Rewrite deserialize as a class to avoid passing options
around.
This commit is contained in:
parent
abd62373e5
commit
9304ca330e
1 changed files with 121 additions and 124 deletions
|
@ -13,29 +13,21 @@ import sanitizeHtml from "../../../../../lib/sanitize-html/index.js"
|
||||||
*/
|
*/
|
||||||
const basicNodes = ["EM", "STRONG", "CODE", "DEL", "P", "DIV", "SPAN" ]
|
const basicNodes = ["EM", "STRONG", "CODE", "DEL", "P", "DIV", "SPAN" ]
|
||||||
|
|
||||||
/**
|
class Deserializer {
|
||||||
* Return a builder function for a particular tag.
|
constructor(result, mediaRepository) {
|
||||||
*/
|
this.result = result;
|
||||||
function basicWrapper(tag) {
|
this.mediaRepository = mediaRepository;
|
||||||
return (_, node, children) => new FormatPart(tag, children);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
parseLink(node, children) {
|
||||||
* Return a builder function for a particular header level.
|
|
||||||
*/
|
|
||||||
function headerWrapper(level) {
|
|
||||||
return (_, node, children) => new HeaderBlock(level, children);
|
|
||||||
}
|
|
||||||
|
|
||||||
function parseLink(options, node, children) {
|
|
||||||
// TODO Not equivalent to `node.href`!
|
// TODO Not equivalent to `node.href`!
|
||||||
// Add another HTMLParseResult method?
|
// Add another HTMLParseResult method?
|
||||||
let href = options.result.getAttributeValue(node, "href");
|
let href = this.result.getAttributeValue(node, "href");
|
||||||
return new LinkPart(href, children);
|
return new LinkPart(href, children);
|
||||||
}
|
}
|
||||||
|
|
||||||
function parseList(options, node) {
|
parseList(node) {
|
||||||
const { result } = options;
|
const result = this.result;
|
||||||
let start = null;
|
let start = null;
|
||||||
if (result.getNodeElementName(node) === "OL") {
|
if (result.getNodeElementName(node) === "OL") {
|
||||||
// Will return 1 for, say, '1A', which may not be intended?
|
// Will return 1 for, say, '1A', which may not be intended?
|
||||||
|
@ -46,14 +38,14 @@ function parseList(options, node) {
|
||||||
if (result.getNodeElementName(child) !== "LI") {
|
if (result.getNodeElementName(child) !== "LI") {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
const item = parseNodes(options, result.getChildNodes(child));
|
const item = this.parseNodes(result.getChildNodes(child));
|
||||||
nodes.push(item);
|
nodes.push(item);
|
||||||
}
|
}
|
||||||
return new ListBlock(start, nodes);
|
return new ListBlock(start, nodes);
|
||||||
}
|
}
|
||||||
|
|
||||||
function parseCodeBlock(options, node) {
|
parseCodeBlock(node) {
|
||||||
const { result } = options;
|
const result = this.result;
|
||||||
let codeNode;
|
let codeNode;
|
||||||
for (const child of result.getChildNodes(node)) {
|
for (const child of result.getChildNodes(node)) {
|
||||||
codeNode = child;
|
codeNode = child;
|
||||||
|
@ -73,10 +65,10 @@ function parseCodeBlock(options, node) {
|
||||||
return new CodeBlock(language, codeNode.textContent);
|
return new CodeBlock(language, codeNode.textContent);
|
||||||
}
|
}
|
||||||
|
|
||||||
function parseImage(options, node) {
|
parseImage(node) {
|
||||||
const { result, mediaRepository } = options;
|
const result = this.result;
|
||||||
const src = result.getAttributeValue(node, "src") || "";
|
const src = result.getAttributeValue(node, "src") || "";
|
||||||
const url = mediaRepository.mxcUrl(src);
|
const url = this.mediaRepository.mxcUrl(src);
|
||||||
// We just ignore non-mxc `src` attributes.
|
// We just ignore non-mxc `src` attributes.
|
||||||
if (!url) {
|
if (!url) {
|
||||||
return null;
|
return null;
|
||||||
|
@ -88,59 +80,63 @@ function parseImage(options, node) {
|
||||||
return new ImagePart(url, { width, height, alt, title });
|
return new ImagePart(url, { width, height, alt, title });
|
||||||
}
|
}
|
||||||
|
|
||||||
function buildNodeMap() {
|
parseElement(node) {
|
||||||
let map = {
|
const result = this.result;
|
||||||
A: { descend: true, parsefn: parseLink },
|
const tag = result.getNodeElementName(node);
|
||||||
UL: { descend: false, parsefn: parseList },
|
switch (tag) {
|
||||||
OL: { descend: false, parsefn: parseList },
|
case "H1":
|
||||||
PRE: { descend: false, parsefn: parseCodeBlock },
|
case "H2":
|
||||||
BR: { descend: false, parsefn: () => new NewLinePart() },
|
case "H3":
|
||||||
HR: { descend: false, parsefn: () => new RulePart() },
|
case "H4":
|
||||||
IMG: { descend: false, parsefn: parseImage }
|
case "H5":
|
||||||
|
case "H6": {
|
||||||
|
const children = this.parseChildNodes(node);
|
||||||
|
return new HeaderBlock(parseInt(tag[1]), children)
|
||||||
}
|
}
|
||||||
for (const tag of basicNodes) {
|
case "A": {
|
||||||
map[tag] = { descend: true, parsefn: basicWrapper(tag) }
|
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;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
parseNode(node) {
|
||||||
* Handlers for various nodes.
|
const result = this.result;
|
||||||
*
|
|
||||||
* 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;
|
|
||||||
if (result.isTextNode(node)) {
|
if (result.isTextNode(node)) {
|
||||||
return new TextPart(result.getNodeText(node));
|
return new TextPart(result.getNodeText(node));
|
||||||
} else if (result.isElementNode(node)) {
|
} else if (result.isElementNode(node)) {
|
||||||
const f = nodes[result.getNodeElementName(node)];
|
return this.parseElement(node);
|
||||||
if (!f) {
|
|
||||||
return null;
|
|
||||||
}
|
|
||||||
const children = f.descend ? parseNodes(options, node.childNodes) : null;
|
|
||||||
return f.parsefn(options, node, children);
|
|
||||||
}
|
}
|
||||||
return null;
|
return null;
|
||||||
}
|
}
|
||||||
|
|
||||||
function parseNodes(options, nodes) {
|
parseChildNodes(node) {
|
||||||
|
const childNodes = this.result.getChildNodes(node);
|
||||||
|
return this.parseNodes(childNodes);
|
||||||
|
}
|
||||||
|
|
||||||
|
parseNodes(nodes) {
|
||||||
const parsed = [];
|
const parsed = [];
|
||||||
for (const htmlNode of nodes) {
|
for (const htmlNode of nodes) {
|
||||||
let node = parseNode(options, htmlNode);
|
let node = this.parseNode(htmlNode);
|
||||||
// Just ignore invalid / unknown tags.
|
// Just ignore invalid / unknown tags.
|
||||||
if (node) {
|
if (node) {
|
||||||
parsed.push(node);
|
parsed.push(node);
|
||||||
|
@ -148,6 +144,7 @@ function parseNodes(options, nodes) {
|
||||||
}
|
}
|
||||||
return parsed;
|
return parsed;
|
||||||
}
|
}
|
||||||
|
}
|
||||||
|
|
||||||
const sanitizeConfig = {
|
const sanitizeConfig = {
|
||||||
allowedTags: [
|
allowedTags: [
|
||||||
|
@ -170,8 +167,8 @@ const sanitizeConfig = {
|
||||||
|
|
||||||
export function parseHTMLBody({ mediaRepository, platform }, html) {
|
export function parseHTMLBody({ mediaRepository, platform }, html) {
|
||||||
const parseResult = platform.parseHTML(sanitizeHtml(html, sanitizeConfig));
|
const parseResult = platform.parseHTML(sanitizeHtml(html, sanitizeConfig));
|
||||||
const options = { result: parseResult, mediaRepository };
|
const deserializer = new Deserializer(parseResult, mediaRepository);
|
||||||
const parts = parseNodes(options, parseResult.rootNodes);
|
const parts = deserializer.parseNodes(parseResult.rootNodes);
|
||||||
return new MessageBody(html, parts);
|
return new MessageBody(html, parts);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
Reference in a new issue