Move HTML parsing into platform.

src/platform/web/HTMLParseResult.js

@@ -0,0 +1,37 @@
+export class HTMLParseResult {
+    constructor(bodyNode) {
+        this._bodyNode = bodyNode;
+    }
+
+    get rootNodes() {
+        return this._bodyNode.childNodes;
+    }
+    
+    getChildNodes(node) {
+        return node.childNodes;
+    }
+
+    getAttributeNames(node) {
+        return node.getAttributeNames();
+    }
+
+    getAttributeValue(node, attr) {
+        return node.getAttribute(attr);
+    }
+
+    isTextNode(node) { 
+        return node.nodeType === Node.TEXT_NODE;
+    }
+
+    getNodeText(node) {
+        return node.nodeValue;
+    }
+
+    isElementNode(node) {
+        return node.nodeType === Node.ELEMENT_NODE;
+    }
+
+    getNodeElementName(node) {
+        return node.tagName;
+    }
+}

src/platform/web/Platform.js

@@ -36,6 +36,7 @@ import {BlobHandle} from "./dom/BlobHandle.js";
 import {hasReadPixelPermission, ImageHandle, VideoHandle} from "./dom/ImageHandle.js";
 import {downloadInIframe} from "./dom/download.js";
 import {Disposables} from "../../utils/Disposables.js";
+import {HTMLParseResult} from "./HTMLParseResult.js";
 
 function addScript(src) {
     return new Promise(function (resolve, reject) {
@@ -236,6 +237,11 @@ export class Platform {
         return promise;
     }
 
+    parseHTML(html) {
+        const bodyNode = new DOMParser().parseFromString(html, "text/html").body;
+        return new HTMLParseResult(bodyNode);
+    }
+
     async loadImage(blob) {
         return ImageHandle.fromBlob(blob);
     }

src/platform/web/dom/deserialize.js

@@ -17,41 +17,49 @@ const basicNodes = ["EM", "STRONG", "CODE", "DEL", "P", "DIV", "SPAN" ]
  * Return a builder function for a particular tag.
  */
 function basicWrapper(tag) {
-    return (_, children) => new FormatPart(tag, children);
+    return (result, node, children) => new FormatPart(tag, children);
 }
 
 /**
  * Return a builder function for a particular header level.
  */
 function headerWrapper(level) {
-    return (_, children) => new HeaderBlock(level, children);
+    return (result, node, children) => new HeaderBlock(level, children);
 }
 
-function parseLink(node, children) {
+function parseLink(result, node, children) {
-    return new LinkPart(node.href, children);
+    // TODO Not equivalent to `node.href`!
+    // Add another HTMLParseResult method?
+    let href = result.getAttributeValue(node, "href");
+    return new LinkPart(href, children);
 }
 
-function parseList(node) {
+function parseList(result, node) {
-    const start = node.getAttribute("start") || 1;
+    // TODO Attribute's a string.
+    const start = result.getAttributeValue(node, "start") || 1;
     const nodes = [];
-    const len = node.childNodes.length;
+    for (const child of result.getChildNodes(node)) {
-    for (let i = 0; i < len; i += 1) {
+        if (result.getNodeElementName(child) !== "LI") {
-        const child = node.childNodes[i];
-        if (child.tagName !== "LI") {
             continue;
         }
-        nodes.push(parseNodes(child.childNodes));
+        const item = parseNodes(result, result.getChildNodes(child));
+        nodes.push(item);
     }
     return new ListBlock(start, nodes);
 }
 
-function parseCodeBlock(node) {
+function parseCodeBlock(result, node) {
     let codeNode;
-    if (!((codeNode = node.firstChild) && codeNode.nodeName === "CODE")) {
+    for (const child of result.getChildNodes(node)) {
+        codeNode = child;
+        break;
+    }
+    if (!(codeNode && result.getNodeElementName(codeNode) === "CODE")) {
         return null;
     }
     let language = "";
-    for (const clname of codeNode.classList) {
+    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;
@@ -60,7 +68,7 @@ function parseCodeBlock(node) {
     return new CodeBlock(language, codeNode.textContent);
 }
 
-function parseImage(node) {
+function parseImage(result, node) {
     return null;
 }
 
@@ -98,25 +106,24 @@ function buildNodeMap() {
  */
 const nodes = buildNodeMap();
 
-function parseNode(node) {
+function parseNode(result, node) {
-    if (node.nodeType === Node.TEXT_NODE) {
+    if (result.isTextNode(node)) {
-        return new TextPart(node.nodeValue);
+        return new TextPart(result.getNodeText(node));
-    } else if (node.nodeType === Node.ELEMENT_NODE) {
+    } else if (result.isElementNode(node)) {
-        const f = nodes[node.nodeName];
+        const f = nodes[result.getNodeElementName(node)];
         if (!f) {
             return null;
         }
-        let result = f.parsefn(node, f.descend ? parseNodes(node.childNodes) : null);
+        const children = f.descend ? parseNodes(result, node.childNodes) : null;
-        return result;
+        return f.parsefn(result, node, children);
     }
     return null;
 }
 
-function parseNodes(nodes) {
+function parseNodes(result, nodes) {
-    const len = nodes.length;
     const parsed = [];
-    for (let i = 0; i < len; i ++) {
+    for (const htmlNode of nodes) {
-        let node = parseNode(nodes[i]);
+        let node = parseNode(result, htmlNode);
         // Just ignore invalid / unknown tags.
         if (node) {
             parsed.push(node);
@@ -125,7 +132,7 @@ function parseNodes(nodes) {
     return parsed;
 }
 
-export function parse(html) {
+export function parse(platform, html) {
-    const rootNode = new DOMParser().parseFromString(html, "text/html").body;
+    const parseResult = platform.parseHTML(html);
-    return parseNodes(rootNode.childNodes);
+    return parseNodes(parseResult, parseResult.rootNodes);
 }