From e2db71192e1d5366db4c280e4471f3126ed8a726 Mon Sep 17 00:00:00 2001
From: Anjo Vahldiek-Oberwagner <anjovahldiek@gmail.com>
Date: Sat, 16 May 2026 00:51:04 +0200
Subject: [PATCH] fix: case-insensitive artifact URL lookup and normalize bare
 DOI links

- Profile page: use lowercase keys in artifactUrlMap so paper titles with
  different casing (e.g. EuroSys 2021/2022) still resolve to artifact URLs
- Search & profile pages: add normalizeUrl() to prepend https://doi.org/
  to bare DOI strings (e.g. '10.5281/zenodo.6337102') used as href values
- Applies to artifact_urls, paper_url, doi_url, and appendix_url fields
---
 src/assets/js/reprodb-profile-page.js | 13 +++++++++----
 src/assets/js/reprodb-search.js       | 17 +++++++++++------
 2 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/src/assets/js/reprodb-profile-page.js b/src/assets/js/reprodb-profile-page.js
index 2c05e5d7..2cace61a 100644
--- a/src/assets/js/reprodb-profile-page.js
+++ b/src/assets/js/reprodb-profile-page.js
@@ -19,6 +19,11 @@
     document.getElementById('inst-profile').classList.add('rdb-hidden');
   }
 
+  function normalizeUrl(u) {
+    if (u && !u.match(/^https?:\/\//i) && /^10\.\d{4,}\//.test(u)) return 'https://doi.org/' + u;
+    return u;
+  }
+
   function availTag(url) {
     if (!url) return '';
     var n = url.replace(/\/+$/, '');
@@ -79,7 +84,7 @@
         columns: [
           { title: '#', formatter: 'rownum', width: 50, headerSort: false },
           { title: 'Title', field: 'title', formatter: function(cell) {
-            var d = cell.getData(), t = d.title.replace(/\.+$/, ''), u = artifactUrlMap[t] || '';
+            var d = cell.getData(), t = d.title.replace(/\.+$/, '').toLowerCase(), u = artifactUrlMap[t] || '';
             return (u ? '<a href="' + escHtml(u) + '" target="_blank" rel="noopener">' + escHtml(d.title) + '</a>' : escHtml(d.title)) + availTag(u);
           }, headerSort: false },
           { title: 'Conference', field: 'conference' },
@@ -246,7 +251,7 @@
       (p.papers || []).forEach(function(paper) {
         if (!paperMap[paper.title]) {
           var t = paper.title.replace(/\.+$/, '');
-          paperMap[paper.title] = { title: paper.title, authors: [], conference: paper.conference, year: paper.year, badges: paper.badges, url: artifactUrlMap[t] || '' };
+          paperMap[paper.title] = { title: paper.title, authors: [], conference: paper.conference, year: paper.year, badges: paper.badges, url: artifactUrlMap[t.toLowerCase()] || '' };
         }
         paperMap[paper.title].authors.push(p.name);
       });
@@ -441,11 +446,11 @@
     citedArtifactsMap = res[2] || {};
     authorRankHistory = res[3] || [];
 
-    // Build artifact URL map
+    // Build artifact URL map (lowercase keys for case-insensitive lookup)
     (res[4] || []).forEach(function(a) {
       var urls = a.artifact_urls || [];
       var u = urls.length ? urls[0] : (a.artifact_url || a.repository_url || '');
-      if (a.title && u) artifactUrlMap[a.title.replace(/\.+$/, '')] = u;
+      if (a.title && u) artifactUrlMap[a.title.replace(/\.+$/, '').toLowerCase()] = normalizeUrl(u);
     });
 
     // Build paper index
diff --git a/src/assets/js/reprodb-search.js b/src/assets/js/reprodb-search.js
index e82bc231..3a5057cf 100644
--- a/src/assets/js/reprodb-search.js
+++ b/src/assets/js/reprodb-search.js
@@ -20,6 +20,11 @@
 
   var escHtml = ReproDB.escHtml;
 
+  function normalizeUrl(u) {
+    if (u && !u.match(/^https?:\/\//i) && /^10\.\d{4,}\//.test(u)) return 'https://doi.org/' + u;
+    return u;
+  }
+
   function normalizeText(s) {
     return (s || '').toLowerCase().replace(/[^a-z0-9 ]/g, ' ');
   }
@@ -199,8 +204,8 @@
       entry.className = 'rdb-result-entry';
 
       // Line 1: Bold title (linked to artifact)
-      var artUrls = d.artifact_urls || [];
-      var titleLink = artUrls.length > 0 ? artUrls[0] : (d.repository_url || d.artifact_url || '');
+      var artUrls = (d.artifact_urls || []).map(normalizeUrl);
+      var titleLink = artUrls.length > 0 ? artUrls[0] : normalizeUrl(d.repository_url || d.artifact_url || '');
       var titleHtml = titleLink
         ? '<a href="' + escHtml(titleLink) + '" target="_blank" rel="noopener">' + escHtml(d.title) + '</a>'
         : escHtml(d.title);
@@ -224,12 +229,12 @@
       var links = [];
       // Paper link: prefer doi_url, fall back to paper_url
       if (d.doi_url) {
-        links.push('<a href="' + escHtml(d.doi_url) + '" target="_blank" rel="noopener">📄 Paper</a>');
+        links.push('<a href="' + escHtml(normalizeUrl(d.doi_url)) + '" target="_blank" rel="noopener">📄 Paper</a>');
       } else if (d.paper_url) {
-        links.push('<a href="' + escHtml(d.paper_url) + '" target="_blank" rel="noopener">📄 Paper</a>');
+        links.push('<a href="' + escHtml(normalizeUrl(d.paper_url)) + '" target="_blank" rel="noopener">📄 Paper</a>');
       }
       // Artifact URLs (unified list)
-      var artUrlList = d.artifact_urls || [];
+      var artUrlList = artUrls;
       if (artUrlList.length === 1) {
         var isGH = artUrlList[0].indexOf('github.com') !== -1;
         var lbl = isGH ? '💻 GitHub' : '📦 Artifact';
@@ -246,7 +251,7 @@
           }
         });
       }
-      if (d.appendix_url) links.push('<a href="' + escHtml(d.appendix_url) + '" target="_blank" rel="noopener">📋 Appendix</a>');
+      if (d.appendix_url) links.push('<a href="' + escHtml(normalizeUrl(d.appendix_url)) + '" target="_blank" rel="noopener">📋 Appendix</a>');
       var linksLine = links.length > 0 ? links.join(' &middot; ') : '';
 
       entry.innerHTML =