Implement a reader for epubs

packages/api/src/generated/graphql.ts

@@ -244,6 +244,7 @@ export enum BulkActionType {
 }
 
 export enum ContentReader {
+  Epub = 'EPUB',
   Pdf = 'PDF',
   Web = 'WEB'
 }

packages/api/src/generated/schema.graphql

@@ -203,6 +203,7 @@ enum BulkActionType {
 }
 
 enum ContentReader {
+  EPUB
   PDF
   WEB
 }

packages/api/src/resolvers/article/index.ts

@@ -98,10 +98,12 @@ import {
 } from '../../utils/parser'
 import { parseSearchQuery, SortBy, SortOrder } from '../../utils/search'
 import {
+  contentReaderForPageType,
   getStorageFileDetails,
   makeStorageFilePublic,
 } from '../../utils/uploads'
 import { WithDataSourcesContext } from '../types'
+import { pageTypeForContentType } from '../upload_files'
 
 enum ArticleFormat {
   Markdown = 'markdown',
@@ -258,7 +260,7 @@ export const createArticleResolver = authorized<
         uploadFileHash = uploadFileDetails.md5Hash
         userArticleUrl = uploadFileDetails.fileUrl
         canonicalUrl = uploadFile.url
-        pageType = PageType.File
+        pageType = pageTypeForContentType(uploadFile.contentType)
         title = titleForFilePath(uploadFile.url)
       } else if (
         source !== 'puppeteer-parse' &&
@@ -950,8 +952,7 @@ export const searchResolver = authorized<
         ...r,
         image: r.image && createImageProxyUrl(r.image, 260, 260),
         isArchived: !!r.archivedAt,
-        contentReader:
-          r.pageType === PageType.File ? ContentReader.Pdf : ContentReader.Web,
+        contentReader: contentReaderForPageType(r.pageType),
         originalArticleUrl: r.url,
         publishedAt: validatedDate(r.publishedAt),
         ownedByViewer: r.userId === claims.uid,
@@ -1054,10 +1055,7 @@ export const updatesSinceResolver = authorized<
           ...p,
           image: p.image && createImageProxyUrl(p.image, 260, 260),
           isArchived: !!p.archivedAt,
-          contentReader:
-            p.pageType === PageType.File
-              ? ContentReader.Pdf
-              : ContentReader.Web,
+          contentReader: contentReaderForPageType(p.pageType),
         } as SearchItem,
         cursor: endCursor,
         itemID: p.id,

packages/api/src/resolvers/function_resolvers.ts

@@ -20,6 +20,7 @@ import {
 import { userDataToUser, validatedDate, wordsCount } from '../utils/helpers'
 import { createImageProxyUrl } from '../utils/imageproxy'
 import {
+  contentReaderForPageType,
   generateDownloadSignedUrl,
   generateUploadFilePathName,
 } from '../utils/uploads'
@@ -375,7 +376,8 @@ export const functionResolvers = {
   Article: {
     async url(article: Article, _: unknown, ctx: WithDataSourcesContext) {
       if (
-        article.pageType == PageType.File &&
+        (article.pageType == PageType.File ||
+          article.pageType == PageType.Book) &&
         ctx.claims &&
         article.uploadFileId
       ) {
@@ -468,9 +470,7 @@ export const functionResolvers = {
       return !!page?.archivedAt || false
     },
     contentReader(article: { pageType: PageType }) {
-      return article.pageType === PageType.File
-        ? ContentReader.Pdf
-        : ContentReader.Web
+      return contentReaderForPageType(article.pageType)
     },
     highlights(
       article: { id: string; userId?: string; highlights?: Highlight[] },
@@ -551,7 +551,11 @@ export const functionResolvers = {
   },
   SearchItem: {
     async url(item: SearchItem, _: unknown, ctx: WithDataSourcesContext) {
-      if (item.pageType == PageType.File && ctx.claims && item.uploadFileId) {
+      if (
+        (item.pageType == PageType.File || item.pageType == PageType.Book) &&
+        ctx.claims &&
+        item.uploadFileId
+      ) {
         const upload = await ctx.models.uploadFile.get(item.uploadFileId)
         if (!upload || !upload.fileName) {
           return undefined

packages/api/src/resolvers/upload_files/index.ts

@@ -27,6 +27,13 @@ const isFileUrl = (url: string): boolean => {
   return parsedUrl.protocol == 'file:'
 }
 
+export const pageTypeForContentType = (contentType: string): PageType => {
+  if (contentType == 'application/epub+zip') {
+    return PageType.Book
+  }
+  return PageType.File
+}
+
 export const uploadFileRequestResolver: ResolverFn<
   UploadFileRequestResult,
   unknown,
@@ -145,7 +152,7 @@ export const uploadFileRequestResolver: ResolverFn<
             title: title,
             hash: uploadFilePathName,
             content: '',
-            pageType: PageType.File,
+            pageType: pageTypeForContentType(input.contentType),
             uploadFileId: uploadFileData.id,
             slug: generateSlug(uploadFilePathName),
             createdAt: new Date(),

packages/api/src/schema.ts

@@ -36,6 +36,7 @@ const schema = gql`
   enum ContentReader {
     WEB
     PDF
+    EPUB
   }
 
   input SortParams {

packages/api/src/services/speech.ts

@@ -1,4 +1,6 @@
 import { Page, PageType } from '../elastic/types'
+import { ContentReader } from '../generated/graphql'
+import { contentReaderForPageType } from '../utils/uploads'
 import { FeatureName, isOptedIn } from './features'
 
 /*
@@ -8,7 +10,10 @@ export const shouldSynthesize = async (
   userId: string,
   page: Page
 ): Promise<boolean> => {
-  if (page.pageType === PageType.File || !page.content) {
+  if (
+    contentReaderForPageType(page.pageType) === ContentReader.Web ||
+    !page.content
+  ) {
     // we don't synthesize files for now
     return false
   }

packages/api/src/utils/uploads.ts

@@ -2,6 +2,19 @@
 /* eslint-disable @typescript-eslint/no-unsafe-assignment */
 import { File, GetSignedUrlConfig, Storage } from '@google-cloud/storage'
 import { env } from '../env'
+import { ContentReader, PageType } from '../generated/graphql'
+
+export const contentReaderForPageType = (pageType: PageType) => {
+  console.log('getting content reader: ', pageType)
+  switch (pageType) {
+    case PageType.Book:
+      return ContentReader.Epub
+    case PageType.File:
+      return ContentReader.Pdf
+    default:
+      return ContentReader.Web
+  }
+}
 
 /* On GAE/Prod, we shall rely on default app engine service account credentials.
  * Two changes needed: 1) add default service account to our uploads GCS Bucket
@@ -39,6 +52,7 @@ export const generateUploadSignedUrl = async (
     expires: Date.now() + 15 * 60 * 1000, // 15 minutes
     contentType: contentType,
   }
+  console.log('signed url for: ', options)
 
   // Get a v4 signed URL for uploading file
   const [url] = await storage
@@ -60,6 +74,7 @@ export const generateDownloadSignedUrl = async (
     .bucket(bucketName)
     .file(filePathName)
     .getSignedUrl(options)
+  console.log('generating download signed url', url)
   return url
 }