diff --git a/web/core/views.py b/web/core/views.py
index b5cf283..3f1ff7a 100644
--- a/web/core/views.py
+++ b/web/core/views.py
@@ -9,7 +9,7 @@ import csv
 
 from .models import Entry
 from .forms import ImportForm, EntryForm
-from .utils import import_csv_bytes, terms, has_wildcards, wildcard_to_regex
+from .utils import terms, has_wildcards, wildcard_to_regex, import_csv_bytes
 
 # Order + labels used in the Search UI
 FIELD_ORDER = [
@@ -73,7 +73,7 @@ def search_page(request):
     Search-first landing. Defaults to Subject, Illustration, Application.
     Supports:
       - quoted phrases
-      - * and ? wildcards (translated to regex when present)
+      - * and ? wildcards (regex); if regex returns zero, falls back to icontains
       - AND across tokens, OR across the selected fields
     """
     default_fields = {
@@ -104,21 +104,40 @@ def search_page(request):
         tokens = terms(q)  # supports quoted phrases
         fields = [f for f, sel in selected.items() if sel] or ["subject"]
 
+        # ---------- Pass A: regex for wildcard tokens ----------
         qs = Entry.objects.all()
-
-        # Build filters: AND across tokens, each token ORs across fields
+        used_regex = False
         for tok in tokens:
-            token_clause = Q()
+            clause = Q()
             if has_wildcards(tok):
-                pattern = wildcard_to_regex(tok)
+                used_regex = True
+                pattern = wildcard_to_regex(tok)  # e.g. "*love*" -> ".*love.*"
                 for f in fields:
-                    token_clause |= Q(**{f + "__iregex": pattern})
+                    clause |= Q(**{f + "__iregex": pattern})
             else:
                 for f in fields:
-                    token_clause |= Q(**{f + "__icontains": tok})
-            qs = qs.filter(token_clause)
+                    clause |= Q(**{f + "__icontains": tok})
+            qs = qs.filter(clause)
 
         ids = list(qs.order_by("-date_added", "-id").values_list("id", flat=True))
+
+        # ---------- Pass B (fallback): plain icontains if regex yielded 0 ----------
+        if used_regex and not ids:
+            qs = Entry.objects.all()
+            for tok in tokens:
+                clause = Q()
+                tok_stripped = tok.replace("*", "").replace("?", "")  # fallback normalize
+                for f in fields:
+                    clause |= Q(**{f + "__icontains": tok_stripped})
+                qs = qs.filter(clause)
+            ids = list(qs.order_by("-date_added", "-id").values_list("id", flat=True))
+
+        # Debug to container logs
+        try:
+            print(f"[search] q={q!r} tokens={tokens} fields={fields} count={len(ids)}")
+        except Exception:
+            pass
+
         request.session["result_ids"] = ids
         count = len(ids)