From e9b31e375f3987c1f50b8aae60ccd8597271437e Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Tue, 27 Dec 2022 21:59:31 +0800
Subject: [PATCH] Update savePage API to accept  parsed content in the param

---
 packages/api/src/generated/graphql.ts     | 18 ++++++++++++++++++
 packages/api/src/generated/schema.graphql | 16 ++++++++++++++++
 packages/api/src/schema.ts                | 16 ++++++++++++++++
 packages/api/src/services/save_page.ts    | 16 ++++++++++------
 packages/api/src/utils/parser.ts          | 15 ++++++++++++---
 5 files changed, 72 insertions(+), 9 deletions(-)

diff --git a/packages/api/src/generated/graphql.ts b/packages/api/src/generated/graphql.ts
index 07b53580a..0c26798f8 100644
--- a/packages/api/src/generated/graphql.ts
+++ b/packages/api/src/generated/graphql.ts
@@ -1612,6 +1612,21 @@ export enum PageType {
   Website = 'WEBSITE'
 }
 
+export type ParseResult = {
+  byline: Scalars['String'];
+  content: Scalars['String'];
+  dir: Scalars['String'];
+  excerpt: Scalars['String'];
+  language?: InputMaybe<Scalars['String']>;
+  length: Scalars['Int'];
+  previewImage?: InputMaybe<Scalars['String']>;
+  publishedDate?: InputMaybe<Scalars['Date']>;
+  siteIcon?: InputMaybe<Scalars['String']>;
+  siteName?: InputMaybe<Scalars['String']>;
+  textContent: Scalars['String'];
+  title: Scalars['String'];
+};
+
 export type PreparedDocumentInput = {
   document: Scalars['String'];
   pageInfo: PageInfoInput;
@@ -2071,6 +2086,7 @@ export type SaveFilterSuccess = {
 export type SavePageInput = {
   clientRequestId: Scalars['ID'];
   originalContent: Scalars['String'];
+  parseResult?: InputMaybe<ParseResult>;
   source: Scalars['String'];
   title?: InputMaybe<Scalars['String']>;
   url: Scalars['String'];
@@ -3291,6 +3307,7 @@ export type ResolversTypes = {
   PageInfo: ResolverTypeWrapper<PageInfo>;
   PageInfoInput: PageInfoInput;
   PageType: PageType;
+  ParseResult: ParseResult;
   PreparedDocumentInput: PreparedDocumentInput;
   Profile: ResolverTypeWrapper<Profile>;
   Query: ResolverTypeWrapper<{}>;
@@ -3704,6 +3721,7 @@ export type ResolversParentTypes = {
   Page: Page;
   PageInfo: PageInfo;
   PageInfoInput: PageInfoInput;
+  ParseResult: ParseResult;
   PreparedDocumentInput: PreparedDocumentInput;
   Profile: Profile;
   Query: {};
diff --git a/packages/api/src/generated/schema.graphql b/packages/api/src/generated/schema.graphql
index 6036c33fc..7d4df8a16 100644
--- a/packages/api/src/generated/schema.graphql
+++ b/packages/api/src/generated/schema.graphql
@@ -1169,6 +1169,21 @@ enum PageType {
   WEBSITE
 }
 
+input ParseResult {
+  byline: String!
+  content: String!
+  dir: String!
+  excerpt: String!
+  language: String
+  length: Int!
+  previewImage: String
+  publishedDate: Date
+  siteIcon: String
+  siteName: String
+  textContent: String!
+  title: String!
+}
+
 input PreparedDocumentInput {
   document: String!
   pageInfo: PageInfoInput!
@@ -1501,6 +1516,7 @@ type SaveFilterSuccess {
 input SavePageInput {
   clientRequestId: ID!
   originalContent: String!
+  parseResult: ParseResult
   source: String!
   title: String
   url: String!
diff --git a/packages/api/src/schema.ts b/packages/api/src/schema.ts
index 081d4b60f..9a17cf019 100755
--- a/packages/api/src/schema.ts
+++ b/packages/api/src/schema.ts
@@ -526,12 +526,28 @@ const schema = gql`
     uploadFileId: ID!
   }
 
+  input ParseResult {
+    title: String!
+    byline: String!
+    dir: String!
+    content: String!
+    textContent: String!
+    length: Int!
+    excerpt: String!
+    siteName: String
+    siteIcon: String
+    previewImage: String
+    publishedDate: Date
+    language: String
+  }
+
   input SavePageInput {
     url: String!
     source: String!
     clientRequestId: ID!
     title: String
     originalContent: String!
+    parseResult: ParseResult
   }
 
   input SaveUrlInput {
diff --git a/packages/api/src/services/save_page.ts b/packages/api/src/services/save_page.ts
index a7997969a..de1d18168 100644
--- a/packages/api/src/services/save_page.ts
+++ b/packages/api/src/services/save_page.ts
@@ -77,13 +77,17 @@ export const savePage = async (
   input: SavePageInput
 ): Promise<SaveResult> => {
   const [slug, croppedPathname] = createSlug(input.url, input.title)
-  const parseResult = await parsePreparedContent(input.url, {
-    document: input.originalContent,
-    pageInfo: {
-      title: input.title,
-      canonicalUrl: input.url,
+  const parseResult = await parsePreparedContent(
+    input.url,
+    {
+      document: input.originalContent,
+      pageInfo: {
+        title: input.title,
+        canonicalUrl: input.url,
+      },
     },
-  })
+    input.parseResult
+  )
 
   const articleToSave = parsedContentToPage({
     url: input.url,
diff --git a/packages/api/src/utils/parser.ts b/packages/api/src/utils/parser.ts
index e19681821..243513667 100644
--- a/packages/api/src/utils/parser.ts
+++ b/packages/api/src/utils/parser.ts
@@ -17,8 +17,8 @@ import { v4 as uuid } from 'uuid'
 import addressparser from 'addressparser'
 import { preParseContent } from '@omnivore/content-handler'
 import {
-  findEmbeddedHighlight,
   EmbeddedHighlightData,
+  findEmbeddedHighlight,
 } from './highlightGenerator'
 
 const logger = buildLogger('utils.parse')
@@ -174,6 +174,7 @@ const getReadabilityResult = async (
 export const parsePreparedContent = async (
   url: string,
   preparedDocument: PreparedDocumentInput,
+  parseResult?: Readability.ParseResult | null,
   isNewsletter?: boolean,
   allowRetry = true
 ): Promise<ParsedContentPuppeteer> => {
@@ -208,13 +209,21 @@ export const parsePreparedContent = async (
   preParsedDom && (dom = preParsedDom)
 
   try {
-    article = await getReadabilityResult(url, document, dom, isNewsletter)
+    article =
+      parseResult ||
+      (await getReadabilityResult(url, document, dom, isNewsletter))
     if (!article?.textContent && allowRetry) {
       const newDocument = {
         ...preparedDocument,
         document: '<html>' + preparedDocument.document + '</html>',
       }
-      return parsePreparedContent(url, newDocument, isNewsletter, false)
+      return parsePreparedContent(
+        url,
+        newDocument,
+        parseResult,
+        isNewsletter,
+        false
+      )
     }
 
     // Format code blocks