mystiq/hydrogen-web
Archived
4
1
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity

Rewrite deserialize as a class to avoid passing options around.

Browse source
This commit is contained in:
Danila Fedorin 2021-07-12 12:19:43 -07:00
parent abd62373e5
commit 9304ca330e
1 changed files with 121 additions and 124 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

123
src/domain/session/room/timeline/deserialize.js
View file

@ -13,29 +13,21 @@ import sanitizeHtml from "../../../../../lib/sanitize-html/index.js"
*/
const basicNodes = ["EM", "STRONG", "CODE", "DEL", "P", "DIV", "SPAN" ]
/**
* Return a builder function for a particular tag.
*/
function basicWrapper(tag) {
return (_, node, children) => new FormatPart(tag, children);
class Deserializer {
constructor(result, mediaRepository) {
this.result = result;
this.mediaRepository = mediaRepository;
}
/**
* Return a builder function for a particular header level.
*/
function headerWrapper(level) {
return (_, node, children) => new HeaderBlock(level, children);
}
function parseLink(options, node, children) {
parseLink(node, children) {
// TODO Not equivalent to `node.href`!
// Add another HTMLParseResult method?
let href = options.result.getAttributeValue(node, "href");
let href = this.result.getAttributeValue(node, "href");
return new LinkPart(href, children);
}
function parseList(options, node) {
const { result } = options;
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?
@ -46,14 +38,14 @@ function parseList(options, node) {
if (result.getNodeElementName(child) !== "LI") {
continue;
}
const item = parseNodes(options, result.getChildNodes(child));
const item = this.parseNodes(result.getChildNodes(child));
nodes.push(item);
}
return new ListBlock(start, nodes);
}
function parseCodeBlock(options, node) {
const { result } = options;
parseCodeBlock(node) {
const result = this.result;
let codeNode;
for (const child of result.getChildNodes(node)) {
codeNode = child;
@ -73,10 +65,10 @@ function parseCodeBlock(options, node) {
return new CodeBlock(language, codeNode.textContent);
}
function parseImage(options, node) {
const { result, mediaRepository } = options;
parseImage(node) {
const result = this.result;
const src = result.getAttributeValue(node, "src") || "";
const url = mediaRepository.mxcUrl(src);
const url = this.mediaRepository.mxcUrl(src);
// We just ignore non-mxc `src` attributes.
if (!url) {
return null;
@ -88,59 +80,63 @@ function parseImage(options, node) {
return new ImagePart(url, { width, height, alt, title });
}
function buildNodeMap() {
let map = {
A: { descend: true, parsefn: parseLink },
UL: { descend: false, parsefn: parseList },
OL: { descend: false, parsefn: parseList },
PRE: { descend: false, parsefn: parseCodeBlock },
BR: { descend: false, parsefn: () => new NewLinePart() },
HR: { descend: false, parsefn: () => new RulePart() },
IMG: { descend: false, parsefn: parseImage }
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)
}
for (const tag of basicNodes) {
map[tag] = { descend: true, parsefn: basicWrapper(tag) }
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);
}
for (let level = 1; level <= 6; level++) {
const tag = "h" + level;
map[tag] = { descend: true, parsefn: headerWrapper(level) }
}
return map;
}
/**
* Handlers for various nodes.
*
* Each handler has two properties: `descend` and `parsefn`.
* If `descend` is true, the node's children should be
* parsed just like any other node, and fed as a second argument
* to `parsefn`. If not, the node's children are either to be ignored
* (as in <pre>) or processed specially (as in <ul>).
*
* The `parsefn` combines a node's data and its children into
* an internal representation node.
*/
const nodes = buildNodeMap();
function parseNode(options, node) {
const { result } = options;
parseNode(node) {
const result = this.result;
if (result.isTextNode(node)) {
return new TextPart(result.getNodeText(node));
} else if (result.isElementNode(node)) {
const f = nodes[result.getNodeElementName(node)];
if (!f) {
return null;
}
const children = f.descend ? parseNodes(options, node.childNodes) : null;
return f.parsefn(options, node, children);
return this.parseElement(node);
}
return null;
}
function parseNodes(options, nodes) {
parseChildNodes(node) {
const childNodes = this.result.getChildNodes(node);
return this.parseNodes(childNodes);
}
parseNodes(nodes) {
const parsed = [];
for (const htmlNode of nodes) {
let node = parseNode(options, htmlNode);
let node = this.parseNode(htmlNode);
// Just ignore invalid / unknown tags.
if (node) {
parsed.push(node);
@ -148,6 +144,7 @@ function parseNodes(options, nodes) {
}
return parsed;
}
}
const sanitizeConfig = {
allowedTags: [
@ -170,8 +167,8 @@ const sanitizeConfig = {
export function parseHTMLBody({ mediaRepository, platform }, html) {
const parseResult = platform.parseHTML(sanitizeHtml(html, sanitizeConfig));
const options = { result: parseResult, mediaRepository };
const parts = parseNodes(options, parseResult.rootNodes);
const deserializer = new Deserializer(parseResult, mediaRepository);
const parts = deserializer.parseNodes(parseResult.rootNodes);
return new MessageBody(html, parts);
}