From 45b7c2b619e64d3095477ef5bb1c908ac379a19a Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Thu, 28 Sep 2023 17:14:17 +0800
Subject: [PATCH] get published date from time element

---
 packages/readabilityjs/Readability.js                | 12 ++++++++++++
 .../nytimes-podcasts/expected-metadata.json          |  2 +-
 .../test/test-pages/nytimes-podcasts/expected.html   |  2 --
 .../test/test-pages/nytimes.com/expected.html        |  2 --
 4 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/packages/readabilityjs/Readability.js b/packages/readabilityjs/Readability.js
index abdbfa8d0..65de68275 100644
--- a/packages/readabilityjs/Readability.js
+++ b/packages/readabilityjs/Readability.js
@@ -1081,6 +1081,18 @@ Readability.prototype = {
     }
     // we don't want to check for dates in the URL's
     if (node.tagName.toLowerCase() === 'a') return
+    // get the datetime from time element
+    if (node.tagName.toLowerCase() === 'time') {
+      const datetime = node.getAttribute('datetime')
+      if (datetime) {
+        const date = new Date(datetime)
+        if (!isNaN(date)) {
+          this._articlePublishedDate = date
+          return true
+        }
+      }
+    }
+        
     // Searching for the real date in the text content
     const content = node.textContent.trim()
     let dateFound
diff --git a/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected-metadata.json b/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected-metadata.json
index 61fe09886..d2bc065e6 100644
--- a/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected-metadata.json
+++ b/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected-metadata.json
@@ -4,7 +4,7 @@
   "dir": null,
   "excerpt": "The Sept. 27, 2022 episode of “The Ezra Klein Show”",
   "siteName": "fakehost",
-  "siteIcon": "/vi-assets/static-assets/favicon-d2483f10ef688e6f89e23806b9700298.ico",
+  "siteIcon": "http://fakehost/vi-assets/static-assets/favicon-d2483f10ef688e6f89e23806b9700298.ico",
   "previewImage": "https://static01.nyt.com/newsgraphics/images/icons/defaultPromoCrop.png",
   "publishedDate": "2022-09-27T16:25:17.221Z",
   "language": "English",
diff --git a/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected.html b/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected.html
index 8d79daf8c..725bef094 100644
--- a/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected.html
+++ b/packages/readabilityjs/test/test-pages/nytimes-podcasts/expected.html
@@ -5,8 +5,6 @@
                 <article id="story">
                     <header>
                         <p> The Ezra Klein Show </p>
-                        <p><time datetime="2022-09-27T12:25:17-04:00">Sept. 27, 2022</time>
-                        </p>
                     </header>
                     <section name="articleBody">
                         <div>
diff --git a/packages/readabilityjs/test/test-pages/nytimes.com/expected.html b/packages/readabilityjs/test/test-pages/nytimes.com/expected.html
index ceb0aa925..9777d24a6 100644
--- a/packages/readabilityjs/test/test-pages/nytimes.com/expected.html
+++ b/packages/readabilityjs/test/test-pages/nytimes.com/expected.html
@@ -30,8 +30,6 @@
                                 </figcaption>
                             </figure>
                         </div>
-                        <p><time datetime="2022-10-28T05:00:25-04:00"><span>Oct. 28, 2022</span></time>
-                        </p>
                     </header>
                     <section name="articleBody">
                         <div>