markmap-html-parser - npm Package Compare versions

Comparing version 0.16.0 to 0.16.1

dist/index.d.ts

@@ -0,1 +1,2 @@
+import { AnyNode, Cheerio, CheerioAPI, Element } from 'cheerio';
 import { IPureNode } from 'markmap-common';
@@ -25,11 +26,20 @@ export declare enum Levels {
 }
+export interface IHtmlParserContext {
+    $node: Cheerio<Element>;
+    $: CheerioAPI;
+    getContent($node: Cheerio<AnyNode>, preserveTag?: boolean): {
+        html?: string;
+        comments?: string[];
+    };
+}
+export interface IHtmlParserResult {
+    html?: string | null;
+    comments?: string[];
+    queue?: Cheerio<Element>;
+    nesting?: boolean;
+}
+export type IHtmlParserSelectorRules = Record<string, (context: IHtmlParserContext) => IHtmlParserResult>;
 export interface IHtmlParserOptions {
-    /** Matches nodes to be displayed on markmap. */
     selector: string;
-    /** 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;
+    selectorRules: IHtmlParserSelectorRules;
 }
@@ -36,0 +46,0 @@ export declare const defaultOptions: IHtmlParserOptions;

dist/index.js

@@ -18,2 +18,39 @@ "use strict";
 })(Levels || {});
+const defaultSelectorRules = {
+  "div,p": ({ $node }) => ({
+    queue: $node.children()
+  }),
+  "h1,h2,h3,h4,h5,h6": ({ $node, getContent }) => ({
+    ...getContent($node.contents())
+  }),
+  "ul,ol": ({ $node }) => ({
+    queue: $node.children(),
+    nesting: true
+  }),
+  li: ({ $node, getContent }) => {
+    const queue = $node.children().filter("ul,ol");
+    let content;
+    if ($node.contents().first().is("div,p")) {
+      content = getContent($node.children().first());
+    } else {
+      let $contents = $node.contents();
+      const i = $contents.index(queue);
+      if (i >= 0)
+        $contents = $contents.slice(0, i);
+      content = getContent($contents);
+    }
+    return {
+      queue,
+      nesting: true,
+      ...content
+    };
+  },
+  "table,pre,p>img:only-child": ({ $node, getContent }) => ({
+    ...getContent($node)
+  })
+};
+const defaultOptions = {
+  selector: "h1,h2,h3,h4,h5,h6,ul,ol,li,table,pre,p>img:only-child",
+  selectorRules: defaultSelectorRules
+};
 const MARKMAP_COMMENT_PREFIX = "markmap: ";
@@ -23,8 +60,2 @@ const SELECTOR_HEADING = /^h[1-6]$/;
 const SELECTOR_LIST_ITEM = /^li$/;
-const defaultOptions = {
-  selector: "h1,h2,h3,h4,h5,h6,ul,ol,li,table,pre,p>img:only-child",
-  selectorWrapper: "div,p",
-  selectorNesting: "ul,ol,li",
-  selectorPreserveTag: "table,pre,img"
-};
 function getLevel(tagName) {
@@ -56,39 +87,24 @@ if (SELECTOR_HEADING.test(tagName))
   };
-  const elMap = {
-    [id]: $("<div>")
-  };
   const headingStack = [];
   let skippingHeading = 0;
-  checkNode($root);
-  markmapCommon.walkTree(rootNode, (node, next) => {
-    next();
-    const $el = elMap[node.id];
-    $el.contents().each((_, child) => {
-      if (child.type === "comment") {
-        const data2 = child.data.trim();
-        if (data2.startsWith(MARKMAP_COMMENT_PREFIX)) {
-          node.comments || (node.comments = []);
-          node.comments.push(data2.slice(MARKMAP_COMMENT_PREFIX.length).trim());
-          $(child).remove();
-        }
-      }
-    });
-    const data = $el.data();
-    if (data && Object.keys(data).length) {
-      node.data = data;
-    }
-    node.html = (($el.is(options.selectorPreserveTag) ? $.html($el) : $el.html()) || "").trimEnd();
-  });
+  checkNodes($root.children());
   return rootNode;
-  function addChild($child, parent, nesting) {
-    const child = $child[0];
+  function addChild(props) {
+    var _a;
+    const { parent } = props;
     const node = {
       id: ++id,
-      tag: child.tagName,
-      html: "",
-      level: getLevel(child.tagName),
+      tag: props.tagName,
+      level: props.level,
+      html: props.html,
       childrenLevel: 0,
-      children: nesting ? [] : void 0,
+      children: props.nesting ? [] : void 0,
       parent: parent.id
     };
+    if ((_a = props.comments) == null ? void 0 : _a.length) {
+      node.comments = props.comments;
+    }
+    if (Object.keys(props.data || {}).length) {
+      node.data = props.data;
+    }
     if (parent.children) {
@@ -103,3 +119,2 @@ if (parent.childrenLevel === 0 || parent.childrenLevel > node.level) {
     }
-    elMap[node.id] = $child;
     return node;
@@ -114,41 +129,60 @@ }
   }
-  function checkNode($el, node) {
-    $el.children().each((_, child) => {
+  function getContent($node) {
+    var _a;
+    const result = extractMagicComments($node);
+    const html2 = (_a = $.html(result.$node)) == null ? void 0 : _a.trimEnd();
+    return { comments: result.comments, html: html2 };
+  }
+  function extractMagicComments($node) {
+    const comments = [];
+    $node = $node.filter((_, child) => {
+      if (child.type === "comment") {
+        const data = child.data.trim();
+        if (data.startsWith(MARKMAP_COMMENT_PREFIX)) {
+          comments.push(data.slice(MARKMAP_COMMENT_PREFIX.length).trim());
+          return false;
+        }
+      }
+      return true;
+    });
+    return { $node, comments };
+  }
+  function checkNodes($els, node) {
+    $els.each((_, child) => {
+      var _a;
       const $child = $(child);
-      if ($child.is(options.selectorWrapper)) {
-        checkNode($child);
+      const rule = (_a = Object.entries(options.selectorRules).find(
+        ([selector]) => $child.is(selector)
+      )) == null ? void 0 : _a[1];
+      const result = rule == null ? void 0 : rule({ $node: $child, $, getContent });
+      if ((result == null ? void 0 : result.queue) && !result.nesting) {
+        checkNodes(result.queue, node);
         return;
       }
-      if (!$child.is(options.selector)) {
-        const level2 = getLevel(child.tagName);
-        if (level2 <= 6) {
-          skippingHeading = level2;
+      const level = getLevel(child.tagName);
+      if (!result) {
+        if (level <= 6) {
+          skippingHeading = level;
         }
         return;
       }
-      const level = getLevel(child.tagName);
-      if (skippingHeading > 0 && level > skippingHeading) {
+      if (skippingHeading > 0 && level > skippingHeading)
         return;
-      }
+      if (!$child.is(options.selector))
+        return;
       skippingHeading = 0;
-      if ($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
-      );
+      const childNode = addChild({
+        parent: node || getCurrentHeading(level),
+        nesting: !!result.queue || isHeading,
+        tagName: child.tagName,
+        level,
+        html: result.html || "",
+        comments: result.comments,
+        data: $child.data()
+      });
       if (isHeading)
         headingStack.push(childNode);
+      if (result.queue)
+        checkNodes(result.queue, childNode);
     });
@@ -155,0 +189,0 @@ }

package.json

 {
   "name": "markmap-html-parser",
-  "version": "0.16.0",
+  "version": "0.16.1",
   "description": "Parse HTML into markmap data structure",
@@ -64,3 +64,3 @@ "author": "Gerald <[e-mail]>",
   },
-  "gitHead": "caf3f840ad39c240ba522bfea83c57de62c6cac4"
+  "gitHead": "f10ad62ef03e02a1843484cca51521f48989fab7"
 }

No alert changes

Improved metrics

Total package byte prevSize

506653

increased by0.71%

Lines of code

14448

increased by0.78%

No dependency changes