From b1f92f4126e4f9b2eab0e04c8d0b69b63d79eabc Mon Sep 17 00:00:00 2001
From: Alexander Zinchuk <alx.zinchuk@gmail.com>
Date: Fri, 3 Feb 2023 03:22:02 +0100
Subject: [PATCH] Paste html: Clean html from Google Docs (#2476)

---
 .../composer/hooks/useClipboardPaste.ts       |  11 +-
 src/lib/cleanDocsHtml.ts                      | 183 ++++++++++++++++++
 src/util/parseMessageInput.ts                 |   1 +
 3 files changed, 194 insertions(+), 1 deletion(-)
 create mode 100644 src/lib/cleanDocsHtml.ts

diff --git a/src/components/middle/composer/hooks/useClipboardPaste.ts b/src/components/middle/composer/hooks/useClipboardPaste.ts
index ea746a682..378a88c40 100644
--- a/src/components/middle/composer/hooks/useClipboardPaste.ts
+++ b/src/components/middle/composer/hooks/useClipboardPaste.ts
@@ -5,9 +5,10 @@ import type { ApiAttachment, ApiFormattedText, ApiMessage } from '../../../../ap
 import { ApiMessageEntityTypes } from '../../../../api/types';
 
 import buildAttachment from '../helpers/buildAttachment';
-import { EDITABLE_INPUT_ID, EDITABLE_INPUT_MODAL_ID } from '../../../../config';
+import { DEBUG, EDITABLE_INPUT_ID, EDITABLE_INPUT_MODAL_ID } from '../../../../config';
 import getFilesFromDataTransferItems from '../helpers/getFilesFromDataTransferItems';
 import parseMessageInput, { ENTITY_CLASS_BY_NODE_NAME } from '../../../../util/parseMessageInput';
+import cleanDocsHtml from '../../../../lib/cleanDocsHtml';
 import { containsCustomEmoji, stripCustomEmoji } from '../../../../global/helpers/symbols';
 
 const MAX_MESSAGE_LENGTH = 4096;
@@ -16,6 +17,14 @@ const STYLE_TAG_REGEX = /<style>(.*?)<\/style>/gs;
 
 function preparePastedHtml(html: string) {
   let fragment = document.createElement('div');
+  try {
+    html = cleanDocsHtml(html);
+  } catch (err) {
+    if (DEBUG) {
+      // eslint-disable-next-line no-console
+      console.error(err);
+    }
+  }
   fragment.innerHTML = html.replace(/\u00a0/g, ' ').replace(STYLE_TAG_REGEX, ''); // Strip &nbsp and styles
 
   const textContents = fragment.querySelectorAll<HTMLDivElement>('.text-content');
diff --git a/src/lib/cleanDocsHtml.ts b/src/lib/cleanDocsHtml.ts
new file mode 100644
index 000000000..2d0b20369
--- /dev/null
+++ b/src/lib/cleanDocsHtml.ts
@@ -0,0 +1,183 @@
+// Utility for cleaning html code from Google Docs.
+// Original source from DocsSoap:
+// https://www.npmjs.com/package/docs-soap
+
+const GDOCS_ELEMENT_ID_REGEXP = /id="docs-internal-guid/i;
+
+const GDOCS_STYLES = {
+  BOLD: '700',
+  ITALIC: 'italic',
+  UNDERLINE: 'underline',
+  STRIKETHROUGH: 'line-through',
+  SUPERSCRIPT: 'super',
+  SUBSCRIPT: 'sub',
+};
+
+const ELEMENTS = {
+  ANCHOR: 'a',
+  BOLD: 'strong',
+  ITALIC: 'em',
+  UNDERLINE: 'u',
+  STRIKETHROUGH: 'del',
+  SUPERSCRIPT: 'sup',
+  SUBSCRIPT: 'sub',
+};
+
+const headers = [
+  'H1',
+  'H2',
+  'H3',
+  'H4',
+  'H5',
+  'H6',
+];
+
+function parseHtml(html: string): HTMLElement {
+  const parser = new DOMParser();
+  const doc = parser.parseFromString(html, 'text/html');
+
+  return doc.body;
+}
+
+function wrapNodeAnchor(node: Node, href: string): HTMLAnchorElement {
+  const anchor = document.createElement(ELEMENTS.ANCHOR) as HTMLAnchorElement;
+  anchor.href = href;
+  anchor.appendChild(node.cloneNode(true));
+
+  return anchor;
+}
+
+function wrapNodeInline(node: Node, style: string): Node {
+  const el = document.createElement(style);
+  el.appendChild(node.cloneNode(true));
+
+  return el;
+}
+
+function wrapNode(inner: HTMLElement, result: Node): Node {
+  let newNode = result.cloneNode(true);
+  if (!inner) {
+    return newNode;
+  }
+  if (inner.style && inner.style.fontWeight === GDOCS_STYLES.BOLD) {
+    newNode = wrapNodeInline(newNode, ELEMENTS.BOLD);
+  }
+  if (inner.style && inner.style.fontStyle === GDOCS_STYLES.ITALIC) {
+    newNode = wrapNodeInline(newNode, ELEMENTS.ITALIC);
+  }
+  if (inner.style && inner.style.textDecoration === GDOCS_STYLES.UNDERLINE) {
+    newNode = wrapNodeInline(newNode, ELEMENTS.UNDERLINE);
+  }
+  if (inner.style && inner.style.textDecoration === GDOCS_STYLES.STRIKETHROUGH) {
+    newNode = wrapNodeInline(newNode, ELEMENTS.STRIKETHROUGH);
+  }
+  if (inner.style && inner.style.verticalAlign === GDOCS_STYLES.SUPERSCRIPT) {
+    newNode = wrapNodeInline(newNode, ELEMENTS.SUPERSCRIPT);
+  }
+  if (inner.style && inner.style.verticalAlign === GDOCS_STYLES.SUBSCRIPT) {
+    newNode = wrapNodeInline(newNode, ELEMENTS.SUBSCRIPT);
+  }
+
+  return newNode;
+}
+
+function applyBlockStyles(dirty: Node): Node {
+  const node = dirty.cloneNode(true);
+  let newNode = document.createTextNode(node.textContent || '') as Node;
+  let styledNode = document.createTextNode('') as Node;
+  if ('style' in node.childNodes[0] && Boolean(node.childNodes[0].style)) {
+    styledNode = node.childNodes[0];
+  }
+  if (node.childNodes[0] && node.childNodes[0].nodeName === 'A') {
+    newNode = wrapNodeAnchor(newNode.cloneNode(true), (node.childNodes[0] as HTMLAnchorElement).href);
+    styledNode = node.childNodes[0].childNodes[0];
+  }
+  newNode = wrapNode(styledNode as HTMLElement, newNode);
+  return newNode;
+}
+
+function applyInlineStyles(dirty: Node): Node {
+  const node = dirty.cloneNode(true);
+  let newNode = document.createTextNode(node.textContent || '') as Node;
+  let styledNode = node;
+  if (node.nodeName === 'A') {
+    newNode = wrapNodeAnchor(newNode, (node as HTMLAnchorElement).href);
+    if ('style' in node.childNodes[0] && Boolean(node.childNodes[0].style)) {
+      styledNode = node.childNodes[0];
+    }
+  }
+  newNode = wrapNode(styledNode as HTMLElement, newNode);
+  return newNode;
+}
+
+function getCleanNode(node: Node): Node[] {
+  if (node.childNodes && (node.childNodes.length <= 1 || node.nodeName === 'OL' || node.nodeName === 'UL')) {
+    let newWrapper: Node | undefined;
+    let newNode = document.createTextNode(node.textContent || '') as Node;
+    if (node.nodeName === 'UL' || node.nodeName === 'OL' || node.nodeName === 'LI') {
+      newWrapper = document.createElement(node.nodeName);
+      newNode = document.createDocumentFragment();
+      const items = [];
+      for (let i = 0; i < node.childNodes.length; i++) {
+        items.push(...getCleanNode(node.childNodes[i]));
+      }
+      items.map((i: Node): Node => newNode.appendChild(i));
+    } else if (headers.indexOf(node.nodeName) !== -1) {
+      newWrapper = document.createElement(node.nodeName);
+      newNode = applyInlineStyles(node.childNodes[0]);
+    } else if (node.nodeName === 'P') {
+      newWrapper = document.createElement('p');
+      newNode = applyBlockStyles(node);
+    } else if (node.nodeName === 'BR') {
+      newNode = node;
+    } else {
+      newWrapper = document.createElement('span');
+      newNode = applyInlineStyles(node);
+    }
+    if (newWrapper) {
+      newWrapper.appendChild(newNode);
+      return [newWrapper];
+    }
+
+    return [node.cloneNode(true)];
+  }
+
+  if (node.childNodes) {
+    const nodes = [];
+    for (let i = 0; i < node.childNodes.length; i++) {
+      nodes.push(...getCleanNode(node.childNodes[i]));
+    }
+    return nodes;
+  }
+
+  return [node];
+}
+
+function filterNode(node: Node): boolean {
+  return node.nodeType !== 8; // Node.COMMENT_NODE = 8
+}
+
+function getCleanDocument(dirty: HTMLElement): HTMLElement {
+  const body = document.createElement('body');
+  const nodes = dirty.childNodes;
+  const filteredNodes = Array.from(nodes).filter(filterNode);
+  const cleanNodes = [];
+
+  for (const node of filteredNodes) {
+    cleanNodes.push(...getCleanNode(node));
+  }
+
+  for (let i = 0; i < cleanNodes.length; i++) {
+    body.appendChild(cleanNodes[i].cloneNode(true));
+  }
+
+  return body;
+}
+
+export default function cleanDocsHtml(clipboardContent: string): string {
+  if (!clipboardContent.match(GDOCS_ELEMENT_ID_REGEXP)) {
+    return parseHtml(clipboardContent.replace(/(\r\n|\n|\r)/, '')).innerHTML;
+  }
+
+  return getCleanDocument(parseHtml(clipboardContent.replace(/(\r\n|\n|\r)/, ''))).innerHTML;
+}
diff --git a/src/util/parseMessageInput.ts b/src/util/parseMessageInput.ts
index 07d075731..c25a4ecf9 100644
--- a/src/util/parseMessageInput.ts
+++ b/src/util/parseMessageInput.ts
@@ -8,6 +8,7 @@ export const ENTITY_CLASS_BY_NODE_NAME: Record<string, ApiMessageEntityTypes> =
   STRONG: ApiMessageEntityTypes.Bold,
   I: ApiMessageEntityTypes.Italic,
   EM: ApiMessageEntityTypes.Italic,
+  INS: ApiMessageEntityTypes.Underline,
   U: ApiMessageEntityTypes.Underline,
   S: ApiMessageEntityTypes.Strike,
   STRIKE: ApiMessageEntityTypes.Strike,