From 15b41560de3018f862ea0069bc9347a5388dad2f Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Fri, 26 May 2023 15:51:54 +0800
Subject: [PATCH] fix: improve the speed of generating highlighted markdown
 content

---
 packages/api/src/utils/highlightGenerator.ts | 32 +++++++++++++++-----
 packages/api/src/utils/parser.ts             |  8 ++++-
 2 files changed, 32 insertions(+), 8 deletions(-)

diff --git a/packages/api/src/utils/highlightGenerator.ts b/packages/api/src/utils/highlightGenerator.ts
index 58bcf325d..cc3ccca0f 100644
--- a/packages/api/src/utils/highlightGenerator.ts
+++ b/packages/api/src/utils/highlightGenerator.ts
@@ -1,4 +1,5 @@
 import { diff_match_patch as DiffMatchPatch } from 'diff-match-patch'
+import { parseHTML } from 'linkedom'
 import { nanoid } from 'nanoid'
 import { v4 as uuidv4 } from 'uuid'
 import { interpolationSearch } from './interpolationSearch'
@@ -381,18 +382,35 @@ const fillHighlight = ({
   }
 }
 
+export function getArticleTextNodes(
+  document: Document
+): ArticleTextContent | null {
+  try {
+    const rootNode = document.getRootNode()
+    return getTextNodesBetween(rootNode, rootNode, rootNode)
+  } catch (error) {
+    console.log(error)
+    return null
+  }
+}
+
 export function makeHighlightNodeAttributes(
   id: string,
   patch: string,
-  document: Document
+  articleTextNodes: ArticleTextContent
 ) {
-  const rootNode = document.getRootNode()
+  const document = parseHTML('').document
+  const textNodes = articleTextNodes.textNodes
+  const { highlightTextStart, highlightTextEnd } = selectionOffsetsFromPatch(
+    articleTextNodes.articleText,
+    patch
+  )
 
-  const allArticleNodes = getTextNodesBetween(rootNode, rootNode, rootNode)
-  const { highlightTextStart, highlightTextEnd, textNodes, textNodeIndex } =
-    getPrefixAndSuffix(allArticleNodes, patch)
-
-  let startingTextNodeIndex = textNodeIndex
+  // Searching for the starting text node using interpolation search algorithm
+  let startingTextNodeIndex = interpolationSearch(
+    textNodes.map(({ startIndex: startIndex }) => startIndex),
+    highlightTextStart
+  )
   let quote = ''
 
   while (
diff --git a/packages/api/src/utils/parser.ts b/packages/api/src/utils/parser.ts
index d3f05c1d2..6d255677a 100644
--- a/packages/api/src/utils/parser.ts
+++ b/packages/api/src/utils/parser.ts
@@ -24,6 +24,7 @@ import { ArticleFormat } from '../resolvers/article'
 import {
   EmbeddedHighlightData,
   findEmbeddedHighlight,
+  getArticleTextNodes,
   highlightIdAttribute,
   makeHighlightNodeAttributes,
 } from './highlightGenerator'
@@ -627,6 +628,11 @@ export const htmlToHighlightedMarkdown = (
     return nhm.translate(/* html */ html)
   }
 
+  const articleTextNodes = getArticleTextNodes(document)
+  if (!articleTextNodes) {
+    return nhm.translate(/* html */ html)
+  }
+
   // wrap highlights in special tags
   highlights
     .filter((h) => h.type == 'HIGHLIGHT' && h.patch)
@@ -635,7 +641,7 @@ export const htmlToHighlightedMarkdown = (
         makeHighlightNodeAttributes(
           highlight.id,
           highlight.patch as string,
-          document
+          articleTextNodes
         )
       } catch (err) {
         console.log(err)