Add glob support and make "noreferrer" configurable

README.md

@@ -21,14 +21,29 @@ $ npm i hexo-filter-nofollow --save
 nofollow:
   enable: true
   field: site
+  elements: 
+    - 'a'
+    - 'img'
   exclude:
-    - 'exclude1.com'
-    - 'exclude2.com'
+    - '*.exclude1.com'
+    - 'exclude2.com/path/*'
+  rel:
+    - 'external'
+    - 'noreferrer'
+    - 'nofollow'
+    - 'noopener'
+  referrerpolicy: 'no-referrer'
 ```
 
 - **enable** - Enable the plugin. Default value is `true`.
 - **field** - The scope you want the plugin to proceed, can be 'site' or 'post'. Default value is `site`.
   - 'post' - Only add nofollow attribute to external links in your post content
   - 'site' - Add nofollow attribute to external links of whole sites
-- **exclude** - Exclude hostname. Specify subdomain when applicable, including `www`.
-  - 'exclude1.com' does not apply to `www.exclude1.com` nor `en.exclude1.com`.
+- **elements** - The tag to be processed, currently only supports `<a>` and `<img>`.
+- **include** - Include hostname. You can use `*` or `?` glob wildcards. If include is configured, other external links will not be processed. 
+- **exclude** - Exclude hostname. You can use `*` or `?` glob wildcards.
+  - `exclude1.com` does not apply to `www.exclude1.com` nor `en.exclude1.com`.
+  - `*.exclude1.com` can be apply to `www.exclude1.com` or `en.exclude1.com`.
+- **minimatch** - The glob wildcard is supported by [minimath](https://github.com/isaacs/minimatch), this field can be configured for it.
+- **rel** - Configurable rel attribute value.
+- **referrerpolicy** - Configurable referrerpolicy attribute value.

index.js

@@ -5,7 +5,11 @@
 hexo.config.nofollow = Object.assign({
   enable: true,
   field: 'site',
-  exclude: []
+  elements: ['a'],
+  include: [],
+  exclude: [],
+  rel: ['noopener', 'external', 'nofollow', 'noreferrer'],
+  referrerpolicy: 'no-referrer'
 }, hexo.config.nofollow);
 
 const config = hexo.config.nofollow;

lib/filter.js

@@ -1,56 +1,116 @@
 'use strict';
 
 const { parse } = require('url');
+const { Minimatch } = require('minimatch');
 
+/**
+ * Check whether the url is an external link
+ * @param {string} url
+ * @param {object} config
+ * @returns boolean
+ */
 function isExternal(url, config) {
-  const exclude = config.nofollow.exclude;
+  const { includeGlobs, excludeGlobs } = config.nofollow;
   const data = parse(url);
-  const host = data.hostname;
+  const { hostname, path } = data;
   const sitehost = parse(config.url).hostname || config.url;
 
-  if (!data.protocol || !sitehost) return false;
+  if (!data.protocol || !hostname || !sitehost) return false;
 
-  if (exclude && exclude.length) {
-    for (const i of exclude) {
-      if (host === i) return false;
+  const target = hostname + path;
+
+  if (excludeGlobs && excludeGlobs.length) {
+    for (const glob of excludeGlobs) {
+      if (glob.match(target)) return false;
     }
   }
 
-  if (host !== sitehost) return true;
+  if (includeGlobs && includeGlobs.length) {
+    for (const glob of includeGlobs) {
+      if (glob.match(target)) return true;
+    }
+  }
 
-  return false;
+  if (includeGlobs && includeGlobs.length) {
+    // If include is configured, other links will not be treated as external links
+    return false;
+  }
+  return hostname != sitehost;
 }
 
-module.exports = function(data) {
-  const hexo = this;
-  const config = hexo.config;
+/**
+ * Add attribute to the tag
+ * @param {string} source tag string
+ * @param {string} attribute string containing the url
+ * @param {string} new attribute key
+ * @param {string | array} new attribute value
+ * @returns new tag string
+ */
+function addAttr(tagStr, urlAttrStr, attrKey, attrValue) {
+  const value = [...toArray(attrValue)];
+  const regexKey = new RegExp(`${attrKey}=`, 'gi');
+  const attrRegex = new RegExp(`\\s${attrKey}="(.*?)"`, 'gi');
+  if (regexKey.test(tagStr)) {
+    tagStr = tagStr.replace(attrRegex, (attrStr, attrStrValue) => {
+      value.push(...attrStrValue.split(' '));
+      return '';
+    });
+  }
+  // De-duplicate
+  const uniqValue = [...new Set(value)];
+  return tagStr.replace(urlAttrStr, `${urlAttrStr} ${attrKey}="${uniqValue.join(' ')}"`);
+}
 
-  const exclude = config.nofollow.exclude;
-  if (exclude && !Array.isArray(exclude)) {
-    config.nofollow.exclude = [exclude];
+function toArray(data) {
+  return data && !Array.isArray(data) ? [data] : data;
+}
+
+function addSitePath(sitePattern) {
+  if (!sitePattern || sitePattern.indexOf('/') >= 0) {
+    return sitePattern;
   }
+  // default wildcard under the site
+  return sitePattern + '/**';
+}
 
-  const filterExternal = data => {
-    return data.replace(/<a.*?(href=['"](.*?)['"]).*?>/gi, (str, hrefStr, href) => {
-      if (!isExternal(href, config)) return str;
+module.exports = function nofollow(data) {
+  const hexo = this;
+  const config = hexo.config;
 
-      let noFollow = ['noopener', 'external', 'nofollow', 'noreferrer'];
+  const { elements, include, exclude, minimatch } = config.nofollow;
+  config.nofollow.elements = toArray(elements);
+  config.nofollow.include = toArray(include);
+  config.nofollow.exclude = toArray(exclude);
 
-      if (/rel=/gi.test(str)) {
-        str = str.replace(/\srel="(.*?)"/gi, (relStr, rel) => {
-          rel = rel.split(' ');
-          noFollow.push(...rel);
-          // De-duplicate
-          noFollow = [...new Set(noFollow)];
+  config.nofollow.includeGlobs = config.nofollow.include.map(pattern => new Minimatch(addSitePath(pattern), minimatch));
+  config.nofollow.excludeGlobs = config.nofollow.exclude.map(pattern => new Minimatch(addSitePath(pattern), minimatch));
 
-          return '';
-        });
-      }
+  const filterATagHrefExternal = data => {
+    return data.replace(/<a.*?(href=['"](.*?)['"]).*?>/gi, (aTagRaw, hrefAttrRaw, href) => {
+      if (!isExternal(href, config)) return aTagRaw;
+      aTagRaw = addAttr(aTagRaw, hrefAttrRaw, 'referrerpolicy', config.nofollow.referrerpolicy);
+      return addAttr(aTagRaw, hrefAttrRaw, 'rel', config.nofollow.rel);
+    });
+  };
 
-      return str.replace(hrefStr, `${hrefStr} rel="${noFollow.join(' ')}"`);
+  const filterImgTagSrcExternal = data => {
+    return data.replace(/<img.*?(src=['"](.*?)['"]).*?>/gi, (imgTagRaw, srcAttrRaw, src) => {
+      if (!isExternal(src, config)) return imgTagRaw;
+      imgTagRaw = addAttr(imgTagRaw, srcAttrRaw, 'referrerpolicy', config.nofollow.referrerpolicy);
+      return addAttr(imgTagRaw, srcAttrRaw, 'rel', config.nofollow.rel);
     });
   };
 
+  const filterExternal = data => {
+    if (config.nofollow.elements.includes('a')) {
+      data = filterATagHrefExternal(data);
+    }
+    if (config.nofollow.elements.includes('img')) {
+      data = filterImgTagSrcExternal(data);
+    }
+    return data;
+  };
+
   if (config.nofollow.field === 'post') {
     data.content = filterExternal(data.content);
   } else {
@@ -59,3 +119,5 @@ module.exports = function(data) {
 
   return data;
 };
+
+

package.json

@@ -37,5 +37,8 @@
     "eslint-config-hexo": "^4.1.0",
     "hexo": "hexojs/hexo",
     "mocha": "^8.0.1"
+  },
+  "dependencies": {
+    "minimatch": "^3.0.4"
   }
 }

test/index.js

@@ -9,49 +9,71 @@ describe('hexo-filter-nofollow', () => {
   const nofollowFilter = require('../lib/filter').bind(hexo);
 
   hexo.config.url = 'https://example.com';
-  hexo.config.nofollow = {};
+  hexo.config.nofollow = { include: [], exclude: [], elements: ['a', 'img'], rel: ['noopener', 'external', 'nofollow', 'noreferrer'], referrerpolicy: 'no-referrer' };
 
   describe('Default', () => {
     const content = [
       '# External link test',
       '1. External link',
       '<a href="https://hexo.io/">Hexo</a>',
+      '<img src="https://hexo.io/">Hexo</img>',
       '2. External link with existed "rel" Attribute',
-      '<a rel="license" href="https://github.com/hexojs/hexo-filter-nofollow/blob/master/LICENSE">Hexo</a>',
-      '<a href="https://github.com/hexojs/hexo-filter-nofollow/blob/master/LICENSE" rel="license">Hexo</a>',
+      '<a rel="license" href="https://hexo.io">Hexo</a>',
+      '<a href="https://hexo.io" rel="license">Hexo</a>',
+      '<img src="https://hexo.io" rel="license">Hexo</img>',
+      '<img rel="license" src="https://hexo.io" rel="license">Hexo</img>',
       '3. External link with existing "rel=noopener", "rel=external" or "rel=noreferrer"',
       '<a rel="noopener" href="https://hexo.io/">Hexo</a>',
       '<a href="https://hexo.io/" rel="noreferrer">Hexo</a>',
       '<a rel="noopener noreferrer" href="https://hexo.io/">Hexo</a>',
       '<a href="https://hexo.io/" rel="external noreferrer">Hexo</a>',
+      '<img rel="noopener" src="https://hexo.io/">Hexo</img>',
+      '<img src="https://hexo.io/" rel="noreferrer">Hexo</img>',
+      '<img rel="noopener noreferrer" src="https://hexo.io/">Hexo</img>',
+      '<img src="https://hexo.io/" rel="external noreferrer">Hexo</img>',
       '4. External link with Other Attributes',
       '<a class="img" href="https://hexo.io/">Hexo</a>',
       '<a href="https://hexo.io/" class="img">Hexo</a>',
+      '<img class="img" src="https://hexo.io/">Hexo</img>',
+      '<img src="https://hexo.io/" class="img">Hexo</img>',
       '5. Internal link',
       '<a href="/archives/foo.html">Link</a>',
+      '<img src="/archives/foo.html">Link</img>',
       '6. Ignore links don\'t have "href" attribute',
-      '<a>Anchor</a>'
+      '<a>Anchor</a>',
+      '<img>Anchor</img>'
     ].join('\n');
 
     const expected = [
       '# External link test',
       '1. External link',
-      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
+      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+      '<img src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</img>',
       '2. External link with existed "rel" Attribute',
-      '<a href="https://github.com/hexojs/hexo-filter-nofollow/blob/master/LICENSE" rel="noopener external nofollow noreferrer license">Hexo</a>',
-      '<a href="https://github.com/hexojs/hexo-filter-nofollow/blob/master/LICENSE" rel="noopener external nofollow noreferrer license">Hexo</a>',
+      '<a href="https://hexo.io" rel="noopener external nofollow noreferrer license" referrerpolicy="no-referrer">Hexo</a>',
+      '<a href="https://hexo.io" rel="noopener external nofollow noreferrer license" referrerpolicy="no-referrer">Hexo</a>',
+      '<img src="https://hexo.io" rel="noopener external nofollow noreferrer license" referrerpolicy="no-referrer">Hexo</img>',
+      '<img src="https://hexo.io" rel="noopener external nofollow noreferrer license" referrerpolicy="no-referrer">Hexo</img>',
       '3. External link with existing "rel=noopener", "rel=external" or "rel=noreferrer"',
-      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
-      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
-      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
-      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
+      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+      '<img src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</img>',
+      '<img src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</img>',
+      '<img src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</img>',
+      '<img src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</img>',
       '4. External link with Other Attributes',
-      '<a class="img" href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
-      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" class="img">Hexo</a>',
+      '<a class="img" href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+      '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer" class="img">Hexo</a>',
+      '<img class="img" src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</img>',
+      '<img src="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer" class="img">Hexo</img>',
       '5. Internal link',
       '<a href="/archives/foo.html">Link</a>',
+      '<img src="/archives/foo.html">Link</img>',
       '6. Ignore links don\'t have "href" attribute',
-      '<a>Anchor</a>'
+      '<a>Anchor</a>',
+      '<img>Anchor</img>'
     ].join('\n');
 
     it('Default to field = "site"', () => {
@@ -72,6 +94,64 @@ describe('hexo-filter-nofollow', () => {
     });
   });
 
+  describe('Include & Pattern', () => {
+    const content = [
+      '# Include & Pattern link test',
+      '1. External link',
+      '<a href="https://hexo.io/">Hexo</a>',
+      '2. External links whose hostname is not match glob pattern',
+      '<a href="https://example.com">Example Domain</a>',
+      '3. External links whose hostname is included in glob pattern',
+      '<a href="https://demo.example.org">Example Domain</a>',
+      '<a href="https://test.example.org">Example Domain</a>',
+      '4. External links whose hostname is included in glob pattern with path',
+      '<a href="https://path.example.org/path/to">Example Domain</a>',
+      '<a href="https://path.example.org/path">Example Domain</a>'
+    ].join('\n');
+
+    it('String', () => {
+      hexo.config.nofollow.include = ['hexo.io', '*.example.org'];
+
+      const result = nofollowFilter(content);
+
+      result.should.eql([
+        '# Include & Pattern link test',
+        '1. External link',
+        '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+        '2. External links whose hostname is not match glob pattern',
+        '<a href="https://example.com">Example Domain</a>',
+        '3. External links whose hostname is included in glob pattern',
+        '<a href="https://demo.example.org" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Example Domain</a>',
+        '<a href="https://test.example.org" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Example Domain</a>',
+        '4. External links whose hostname is included in glob pattern with path',
+        '<a href="https://path.example.org/path/to" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Example Domain</a>',
+        '<a href="https://path.example.org/path" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Example Domain</a>'
+      ].join('\n'));
+    });
+
+    it('Array', () => {
+      hexo.config.nofollow.include = 'hexo.io';
+      hexo.config.nofollow.exclude = ['example.org', '*.example.org', 'path.example.org/**'];
+
+      const result = nofollowFilter(content);
+
+      result.should.eql([
+        '# Include & Pattern link test',
+        '1. External link',
+        '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
+        '2. External links whose hostname is not match glob pattern',
+        '<a href="https://example.com">Example Domain</a>',
+        '3. External links whose hostname is included in glob pattern',
+        '<a href="https://demo.example.org">Example Domain</a>',
+        '<a href="https://test.example.org">Example Domain</a>',
+        '4. External links whose hostname is included in glob pattern with path',
+        '<a href="https://path.example.org/path/to">Example Domain</a>',
+        '<a href="https://path.example.org/path">Example Domain</a>'
+      ].join('\n'));
+    });
+  });
+
+
   describe('Exclude', () => {
     const content = [
       '# Exclude link test',
@@ -85,19 +165,20 @@ describe('hexo-filter-nofollow', () => {
     ].join('\n');
 
     it('String', () => {
+      hexo.config.nofollow.include = [];
       hexo.config.nofollow.exclude = 'example.org';
 
       const result = nofollowFilter(content);
 
       result.should.eql([
         '# Exclude link test',
         '1. External link',
-        '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
+        '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
         '2. Ignore links whose hostname is same as config',
         '<a href="https://example.com">Example Domain</a>',
         '3. Ignore links whose hostname is included in exclude',
         '<a href="https://example.org">Example Domain</a>',
-        '<a href="https://test.example.org" rel="noopener external nofollow noreferrer">Example Domain</a>'
+        '<a href="https://test.example.org" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Example Domain</a>'
       ].join('\n'));
     });
 
@@ -109,7 +190,7 @@ describe('hexo-filter-nofollow', () => {
       result.should.eql([
         '# Exclude link test',
         '1. External link',
-        '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer">Hexo</a>',
+        '<a href="https://hexo.io/" rel="noopener external nofollow noreferrer" referrerpolicy="no-referrer">Hexo</a>',
         '2. Ignore links whose hostname is same as config',
         '<a href="https://example.com">Example Domain</a>',
         '3. Ignore links whose hostname is included in exclude',