Merge pull request #717 from masylum/date-title

chore(date): get value from title attribute

package.json

@@ -92,6 +92,10 @@
       "name": "Laurent Goderre",
       "email": "[email protected]"
     },
+    {
+      "name": "Pau Ramon Revilla",
+      "email": "[email protected]"
+    },
     {
       "name": "Omar Diab",
       "email": "[email protected]"

packages/metascraper-date/src/index.js

@@ -8,6 +8,7 @@ const dateRules = () => {
   return [
     toDate($ => $('meta[name="date" i]').attr('content')),
     toDate($ => $('[itemprop*="date" i]').attr('content')),
+    toDate($ => $('[itemprop*="date" i]').attr('title')),
     toDate($ => $('time[itemprop*="date" i]').attr('datetime')),
     toDate($ => $('time[datetime]').attr('datetime')),
     toDate($ => $filter($, $('[class*="byline" i]'))),
@@ -23,6 +24,7 @@ const datePublishedRules = () => {
     toDate($jsonld('dateCreated')),
     toDate($ => $('meta[property*="published_time" i]').attr('content')),
     toDate($ => $('[itemprop="datepublished" i]').attr('content')),
+    toDate($ => $('[itemprop="datepublished" i]').attr('title')),
     toDate($ => $filter($, $('[class*="publish" i]')))
   ]
 }

packages/metascraper/test/integration/blogger/index.js

@@ -0,0 +1,30 @@
+'use strict'
+
+const { readFile } = require('fs/promises')
+const { resolve } = require('path')
+const test = require('ava')
+
+const metascraper = require('../../..')([
+  require('metascraper-author')(),
+  require('metascraper-date')(),
+  require('metascraper-description')(),
+  require('metascraper-audio')(),
+  require('metascraper-video')(),
+  require('metascraper-image')(),
+  require('metascraper-lang')(),
+  require('metascraper-logo')(),
+  require('metascraper-logo-favicon')(),
+  require('metascraper-manifest')(),
+  require('metascraper-publisher')(),
+  require('metascraper-title')(),
+  require('metascraper-url')(),
+  require('metascraper-readability')()
+])
+
+const url = 'https://blog.gardeviance.org/2024/02/a-good-enough-map.html'
+
+test('blogger', async t => {
+  const html = await readFile(resolve(__dirname, 'input.html'))
+  const metadata = await metascraper({ html, url })
+  t.snapshot(metadata)
+})