Fixed 267U.pre2
This commit is contained in:
@@ -0,0 +1,194 @@
|
||||
import * as parse5 from "parse5";
|
||||
import { defaultTreeAdapter as treeAdapter } from "parse5";
|
||||
import { getSubjectHostFromUriAndUsernameCached } from "../remote/resolve-user.js";
|
||||
const urlRegex = /^https?:\/\/[\w\/:%#@$&?!()\[\]~.,=+\-]+/;
|
||||
const urlRegexFull = /^https?:\/\/[\w\/:%#@$&?!()\[\]~.,=+\-]+$/;
|
||||
export async function fromHtml(html, hashtagNames) {
|
||||
// some AP servers like Pixelfed use br tags as well as newlines
|
||||
html = html.replace(/<br\s?\/?>\r?\n/gi, "\n");
|
||||
const dom = parse5.parseFragment(html);
|
||||
let text = "";
|
||||
for (const n of dom.childNodes){
|
||||
await analyze(n);
|
||||
}
|
||||
return text.trim();
|
||||
function getText(node) {
|
||||
if (treeAdapter.isTextNode(node)) return node.value;
|
||||
if (!treeAdapter.isElementNode(node)) return "";
|
||||
if (node.nodeName === "br") return "\n";
|
||||
if (node.childNodes) {
|
||||
return node.childNodes.map((n)=>getText(n)).join("");
|
||||
}
|
||||
return "";
|
||||
}
|
||||
async function appendChildren(childNodes) {
|
||||
if (childNodes) {
|
||||
for (const n of childNodes){
|
||||
await analyze(n);
|
||||
}
|
||||
}
|
||||
}
|
||||
async function analyze(node) {
|
||||
if (treeAdapter.isTextNode(node)) {
|
||||
text += node.value;
|
||||
return;
|
||||
}
|
||||
// Skip comment or document type node
|
||||
if (!treeAdapter.isElementNode(node)) return;
|
||||
// Strip quote marker
|
||||
const classes = node.attrs.find((x)=>x.name === "class");
|
||||
if (classes && classes.value.split(" ").includes("quote-inline")) {
|
||||
return;
|
||||
}
|
||||
switch(node.nodeName){
|
||||
case "br":
|
||||
{
|
||||
text += "\n";
|
||||
break;
|
||||
}
|
||||
case "a":
|
||||
{
|
||||
const txt = getText(node);
|
||||
const rel = node.attrs.find((x)=>x.name === "rel");
|
||||
const href = node.attrs.find((x)=>x.name === "href");
|
||||
// ハッシュタグ
|
||||
if (hashtagNames && href && hashtagNames.map((x)=>x.toLowerCase()).includes(txt.toLowerCase())) {
|
||||
text += txt;
|
||||
// メンション
|
||||
} else if (txt.startsWith("@") && !rel?.value.match(/^me /)) {
|
||||
const part = txt.split("@");
|
||||
if (part.length === 2 && href) {
|
||||
//#region ホスト名部分が省略されているので復元する
|
||||
const acct = `${txt}@${await getSubjectHostFromUriAndUsernameCached(href.value, txt)}`;
|
||||
text += acct;
|
||||
//#endregion
|
||||
} else if (part.length === 3) {
|
||||
text += txt;
|
||||
}
|
||||
// その他
|
||||
} else {
|
||||
const generateLink = ()=>{
|
||||
if (!(href || txt)) {
|
||||
return "";
|
||||
}
|
||||
if (!href) {
|
||||
return txt;
|
||||
}
|
||||
if (!txt || txt === href.value) {
|
||||
// #6383: Missing text node
|
||||
if (href.value.match(urlRegexFull)) {
|
||||
return href.value;
|
||||
} else {
|
||||
return `<${href.value}>`;
|
||||
}
|
||||
}
|
||||
if (href.value.match(urlRegex) && !href.value.match(urlRegexFull)) {
|
||||
return `[${txt}](<${href.value}>)`; // #6846
|
||||
} else {
|
||||
return `[${txt}](${href.value})`;
|
||||
}
|
||||
};
|
||||
text += generateLink();
|
||||
}
|
||||
break;
|
||||
}
|
||||
case "h1":
|
||||
{
|
||||
text += "【";
|
||||
await appendChildren(node.childNodes);
|
||||
text += "】\n";
|
||||
break;
|
||||
}
|
||||
case "b":
|
||||
case "strong":
|
||||
{
|
||||
text += "**";
|
||||
await appendChildren(node.childNodes);
|
||||
text += "**";
|
||||
break;
|
||||
}
|
||||
case "small":
|
||||
{
|
||||
text += "<small>";
|
||||
await appendChildren(node.childNodes);
|
||||
text += "</small>";
|
||||
break;
|
||||
}
|
||||
case "s":
|
||||
case "del":
|
||||
{
|
||||
text += "~~";
|
||||
await appendChildren(node.childNodes);
|
||||
text += "~~";
|
||||
break;
|
||||
}
|
||||
case "i":
|
||||
case "em":
|
||||
{
|
||||
text += "<i>";
|
||||
await appendChildren(node.childNodes);
|
||||
text += "</i>";
|
||||
break;
|
||||
}
|
||||
// block code (<pre><code>)
|
||||
case "pre":
|
||||
{
|
||||
if (node.childNodes.length === 1 && node.childNodes[0].nodeName === "code") {
|
||||
text += "\n```\n";
|
||||
text += getText(node.childNodes[0]);
|
||||
text += "\n```\n";
|
||||
} else {
|
||||
await appendChildren(node.childNodes);
|
||||
}
|
||||
break;
|
||||
}
|
||||
// inline code (<code>)
|
||||
case "code":
|
||||
{
|
||||
text += "`";
|
||||
await appendChildren(node.childNodes);
|
||||
text += "`";
|
||||
break;
|
||||
}
|
||||
case "blockquote":
|
||||
{
|
||||
const t = getText(node);
|
||||
if (t) {
|
||||
text += "\n> ";
|
||||
text += t.split("\n").join("\n> ");
|
||||
}
|
||||
break;
|
||||
}
|
||||
case "p":
|
||||
case "h2":
|
||||
case "h3":
|
||||
case "h4":
|
||||
case "h5":
|
||||
case "h6":
|
||||
{
|
||||
text += "\n\n";
|
||||
await appendChildren(node.childNodes);
|
||||
break;
|
||||
}
|
||||
// other block elements
|
||||
case "div":
|
||||
case "header":
|
||||
case "footer":
|
||||
case "article":
|
||||
case "li":
|
||||
case "dt":
|
||||
case "dd":
|
||||
{
|
||||
text += "\n";
|
||||
await appendChildren(node.childNodes);
|
||||
break;
|
||||
}
|
||||
default:
|
||||
{
|
||||
// includes inline elements
|
||||
await appendChildren(node.childNodes);
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,161 @@
|
||||
import { Window as HappyDom } from "happy-dom";
|
||||
import config from "../config/index.js";
|
||||
import { intersperse } from "../prelude/array.js";
|
||||
import { resolveMentionFromCache } from "../remote/resolve-user.js";
|
||||
export async function toHtml(nodes, mentionedRemoteUsers = [], objectHost) {
|
||||
if (nodes == null) {
|
||||
return null;
|
||||
}
|
||||
const window = new HappyDom();
|
||||
const doc = window.document;
|
||||
function appendTextWithGlyphs(text, targetElement) {
|
||||
const regexp = /;([^:;\s]{1,100});/g;
|
||||
let last = 0;
|
||||
for (const match of text.matchAll(regexp)){
|
||||
if (match.index > last) {
|
||||
targetElement.appendChild(doc.createTextNode(text.slice(last, match.index)));
|
||||
}
|
||||
targetElement.appendChild(doc.createTextNode(`\u200B:${match[1]}:\u200B`));
|
||||
last = match.index + match[0].length;
|
||||
}
|
||||
if (last < text.length) {
|
||||
targetElement.appendChild(doc.createTextNode(text.slice(last)));
|
||||
}
|
||||
}
|
||||
async function appendChildren(children, targetElement) {
|
||||
if (children) {
|
||||
for (const child of (await Promise.all(children.map(async (x)=>await handlers[x.type](x)))))targetElement.appendChild(child);
|
||||
}
|
||||
}
|
||||
const handlers = {
|
||||
async bold (node) {
|
||||
const el = doc.createElement("b");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
async small (node) {
|
||||
const el = doc.createElement("small");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
async strike (node) {
|
||||
const el = doc.createElement("del");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
async italic (node) {
|
||||
const el = doc.createElement("i");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
async fn (node) {
|
||||
const el = doc.createElement("i");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
blockCode (node) {
|
||||
const pre = doc.createElement("pre");
|
||||
const inner = doc.createElement("code");
|
||||
inner.textContent = node.props.code;
|
||||
pre.appendChild(inner);
|
||||
return pre;
|
||||
},
|
||||
async center (node) {
|
||||
const el = doc.createElement("div");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
emojiCode (node) {
|
||||
return doc.createTextNode(`\u200B:${node.props.name}:\u200B`);
|
||||
},
|
||||
unicodeEmoji (node) {
|
||||
return doc.createTextNode(node.props.emoji);
|
||||
},
|
||||
hashtag (node) {
|
||||
const a = doc.createElement("a");
|
||||
a.setAttribute('href', `${config.url}/tags/${node.props.hashtag}`);
|
||||
a.textContent = `#${node.props.hashtag}`;
|
||||
a.setAttribute("rel", "tag");
|
||||
return a;
|
||||
},
|
||||
inlineCode (node) {
|
||||
const el = doc.createElement("code");
|
||||
el.textContent = node.props.code;
|
||||
return el;
|
||||
},
|
||||
mathInline (node) {
|
||||
const el = doc.createElement("code");
|
||||
el.textContent = node.props.formula;
|
||||
return el;
|
||||
},
|
||||
mathBlock (node) {
|
||||
const el = doc.createElement("code");
|
||||
el.textContent = node.props.formula;
|
||||
return el;
|
||||
},
|
||||
async link (node) {
|
||||
const a = doc.createElement("a");
|
||||
a.setAttribute('href', node.props.url);
|
||||
await appendChildren(node.children, a);
|
||||
return a;
|
||||
},
|
||||
async mention (node) {
|
||||
const { username, host, acct } = node.props;
|
||||
const resolved = await resolveMentionFromCache(username, host, objectHost, mentionedRemoteUsers);
|
||||
const el = doc.createElement("span");
|
||||
if (resolved === null) {
|
||||
el.textContent = acct;
|
||||
} else {
|
||||
el.setAttribute("class", "h-card");
|
||||
el.setAttribute("translate", "no");
|
||||
const a = doc.createElement("a");
|
||||
a.setAttribute('href', resolved.href);
|
||||
a.className = "u-url mention";
|
||||
const span = doc.createElement("span");
|
||||
span.textContent = resolved.username;
|
||||
a.textContent = '@';
|
||||
a.appendChild(span);
|
||||
el.appendChild(a);
|
||||
}
|
||||
return el;
|
||||
},
|
||||
async quote (node) {
|
||||
const el = doc.createElement("blockquote");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
},
|
||||
text (node) {
|
||||
const el = doc.createElement("span");
|
||||
const lines = node.props.text.split(/\r\n|\r|\n/);
|
||||
for (const x of intersperse("br", lines)){
|
||||
if (x === "br") {
|
||||
el.appendChild(doc.createElement("br"));
|
||||
continue;
|
||||
}
|
||||
appendTextWithGlyphs(x, el);
|
||||
}
|
||||
return el;
|
||||
},
|
||||
url (node) {
|
||||
const a = doc.createElement("a");
|
||||
a.setAttribute('href', node.props.url);
|
||||
a.textContent = node.props.url.replace(/^https?:\/\//, '');
|
||||
return a;
|
||||
},
|
||||
search (node) {
|
||||
const a = doc.createElement("a");
|
||||
a.setAttribute('href', `${config.searchEngine}${node.props.query}`);
|
||||
a.textContent = node.props.content;
|
||||
return a;
|
||||
},
|
||||
async plain (node) {
|
||||
const el = doc.createElement("span");
|
||||
await appendChildren(node.children, el);
|
||||
return el;
|
||||
}
|
||||
};
|
||||
await appendChildren(nodes, doc.body);
|
||||
const html = `<p>${doc.body.innerHTML}</p>`;
|
||||
await window.happyDOM.close();
|
||||
return html;
|
||||
}
|
||||
Reference in New Issue
Block a user