Merge pull request #418 from extractus/8.0.19

v8.0.19

Author: ndaidong

.github/workflows/ci-test.yml

@@ -12,7 +12,7 @@ jobs:
 
     strategy:
       matrix:
-        node_version: [20.x, 22.x, 23.x]
+        node_version: [20.x, 22.x, 24.x]
 
     steps:
     - uses: actions/checkout@v4

eval.js

@@ -3,7 +3,7 @@
 import { execSync } from 'node:child_process'
 import { readFileSync, writeFileSync, existsSync } from 'node:fs'
 
-import { slugify } from 'bellajs'
+import { slugify } from '@ndaidong/bellajs'
 
 import { isValid as isValidUrl } from './src/utils/linker.js'
 import { extract, extractFromHtml } from './src/main.js'

package.json

@@ -1,5 +1,5 @@
 {
-  "version": "8.0.18",
+  "version": "8.0.19",
   "name": "@extractus/article-extractor",
   "description": "To extract main article from given URL",
   "homepage": "https://github.com/extractus/article-extractor",
@@ -31,16 +31,16 @@
   },
   "dependencies": {
     "@mozilla/readability": "^0.6.0",
-    "bellajs": "^11.2.0",
+    "@ndaidong/bellajs": "^12.0.1",
     "cross-fetch": "^4.1.0",
     "linkedom": "^0.18.10",
     "sanitize-html": "2.16.0"
   },
   "devDependencies": {
     "@eslint/js": "^9.26.0",
-    "@types/sanitize-html": "^2.15.0",
+    "@types/sanitize-html": "^2.16.0",
     "eslint": "^9.26.0",
-    "globals": "^16.0.0",
+    "globals": "^16.1.0",
     "https-proxy-agent": "^7.0.6",
     "nock": "^14.0.4"
   },

src/config.js

@@ -1,6 +1,6 @@
 // config.js
 
-import { clone } from 'bellajs'
+import { clone } from '@ndaidong/bellajs'
 
 const sanitizeHtmlOptions = {
   allowedTags: [

src/main.js

@@ -2,7 +2,7 @@
 
 import {
   isString
-} from 'bellajs'
+} from '@ndaidong/bellajs'
 
 import retrieve from './utils/retrieve.js'
 import parseFromHtml from './utils/parseFromHtml.js'

src/utils/extractLdSchema.js

@@ -1,6 +1,6 @@
 // utils -> extractLdSchema.js
 
-import { isArray, isObject, isString } from 'bellajs'
+import { isArray, isObject, isString } from '@ndaidong/bellajs'
 
 const typeSchemas = [
   'aboutpage',
@@ -74,7 +74,7 @@ export default (document, entry) => {
 
         const keyValue = ldJson[attr]
         const val = isArray(keyValue) ? keyValue[0] : isObject(keyValue) ? keyValue?.name || '' : keyValue
-        if (isString(val)) {
+        if (isString(val) && val !== '') {
           entry[key] = val.trim()
         }
       })

src/utils/extractMetaData.js

@@ -138,8 +138,9 @@ export default (html) => {
 
   Array.from(doc.getElementsByTagName('meta')).forEach(node => {
     const result = getMetaContentByNameOrProperty(node, attributeLists)
-    if (result) {
-      entry[result.key] = result.content
+    const val = result?.content || ''
+    if (val !== '') {
+      entry[result.key] = val
     }
   })
 

src/utils/extractMetaData.test.js

@@ -4,7 +4,7 @@ import assert from 'node:assert'
 
 import { readFileSync } from 'node:fs'
 
-import { isObject, hasProperty } from 'bellajs'
+import { isObject, hasProperty } from '@ndaidong/bellajs'
 
 import extractMetaData from './extractMetaData.js'
 

src/utils/extractWithReadability.js

@@ -2,7 +2,7 @@
 
 import { Readability } from '@mozilla/readability'
 import { DOMParser } from 'linkedom'
-import { isString } from 'bellajs'
+import { isString } from '@ndaidong/bellajs'
 
 export default (html, url = '') => {
   if (!isString(html)) {

src/utils/extractWithReadability.test.js

@@ -5,7 +5,7 @@ import assert from 'node:assert'
 
 import { readFileSync } from 'node:fs'
 
-import { isString } from 'bellajs'
+import { isString } from '@ndaidong/bellajs'
 
 import extractWithReadability, { extractTitleWithReadability } from './extractWithReadability.js'