From 0f489d54c3513e8abab09fd261a12367cc922b81 Mon Sep 17 00:00:00 2001
From: Tom Moor <tom.moor@gmail.com>
Date: Fri, 30 Dec 2022 14:49:53 -0500
Subject: [PATCH] Add DocumentHelper.toPlainText

---
 server/models/helpers/DocumentHelper.test.ts | 63 +++++++++++++++++++-
 server/models/helpers/DocumentHelper.tsx     | 21 ++++++-
 2 files changed, 82 insertions(+), 2 deletions(-)

diff --git a/server/models/helpers/DocumentHelper.test.ts b/server/models/helpers/DocumentHelper.test.ts
index dda1f2ce2..9dec809a3 100644
--- a/server/models/helpers/DocumentHelper.test.ts
+++ b/server/models/helpers/DocumentHelper.test.ts
@@ -1,7 +1,7 @@
 import Revision from "@server/models/Revision";
 import DocumentHelper from "./DocumentHelper";
 
-describe("toEmailDiff", () => {
+describe("DocumentHelper", () => {
   test("toEmailDiff", async () => {
     const before = new Revision({
       title: "Title",
@@ -80,4 +80,65 @@ same on both sides`,
     expect(html).not.toContain("same on both sides");
     expect(html).not.toContain("this is a highlight");
   });
+
+  test("toPlainText", async () => {
+    const revision = new Revision({
+      title: "Title",
+      text: `
+This is a test paragraph
+
+A new [link](https://www.google.com)
+
+- list item 1
+
+This is a new paragraph.
+
+!!This is a placeholder!!
+
+==this is a highlight==
+
+- [x] checklist item 1
+- [x] checklist item 2
+- [ ] checklist item 3
+- [ ] checklist item 4
+- [x] checklist item 5
+
+same on both sides
+
+same on both sides
+
+same on both sides`,
+    });
+
+    const text = await DocumentHelper.toPlainText(revision);
+
+    // Strip all formatting
+    expect(text).toEqual(`This is a test paragraph
+
+A new link
+
+list item 1
+
+This is a new paragraph.
+
+This is a placeholder
+
+this is a highlight
+
+checklist item 1
+
+checklist item 2
+
+checklist item 3
+
+checklist item 4
+
+checklist item 5
+
+same on both sides
+
+same on both sides
+
+same on both sides`);
+  });
 });
diff --git a/server/models/helpers/DocumentHelper.tsx b/server/models/helpers/DocumentHelper.tsx
index d8af4f1d0..55ded7696 100644
--- a/server/models/helpers/DocumentHelper.tsx
+++ b/server/models/helpers/DocumentHelper.tsx
@@ -10,6 +10,7 @@ import { renderToString } from "react-dom/server";
 import styled, { ServerStyleSheet, ThemeProvider } from "styled-components";
 import * as Y from "yjs";
 import EditorContainer from "@shared/editor/components/Styles";
+import textBetween from "@shared/editor/lib/textBetween";
 import GlobalStyles from "@shared/styles/globals";
 import light from "@shared/styles/theme";
 import {
@@ -44,7 +45,7 @@ type HTMLOptions = {
 export default class DocumentHelper {
   /**
    * Returns the document as a Prosemirror Node. This method uses the
-   * collaborative state if available, otherwise it falls back to Markdown->HTML.
+   * collaborative state if available, otherwise it falls back to Markdown.
    *
    * @param document The document or revision to convert
    * @returns The document content as a Prosemirror Node
@@ -58,6 +59,24 @@ export default class DocumentHelper {
     return parser.parse(document.text);
   }
 
+  /**
+   * Returns the document as plain text. This method uses the
+   * collaborative state if available, otherwise it falls back to Markdown.
+   *
+   * @param document The document or revision to convert
+   * @returns The document content as plain text without formatting.
+   */
+  static toPlainText(document: Document | Revision) {
+    const node = DocumentHelper.toProsemirror(document);
+    const textSerializers = Object.fromEntries(
+      Object.entries(schema.nodes)
+        .filter(([, node]) => node.spec.toPlainText)
+        .map(([name, node]) => [name, node.spec.toPlainText])
+    );
+
+    return textBetween(node, 0, node.content.size, textSerializers);
+  }
+
   /**
    * Returns the document as Markdown. This is a lossy conversion and should
    * only be used for export.