feat: support svg elements

packages/html-data/bin/aria.ts

@@ -11,7 +11,7 @@ import {
 } from "@webstudio-is/sdk";
 import { generateWebstudioComponent } from "@webstudio-is/react-sdk";
 import {
-  findTags,
+  findByTags,
   getAttr,
   getTextContent,
   loadPage,
@@ -34,11 +34,11 @@ const overrides: Record<string, Partial<Attribute>> = {
 
 const html = await loadPage("aria1.3", "https://www.w3.org/TR/wai-aria-1.3");
 const document = parseHtml(html);
-const list = findTags(document, "dl").find(
+const list = findByTags(document, "dl").find(
   (table) => getAttr(table, "id")?.value === "index_state_prop"
 );
-const terms = findTags(list, "dt");
-const details = findTags(list, "dd");
+const terms = findByTags(list, "dt");
+const details = findByTags(list, "dd");
 const descriptions = new Map<string, string>();
 for (let index = 0; index < terms.length; index += 1) {
   const term = getTextContent(terms[index]);

packages/html-data/bin/attributes.ts

@@ -10,7 +10,7 @@ import {
 } from "@webstudio-is/sdk";
 import { generateWebstudioComponent } from "@webstudio-is/react-sdk";
 import {
-  findTags,
+  findByTags,
   getAttr,
   getTextContent,
   loadHtmlIndices,
@@ -123,11 +123,11 @@ const overrides: Record<
 // Crawl WHATWG HTML.
 const html = await loadHtmlIndices();
 const document = parseHtml(html);
-const table = findTags(document, "table").find(
+const table = findByTags(document, "table").find(
   (table) => getAttr(table, "id")?.value === "attributes-1"
 );
-const [tbody] = findTags(table, "tbody");
-const rows = findTags(tbody, "tr");
+const [tbody] = findByTags(table, "tbody");
+const rows = findByTags(tbody, "tr");
 
 const attributesByTag: Record<string, Attribute[]> = {};
 // textarea does not have value attribute and text content is used as initial value

packages/html-data/bin/crawler.ts

@@ -13,7 +13,7 @@ type Element = DefaultTreeAdapterMap["element"];
 
 type Attribute = Element["attrs"][number];
 
-export const findTags = (
+export const findByTags = (
   node: undefined | Node,
   tagName: string,
   result: NodeWithChildren[] = []
@@ -23,7 +23,28 @@ export const findTags = (
       result.push(node);
     }
     for (const child of node.childNodes) {
-      findTags(child, tagName, result);
+      findByTags(child, tagName, result);
+    }
+  }
+  return result;
+};
+
+export const findByClasses = (
+  node: undefined | Node,
+  className: string,
+  result: NodeWithChildren[] = []
+): NodeWithChildren[] => {
+  if (node && "childNodes" in node) {
+    if (
+      "tagName" in node &&
+      node.attrs.some(
+        (item) => item.name === "class" && item.value === className
+      )
+    ) {
+      result.push(node);
+    }
+    for (const child of node.childNodes) {
+      findByClasses(child, className, result);
     }
   }
   return result;
@@ -72,3 +93,6 @@ export const loadHtmlIndices = () =>
     "html-spec-indices",
     "https://html.spec.whatwg.org/multipage/indices.html"
   );
+
+export const loadSvgSinglePage = () =>
+  loadPage("svg-spec", "https://www.w3.org/TR/SVG11/single-page.html");

packages/html-data/bin/elements.ts

@@ -1,17 +1,16 @@
 import { dirname } from "node:path";
 import { mkdir, writeFile } from "node:fs/promises";
 import {
-  findTags,
+  findByClasses,
+  findByTags,
   getTextContent,
   loadHtmlIndices,
+  loadSvgSinglePage,
   parseHtml,
 } from "./crawler";
 
 // Crawl WHATWG HTML.
 
-const html = await loadHtmlIndices();
-const document = parseHtml(html);
-
 type Element = {
   description: string;
   categories: string[];
@@ -24,12 +23,15 @@ const elementsByTag: Record<string, Element> = {};
  * scrape elements table with content model
  */
 {
-  const table = findTags(document, "table").find((table) => {
-    const [caption] = findTags(table, "caption");
+  const html = await loadHtmlIndices();
+  const document = parseHtml(html);
+
+  const table = findByTags(document, "table").find((table) => {
+    const [caption] = findByTags(table, "caption");
     return getTextContent(caption).toLowerCase().includes("list of elements");
   });
-  const [tbody] = findTags(table, "tbody");
-  const rows = findTags(tbody, "tr");
+  const [tbody] = findByTags(table, "tbody");
+  const rows = findByTags(tbody, "tr");
   const parseList = (text: string) => {
     return text
       .trim()
@@ -54,6 +56,7 @@ const elementsByTag: Record<string, Element> = {};
         return item;
       }
     );
+    categories.unshift("html-element");
     let children = parseList(getTextContent(row.childNodes[4]));
     for (const tag of elements) {
       // textarea does not have value attribute and text content is used as initial value
@@ -78,6 +81,39 @@ const elementsByTag: Record<string, Element> = {};
   }
 }
 
+{
+  const svg = await loadSvgSinglePage();
+  const document = parseHtml(svg);
+  const summaries = findByClasses(document, "element-summary");
+  for (const summary of summaries) {
+    const [name] = findByClasses(summary, "element-summary-name").map((item) =>
+      getTextContent(item).slice(1, -1)
+    );
+    const children: string[] = [];
+    const [dl] = findByTags(summary, "dl");
+    for (let index = 0; index < dl.childNodes.length; index += 1) {
+      const child = dl.childNodes[index];
+      if (getTextContent(child).toLowerCase().includes("content model")) {
+        const dd = dl.childNodes[index + 1];
+        for (const elementName of findByClasses(dd, "element-name")) {
+          children.push(getTextContent(elementName).slice(1, -1));
+        }
+      }
+    }
+    if (elementsByTag[name]) {
+      console.info(`${name} element from SVG specification is skipped`);
+      continue;
+    }
+    const categories = name === "svg" ? ["flow", "phrasing"] : ["none"];
+    categories.unshift("svg-element");
+    elementsByTag[name] = {
+      description: "",
+      categories,
+      children,
+    };
+  }
+}
+
 const contentModel = `type Element = {
   description: string;
   categories: string[];