From b11223277c4ab24378682f5f174c76f7aee62dab Mon Sep 17 00:00:00 2001
From: kempersc <sckemper@mailfence.com>
Date: Sun, 18 Jan 2026 15:27:04 +0100
Subject: [PATCH] fix(entity-review): persist source URLs for WCMS-only
 profiles

- Add source_urls to WCMS-only profile detail response
- Update _candidates_by_wcms cache when creating new WCMS-only entries
- Use correct refresh method (fetchWcmsOnlyProfileDetail) after adding source URL

Fixes issue where source URLs added to WCMS-only profiles were not
displayed after page refresh because:
1. The wcms-only-profile/{email} endpoint wasn't returning source_urls
2. The frontend was calling fetchProfileDetail instead of
   fetchWcmsOnlyProfileDetail after adding a source URL
3. New WCMS-only entries weren't added to the lookup cache
---
 frontend/src/pages/EntityReviewPage.tsx | 11 ++++++++---
 src/glam_extractor/api/entity_review.py | 20 ++++++++++++++++++--
 2 files changed, 26 insertions(+), 5 deletions(-)

diff --git a/frontend/src/pages/EntityReviewPage.tsx b/frontend/src/pages/EntityReviewPage.tsx
index 5759e3ff6a..335589c83c 100644
--- a/frontend/src/pages/EntityReviewPage.tsx
+++ b/frontend/src/pages/EntityReviewPage.tsx
@@ -752,15 +752,20 @@ export default function EntityReviewPage() {
       // Hide success message after 3 seconds
       setTimeout(() => setSourceUrlSuccess(null), 3000);
       
-      // Optionally refresh profile (the source URLs aren't displayed yet, but could be in the future)
-      await fetchProfileDetail(selectedProfile.ppid);
+      // Refresh profile to show the newly added source URL
+      // Use appropriate fetch method based on profile type
+      if (selectedProfile.is_wcms_only && selectedProfile.email) {
+        await fetchWcmsOnlyProfileDetail(selectedProfile.email);
+      } else {
+        await fetchProfileDetail(selectedProfile.ppid);
+      }
       
     } catch (err) {
       setSourceUrlError(err instanceof Error ? err.message : 'Failed to add source');
     } finally {
       setAddingSourceUrl(false);
     }
-  }, [selectedProfile, sourceUrl, sourceComment, language, fetchProfileDetail]);
+  }, [selectedProfile, sourceUrl, sourceComment, language, fetchProfileDetail, fetchWcmsOnlyProfileDetail]);
 
   // Save review decision
   const saveDecision = useCallback(async (decision: 'match' | 'not_match' | 'uncertain') => {
diff --git a/src/glam_extractor/api/entity_review.py b/src/glam_extractor/api/entity_review.py
index d3b96b6e36..74e69fb441 100644
--- a/src/glam_extractor/api/entity_review.py
+++ b/src/glam_extractor/api/entity_review.py
@@ -938,8 +938,19 @@ async def get_wcms_only_profile_detail(email: str):
         "last_access": wcms_data.get("last_access"),
     }
     
+    # Build PPID for WCMS-only profiles
+    wcms_only_ppid = f"wcms-only-{wcms_data.get('user_id', email_lower)}"
+    
+    # Load candidates to check for source_urls
+    # WCMS-only profiles that have had sources added will be in candidates file
+    load_candidates()
+    source_urls = []
+    if _candidates_by_wcms is not None and wcms_only_ppid in _candidates_by_wcms:
+        source_urls_raw = _candidates_by_wcms[wcms_only_ppid].get('source_urls', [])
+        source_urls = source_urls_raw  # Already list of dicts
+    
     return {
-        "ppid": f"wcms-only-{wcms_data.get('user_id', email_lower)}",
+        "ppid": wcms_only_ppid,
         "name": name,
         "email": wcms_data.get("email", email),
         "email_domain": email_domain,
@@ -949,6 +960,7 @@ async def get_wcms_only_profile_detail(email: str):
         "match_candidates": [],  # Empty - ready for manual LinkedIn addition
         "annotation_date": None,
         "is_wcms_only": True,  # Flag to indicate this is a WCMS-only profile
+        "source_urls": source_urls,  # Include any added source URLs
     }
 
 
@@ -2702,8 +2714,12 @@ async def add_source_url(request: AddSourceUrlRequest):
             'source_urls': [source_observation]
         }
         _candidates_cache['candidates'].append(new_entry)
+        
+        # Also add to the in-memory lookup cache so it can be found on subsequent requests
+        if _candidates_by_wcms is not None:
+            _candidates_by_wcms[wcms_ppid] = new_entry
     
-    # Also update the wcms lookup cache
+    # Also update the wcms lookup cache (for existing entries)
     if _candidates_by_wcms is not None and wcms_ppid in _candidates_by_wcms:
         if 'source_urls' not in _candidates_by_wcms[wcms_ppid]:
             _candidates_by_wcms[wcms_ppid]['source_urls'] = []