markmap-html-parser - npm Package Compare versions

Comparing version 0.15.9-alpha.0 to 0.15.9-alpha.5

dist/index.d.ts

@@ -19,11 +19,15 @@ import { IPureNode } from 'markmap-common';
     level: Levels;
+    parent: number;
     childrenLevel: Levels;
     children?: IHtmlNode[];
-    parent: number;
+    comments?: string[];
 }
 export interface IHtmlParserOptions {
+    /** Matches nodes to be displayed on markmap. */
     selector: string;
-    /** If matches, the whole content will be preserved and no any other nodes will be selected. */
-    selectorPreserveContent: string;
-    /** If matches, the current tag will be preserved along with its content. */
+    /** Matches wrapper elements that should be ignored. */
+    selectorWrapper: string;
+    /** Matches elements that can be nested, such as `li`. */
+    selectorNesting: string;
+    /** Matches elements whose tag name should be preserved, such as `<pre>`, `<img>`. */
     selectorPreserveTag: string;
@@ -33,2 +37,3 @@ }
 export declare function parseHtml(html: string, opts?: Partial<IHtmlParserOptions>): IHtmlNode;
+export declare function convertNode(htmlRoot: IHtmlNode): IPureNode;
 export declare function buildTree(html: string, opts?: Partial<IHtmlParserOptions>): IPureNode;

dist/index.js

@@ -18,2 +18,3 @@ "use strict";
 })(Levels || {});
+const MARKMAP_COMMENT_PREFIX = "markmap: ";
 const SELECTOR_HEADING = /^h[1-6]$/;
@@ -24,3 +25,4 @@ const SELECTOR_LIST = /^[uo]l$/;
   selector: "h1,h2,h3,h4,h5,h6,ul,ol,li,table,pre,p>img:first-child:last-child",
-  selectorPreserveContent: "table,pre",
+  selectorWrapper: "div,p",
+  selectorNesting: "ul,ol,li",
   selectorPreserveTag: "table,pre,img"
@@ -43,7 +45,5 @@ };
   const $ = cheerio.load(html);
+  const $root = $("body");
   let id = 0;
-  const elMap = {
-    [id]: $("<div>")
-  };
-  const root = {
+  const rootNode = {
     id,
@@ -53,62 +53,124 @@ tag: "",
     level: 0,
+    parent: 0,
     childrenLevel: 0,
-    children: [],
-    parent: 0
+    children: []
   };
-  const nodeMap = {
-    [id]: root
+  const elMap = {
+    [id]: $("<div>")
   };
-  let cur = root;
-  $(options.selector).each((_, el) => {
+  const headingStack = [];
+  let skippingHeading = 0;
+  checkNode($root);
+  markmapCommon.walkTree(rootNode, (node, next) => {
+    next();
+    const $el = elMap[node.id];
+    $el.contents().each((_, child) => {
+      if (child.type === "comment") {
+        const data = child.data.trim();
+        if (data.startsWith(MARKMAP_COMMENT_PREFIX)) {
+          node.comments || (node.comments = []);
+          node.comments.push(data.slice(MARKMAP_COMMENT_PREFIX.length).trim());
+          $(child).remove();
+        }
+      }
+    });
+    node.html = (($el.is(options.selectorPreserveTag) ? $.html($el) : $el.html()) || "").trimEnd();
+  });
+  return rootNode;
+  function addChild($child, parent, nesting) {
+    const child = $child[0];
     const node = {
       id: ++id,
-      tag: el.tagName,
+      tag: child.tagName,
       html: "",
-      level: getLevel(el.tagName),
+      level: getLevel(child.tagName),
       childrenLevel: 0,
-      children: $(el).is(options.selectorPreserveContent) ? void 0 : [],
-      parent: 0
+      children: nesting ? [] : void 0,
+      parent: parent.id
     };
-    elMap[node.id] = $(el);
-    nodeMap[node.id] = node;
-    if (node.level <= 6) {
-      while (cur !== root && (cur.level > 6 || cur.tag >= node.tag)) {
-        cur = nodeMap[cur.parent];
+    if (parent.children) {
+      if (parent.childrenLevel === 0 || parent.childrenLevel > node.level) {
+        parent.children = [];
+        parent.childrenLevel = node.level;
       }
-    } else {
-      while (cur !== root && cur.level > 6 && !elMap[cur.id].find(el).length) {
-        cur = nodeMap[cur.parent];
+      if (parent.childrenLevel === node.level) {
+        parent.children.push(node);
       }
     }
-    if (cur.children) {
-      node.parent = cur.id;
-      const level = getLevel(node.tag);
-      if (!cur.childrenLevel || cur.childrenLevel > level) {
-        cur.childrenLevel = level;
-        cur.children = [];
+    elMap[node.id] = $child;
+    return node;
+  }
+  function getCurrentHeading(level) {
+    let heading;
+    while ((heading = headingStack.at(-1)) && heading.level >= level) {
+      headingStack.pop();
+    }
+    return heading || rootNode;
+  }
+  function checkNode($el, node) {
+    $el.children().each((_, child) => {
+      const $child = $(child);
+      if ($child.is(options.selectorWrapper)) {
+        checkNode($child);
+        return;
       }
-      if (cur.childrenLevel === level) {
-        cur.children.push(node);
+      if (!$child.is(options.selector)) {
+        const level2 = getLevel(child.tagName);
+        if (level2 <= 6) {
+          skippingHeading = level2;
+        }
+        return;
       }
-      cur = node;
+      const level = getLevel(child.tagName);
+      if (skippingHeading > 0 && level > skippingHeading) {
+        return;
+      }
+      skippingHeading = 0;
+      if ($child.is(options.selectorNesting)) {
+        const childNode2 = addChild(
+          $child,
+          node || getCurrentHeading(level),
+          true
+        );
+        if (childNode2) {
+          checkNode($child, childNode2);
+          $child.remove();
+        }
+        return;
+      }
+      const isHeading = level <= 6;
+      const childNode = addChild(
+        $child,
+        node || getCurrentHeading(level),
+        isHeading
+      );
+      if (isHeading)
+        headingStack.push(childNode);
+    });
+  }
+}
+function convertNode(htmlRoot) {
+  return markmapCommon.walkTree(htmlRoot, (htmlNode, next) => {
+    const node = {
+      content: htmlNode.html,
+      children: next() || []
+    };
+    if (htmlNode.comments) {
+      if (htmlNode.comments.includes("foldAll")) {
+        node.payload = { fold: 2 };
+      } else if (htmlNode.comments.includes("fold")) {
+        node.payload = { fold: 1 };
+      }
     }
+    return node;
   });
-  markmapCommon.walkTree(root, (node, next) => {
-    next();
-    const $el = elMap[node.id];
-    $el.remove();
-    node.html = (($el.is(options.selectorPreserveTag) ? $.html($el) : $el.html()) || "").trimEnd();
-  });
-  return root;
 }
 function buildTree(html, opts) {
   const htmlRoot = parseHtml(html, opts);
-  return markmapCommon.walkTree(htmlRoot, (node, next) => ({
-    content: node.html,
-    children: next() || []
-  }));
+  return convertNode(htmlRoot);
 }
 exports.Levels = Levels;
 exports.buildTree = buildTree;
+exports.convertNode = convertNode;
 exports.defaultOptions = defaultOptions;
 exports.parseHtml = parseHtml;

package.json

 {
   "name": "markmap-html-parser",
-  "version": "0.15.9-alpha.0+9cb40e7",
+  "version": "0.15.9-alpha.5+c64866f",
   "description": "Parse HTML into markmap data structure",
@@ -44,3 +44,3 @@ "author": "Gerald <[e-mail]>",
   "devDependencies": {
-    "markmap-common": "0.15.9-alpha.0+9cb40e7"
+    "markmap-common": "0.15.9-alpha.5+c64866f"
   },
@@ -64,3 +64,3 @@ "dependencies": {
   },
-  "gitHead": "9cb40e7e9ec0aa1f25a2de6a74912cd0c21f52ed"
+  "gitHead": "c64866fb4873a99923be18ea70715aa49a6dabef"
 }

New alerts

Manifest confusion

Supply chain risk

This package has inconsistent metadata. This could be malicious or caused by an error when publishing the package.

Found 1 instance in 1 package

Fixed alerts

Manifest confusion

Supply chain risk

This package has inconsistent metadata. This could be malicious or caused by an error when publishing the package.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

502392

increased by1.14%

Lines of code

14308

increased by1.35%

No dependency changes