vector-im-hydrogen-web/src/domain/session/room/timeline/deserialize.js

271 lines
8.7 KiB
JavaScript
Raw Normal View History

2021-07-08 08:40:16 +02:00
import { MessageBody, HeaderBlock, ListBlock, CodeBlock, FormatPart, NewLinePart, RulePart, TextPart, LinkPart, ImagePart } from "./MessageBody.js"
/* At the time of writing (Jul 1 2021), Matrix Spec recommends
* allowing the following HTML tags:
* font, del, h1, h2, h3, h4, h5, h6, blockquote, p, a, ul, ol, sup, sub, li, b, i, u,
* strong, em, strike, code, hr, br, div, table, thead, tbody, tr, th, td, caption, pre, span, img
*/
2021-07-02 09:23:59 +02:00
/**
* Nodes that don't have any properties to them other than their tag.
* While <a> has `href`, and <img> has `src`, these have... themselves.
*/
const basicNodes = ["EM", "STRONG", "CODE", "DEL", "P", "DIV", "SPAN" ]
class Deserializer {
constructor(result, mediaRepository) {
this.result = result;
this.mediaRepository = mediaRepository;
}
parseLink(node, children) {
// TODO Not equivalent to `node.href`!
// Add another HTMLParseResult method?
let href = this.result.getAttributeValue(node, "href");
return new LinkPart(href, children);
2021-07-07 01:14:35 +02:00
}
parseList(node) {
const result = this.result;
let start = null;
if (result.getNodeElementName(node) === "OL") {
// Will return 1 for, say, '1A', which may not be intended?
start = parseInt(result.getAttributeValue(node, "start")) || 1;
}
const nodes = [];
for (const child of result.getChildNodes(node)) {
if (result.getNodeElementName(child) !== "LI") {
continue;
}
const item = this.parseNodes(result.getChildNodes(child));
nodes.push(item);
}
return new ListBlock(start, nodes);
}
parseCodeBlock(node) {
const result = this.result;
let codeNode;
for (const child of result.getChildNodes(node)) {
codeNode = child;
break;
}
if (!(codeNode && result.getNodeElementName(codeNode) === "CODE")) {
return null;
}
let language = "";
const cl = result.getAttributeValue(codeNode, "class") || ""
for (const clname of cl.split(" ")) {
if (clname.startsWith("language-") && !clname.startsWith("language-_")) {
language = clname.substring(9) // "language-".length
break;
}
}
return new CodeBlock(language, codeNode.textContent);
}
parseImage(node) {
const result = this.result;
const src = result.getAttributeValue(node, "src") || "";
const url = this.mediaRepository.mxcUrl(src);
// We just ignore non-mxc `src` attributes.
if (!url) {
return null;
}
const width = parseInt(result.getAttributeValue(node, "width")) || null;
const height = parseInt(result.getAttributeValue(node, "height")) || null;
const alt = result.getAttributeValue(node, "alt");
const title = result.getAttributeValue(node, "title");
return new ImagePart(url, width, height, alt, title);
2021-07-08 08:40:16 +02:00
}
parseElement(node) {
const result = this.result;
const tag = result.getNodeElementName(node);
switch (tag) {
case "H1":
case "H2":
case "H3":
case "H4":
case "H5":
case "H6": {
const children = this.parseChildNodes(node);
return new HeaderBlock(parseInt(tag[1]), children)
}
case "A": {
const children = this.parseChildNodes(node);
return this.parseLink(node, children);
}
case "UL":
case "OL":
return this.parseList(node);
case "PRE":
return this.parseCodeBlock(node);
case "BR":
return new NewLinePart();
case "HR":
return new RulePart();
case "IMG":
return this.parseImage(node);
default: {
if (!basicNodes.includes(tag)) {
return null;
}
const children = this.parseChildNodes(node);
return new FormatPart(tag, children);
}
}
}
parseNode(node) {
const result = this.result;
if (result.isTextNode(node)) {
return new TextPart(result.getNodeText(node));
} else if (result.isElementNode(node)) {
return this.parseElement(node);
}
return null;
}
parseChildNodes(node) {
const childNodes = this.result.getChildNodes(node);
return this.parseNodes(childNodes);
}
parseNodes(nodes) {
const parsed = [];
for (const htmlNode of nodes) {
let node = this.parseNode(htmlNode);
// Just ignore invalid / unknown tags.
if (node) {
parsed.push(node);
}
}
return parsed;
}
}
export function parseHTMLBody(platform, mediaRepository, html) {
const parseResult = platform.parseHTML(html);
const deserializer = new Deserializer(parseResult, mediaRepository);
const parts = deserializer.parseNodes(parseResult.rootNodes);
return new MessageBody(html, parts);
}
2021-07-07 04:11:05 +02:00
import parse from '../../../../../lib/node-html-parser/index.js';
2021-07-07 04:11:05 +02:00
export function tests() {
class HTMLParseResult {
constructor(bodyNode) {
this._bodyNode = bodyNode;
}
2021-07-07 04:11:05 +02:00
get rootNodes() {
return this._bodyNode.childNodes;
}
2021-07-07 04:11:05 +02:00
getChildNodes(node) {
return node.childNodes;
}
2021-07-07 04:11:05 +02:00
getAttributeNames(node) {
return node.getAttributeNames();
}
2021-07-07 04:11:05 +02:00
getAttributeValue(node, attr) {
return node.getAttribute(attr);
}
2021-07-07 04:11:05 +02:00
isTextNode(node) {
return !node.tagName;
}
2021-07-07 04:11:05 +02:00
getNodeText(node) {
return node.text;
}
2021-07-07 04:11:05 +02:00
isElementNode(node) {
return !!node.tagName;
}
2021-07-07 04:11:05 +02:00
getNodeElementName(node) {
return node.tagName;
}
2021-07-07 04:11:05 +02:00
}
const platform = {
parseHTML: (html) => new HTMLParseResult(parse(html))
};
2021-07-07 04:11:05 +02:00
function test(assert, input, output) {
assert.deepEqual(parseHTMLBody(platform, null, input), new MessageBody(input, output));
2021-07-07 04:11:05 +02:00
}
return {
"Text only": assert => {
const input = "This is a sentence";
const output = [new TextPart(input)];
test(assert, input, output);
},
"Text with inline code format": assert => {
const input = "Here's <em>some</em> <code>code</code>!";
const output = [
new TextPart("Here's "),
new FormatPart("em", [new TextPart("some")]),
new TextPart(" "),
new FormatPart("code", [new TextPart("code")]),
new TextPart("!")
];
test(assert, input, output);
},
"Text with ordered list with no attributes": assert => {
const input = "<ol><li>Lorem</li><li>Ipsum</li></ol>";
const output = [
new ListBlock(1, [
[ new TextPart("Lorem") ],
[ new TextPart("Ipsum") ]
])
];
test(assert, input, output);
},
"Text with ordered list starting at 3": assert => {
const input = '<ol start="3"><li>Lorem</li><li>Ipsum</li></ol>';
const output = [
new ListBlock(3, [
[ new TextPart("Lorem") ],
[ new TextPart("Ipsum") ]
])
];
test(assert, input, output);
},
"Text with unordered list": assert => {
const input = '<ul start="3"><li>Lorem</li><li>Ipsum</li></ul>';
const output = [
new ListBlock(null, [
[ new TextPart("Lorem") ],
[ new TextPart("Ipsum") ]
])
];
test(assert, input, output);
},
2021-07-13 00:16:04 +02:00
"Auto-closed tags": assert => {
const input = '<p>hello<p>world</p></p>';
const output = [
new FormatPart("p", [new TextPart("hello")]),
new FormatPart("p", [new TextPart("world")])
];
test(assert, input, output);
},
2021-07-07 04:11:05 +02:00
/* Doesnt work: HTML library doesn't handle <pre><code> properly.
"Text with code block": assert => {
const code = 'main :: IO ()\nmain = putStrLn "Hello"'
const input = `<pre><code>${code}</code></pre>`;
const output = [
new CodeBlock(null, code)
];
test(assert, input, output);
}
*/
};
}