Merge pull request #408 from extractus/8.0.14

v8.0.14

package.json

@@ -1,5 +1,5 @@
 {
-  "version": "8.0.13",
+  "version": "8.0.14",
   "name": "@extractus/article-extractor",
   "description": "To extract main article from given URL",
   "homepage": "https://github.com/extractus/article-extractor",

src/utils/extractLdSchema.js

@@ -1,3 +1,7 @@
+// utils -> extractLdSchema.js
+
+import { isArray, isObject, isString } from 'bellajs'
+
 const typeSchemas = [
   'aboutpage',
   'checkoutpage',
@@ -53,25 +57,20 @@ const parseJson = (text) => {
  */
 export default (document, entry) => {
   const ldSchemas = document.querySelectorAll('script[type="application/ld+json"]')
-
   ldSchemas.forEach(ldSchema => {
     const ldJson = parseJson(ldSchema.textContent.replace(/[\n\r\t]/g, ''))
     const isAllowedLdJsonType = typeSchemas.includes(ldJson['@type']?.toLowerCase())
 
     if (ldJson && isAllowedLdJsonType) {
       Object.entries(attributeLists).forEach(([key, attr]) => {
-        const isEntryAlreadyPopulated = typeof entry[key] !== 'undefined' && entry[key] !== ''
-
-        if (isEntryAlreadyPopulated || !ldJson[attr]) {
+        if (!entry[key] || !ldJson[attr]) {
           return
         }
 
         const keyValue = ldJson[attr]
-        if (keyValue) {
-          entry[key] = Array.isArray(keyValue) ? keyValue[0] : keyValue
-          if (typeof entry[key] === 'string') {
-            entry[key] = entry[key].toLowerCase().trim()
-          }
+        const val = isArray(keyValue) ? keyValue[0] : isObject(keyValue) ? keyValue?.name || '' : keyValue
+        if (isString(val)) {
+          entry[key] = val.trim()
         }
       })
     }

src/utils/extractMetaData.js

@@ -143,11 +143,11 @@ export default (html) => {
     }
   })
 
-  const entries = extractLdSchema(doc, entry)
+  const metadata = extractLdSchema(doc, entry)
 
-  if (!entries.published) {
-    entries.published = findDate(doc)
+  if (!metadata.published) {
+    metadata.published = findDate(doc) || ''
   }
 
-  return entries
+  return metadata
 }

src/utils/findDate.js

@@ -38,7 +38,7 @@ export default function (doc) {
       const match = element.textContent.match(pattern)
       if (match) return convertDateFormat(match[0])
     }
-    return null
+    return ''
   }
 
   const priorityElements = doc.querySelectorAll('time, [datetime], [itemprop~=datePublished], [itemprop~=dateCreated]')
@@ -53,5 +53,5 @@ export default function (doc) {
     if (date) return date
   }
 
-  return null
+  return ''
 }