From 15ed660975907f659472ae7ab309f153cf902231 Mon Sep 17 00:00:00 2001
From: Michael Telatynski <7t3chguy@gmail.com>
Date: Fri, 12 May 2023 12:13:08 +0100
Subject: [PATCH] Improve bundle size, dynamic imports & remove parse5 (#10865)

* Remove unused import

* Lazy load tar-js and pako for rageshakes

* Update cheerio imports

* Replace parse5 with DOMParser

* Remove stale comment
---
 __mocks__/maplibre-gl.js          |  2 +-
 package.json                      |  2 --
 src/HtmlUtils.tsx                 |  4 ++--
 src/SlashCommands.tsx             | 16 +++++++---------
 src/editor/serialize.ts           |  6 +++---
 src/rageshake/submit-rageshake.ts |  9 +++++++--
 yarn.lock                         | 10 ----------
 7 files changed, 20 insertions(+), 29 deletions(-)

diff --git a/__mocks__/maplibre-gl.js b/__mocks__/maplibre-gl.js
index 77ee0e9a02..6c94f7c77b 100644
--- a/__mocks__/maplibre-gl.js
+++ b/__mocks__/maplibre-gl.js
@@ -15,7 +15,7 @@ limitations under the License.
 */
 
 const EventEmitter = require("events");
-const { LngLat, NavigationControl, LngLatBounds, AttributionControl } = require("maplibre-gl");
+const { LngLat, NavigationControl, LngLatBounds } = require("maplibre-gl");
 
 class MockMap extends EventEmitter {
     addControl = jest.fn();
diff --git a/package.json b/package.json
index ac5657372e..4a48d01b3d 100644
--- a/package.json
+++ b/package.json
@@ -102,7 +102,6 @@
         "minimist": "^1.2.5",
         "opus-recorder": "^8.0.3",
         "pako": "^2.0.3",
-        "parse5": "^6.0.1",
         "png-chunks-extract": "^1.0.0",
         "posthog-js": "1.53.2",
         "proposal-temporal": "^0.9.0",
@@ -164,7 +163,6 @@
         "@types/node": "^16",
         "@types/node-fetch": "^2.6.2",
         "@types/pako": "^2.0.0",
-        "@types/parse5": "^6.0.0",
         "@types/qrcode": "^1.3.5",
         "@types/react": "17.0.58",
         "@types/react-beautiful-dnd": "^13.0.0",
diff --git a/src/HtmlUtils.tsx b/src/HtmlUtils.tsx
index 71f3c9e079..f8dd52a4f0 100644
--- a/src/HtmlUtils.tsx
+++ b/src/HtmlUtils.tsx
@@ -19,7 +19,7 @@ limitations under the License.
 
 import React, { LegacyRef, ReactElement, ReactNode } from "react";
 import sanitizeHtml from "sanitize-html";
-import cheerio from "cheerio";
+import { load as cheerio } from "cheerio";
 import classNames from "classnames";
 import EMOJIBASE_REGEX from "emojibase-regex";
 import { merge, split } from "lodash";
@@ -549,7 +549,7 @@ export function bodyToHtml(content: IContent, highlights: Optional<string[]>, op
             }
 
             safeBody = sanitizeHtml(formattedBody!, sanitizeParams);
-            const phtml = cheerio.load(safeBody, {
+            const phtml = cheerio(safeBody, {
                 // @ts-ignore: The `_useHtmlParser2` internal option is the
                 // simplest way to both parse and render using `htmlparser2`.
                 _useHtmlParser2: true,
diff --git a/src/SlashCommands.tsx b/src/SlashCommands.tsx
index f3355388f2..3615e1a8a3 100644
--- a/src/SlashCommands.tsx
+++ b/src/SlashCommands.tsx
@@ -22,7 +22,6 @@ import { User } from "matrix-js-sdk/src/models/user";
 import { Direction } from "matrix-js-sdk/src/models/event-timeline";
 import { EventType } from "matrix-js-sdk/src/@types/event";
 import * as ContentHelpers from "matrix-js-sdk/src/content-helpers";
-import { Element as ChildElement, parseFragment as parseHtml } from "parse5";
 import { logger } from "matrix-js-sdk/src/logger";
 import { IContent } from "matrix-js-sdk/src/models/event";
 import { MRoomTopicEventContent } from "matrix-js-sdk/src/@types/topic";
@@ -1003,16 +1002,15 @@ export const Commands = [
 
             // Try and parse out a widget URL from iframes
             if (widgetUrl.toLowerCase().startsWith("<iframe ")) {
-                // We use parse5, which doesn't render/create a DOM node. It instead runs
-                // some superfast regex over the text so we don't have to.
-                const embed = parseHtml(widgetUrl);
+                const embed = new DOMParser().parseFromString(widgetUrl, "text/html").body;
                 if (embed?.childNodes?.length === 1) {
-                    const iframe = embed.childNodes[0] as ChildElement;
-                    if (iframe.tagName.toLowerCase() === "iframe" && iframe.attrs) {
-                        const srcAttr = iframe.attrs.find((a) => a.name === "src");
+                    const iframe = embed.firstElementChild;
+                    if (iframe.tagName.toLowerCase() === "iframe") {
                         logger.log("Pulling URL out of iframe (embed code)");
-                        if (!srcAttr) return reject(new UserFriendlyError("iframe has no src attribute"));
-                        widgetUrl = srcAttr.value;
+                        if (!iframe.hasAttribute("src")) {
+                            return reject(new UserFriendlyError("iframe has no src attribute"));
+                        }
+                        widgetUrl = iframe.getAttribute("src");
                     }
                 }
             }
diff --git a/src/editor/serialize.ts b/src/editor/serialize.ts
index 20ed9ab67e..b01719afb8 100644
--- a/src/editor/serialize.ts
+++ b/src/editor/serialize.ts
@@ -16,7 +16,7 @@ limitations under the License.
 */
 
 import { encode } from "html-entities";
-import cheerio from "cheerio";
+import { load as cheerio } from "cheerio";
 import escapeHtml from "escape-html";
 
 import Markdown from "../Markdown";
@@ -143,7 +143,7 @@ export function htmlSerializeFromMdIfNeeded(md: string, { forceHTML = false } =
     const parser = new Markdown(md);
     if (!parser.isPlainText() || forceHTML) {
         // feed Markdown output to HTML parser
-        const phtml = cheerio.load(parser.toHTML(), {
+        const phtml = cheerio(parser.toHTML(), {
             // @ts-ignore: The `_useHtmlParser2` internal option is the
             // simplest way to both parse and render using `htmlparser2`.
             _useHtmlParser2: true,
@@ -153,7 +153,7 @@ export function htmlSerializeFromMdIfNeeded(md: string, { forceHTML = false } =
         if (SettingsStore.getValue("feature_latex_maths")) {
             // original Markdown without LaTeX replacements
             const parserOrig = new Markdown(orig);
-            const phtmlOrig = cheerio.load(parserOrig.toHTML(), {
+            const phtmlOrig = cheerio(parserOrig.toHTML(), {
                 // @ts-ignore: The `_useHtmlParser2` internal option is the
                 // simplest way to both parse and render using `htmlparser2`.
                 _useHtmlParser2: true,
diff --git a/src/rageshake/submit-rageshake.ts b/src/rageshake/submit-rageshake.ts
index bd40aefe7a..b62688325f 100644
--- a/src/rageshake/submit-rageshake.ts
+++ b/src/rageshake/submit-rageshake.ts
@@ -16,10 +16,9 @@ See the License for the specific language governing permissions and
 limitations under the License.
 */
 
-import pako from "pako";
-import Tar from "tar-js";
 import { logger } from "matrix-js-sdk/src/logger";
 
+import type * as Pako from "pako";
 import { MatrixClientPeg } from "../MatrixClientPeg";
 import PlatformPeg from "../PlatformPeg";
 import { _t } from "../languageHandler";
@@ -177,6 +176,11 @@ async function collectBugReport(opts: IOpts = {}, gzipLogs = true): Promise<Form
     body.append("mx_local_settings", localStorage.getItem("mx_local_settings")!);
 
     if (opts.sendLogs) {
+        let pako: typeof Pako | undefined;
+        if (gzipLogs) {
+            pako = await import("pako");
+        }
+
         progressCallback(_t("Collecting logs"));
         const logs = await rageshake.getLogsForReport();
         for (const entry of logs) {
@@ -237,6 +241,7 @@ export default async function sendBugReport(bugReportEndpoint?: string, opts: IO
  * @return {Promise} Resolved when the bug report is downloaded (or started).
  */
 export async function downloadBugReport(opts: IOpts = {}): Promise<void> {
+    const Tar = (await import("tar-js")).default;
     const progressCallback = opts.progressCallback || ((): void => {});
     const body = await collectBugReport(opts, false);
 
diff --git a/yarn.lock b/yarn.lock
index b5152d24d1..b2b35faeab 100644
--- a/yarn.lock
+++ b/yarn.lock
@@ -2293,11 +2293,6 @@
   resolved "https://registry.yarnpkg.com/@types/parse-json/-/parse-json-4.0.0.tgz#2f8bb441434d163b35fb8ffdccd7138927ffb8c0"
   integrity sha512-//oorEZjL6sbPcKUaCdIGlIUeH26mgzimjBB77G6XRgnDl/L5wOnpyBGRe/Mmf5CVW3PwEBE1NjiMZ/ssFh4wA==
 
-"@types/parse5@^6.0.0":
-  version "6.0.3"
-  resolved "https://registry.yarnpkg.com/@types/parse5/-/parse5-6.0.3.tgz#705bb349e789efa06f43f128cef51240753424cb"
-  integrity sha512-SuT16Q1K51EAVPz1K29DJ/sXjhSQ0zjvsypYJ6tlwVsRV9jwW5Adq2ch8Dq8kDBCkYnELS7N7VNCSB5nC56t/g==
-
 "@types/pbf@*", "@types/pbf@^3.0.2":
   version "3.0.2"
   resolved "https://registry.yarnpkg.com/@types/pbf/-/pbf-3.0.2.tgz#8d291ad68b4b8c533e96c174a2e3e6399a59ed61"
@@ -6990,11 +6985,6 @@ parse5-htmlparser2-tree-adapter@^7.0.0:
     domhandler "^5.0.2"
     parse5 "^7.0.0"
 
-parse5@^6.0.1:
-  version "6.0.1"
-  resolved "https://registry.yarnpkg.com/parse5/-/parse5-6.0.1.tgz#e1a1c085c569b3dc08321184f19a39cc27f7c30b"
-  integrity sha512-Ofn/CTFzRGTTxwpNEs9PP93gXShHcTq255nzRYSKe8AkVpZY7e1fpmTfOyoIvjP5HG7Z2ZM7VS9PPhQGW2pOpw==
-
 parse5@^7.0.0, parse5@^7.1.1:
   version "7.1.2"
   resolved "https://registry.yarnpkg.com/parse5/-/parse5-7.1.2.tgz#0736bebbfd77793823240a23b7fc5e010b7f8e32"