summaryrefslogtreecommitdiff
path: root/packages/backend/src
diff options
context:
space:
mode:
authorHazelnoot <acomputerdog@gmail.com>2025-05-21 19:17:51 -0400
committerHazelnoot <acomputerdog@gmail.com>2025-06-12 21:11:16 -0400
commit1a964cb6c037efe8e71c6e91e5c2ce032d22e107 (patch)
treec3332902f6666a256b62b791b5a6df7ed0598b5f /packages/backend/src
parentmerge: Emit log messages with correct level (!1097) (diff)
downloadsharkey-1a964cb6c037efe8e71c6e91e5c2ce032d22e107.tar.gz
sharkey-1a964cb6c037efe8e71c6e91e5c2ce032d22e107.tar.bz2
sharkey-1a964cb6c037efe8e71c6e91e5c2ce032d22e107.zip
pcleanup dependencies:
* Consolidate multiple different HTML/XML/RSS libraries to use the Cheerio stack * Remove unused deps * Move dev dependencies to correct section * Pin versions where missing
Diffstat (limited to 'packages/backend/src')
-rw-r--r--packages/backend/src/core/FetchInstanceMetadataService.ts4
-rw-r--r--packages/backend/src/core/MfmService.ts483
-rw-r--r--packages/backend/src/core/WebfingerService.ts16
-rw-r--r--packages/backend/src/core/activitypub/ApRendererService.ts45
-rw-r--r--packages/backend/src/core/activitypub/ApRequestService.ts73
-rw-r--r--packages/backend/src/misc/truncate.ts4
-rw-r--r--packages/backend/src/misc/verify-field-link.ts2
-rw-r--r--packages/backend/src/server/api/endpoints/fetch-rss.ts213
8 files changed, 397 insertions, 443 deletions
diff --git a/packages/backend/src/core/FetchInstanceMetadataService.ts b/packages/backend/src/core/FetchInstanceMetadataService.ts
index 9bfd7381f1..6fcfdfb596 100644
--- a/packages/backend/src/core/FetchInstanceMetadataService.ts
+++ b/packages/backend/src/core/FetchInstanceMetadataService.ts
@@ -7,7 +7,7 @@ import { URL } from 'node:url';
import { Inject, Injectable } from '@nestjs/common';
import tinycolor from 'tinycolor2';
import * as Redis from 'ioredis';
-import { load as cheerio } from 'cheerio';
+import { load as cheerio } from 'cheerio/slim';
import type { MiInstance } from '@/models/Instance.js';
import type Logger from '@/logger.js';
import { DI } from '@/di-symbols.js';
@@ -16,7 +16,7 @@ import { HttpRequestService } from '@/core/HttpRequestService.js';
import { bindThis } from '@/decorators.js';
import { FederatedInstanceService } from '@/core/FederatedInstanceService.js';
import { renderInlineError } from '@/misc/render-inline-error.js';
-import type { CheerioAPI } from 'cheerio';
+import type { CheerioAPI } from 'cheerio/slim';
type NodeInfo = {
openRegistrations?: unknown;
diff --git a/packages/backend/src/core/MfmService.ts b/packages/backend/src/core/MfmService.ts
index 1ee3bd2275..d85ac7c807 100644
--- a/packages/backend/src/core/MfmService.ts
+++ b/packages/backend/src/core/MfmService.ts
@@ -5,25 +5,22 @@
import { URL } from 'node:url';
import { Inject, Injectable } from '@nestjs/common';
-import * as parse5 from 'parse5';
-import { type Document, type HTMLParagraphElement, Window } from 'happy-dom';
+import { isText, isTag, Text } from 'domhandler';
+import * as htmlparser2 from 'htmlparser2';
+import { Node, Document, ChildNode, Element, ParentNode } from 'domhandler';
+import * as domserializer from 'dom-serializer';
import { DI } from '@/di-symbols.js';
import type { Config } from '@/config.js';
import { intersperse } from '@/misc/prelude/array.js';
import { normalizeForSearch } from '@/misc/normalize-for-search.js';
import type { IMentionedRemoteUsers } from '@/models/Note.js';
import { bindThis } from '@/decorators.js';
-import type { DefaultTreeAdapterMap } from 'parse5';
import type * as mfm from '@transfem-org/sfm-js';
-const treeAdapter = parse5.defaultTreeAdapter;
-type Node = DefaultTreeAdapterMap['node'];
-type ChildNode = DefaultTreeAdapterMap['childNode'];
-
const urlRegex = /^https?:\/\/[\w\/:%#@$&?!()\[\]~.,=+\-]+/;
const urlRegexFull = /^https?:\/\/[\w\/:%#@$&?!()\[\]~.,=+\-]+$/;
-export type Appender = (document: Document, body: HTMLParagraphElement) => void;
+export type Appender = (document: Document, body: Element) => void;
@Injectable()
export class MfmService {
@@ -40,7 +37,7 @@ export class MfmService {
const normalizedHashtagNames = hashtagNames == null ? undefined : new Set<string>(hashtagNames.map(x => normalizeForSearch(x)));
- const dom = parse5.parseFragment(html);
+ const dom = htmlparser2.parseDocument(html);
let text = '';
@@ -51,37 +48,31 @@ export class MfmService {
return text.trim();
function getText(node: Node): string {
- if (treeAdapter.isTextNode(node)) return node.value;
- if (!treeAdapter.isElementNode(node)) return '';
- if (node.nodeName === 'br') return '\n';
-
- if (node.childNodes) {
- return node.childNodes.map(n => getText(n)).join('');
- }
+ if (isText(node)) return node.data;
+ if (!isTag(node)) return '';
+ if (node.tagName === 'br') return '\n';
- return '';
+ return node.childNodes.map(n => getText(n)).join('');
}
function appendChildren(childNodes: ChildNode[]): void {
- if (childNodes) {
- for (const n of childNodes) {
- analyze(n);
- }
+ for (const n of childNodes) {
+ analyze(n);
}
}
function analyze(node: Node) {
- if (treeAdapter.isTextNode(node)) {
- text += node.value;
+ if (isText(node)) {
+ text += node.data;
return;
}
// Skip comment or document type node
- if (!treeAdapter.isElementNode(node)) {
+ if (!isTag(node)) {
return;
}
- switch (node.nodeName) {
+ switch (node.tagName) {
case 'br': {
text += '\n';
break;
@@ -89,19 +80,19 @@ export class MfmService {
case 'a': {
const txt = getText(node);
- const rel = node.attrs.find(x => x.name === 'rel');
- const href = node.attrs.find(x => x.name === 'href');
+ const rel = node.attribs.rel;
+ const href = node.attribs.href;
// ハッシュタグ
if (normalizedHashtagNames && href && normalizedHashtagNames.has(normalizeForSearch(txt))) {
text += txt;
// メンション
- } else if (txt.startsWith('@') && !(rel && rel.value.startsWith('me '))) {
+ } else if (txt.startsWith('@') && !(rel && rel.startsWith('me '))) {
const part = txt.split('@');
if (part.length === 2 && href) {
//#region ホスト名部分が省略されているので復元する
- const acct = `${txt}@${(new URL(href.value)).hostname}`;
+ const acct = `${txt}@${(new URL(href)).hostname}`;
text += acct;
//#endregion
} else if (part.length === 3) {
@@ -116,17 +107,17 @@ export class MfmService {
if (!href) {
return txt;
}
- if (!txt || txt === href.value) { // #6383: Missing text node
- if (href.value.match(urlRegexFull)) {
- return href.value;
+ if (!txt || txt === href) { // #6383: Missing text node
+ if (href.match(urlRegexFull)) {
+ return href;
} else {
- return `<${href.value}>`;
+ return `<${href}>`;
}
}
- if (href.value.match(urlRegex) && !href.value.match(urlRegexFull)) {
- return `[${txt}](<${href.value}>)`; // #6846
+ if (href.match(urlRegex) && !href.match(urlRegexFull)) {
+ return `[${txt}](<${href}>)`; // #6846
} else {
- return `[${txt}](${href.value})`;
+ return `[${txt}](${href})`;
}
};
@@ -185,14 +176,17 @@ export class MfmService {
case 'ruby--': {
let ruby: [string, string][] = [];
for (const child of node.childNodes) {
- if (child.nodeName === 'rp') {
+ if (isText(child) && !/\s|\[|\]/.test(child.data)) {
+ ruby.push([child.data, '']);
+ continue;
+ }
+ if (!isTag(child)) {
continue;
}
- if (treeAdapter.isTextNode(child) && !/\s|\[|\]/.test(child.value)) {
- ruby.push([child.value, '']);
+ if (child.tagName === 'rp') {
continue;
}
- if (child.nodeName === 'rt' && ruby.length > 0) {
+ if (child.tagName === 'rt' && ruby.length > 0) {
const rt = getText(child);
if (/\s|\[|\]/.test(rt)) {
// If any space is included in rt, it is treated as a normal text
@@ -217,7 +211,7 @@ export class MfmService {
// block code (<pre><code>)
case 'pre': {
- if (node.childNodes.length === 1 && node.childNodes[0].nodeName === 'code') {
+ if (node.childNodes.length === 1 && isTag(node.childNodes[0]) && node.childNodes[0].tagName === 'code') {
text += '\n```\n';
text += getText(node.childNodes[0]);
text += '\n```\n';
@@ -302,17 +296,17 @@ export class MfmService {
let nonRtNodes = [];
// scan children, ignore `rp`, split on `rt`
for (const child of node.childNodes) {
- if (treeAdapter.isTextNode(child)) {
+ if (isText(child)) {
nonRtNodes.push(child);
continue;
}
- if (!treeAdapter.isElementNode(child)) {
+ if (!isTag(child)) {
continue;
}
- if (child.nodeName === 'rp') {
+ if (child.tagName === 'rp') {
continue;
}
- if (child.nodeName === 'rt') {
+ if (child.tagName === 'rt') {
// the only case in which we don't need a `$[group ]`
// is when both sides of the ruby are simple words
const needsGroup = nonRtNodes.length > 1 ||
@@ -350,45 +344,44 @@ export class MfmService {
return null;
}
- const { happyDOM, window } = new Window();
-
- const doc = window.document;
+ const doc = new Document([]);
- const body = doc.createElement('p');
+ const body = new Element('p', {});
+ doc.childNodes.push(body);
- function appendChildren(children: mfm.MfmNode[], targetElement: any): void {
- if (children) {
- for (const child of children.map(x => (handlers as any)[x.type](x))) targetElement.appendChild(child);
+ function appendChildren(children: mfm.MfmNode[], targetElement: ParentNode): void {
+ for (const child of children.map(x => handle(x))) {
+ targetElement.childNodes.push(child);
}
}
function fnDefault(node: mfm.MfmFn) {
- const el = doc.createElement('i');
+ const el = new Element('i', {});
appendChildren(node.children, el);
return el;
}
- const handlers: { [K in mfm.MfmNode['type']]: (node: mfm.NodeType<K>) => any } = {
+ const handlers: { [K in mfm.MfmNode['type']]: (node: mfm.NodeType<K>) => ChildNode } = {
bold: (node) => {
- const el = doc.createElement('b');
+ const el = new Element('b', {});
appendChildren(node.children, el);
return el;
},
small: (node) => {
- const el = doc.createElement('small');
+ const el = new Element('small', {});
appendChildren(node.children, el);
return el;
},
strike: (node) => {
- const el = doc.createElement('del');
+ const el = new Element('del', {});
appendChildren(node.children, el);
return el;
},
italic: (node) => {
- const el = doc.createElement('i');
+ const el = new Element('i', {});
appendChildren(node.children, el);
return el;
},
@@ -399,11 +392,12 @@ export class MfmService {
const text = node.children[0].type === 'text' ? node.children[0].props.text : '';
try {
const date = new Date(parseInt(text, 10) * 1000);
- const el = doc.createElement('time');
- el.setAttribute('datetime', date.toISOString());
- el.textContent = date.toISOString();
+ const el = new Element('time', {
+ datetime: date.toISOString(),
+ });
+ el.childNodes.push(new Text(date.toISOString()));
return el;
- } catch (err) {
+ } catch {
return fnDefault(node);
}
}
@@ -412,20 +406,20 @@ export class MfmService {
if (node.children.length === 1) {
const child = node.children[0];
const text = child.type === 'text' ? child.props.text : '';
- const rubyEl = doc.createElement('ruby');
- const rtEl = doc.createElement('rt');
+ const rubyEl = new Element('ruby', {});
+ const rtEl = new Element('rt', {});
// ruby未対応のHTMLサニタイザーを通したときにルビが「劉備(りゅうび)」となるようにする
- const rpStartEl = doc.createElement('rp');
- rpStartEl.appendChild(doc.createTextNode('('));
- const rpEndEl = doc.createElement('rp');
- rpEndEl.appendChild(doc.createTextNode(')'));
+ const rpStartEl = new Element('rp', {});
+ rpStartEl.childNodes.push(new Text('('));
+ const rpEndEl = new Element('rp', {});
+ rpEndEl.childNodes.push(new Text(')'));
- rubyEl.appendChild(doc.createTextNode(text.split(' ')[0]));
- rtEl.appendChild(doc.createTextNode(text.split(' ')[1]));
- rubyEl.appendChild(rpStartEl);
- rubyEl.appendChild(rtEl);
- rubyEl.appendChild(rpEndEl);
+ rubyEl.childNodes.push(new Text(text.split(' ')[0]));
+ rtEl.childNodes.push(new Text(text.split(' ')[1]));
+ rubyEl.childNodes.push(rpStartEl);
+ rubyEl.childNodes.push(rtEl);
+ rubyEl.childNodes.push(rpEndEl);
return rubyEl;
} else {
const rt = node.children.at(-1);
@@ -435,20 +429,20 @@ export class MfmService {
}
const text = rt.type === 'text' ? rt.props.text : '';
- const rubyEl = doc.createElement('ruby');
- const rtEl = doc.createElement('rt');
+ const rubyEl = new Element('ruby', {});
+ const rtEl = new Element('rt', {});
// ruby未対応のHTMLサニタイザーを通したときにルビが「劉備(りゅうび)」となるようにする
- const rpStartEl = doc.createElement('rp');
- rpStartEl.appendChild(doc.createTextNode('('));
- const rpEndEl = doc.createElement('rp');
- rpEndEl.appendChild(doc.createTextNode(')'));
+ const rpStartEl = new Element('rp', {});
+ rpStartEl.childNodes.push(new Text('('));
+ const rpEndEl = new Element('rp', {});
+ rpEndEl.childNodes.push(new Text(')'));
appendChildren(node.children.slice(0, node.children.length - 1), rubyEl);
- rtEl.appendChild(doc.createTextNode(text.trim()));
- rubyEl.appendChild(rpStartEl);
- rubyEl.appendChild(rtEl);
- rubyEl.appendChild(rpEndEl);
+ rtEl.childNodes.push(new Text(text.trim()));
+ rubyEl.childNodes.push(rpStartEl);
+ rubyEl.childNodes.push(rtEl);
+ rubyEl.childNodes.push(rpEndEl);
return rubyEl;
}
}
@@ -456,7 +450,7 @@ export class MfmService {
// hack for ruby, should never be needed because we should
// never send this out to other instances
case 'group': {
- const el = doc.createElement('span');
+ const el = new Element('span', {});
appendChildren(node.children, el);
return el;
}
@@ -468,125 +462,135 @@ export class MfmService {
},
blockCode: (node) => {
- const pre = doc.createElement('pre');
- const inner = doc.createElement('code');
- inner.textContent = node.props.code;
- pre.appendChild(inner);
+ const pre = new Element('pre', {});
+ const inner = new Element('code', {});
+ inner.childNodes.push(new Text(node.props.code));
+ pre.childNodes.push(inner);
return pre;
},
center: (node) => {
- const el = doc.createElement('div');
+ const el = new Element('div', {});
appendChildren(node.children, el);
return el;
},
emojiCode: (node) => {
- return doc.createTextNode(`\u200B:${node.props.name}:\u200B`);
+ return new Text(`\u200B:${node.props.name}:\u200B`);
},
unicodeEmoji: (node) => {
- return doc.createTextNode(node.props.emoji);
+ return new Text(node.props.emoji);
},
hashtag: (node) => {
- const a = doc.createElement('a');
- a.setAttribute('href', `${this.config.url}/tags/${node.props.hashtag}`);
- a.textContent = `#${node.props.hashtag}`;
- a.setAttribute('rel', 'tag');
+ const a = new Element('a', {
+ href: `${this.config.url}/tags/${node.props.hashtag}`,
+ rel: 'tag',
+ });
+ a.childNodes.push(new Text(`#${node.props.hashtag}`));
return a;
},
inlineCode: (node) => {
- const el = doc.createElement('code');
- el.textContent = node.props.code;
+ const el = new Element('code', {});
+ el.childNodes.push(new Text(node.props.code));
return el;
},
mathInline: (node) => {
- const el = doc.createElement('code');
- el.textContent = node.props.formula;
+ const el = new Element('code', {});
+ el.childNodes.push(new Text(node.props.formula));
return el;
},
mathBlock: (node) => {
- const el = doc.createElement('code');
- el.textContent = node.props.formula;
+ const el = new Element('code', {});
+ el.childNodes.push(new Text(node.props.formula));
return el;
},
link: (node) => {
- const a = doc.createElement('a');
- a.setAttribute('href', node.props.url);
+ const a = new Element('a', {
+ href: node.props.url,
+ });
appendChildren(node.children, a);
return a;
},
mention: (node) => {
- const a = doc.createElement('a');
const { username, host, acct } = node.props;
const remoteUserInfo = mentionedRemoteUsers.find(remoteUser => remoteUser.username.toLowerCase() === username.toLowerCase() && remoteUser.host?.toLowerCase() === host?.toLowerCase());
- a.setAttribute('href', remoteUserInfo
- ? (remoteUserInfo.url ? remoteUserInfo.url : remoteUserInfo.uri)
- : `${this.config.url}/${acct.endsWith(`@${this.config.url}`) ? acct.substring(0, acct.length - this.config.url.length - 1) : acct}`);
- a.className = 'u-url mention';
- a.textContent = acct;
+
+ const a = new Element('a', {
+ href: remoteUserInfo
+ ? (remoteUserInfo.url ? remoteUserInfo.url : remoteUserInfo.uri)
+ : `${this.config.url}/${acct.endsWith(`@${this.config.url}`) ? acct.substring(0, acct.length - this.config.url.length - 1) : acct}`,
+ class: 'u-url mention',
+ });
+ a.childNodes.push(new Text(acct));
return a;
},
quote: (node) => {
- const el = doc.createElement('blockquote');
+ const el = new Element('blockquote', {});
appendChildren(node.children, el);
return el;
},
text: (node) => {
if (!node.props.text.match(/[\r\n]/)) {
- return doc.createTextNode(node.props.text);
+ return new Text(node.props.text);
}
- const el = doc.createElement('span');
- const nodes = node.props.text.split(/\r\n|\r|\n/).map(x => doc.createTextNode(x));
+ const el = new Element('span', {});
+ const nodes = node.props.text.split(/\r\n|\r|\n/).map(x => new Text(x));
for (const x of intersperse<FIXME | 'br'>('br', nodes)) {
- el.appendChild(x === 'br' ? doc.createElement('br') : x);
+ el.childNodes.push(x === 'br' ? new Element('br', {}) : x);
}
return el;
},
url: (node) => {
- const a = doc.createElement('a');
- a.setAttribute('href', node.props.url);
- a.textContent = node.props.url;
+ const a = new Element('a', {
+ href: node.props.url,
+ });
+ a.childNodes.push(new Text(node.props.url));
return a;
},
search: (node) => {
- const a = doc.createElement('a');
- a.setAttribute('href', `https://www.google.com/search?q=${node.props.query}`);
- a.textContent = node.props.content;
+ const a = new Element('a', {
+ href: `https://www.google.com/search?q=${node.props.query}`,
+ });
+ a.childNodes.push(new Text(node.props.content));
return a;
},
plain: (node) => {
- const el = doc.createElement('span');
+ const el = new Element('span', {});
appendChildren(node.children, el);
return el;
},
};
+ // Utility function to make TypeScript behave
+ function handle<T extends mfm.MfmNode>(node: T): ChildNode {
+ const handler = handlers[node.type] as (node: T) => ChildNode;
+ return handler(node);
+ }
+
appendChildren(nodes, body);
for (const additionalAppender of additionalAppenders) {
additionalAppender(doc, body);
}
- const serialized = body.outerHTML;
-
- happyDOM.close().catch(err => {});
-
- return serialized;
+ return domserializer.render(body, {
+ encodeEntities: 'utf8'
+ });
}
// the toMastoApiHtml function was taken from Iceshrimp and written by zotan and modified by marie to work with the current MK version
@@ -598,55 +602,55 @@ export class MfmService {
return null;
}
- const { happyDOM, window } = new Window();
-
- const doc = window.document;
+ const doc = new Document([]);
- const body = doc.createElement('p');
+ const body = new Element('p', {});
+ doc.childNodes.push(body);
- function appendChildren(children: mfm.MfmNode[], targetElement: any): void {
- if (children) {
- for (const child of children.map((x) => (handlers as any)[x.type](x))) targetElement.appendChild(child);
+ function appendChildren(children: mfm.MfmNode[], targetElement: ParentNode): void {
+ for (const child of children) {
+ const result = handle(child);
+ targetElement.childNodes.push(result);
}
}
const handlers: {
- [K in mfm.MfmNode['type']]: (node: mfm.NodeType<K>) => any;
+ [K in mfm.MfmNode['type']]: (node: mfm.NodeType<K>) => ChildNode;
} = {
bold(node) {
- const el = doc.createElement('span');
- el.textContent = '**';
+ const el = new Element('span', {});
+ el.childNodes.push(new Text('**'));
appendChildren(node.children, el);
- el.textContent += '**';
+ el.childNodes.push(new Text('**'));
return el;
},
small(node) {
- const el = doc.createElement('small');
+ const el = new Element('small', {});
appendChildren(node.children, el);
return el;
},
strike(node) {
- const el = doc.createElement('span');
- el.textContent = '~~';
+ const el = new Element('span', {});
+ el.childNodes.push(new Text('~~'));
appendChildren(node.children, el);
- el.textContent += '~~';
+ el.childNodes.push(new Text('~~'));
return el;
},
italic(node) {
- const el = doc.createElement('span');
- el.textContent = '*';
+ const el = new Element('span', {});
+ el.childNodes.push(new Text('*'));
appendChildren(node.children, el);
- el.textContent += '*';
+ el.childNodes.push(new Text('*'));
return el;
},
fn(node) {
switch (node.props.name) {
case 'group': { // hack for ruby
- const el = doc.createElement('span');
+ const el = new Element('span', {});
appendChildren(node.children, el);
return el;
}
@@ -654,119 +658,121 @@ export class MfmService {
if (node.children.length === 1) {
const child = node.children[0];
const text = child.type === 'text' ? child.props.text : '';
- const rubyEl = doc.createElement('ruby');
- const rtEl = doc.createElement('rt');
+ const rubyEl = new Element('ruby', {});
+ const rtEl = new Element('rt', {});
- const rpStartEl = doc.createElement('rp');
- rpStartEl.appendChild(doc.createTextNode('('));
- const rpEndEl = doc.createElement('rp');
- rpEndEl.appendChild(doc.createTextNode(')'));
+ const rpStartEl = new Element('rp', {});
+ rpStartEl.childNodes.push(new Text('('));
+ const rpEndEl = new Element('rp', {});
+ rpEndEl.childNodes.push(new Text(')'));
- rubyEl.appendChild(doc.createTextNode(text.split(' ')[0]));
- rtEl.appendChild(doc.createTextNode(text.split(' ')[1]));
- rubyEl.appendChild(rpStartEl);
- rubyEl.appendChild(rtEl);
- rubyEl.appendChild(rpEndEl);
+ rubyEl.childNodes.push(new Text(text.split(' ')[0]));
+ rtEl.childNodes.push(new Text(text.split(' ')[1]));
+ rubyEl.childNodes.push(rpStartEl);
+ rubyEl.childNodes.push(rtEl);
+ rubyEl.childNodes.push(rpEndEl);
return rubyEl;
} else {
const rt = node.children.at(-1);
if (!rt) {
- const el = doc.createElement('span');
+ const el = new Element('span', {});
appendChildren(node.children, el);
return el;
}
const text = rt.type === 'text' ? rt.props.text : '';
- const rubyEl = doc.createElement('ruby');
- const rtEl = doc.createElement('rt');
+ const rubyEl = new Element('ruby', {});
+ const rtEl = new Element('rt', {});
- const rpStartEl = doc.createElement('rp');
- rpStartEl.appendChild(doc.createTextNode('('));
- const rpEndEl = doc.createElement('rp');
- rpEndEl.appendChild(doc.createTextNode(')'));
+ const rpStartEl = new Element('rp', {});
+ rpStartEl.childNodes.push(new Text('('));
+ const rpEndEl = new Element('rp', {});
+ rpEndEl.childNodes.push(new Text(')'));
appendChildren(node.children.slice(0, node.children.length - 1), rubyEl);
- rtEl.appendChild(doc.createTextNode(text.trim()));
- rubyEl.appendChild(rpStartEl);
- rubyEl.appendChild(rtEl);
- rubyEl.appendChild(rpEndEl);
+ rtEl.childNodes.push(new Text(text.trim()));
+ rubyEl.childNodes.push(rpStartEl);
+ rubyEl.childNodes.push(rtEl);
+ rubyEl.childNodes.push(rpEndEl);
return rubyEl;
}
}
default: {
- const el = doc.createElement('span');
- el.textContent = '*';
+ const el = new Element('span', {});
+ el.childNodes.push(new Text('*'));
appendChildren(node.children, el);
- el.textContent += '*';
+ el.childNodes.push(new Text('*'));
return el;
}
}
},
blockCode(node) {
- const pre = doc.createElement('pre');
- const inner = doc.createElement('code');
+ const pre = new Element('pre', {});
+ const inner = new Element('code', {});
const nodes = node.props.code
.split(/\r\n|\r|\n/)
- .map((x) => doc.createTextNode(x));
+ .map((x) => new Text(x));
for (const x of intersperse<FIXME | 'br'>('br', nodes)) {
- inner.appendChild(x === 'br' ? doc.createElement('br') : x);
+ inner.childNodes.push(x === 'br' ? new Element('br', {}) : x);
}
- pre.appendChild(inner);
+ pre.childNodes.push(inner);
return pre;
},
center(node) {
- const el = doc.createElement('div');
+ const el = new Element('div', {});
appendChildren(node.children, el);
return el;
},
emojiCode(node) {
- return doc.createTextNode(`\u200B:${node.props.name}:\u200B`);
+ return new Text(`\u200B:${node.props.name}:\u200B`);
},
unicodeEmoji(node) {
- return doc.createTextNode(node.props.emoji);
+ return new Text(node.props.emoji);
},
hashtag: (node) => {
- const a = doc.createElement('a');
- a.setAttribute('href', `${this.config.url}/tags/${node.props.hashtag}`);
- a.textContent = `#${node.props.hashtag}`;
- a.setAttribute('rel', 'tag');
- a.setAttribute('class', 'hashtag');
+ const a = new Element('a', {
+ href: `${this.config.url}/tags/${node.props.hashtag}`,
+ rel: 'tag',
+ class: 'hashtag',
+ });
+ a.childNodes.push(new Text(`#${node.props.hashtag}`));
return a;
},
inlineCode(node) {
- const el = doc.createElement('code');
- el.textContent = node.props.code;
+ const el = new Element('code', {});
+ el.childNodes.push(new Text(node.props.code));
return el;
},
mathInline(node) {
- const el = doc.createElement('code');
- el.textContent = node.props.formula;
+ const el = new Element('code', {});
+ el.childNodes.push(new Text(node.props.formula));
return el;
},
mathBlock(node) {
- const el = doc.createElement('code');
- el.textContent = node.props.formula;
+ const el = new Element('code', {});
+ el.childNodes.push(new Text(node.props.formula));
return el;
},
link(node) {
- const a = doc.createElement('a');
- a.setAttribute('rel', 'nofollow noopener noreferrer');
- a.setAttribute('target', '_blank');
- a.setAttribute('href', node.props.url);
+ const a = new Element('a', {
+ rel: 'nofollow noopener noreferrer',
+ target: '_blank',
+ href: node.props.url,
+ });
appendChildren(node.children, a);
return a;
},
@@ -775,92 +781,107 @@ export class MfmService {
const { username, host, acct } = node.props;
const resolved = mentionedRemoteUsers.find(remoteUser => remoteUser.username === username && remoteUser.host === host);
- const el = doc.createElement('span');
+ const el = new Element('span', {});
if (!resolved) {
- el.textContent = acct;
+ el.childNodes.push(new Text(acct));
} else {
- el.setAttribute('class', 'h-card');
- el.setAttribute('translate', 'no');
- const a = doc.createElement('a');
- a.setAttribute('href', resolved.url ? resolved.url : resolved.uri);
- a.className = 'u-url mention';
- const span = doc.createElement('span');
- span.textContent = resolved.username || username;
- a.textContent = '@';
- a.appendChild(span);
- el.appendChild(a);
+ el.attribs.class = 'h-card';
+ el.attribs.translate = 'no';
+ const a = new Element('a', {
+ href: resolved.url ? resolved.url : resolved.uri,
+ class: 'u-url mention',
+ });
+ const span = new Element('span', {});
+ span.childNodes.push(new Text(resolved.username || username));
+ a.childNodes.push(new Text('@'));
+ a.childNodes.push(span);
+ el.childNodes.push(a);
}
return el;
},
quote(node) {
- const el = doc.createElement('blockquote');
+ const el = new Element('blockquote', {});
appendChildren(node.children, el);
return el;
},
text(node) {
- const el = doc.createElement('span');
+ if (!node.props.text.match(/[\r\n]/)) {
+ return new Text(node.props.text);
+ }
+
+ const el = new Element('span', {});
const nodes = node.props.text
.split(/\r\n|\r|\n/)
- .map((x) => doc.createTextNode(x));
+ .map((x) => new Text(x));
for (const x of intersperse<FIXME | 'br'>('br', nodes)) {
- el.appendChild(x === 'br' ? doc.createElement('br') : x);
+ el.childNodes.push(x === 'br' ? new Element('br', {}) : x);
}
return el;
},
url(node) {
- const a = doc.createElement('a');
- a.setAttribute('rel', 'nofollow noopener noreferrer');
- a.setAttribute('target', '_blank');
- a.setAttribute('href', node.props.url);
- a.textContent = node.props.url.replace(/^https?:\/\//, '');
+ const a = new Element('a', {
+ rel: 'nofollow noopener noreferrer',
+ target: '_blank',
+ href: node.props.url,
+ });
+ a.childNodes.push(new Text(node.props.url.replace(/^https?:\/\//, '')));
return a;
},
search: (node) => {
- const a = doc.createElement('a');
- a.setAttribute('href', `https://www.google.com/search?q=${node.props.query}`);
- a.textContent = node.props.content;
+ const a = new Element('a', {
+ href: `https://www.google.com/search?q=${node.props.query}`,
+ });
+ a.childNodes.push(new Text(node.props.content));
return a;
},
plain(node) {
- const el = doc.createElement('span');
+ const el = new Element('span', {});
appendChildren(node.children, el);
return el;
},
};
+ // Utility function to make TypeScript behave
+ function handle<T extends mfm.MfmNode>(node: T): ChildNode {
+ const handler = handlers[node.type] as (node: T) => ChildNode;
+ return handler(node);
+ }
+
appendChildren(nodes, body);
if (quoteUri !== null) {
- const a = doc.createElement('a');
- a.setAttribute('href', quoteUri);
- a.textContent = quoteUri.replace(/^https?:\/\//, '');
+ const a = new Element('a', {
+ href: quoteUri,
+ });
+ a.childNodes.push(new Text(quoteUri.replace(/^https?:\/\//, '')));
- const quote = doc.createElement('span');
- quote.setAttribute('class', 'quote-inline');
- quote.appendChild(doc.createElement('br'));
- quote.appendChild(doc.createElement('br'));
- quote.innerHTML += 'RE: ';
- quote.appendChild(a);
+ const quote = new Element('span', {
+ class: 'quote-inline',
+ });
+ quote.childNodes.push(new Element('br', {}));
+ quote.childNodes.push(new Element('br', {}));
+ quote.childNodes.push(new Text('RE: '));
+ quote.childNodes.push(a);
- body.appendChild(quote);
+ body.childNodes.push(quote);
}
- let result = body.outerHTML;
+ let result = domserializer.render(body, {
+ encodeEntities: 'utf8'
+ });
if (inline) {
result = result.replace(/^<p>/, '').replace(/<\/p>$/, '');
}
- happyDOM.close().catch(() => {});
-
return result;
}
}
diff --git a/packages/backend/src/core/WebfingerService.ts b/packages/backend/src/core/WebfingerService.ts
index 664963f3a3..bb9f0be4c6 100644
--- a/packages/backend/src/core/WebfingerService.ts
+++ b/packages/backend/src/core/WebfingerService.ts
@@ -5,7 +5,7 @@
import { URL } from 'node:url';
import { Injectable } from '@nestjs/common';
-import { XMLParser } from 'fast-xml-parser';
+import { load as cheerio } from 'cheerio/slim';
import { HttpRequestService } from '@/core/HttpRequestService.js';
import { bindThis } from '@/decorators.js';
import type Logger from '@/logger.js';
@@ -101,14 +101,12 @@ export class WebfingerService {
private async fetchWebFingerTemplateFromHostMeta(url: string): Promise<string | null> {
try {
const res = await this.httpRequestService.getHtml(url, 'application/xrd+xml');
- const options = {
- ignoreAttributes: false,
- isArray: (_name: string, jpath: string) => jpath === 'XRD.Link',
- };
- const parser = new XMLParser(options);
- const hostMeta = parser.parse(res);
- const template = (hostMeta['XRD']['Link'] as Array<any>).filter(p => p['@_rel'] === 'lrdd')[0]['@_template'];
- return template.indexOf('{uri}') < 0 ? null : template;
+ const hostMeta = cheerio(res, {
+ xml: true,
+ });
+
+ const template = hostMeta('XRD > Link[rel="lrdd"][template*="{uri}"]').attr('template');
+ return template ?? null;
} catch (err) {
this.logger.error(`error while request host-meta for ${url}: ${renderInlineError(err)}`);
return null;
diff --git a/packages/backend/src/core/activitypub/ApRendererService.ts b/packages/backend/src/core/activitypub/ApRendererService.ts
index 6068d707de..789611fd97 100644
--- a/packages/backend/src/core/activitypub/ApRendererService.ts
+++ b/packages/backend/src/core/activitypub/ApRendererService.ts
@@ -8,6 +8,7 @@ import { Inject, Injectable } from '@nestjs/common';
import { In } from 'typeorm';
import * as mfm from '@transfem-org/sfm-js';
import { UnrecoverableError } from 'bullmq';
+import { Element, Text } from 'domhandler';
import { DI } from '@/di-symbols.js';
import type { Config } from '@/config.js';
import type { MiPartialLocalUser, MiLocalUser, MiPartialRemoteUser, MiRemoteUser, MiUser } from '@/models/User.js';
@@ -475,16 +476,18 @@ export class ApRendererService {
// the claas name `quote-inline` is used in non-misskey clients for styling quote notes.
// For compatibility, the span part should be kept as possible.
apAppend.push((doc, body) => {
- body.appendChild(doc.createElement('br'));
- body.appendChild(doc.createElement('br'));
- const span = doc.createElement('span');
- span.className = 'quote-inline';
- span.appendChild(doc.createTextNode('RE: '));
- const link = doc.createElement('a');
- link.setAttribute('href', quote);
- link.textContent = quote;
- span.appendChild(link);
- body.appendChild(span);
+ body.childNodes.push(new Element('br', {}));
+ body.childNodes.push(new Element('br', {}));
+ const span = new Element('span', {
+ class: 'quote-inline',
+ });
+ span.childNodes.push(new Text('RE: '));
+ const link = new Element('a', {
+ href: quote,
+ });
+ link.childNodes.push(new Text(quote));
+ span.childNodes.push(link);
+ body.childNodes.push(span);
});
}
@@ -839,16 +842,18 @@ export class ApRendererService {
// the claas name `quote-inline` is used in non-misskey clients for styling quote notes.
// For compatibility, the span part should be kept as possible.
apAppend.push((doc, body) => {
- body.appendChild(doc.createElement('br'));
- body.appendChild(doc.createElement('br'));
- const span = doc.createElement('span');
- span.className = 'quote-inline';
- span.appendChild(doc.createTextNode('RE: '));
- const link = doc.createElement('a');
- link.setAttribute('href', quote);
- link.textContent = quote;
- span.appendChild(link);
- body.appendChild(span);
+ body.childNodes.push(new Element('br', {}));
+ body.childNodes.push(new Element('br', {}));
+ const span = new Element('span', {
+ class: 'quote-inline',
+ });
+ span.childNodes.push(new Text('RE: '));
+ const link = new Element('a', {
+ href: quote,
+ });
+ link.childNodes.push(new Text(quote));
+ span.childNodes.push(link);
+ body.childNodes.push(span);
});
}
diff --git a/packages/backend/src/core/activitypub/ApRequestService.ts b/packages/backend/src/core/activitypub/ApRequestService.ts
index 4c7cac2169..e4db9b237c 100644
--- a/packages/backend/src/core/activitypub/ApRequestService.ts
+++ b/packages/backend/src/core/activitypub/ApRequestService.ts
@@ -6,7 +6,7 @@
import * as crypto from 'node:crypto';
import { URL } from 'node:url';
import { Inject, Injectable } from '@nestjs/common';
-import { Window } from 'happy-dom';
+import { load as cheerio } from 'cheerio/slim';
import { DI } from '@/di-symbols.js';
import type { Config } from '@/config.js';
import type { MiUser } from '@/models/User.js';
@@ -18,6 +18,8 @@ import { bindThis } from '@/decorators.js';
import type Logger from '@/logger.js';
import { validateContentTypeSetAsActivityPub } from '@/core/activitypub/misc/validator.js';
import type { IObject, IObjectWithId } from './type.js';
+import type { Cheerio, CheerioAPI } from 'cheerio/slim';
+import type { AnyNode } from 'domhandler';
type Request = {
url: string;
@@ -219,53 +221,33 @@ export class ApRequestService {
(contentType ?? '').split(';')[0].trimEnd().toLowerCase() === 'text/html' &&
_followAlternate === true
) {
- const html = await res.text();
- const { window, happyDOM } = new Window({
- settings: {
- disableJavaScriptEvaluation: true,
- disableJavaScriptFileLoading: true,
- disableCSSFileLoading: true,
- disableComputedStyleRendering: true,
- handleDisabledFileLoadingAsSuccess: true,
- navigation: {
- disableMainFrameNavigation: true,
- disableChildFrameNavigation: true,
- disableChildPageNavigation: true,
- disableFallbackToSetURL: true,
- },
- timer: {
- maxTimeout: 0,
- maxIntervalTime: 0,
- maxIntervalIterations: 0,
- },
- },
- });
- const document = window.document;
+ let alternate: Cheerio<AnyNode> | null;
try {
- document.documentElement.innerHTML = html;
+ const html = await res.text();
+ const document = cheerio(html);
// Search for any matching value in priority order:
// 1. Type=AP > Type=none > Type=anything
// 2. Alternate > Canonical
// 3. Page order (fallback)
- const alternate =
- document.querySelector('head > link[href][rel="alternate"][type="application/activity+json"]') ??
- document.querySelector('head > link[href][rel="canonical"][type="application/activity+json"]') ??
- document.querySelector('head > link[href][rel="alternate"]:not([type])') ??
- document.querySelector('head > link[href][rel="canonical"]:not([type])') ??
- document.querySelector('head > link[href][rel="alternate"]') ??
- document.querySelector('head > link[href][rel="canonical"]');
-
- if (alternate) {
- const href = alternate.getAttribute('href');
- if (href && this.apUtilityService.haveSameAuthority(url, href)) {
- return await this.signedGet(href, user, allowAnonymous, false);
- }
- }
+ alternate = selectFirst(document, [
+ 'head > link[href][rel="alternate"][type="application/activity+json"]',
+ 'head > link[href][rel="canonical"][type="application/activity+json"]',
+ 'head > link[href][rel="alternate"]:not([type])',
+ 'head > link[href][rel="canonical"]:not([type])',
+ 'head > link[href][rel="alternate"]',
+ 'head > link[href][rel="canonical"]',
+ ]);
} catch {
// something went wrong parsing the HTML, ignore the whole thing
- } finally {
- happyDOM.close().catch(err => {});
+ alternate = null;
+ }
+
+ if (alternate) {
+ const href = alternate.attr('href');
+ if (href && this.apUtilityService.haveSameAuthority(url, href)) {
+ return await this.signedGet(href, user, allowAnonymous, false);
+ }
}
}
//#endregion
@@ -285,3 +267,14 @@ export class ApRequestService {
return activity as IObjectWithId;
}
}
+
+function selectFirst($: CheerioAPI, selectors: string[]): Cheerio<AnyNode> | null {
+ for (const selector of selectors) {
+ const selection = $(selector);
+ if (selection.length > 0) {
+ return selection;
+ }
+ }
+
+ return null;
+}
diff --git a/packages/backend/src/misc/truncate.ts b/packages/backend/src/misc/truncate.ts
index 1c8a274609..a313ab7854 100644
--- a/packages/backend/src/misc/truncate.ts
+++ b/packages/backend/src/misc/truncate.ts
@@ -3,14 +3,12 @@
* SPDX-License-Identifier: AGPL-3.0-only
*/
-import { substring } from 'stringz';
-
export function truncate(input: string, size: number): string;
export function truncate(input: string | undefined, size: number): string | undefined;
export function truncate(input: string | undefined, size: number): string | undefined {
if (!input) {
return input;
} else {
- return substring(input, 0, size);
+ return input.slice(0, size);
}
}
diff --git a/packages/backend/src/misc/verify-field-link.ts b/packages/backend/src/misc/verify-field-link.ts
index 62542eaaa0..f9fc352806 100644
--- a/packages/backend/src/misc/verify-field-link.ts
+++ b/packages/backend/src/misc/verify-field-link.ts
@@ -3,7 +3,7 @@
* SPDX-License-Identifier: AGPL-3.0-only
*/
-import { load as cheerio } from 'cheerio';
+import { load as cheerio } from 'cheerio/slim';
import type { HttpRequestService } from '@/core/HttpRequestService.js';
type Field = { name: string, value: string };
diff --git a/packages/backend/src/server/api/endpoints/fetch-rss.ts b/packages/backend/src/server/api/endpoints/fetch-rss.ts
index 03f35f16a5..11244b30f6 100644
--- a/packages/backend/src/server/api/endpoints/fetch-rss.ts
+++ b/packages/backend/src/server/api/endpoints/fetch-rss.ts
@@ -3,12 +3,12 @@
* SPDX-License-Identifier: AGPL-3.0-only
*/
-import Parser from 'rss-parser';
import { Injectable } from '@nestjs/common';
+import { parseFeed } from 'htmlparser2';
import { Endpoint } from '@/server/api/endpoint-base.js';
import { HttpRequestService } from '@/core/HttpRequestService.js';
-
-const rssParser = new Parser();
+import { ApiError } from '../error.js';
+import type { FeedItem } from 'domutils';
export const meta = {
tags: ['meta'],
@@ -17,52 +17,32 @@ export const meta = {
allowGet: true,
cacheSec: 60 * 3,
+ errors: {
+ fetchFailed: {
+ id: '88f4356f-719d-4715-b4fc-703a10a812d2',
+ code: 'FETCH_FAILED',
+ message: 'Failed to fetch RSS feed',
+ },
+ },
+
res: {
type: 'object',
properties: {
- image: {
- type: 'object',
+ type: {
+ type: 'string',
+ optional: false,
+ },
+ id: {
+ type: 'string',
optional: true,
- properties: {
- link: {
- type: 'string',
- optional: true,
- },
- url: {
- type: 'string',
- optional: false,
- },
- title: {
- type: 'string',
- optional: true,
- },
- },
},
- paginationLinks: {
- type: 'object',
+ updated: {
+ type: 'string',
+ optional: true,
+ },
+ author: {
+ type: 'string',
optional: true,
- properties: {
- self: {
- type: 'string',
- optional: true,
- },
- first: {
- type: 'string',
- optional: true,
- },
- next: {
- type: 'string',
- optional: true,
- },
- last: {
- type: 'string',
- optional: true,
- },
- prev: {
- type: 'string',
- optional: true,
- },
- },
},
link: {
type: 'string',
@@ -94,113 +74,42 @@ export const meta = {
type: 'string',
optional: true,
},
- creator: {
- type: 'string',
- optional: true,
- },
- summary: {
- type: 'string',
- optional: true,
- },
- content: {
- type: 'string',
- optional: true,
- },
- isoDate: {
+ description: {
type: 'string',
optional: true,
},
- categories: {
+ media: {
type: 'array',
- optional: true,
+ optional: false,
items: {
- type: 'string',
- },
- },
- contentSnippet: {
- type: 'string',
- optional: true,
- },
- enclosure: {
- type: 'object',
- optional: true,
- properties: {
- url: {
- type: 'string',
- optional: false,
- },
- length: {
- type: 'number',
- optional: true,
- },
- type: {
- type: 'string',
- optional: true,
+ type: 'object',
+ properties: {
+ medium: {
+ type: 'string',
+ optional: true,
+ },
+ url: {
+ type: 'string',
+ optional: true,
+ },
+ type: {
+ type: 'string',
+ optional: true,
+ },
+ lang: {
+ type: 'string',
+ optional: true,
+ },
},
},
},
},
},
},
- feedUrl: {
- type: 'string',
- optional: true,
- },
description: {
type: 'string',
optional: true,
},
- itunes: {
- type: 'object',
- optional: true,
- additionalProperties: true,
- properties: {
- image: {
- type: 'string',
- optional: true,
- },
- owner: {
- type: 'object',
- optional: true,
- properties: {
- name: {
- type: 'string',
- optional: true,
- },
- email: {
- type: 'string',
- optional: true,
- },
- },
- },
- author: {
- type: 'string',
- optional: true,
- },
- summary: {
- type: 'string',
- optional: true,
- },
- explicit: {
- type: 'string',
- optional: true,
- },
- categories: {
- type: 'array',
- optional: true,
- items: {
- type: 'string',
- },
- },
- keywords: {
- type: 'array',
- optional: true,
- items: {
- type: 'string',
- },
- },
- },
- },
},
},
@@ -224,7 +133,7 @@ export default class extends Endpoint<typeof meta, typeof paramDef> { // eslint-
constructor(
private httpRequestService: HttpRequestService,
) {
- super(meta, paramDef, async (ps, me) => {
+ super(meta, paramDef, async (ps) => {
const res = await this.httpRequestService.send(ps.url, {
method: 'GET',
headers: {
@@ -234,8 +143,38 @@ export default class extends Endpoint<typeof meta, typeof paramDef> { // eslint-
});
const text = await res.text();
+ const feed = parseFeed(text, {
+ xmlMode: true,
+ });
+
+ if (!feed) {
+ throw new ApiError(meta.errors.fetchFailed);
+ }
- return rssParser.parseString(text);
+ return {
+ type: feed.type,
+ id: feed.id,
+ title: feed.title,
+ link: feed.link,
+ description: feed.description,
+ updated: feed.updated?.toISOString(),
+ author: feed.author,
+ items: feed.items
+ .filter((item): item is FeedItem & { link: string, title: string } => !!item.link && !!item.title)
+ .map(item => ({
+ guid: item.id,
+ title: item.title,
+ link: item.link,
+ description: item.description,
+ pubDate: item.pubDate?.toISOString(),
+ media: item.media.map(media => ({
+ medium: media.medium,
+ url: media.url,
+ type: media.type,
+ lang: media.lang,
+ })),
+ })),
+ };
});
}
}