Add an option for the original html to be parsed by distiller

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

@@ -68,6 +68,7 @@ import {
   validatedDate,
 } from '../../utils/helpers'
 import {
+  getDistillerResult,
   htmlToMarkdown,
   ParsedContentPuppeteer,
   parsePreparedContent,
@@ -106,6 +107,12 @@ import { saveSearchHistory } from '../../services/search_history'
 import { parsedContentToPage } from '../../services/save_page'
 import * as httpContext from 'express-http-context'
 
+enum ArticleFormat {
+  Markdown = 'markdown',
+  Html = 'html',
+  Distiller = 'distiller',
+}
+
 export type PartialArticle = Omit<
   Article,
   | 'updatedAt'
@@ -413,7 +420,9 @@ export const getArticleResolver: ResolverFn<
       return { errorCodes: [ArticleErrorCode.Unauthorized] }
     }
 
-    const includeOriginalHtml = !!graphqlFields(info).article.originalHtml
+    const includeOriginalHtml =
+      format === ArticleFormat.Distiller ||
+      !!graphqlFields(info).article.originalHtml
 
     analytics.track({
       userId: claims?.uid,
@@ -449,8 +458,17 @@ export const getArticleResolver: ResolverFn<
       page.content = UNPARSEABLE_CONTENT
     }
 
-    if (format === 'markdown') {
+    if (format === ArticleFormat.Markdown) {
       page.content = htmlToMarkdown(page.content)
+    } else if (format === ArticleFormat.Distiller) {
+      if (!page.originalHtml) {
+        return { errorCodes: [ArticleErrorCode.BadData] }
+      }
+      const distillerResult = await getDistillerResult(page.originalHtml)
+      if (!distillerResult) {
+        return { errorCodes: [ArticleErrorCode.BadData] }
+      }
+      page.content = distillerResult
     }
 
     return {

packages/api/src/utils/parser.ts

@@ -485,3 +485,23 @@ const nhm = new NodeHtmlMarkdown(
 export const htmlToMarkdown = (html: string) => {
   return nhm.translate(/* html */ html)
 }
+
+export const getDistillerResult = async (
+  html: string
+): Promise<string | undefined> => {
+  try {
+    const url = process.env.DISTILLER_URL
+    if (!url) {
+      console.log('No distiller url')
+      return undefined
+    }
+
+    const response = await axios.post<string>(url, html, {
+      timeout: 5000,
+    })
+    return response.data
+  } catch (e) {
+    console.log('Error parsing by distiller', e)
+    return undefined
+  }
+}