refactor: extract simplifyHtmlElement to lib/html.js and add toSimplifiedHTML to WebElement

DavertMik · claude · DavertMik · commit ca7ba6b70b8b · 2026-03-17T03:30:33.000+02:00
Move HTML simplification logic (removeNonInteractiveElements + unwrap + truncate)
into a reusable simplifyHtmlElement() function in lib/html.js. Add toSimplifiedHTML()
method to WebElement that combines toOuterHTML() with simplifyHtmlElement().
MultipleElementsFound.fetchDetails() now uses webEl.toSimplifiedHTML() directly.

Co-Authored-By: Claude Opus 4.6 (1M context) &lt;noreply@anthropic.com&gt;
diff --git a/lib/element/WebElement.js b/lib/element/WebElement.js
@@ -1,4 +1,5 @@
 import assert from 'assert'
+import { simplifyHtmlElement } from '../html.js'
 
 /**
  * Unified WebElement class that wraps native element instances from different helpers
@@ -352,6 +353,11 @@ class WebElement {
     }
   }
 
+  async toSimplifiedHTML(maxLength = 300) {
+    const outerHTML = await this.toOuterHTML()
+    return simplifyHtmlElement(outerHTML, maxLength)
+  }
+
   _normalizeLocator(locator) {
     if (typeof locator === 'string') {
       return locator
diff --git a/lib/helper/errors/MultipleElementsFound.js b/lib/helper/errors/MultipleElementsFound.js
@@ -1,5 +1,4 @@
 import Locator from '../../locator.js'
-import { removeNonInteractiveElements } from '../../html.js'
 
 class MultipleElementsFound extends Error {
   constructor(locator, webElements) {
@@ -25,17 +24,8 @@ class MultipleElementsFound extends Error {
         const webEl = this.webElements[i]
         try {
           const xpath = await webEl.toAbsoluteXPath()
-          let outerHTML = await webEl.toOuterHTML()
-          try {
-            outerHTML = removeNonInteractiveElements(outerHTML)
-            outerHTML = outerHTML.replace(/<html><head><\/head><body>(.*)<\/body><\/html>/s, '$1').trim()
-          } catch (e) {
-            // keep raw outerHTML if minification fails
-          }
-          if (outerHTML.length > 300) {
-            outerHTML = outerHTML.slice(0, 300) + '...'
-          }
-          items.push(`  ${i + 1}. > ${xpath}\n     ${outerHTML}`)
+          const html = await webEl.toSimplifiedHTML()
+          items.push(`  ${i + 1}. > ${xpath}\n     ${html}`)
         } catch (err) {
           items.push(`  ${i + 1}. [Unable to get element info: ${err.message}]`)
         }
diff --git a/lib/html.js b/lib/html.js
@@ -245,4 +245,17 @@ function splitByChunks(text, chunkSize) {
   return chunks.map(chunk => chunk.trim())
 }
 
-export { scanForErrorMessages, removeNonInteractiveElements, splitByChunks, minifyHtml }
+function simplifyHtmlElement(html, maxLength = 300) {
+  try {
+    html = removeNonInteractiveElements(html)
+    html = html.replace(/<html><head><\/head><body>(.*)<\/body><\/html>/s, '$1').trim()
+  } catch (e) {
+    // keep raw html if minification fails
+  }
+  if (html.length > maxLength) {
+    html = html.slice(0, maxLength) + '...'
+  }
+  return html
+}
+
+export { scanForErrorMessages, removeNonInteractiveElements, splitByChunks, minifyHtml, simplifyHtmlElement }
