From 665eef981bc71635a8c4ce579fb740ee86a541f2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=D0=9C=D0=B0=D0=BA=D1=81=D0=B8=D0=BC=20=D0=A2=D1=80=D0=B0?=
 =?UTF-8?q?=D0=B2=D0=B8=D0=BD=D1=86=D0=B5=D0=B2?= <smsochneg@yandex-team.ru>
Date: Wed, 15 Feb 2023 18:39:45 +0530
Subject: [PATCH] feat: get markup from html that has only text

---
 .../behavior/Clipboard/clipboard.ts           | 27 ++++++++++++++++---
 src/extensions/behavior/Clipboard/utils.ts    | 22 +++++++++++++++
 2 files changed, 45 insertions(+), 4 deletions(-)

diff --git a/src/extensions/behavior/Clipboard/clipboard.ts b/src/extensions/behavior/Clipboard/clipboard.ts
index 61d79b140..844bbf77d 100644
--- a/src/extensions/behavior/Clipboard/clipboard.ts
+++ b/src/extensions/behavior/Clipboard/clipboard.ts
@@ -19,7 +19,7 @@ import {isTextSelection, isNodeSelection, isWholeSelection} from '../../../utils
 import {BaseNode, pType} from '../../base/BaseSchema';
 
 import {isInsideCode} from './code';
-import {DataTransferType, isIosSafariShare} from './utils';
+import {DataTransferType, extractTextContentFromHtml, isIosSafariShare} from './utils';
 
 export type ClipboardPluginOptions = {
     yfmParser: Parser;
@@ -99,12 +99,31 @@ export const clipboard = ({
                         !e.clipboardData.types.includes(DataTransferType.Yfm) &&
                         (data = e.clipboardData.getData(DataTransferType.Html))
                     ) {
-                        return false; // default html pasting
+                        const textFromHtml = extractTextContentFromHtml(data);
+                        if (textFromHtml) {
+                            const res = tryCatch(() => textParser.parse(textFromHtml));
+                            if (res.success) {
+                                const docNode = res.result;
+                                const slice = getSliceFromMarkupFragment(docNode.content);
+                                view.dispatch(
+                                    trackTransactionMetrics(
+                                        view.state.tr.replaceSelection(slice),
+                                        'paste',
+                                        {clipboardDataFormat: DataTransferType.Html},
+                                    ),
+                                );
+                                isPasteHandled = true;
+                            } else {
+                                logger.error(res.error);
+                                console.error(res.error);
+                            }
+                        } else return false; // default html pasting
                     }
 
                     if (
-                        (data = e.clipboardData.getData(DataTransferType.Yfm)) ||
-                        (data = e.clipboardData.getData(DataTransferType.Text))
+                        !isPasteHandled &&
+                        ((data = e.clipboardData.getData(DataTransferType.Yfm)) ||
+                            (data = e.clipboardData.getData(DataTransferType.Text)))
                     ) {
                         let parser: Parser;
                         let dataFormat: string;
diff --git a/src/extensions/behavior/Clipboard/utils.ts b/src/extensions/behavior/Clipboard/utils.ts
index 0f36f6444..abb010d1d 100644
--- a/src/extensions/behavior/Clipboard/utils.ts
+++ b/src/extensions/behavior/Clipboard/utils.ts
@@ -33,3 +33,25 @@ export function isFilesFromHtml({types}: DataTransfer): boolean {
 export function isImageFile(file: File): boolean {
     return file.type.startsWith('image/');
 }
+
+export function extractTextContentFromHtml(html: string) {
+    const element = document.createElement('div');
+    element.innerHTML = html;
+    element.replaceChildren(...Array.from(element.children).filter((v) => v.nodeName !== 'META'));
+
+    // Look if all nodes are div or span. That they don't have any classname and have only text child.
+    if (
+        Array.from(element.children).every(({nodeName, classList, childNodes}) => {
+            return (
+                (nodeName === 'DIV' || nodeName === 'SPAN') &&
+                !classList.length &&
+                childNodes.length === 1 &&
+                (childNodes[0].nodeName === '#text' || childNodes[0].nodeName === 'BR')
+            );
+        })
+    ) {
+        return Array.from(element.children).reduce((a, v) => a + v.textContent + '\n', '');
+    }
+
+    return null;
+}