Merge pull request #143 from extractus/7.1.4

v7.1.4

build.js

@@ -1,5 +1,6 @@
+import fs from 'node:fs'
+
 import { build } from 'esbuild'
-import fs from 'fs'
 
 const { dependencies } = JSON.parse(fs.readFileSync('./package.json', 'utf8'))
 // we need esbuild to process esm dependencies while leaving cjs compatible ones

package.json

@@ -1,5 +1,5 @@
 {
-  "version": "7.1.3",
+  "version": "7.1.4",
   "name": "@extractus/feed-extractor",
   "description": "To read and normalize RSS/ATOM/JSON feed data",
   "homepage": "https://extractor-demos.pages.dev",
@@ -39,18 +39,18 @@
     "reset": "node reset"
   },
   "dependencies": {
-    "bellajs": "^11.1.3",
-    "cross-fetch": "^4.0.0",
-    "fast-xml-parser": "^4.3.6",
+    "bellajs": "^11.2.0",
+    "cross-fetch": "^4.1.0",
+    "fast-xml-parser": "^4.5.1",
     "html-entities": "^2.5.2"
   },
   "devDependencies": {
-    "esbuild": "^0.21.0",
-    "eslint": "^9.2.0",
-    "globals": "^15.1.0",
-    "https-proxy-agent": "^7.0.4",
+    "esbuild": "^0.25.0",
+    "eslint": "^9.20.0",
+    "globals": "^15.14.0",
+    "https-proxy-agent": "^7.0.6",
     "jest": "^29.7.0",
-    "nock": "^13.5.4"
+    "nock": "^14.0.1"
   },
   "keywords": [
     "extractor",

src/utils/parseJsonFeed.js

@@ -7,7 +7,8 @@ import { isArray } from 'bellajs'
 import {
   toISODateString,
   buildDescription,
-  getEntryId
+  getEntryId,
+  getPureUrl
 } from './normalizer.js'
 
 import { absolutify, purify as purifyUrl } from './linker.js'
@@ -36,7 +37,7 @@ const transform = (item, options) => {
   const entry = {
     id: getEntryId(id, link, pubDate),
     title,
-    link: purifyUrl(link) || absolutify(baseUrl, link),
+    link: getPureUrl(link, '', baseUrl),
     published,
     description: buildDescription(textContent || htmlContent || summary, descriptionMaxLen),
   }