Preserve content when pasting multiple HTML tables

src/paste-markdown-html.ts

@@ -10,6 +10,7 @@ export function uninstall(el: HTMLElement): void {
 }
 
 function onPaste(event: ClipboardEvent) {
+  if (event.defaultPrevented) return
   const transfer = event.clipboardData
   const {currentTarget: el} = event
   if (shouldSkipFormatting(el as HTMLElement)) return

src/paste-markdown-image-link.ts

@@ -14,6 +14,7 @@ export function uninstall(el: HTMLElement): void {
 }
 
 function onDrop(event: DragEvent) {
+  if (event.defaultPrevented) return
   const transfer = event.dataTransfer
   if (!transfer) return
 
@@ -37,6 +38,7 @@ function onDragover(event: DragEvent) {
 }
 
 function onPaste(event: ClipboardEvent) {
+  if (event.defaultPrevented) return
   const {currentTarget: el} = event
   if (shouldSkipFormatting(el as HTMLElement)) return
 

src/paste-markdown-link.ts

@@ -14,6 +14,7 @@ export function uninstall(el: HTMLElement): void {
 }
 
 function onPaste(event: ClipboardEvent) {
+  if (event.defaultPrevented) return
   const {currentTarget: el} = event
   const element = el as HTMLElement
   const shouldPasteAsPlainText = pasteLinkAsPlainTextOverSelectedTextMap.get(element) ?? false

src/paste-markdown-table.ts

@@ -14,6 +14,8 @@ export function uninstall(el: HTMLElement): void {
 }
 
 function onDrop(event: DragEvent) {
+  if (event.defaultPrevented) return
+  if (!(event.currentTarget instanceof HTMLTextAreaElement)) return
   const transfer = event.dataTransfer
   if (!transfer) return
   if (hasFile(transfer)) return
@@ -36,7 +38,9 @@ function onDragover(event: DragEvent) {
 }
 
 function onPaste(event: ClipboardEvent) {
+  if (event.defaultPrevented) return
   const {currentTarget: el} = event
+  if (!(el instanceof HTMLTextAreaElement)) return
   if (shouldSkipFormatting(el as HTMLElement)) return
 
   if (!event.clipboardData) return
@@ -90,16 +94,24 @@ function generateText(transfer: DataTransfer): string | undefined {
 
   const html = transfer.getData('text/html')
   if (!/<table/i.test(html)) return
+  const parsedDocument = new DOMParser().parseFromString(html, 'text/html')
+  const tables = parsedDocument.querySelectorAll('table')
+  if (!tables.length) return
+  const openingTables = countMatches(html, /<table(?=[\s>])/gi)
+  const closingTables = countMatches(html, /<\/table\s*>/gi)
+  if (tables.length !== 1 || openingTables !== 1 || closingTables !== 1) {
+    return hasCompletePlaintext(parsedDocument, transfer.getData('text/plain')) ? transfer.getData('text/plain') : html
+  }
+  // The existing source slicing is case-sensitive. Preserve other valid casing.
+  // eslint-disable-next-line github/unescaped-html-literal
+  if (!html.includes('<table') || !html.includes('</table>')) return html
 
   // eslint-disable-next-line github/unescaped-html-literal
   const start = html.substring(0, html.indexOf('<table'))
   const tableCloseIndex = html.lastIndexOf('</table>')
   if (!start || !tableCloseIndex) return
   const end = html.substring(tableCloseIndex + 8)
 
-  const parser = new DOMParser()
-  const parsedDocument = parser.parseFromString(html, 'text/html')
-
   let table = parsedDocument.querySelector('table')
   table = !table || table.closest('[data-paste-markdown-skip]') ? null : table
   if (!table) return
@@ -110,3 +122,121 @@ function generateText(transfer: DataTransfer): string | undefined {
 
   return [start, formattedTable, end].join('').replace(/<meta.*?>/, '')
 }
+
+function countMatches(text: string, pattern: RegExp): number {
+  let count = 0
+  while (pattern.exec(text)) {
+    count++
+    if (count > 1) break
+  }
+  return count
+}
+
+// Accept only simple paragraphs and independent tables. Unknown markup falls back to
+// the original HTML, so an incomplete plain-text representation is never trusted.
+function hasCompletePlaintext(document: Document, plain: string): boolean {
+  if (!plain || plain.length > 1_000_000) return false
+  if (document.createTreeWalker(document, NodeFilter.SHOW_COMMENT).nextNode()) return false
+  if (document.documentElement.hasAttributes() || document.body.hasAttributes()) return false
+  if (Array.from(document.head.children).some(element => element.tagName !== 'META')) return false
+  const body = document.body
+  if (Array.from(body.childNodes).some(node => node.nodeType === Node.TEXT_NODE && node.textContent?.trim()))
+    return false
+
+  const blocks = plain.trim().split(/\n\s*\n/)
+  const elements = Array.from(body.children).filter(element => element.tagName !== 'META')
+  if (blocks.length !== elements.length || elements.length > 1000) return false
+
+  return elements.every((element, index) => {
+    const block = blocks[index]
+    if (element.hasAttributes()) return false
+    if (element.tagName === 'P' && element.childNodes.length === 1 && element.firstChild?.nodeType === Node.TEXT_NODE) {
+      return (
+        element.textContent === element.textContent?.trim() &&
+        !hasMarkdownSyntax(element.textContent || '') &&
+        block === element.textContent
+      )
+    }
+    if (element.tagName !== 'TABLE' || element.querySelector('table, [data-paste-markdown-skip]')) return false
+    if (
+      Array.from(element.querySelectorAll('*')).some(
+        node => node.hasAttributes() || !['THEAD', 'TBODY', 'TR', 'TD', 'TH'].includes(node.tagName),
+      )
+    )
+      return false
+    if (
+      Array.from(element.querySelectorAll('table, thead, tbody, tfoot, tr')).some(node =>
+        Array.from(node.childNodes).some(child => child.nodeType === Node.TEXT_NODE && child.textContent?.trim()),
+      )
+    )
+      return false
+
+    const rows = Array.from(element.querySelectorAll('tr'))
+    const lines = block.split('\n')
+    if (!rows.length || rows.length > 1000 || lines.length !== rows.length + 1) return false
+    const cells = rows.map(row => Array.from(row.children))
+    if (
+      cells.some(
+        row =>
+          !row.length ||
+          row.length > 100 ||
+          row.some(cell => {
+            return (
+              !['TD', 'TH'].includes(cell.tagName) ||
+              cell.childNodes.length > 1 ||
+              (cell.firstChild !== null && cell.firstChild.nodeType !== Node.TEXT_NODE) ||
+              cell.textContent !== cell.textContent?.trim() ||
+              hasMarkdownSyntax(cell.textContent || '') ||
+              /[|\n]/.test(cell.textContent || '')
+            )
+          }),
+      )
+    )
+      return false
+    const columnCount = cells[0].length
+    if (
+      cells.some(
+        (row, rowIndex) =>
+          row.length !== columnCount ||
+          row.some(cell => cell.tagName !== (rowIndex === 0 ? 'TH' : 'TD')) ||
+          (rowIndex === 0 && row.some(cell => !cell.textContent)),
+      )
+    )
+      return false
+
+    const parseCells = (line: string): string[] | null => {
+      if (!/^\|.*\|$/.test(line.trim())) return null
+      return line
+        .trim()
+        .slice(1, -1)
+        .split('|')
+        .map(cell => cell.trim())
+    }
+    const header = parseCells(lines[0])
+    const separator = parseCells(lines[1])
+    if (!header || !separator || separator.length !== header.length || !separator.every(cell => /^-{3,}$/.test(cell)))
+      return false
+    return cells.every((row, rowIndex) => {
+      const pasted = rowIndex === 0 ? header : parseCells(lines[rowIndex + 1])
+      return (
+        pasted?.length === row.length && row.every((cell, cellIndex) => cell.textContent?.trim() === pasted[cellIndex])
+      )
+    })
+  })
+}
+
+// Only unambiguous, single-line literal text is eligible for the plain-text path.
+// The prefix checks exclude headings, rules and list markers without parsing Markdown.
+const simpleLiteralText = new RegExp('^[\\p{L}\\p{N}\\p{M} _.,#/-]+$', 'u')
+// Dotted alphanumeric words can become GFM autolinks. Numeric decimals stay literal.
+const dottedWord = new RegExp('(?:\\p{L}\\.[\\p{L}\\p{N}]|[\\p{L}\\p{N}]\\.\\p{L})', 'u')
+
+function hasMarkdownSyntax(text: string): boolean {
+  return (
+    !simpleLiteralText.test(text) ||
+    dottedWord.test(text) ||
+    /^[#_-]/.test(text) ||
+    /^\d+[.)]\s/.test(text) ||
+    /(^|[^a-zA-Z0-9])_|_($|[^a-zA-Z0-9])/.test(text)
+  )
+}

src/paste-markdown-text.ts

@@ -10,6 +10,7 @@ export function uninstall(el: HTMLElement): void {
 }
 
 function onPaste(event: ClipboardEvent) {
+  if (event.defaultPrevented) return
   const {currentTarget: el} = event
   if (shouldSkipFormatting(el as HTMLElement)) return