegmentInfo> {
-    const { cfi, url } = await this._getPageInfo(false /* includeTitle */);
+    const { cfi, url } = await this._getPageInfo();
     return { cfi, url };
   }
 

src/types/vitalsource.ts

@@ -164,6 +164,9 @@ export type MosaicBookElement = HTMLElement & {
    */
   getCurrentPage(): Promise<PageInfo>;
 
+  /** Get the list of pages in the book. */
+  getPages(): Promise<DataResponse<PageInfo[]>>;
+
   /** Retrieve the book's table of contents. */
   getTOC(): Promise<DataResponse<TableOfContentsEntry[]>>;