🐛 Fix Libération published date time used (#264)

lovasoa · Dec 1, 2024 · baf7f3b · baf7f3b
1 parent a142600
commit baf7f3b
Show file tree

Hide file tree

Showing 2 changed files with 40 additions and 5 deletions.
diff --git a/ophirofox/content_scripts/config.js b/ophirofox/content_scripts/config.js
@@ -69,16 +69,26 @@ const ophirofox_config = getOphirofoxConfig();
 
 /**
  * Crée un lien vers Europresse avec les keywords donnés
- * @param {string} keywords
+ * @param {string} keywords - input for Europress search engine
+ * @param {string} publishedTime - article publication date (i.e. 2024-08-27T18:18:55.663Z or 2024-08-27)
  * @returns {Promise<HTMLAnchorElement>}
  */
 async function ophirofoxEuropresseLink(keywords, { publishedTime } = {}) {
   // Keywords is the article name
   keywords = keywords ? keywords.trim() : document.querySelector("h1").textContent;
 
-  // Trying to determine published time with meta tags (Open Graph values)
+  // Trying generically to determine published time with meta tags (Open Graph values)
+  // Heuristics would suggest that it's normally UTC in the media world. 
   publishedTime = publishedTime || document.querySelector( "meta[property='article:published_time'], meta[property='og:article:published_time'], meta[property='date:published_time']")
   ?.getAttribute("content") || '';
+  let publishedTimeInstance = new Date(publishedTime);
+
+  if (!isNaN(publishedTimeInstance)) {
+    // JavaScript, what a pleasure.
+    publishedTime = publishedTimeInstance.toISOString().slice(0, 10);
+  } else {
+    publishedTime = ''
+  }
 
   // Creating HTML anchor element
   const a = document.createElement("a");

diff --git a/ophirofox/content_scripts/liberation.js b/ophirofox/content_scripts/liberation.js
@@ -4,8 +4,13 @@ function extractKeywords() {
     .getAttribute("content");
 }
 
-async function createLink() {
-  const a = await ophirofoxEuropresseLink(extractKeywords());
+/**
+ * Crée un lien vers Europresse avec les keywords donnés
+ * @param {string} publishedTime - article publication date (2024-01-01)
+ * @returns {Promise<HTMLAnchorElement>}
+ */
+async function createLink(publishedTime) {
+  const a = await ophirofoxEuropresseLink(extractKeywords(), { publishedTime: publishedTime });
   return a;
 }
 
@@ -40,7 +45,27 @@ async function onLoad(premiumBanner) {
 
             // Not sure if premiumBanner is (and will be) still valid after DOM rewrite
             if (!document.querySelector('div.TypologyArticle__BlockPremium-sc-1vro4tp-2 + a.ophirofox-europresse')) {
-              findPremiumBanner().after(await createLink());
+              // See #239, Libération replaces date:published_time with the date of edit, which means that a search limited by the time of publication may be too restrictive
+              // We need to specify the date to use for the generic ophirofoxEuropresseLink function 
+              // Might need refactor if other medias have the same problem, more properties for fail-safe
+              let publishedDate = document.querySelector( "meta[property='article:published_time'], meta[property='og:article:published_time'], meta[property='date:published_time']")
+  ?.getAttribute("content") || '';
+              let firstPublishedDate = /\"first_publish_date\":\"(\d{4}-\d{2}-\d{2}[A-Z]+\d{2}:\d{2}:\d{2}.[0-9+-:]+Z)/.exec(document.getElementById('fusion-metadata').textContent)[1]  // 2024-08-27T18:18:55.663Z => UTC
+              let firstPublishedDateInstance = new Date(firstPublishedDate);
+
+              // date:published_time is used by default and when firstPublishedDate is not older 
+              if (publishedDate && !publishedDate.trim()) {
+                // If the first published date is valid and older
+                if (!isNaN(firstPublishedDateInstance) && (firstPublishedDateInstance < new Date(publishedDate)))
+                  publishedDate = firstPublishedDate;   
+              } else {
+                  // If we are here, Libération did big shit or just changed their Open Grahs properties
+                  if (!isNaN(firstPublishedDateInstance)) {
+                    publishedDate = firstPublishedDate;
+                  }
+              }
+
+              findPremiumBanner().after(await createLink(publishedDate));
               console.log('Ophirofox injected after React DOM rewrite');
               break;
             }