Merge pull request #340 from extractus/7.2.14

v7.2.14
extractus · Apr 18, 2023 · 87a9708 · 87a9708
2 parents 2715cd9 + 8b3d594
commit 87a9708
Show file tree

Hide file tree

Showing 6 changed files with 14 additions and 7 deletions.
diff --git a/dist/article-extractor.esm.js b/dist/article-extractor.esm.js
diff --git a/dist/cjs/article-extractor.js b/dist/cjs/article-extractor.js
diff --git a/dist/cjs/package.json b/dist/cjs/package.json
@@ -1,5 +1,5 @@
 {
   "name": "@extractus/article-extractor",
-  "version": "7.2.13",
+  "version": "7.2.14",
   "main": "./article-extractor.js"
 }
diff --git a/package.json b/package.json
@@ -1,5 +1,5 @@
 {
-  "version": "7.2.13",
+  "version": "7.2.14",
   "name": "@extractus/article-extractor",
   "description": "To extract main article from given URL",
   "homepage": "https://github.com/extractus/article-extractor",
@@ -40,7 +40,7 @@
   },
   "devDependencies": {
     "@types/sanitize-html": "^2.9.0",
-    "esbuild": "^0.17.16",
+    "esbuild": "^0.17.17",
     "eslint": "^8.38.0",
     "jest": "^29.5.0",
     "nock": "^13.3.0"

diff --git a/src/utils/extractMetaData.js b/src/utils/extractMetaData.js
@@ -29,16 +29,19 @@ export default (html) => {
   const urlAttrs = [
     'og:url',
     'twitter:url',
+    'parsely-link',
   ]
   const titleAttrs = [
     'title',
     'og:title',
     'twitter:title',
+    'parsely-title',
   ]
   const descriptionAttrs = [
     'description',
     'og:description',
     'twitter:description',
+    'parsely-description',
   ]
   const imageAttrs = [
     'image',
@@ -47,6 +50,7 @@ export default (html) => {
     'og:image:secure_url',
     'twitter:image',
     'twitter:image:src',
+    'parsely-image-url',
   ]
   const authorAttrs = [
     'author',
@@ -55,6 +59,7 @@ export default (html) => {
     'article:author',
     'twitter:creator',
     'dc.creator',
+    'parsely-author',
   ]
   const publishedTimeAttrs = [
     'article:published_time',
@@ -72,6 +77,7 @@ export default (html) => {
     'published_time',
     'release_date',
     'date',
+    'parsely-pub-date',
   ]
 
   const document = new DOMParser().parseFromString(html, 'text/html')

diff --git a/src/utils/parseFromHtml.js b/src/utils/parseFromHtml.js
@@ -32,6 +32,7 @@ const summarize = (desc, txt, threshold, maxlen) => { // eslint-disable-line
 export default async (inputHtml, inputUrl = '', parserOptions = {}) => {
   const html = purify(inputHtml)
   const meta = extractMetaData(html)
+
   let title = meta.title
 
   const {