Enhance LLM question import process with preview and confirmation steps, including error handling and M2M record creation

2025-10-20 11:01:47 +01:00
parent 06b7239daa
commit 01f67f3e32
6 changed files with 211 additions and 4 deletions
@@ -2,10 +2,14 @@

 {% block content %}
  <h1>Import LLM Questions</h1>
-  <form method="post" enctype="multipart/form-data">
+  <form id="llm-import-form" method="post" enctype="multipart/form-data" hx-post="{% url 'sbas:import_llm_questions' %}" hx-target="#import-preview" hx-swap="innerHTML">
    {% csrf_token %}
    {{ form.as_p }}
-    <button class="btn btn-primary" type="submit">Upload and Import</button>
+    <div class="d-flex gap-2">
+      <button class="btn btn-outline-primary" type="submit">Preview (dry run)</button>
+      <button class="btn btn-secondary" type="button" onclick="document.getElementById('llm-import-form').reset()">Reset</button>
+    </div>
  </form>
-  <p>Upload a single JSON array, a single object, or JSONL (one JSON object per line) matching the agreed schema.</p>
+  <p>Upload a single JSON array, a single object, or JSONL (one JSON object per line) matching the agreed schema. Use the Preview button to inspect results before importing.</p>
+  <div id="import-preview" class="mt-3"></div>
 {% endblock %}
@@ -0,0 +1,8 @@
+<div class="alert alert-danger">
+  <h5>Import error</h5>
+  <ul>
+    {% for e in errors %}
+      <li>{{ e }}</li>
+    {% endfor %}
+  </ul>
+</div>
@@ -0,0 +1,42 @@
+{% for item in items %}
+  <div class="card mb-2">
+    <div class="card-body">
+      <h5 class="card-title">Question {{ item.index }}</h5>
+      {% with payload=payloads|slice:item.index|first %}
+        {% if payload %}
+          <p><strong>Stem:</strong> {{ payload.stem|default:''|safe }}</p>
+          <p><strong>Answers:</strong>
+            <ul>
+              <li>A: {{ payload.a_answer|default:''|safe }}</li>
+              <li>B: {{ payload.b_answer|default:''|safe }}</li>
+              <li>C: {{ payload.c_answer|default:''|safe }}</li>
+              <li>D: {{ payload.d_answer|default:''|safe }}</li>
+              <li>E: {{ payload.e_answer|default:''|safe }}</li>
+            </ul>
+          </p>
+        {% endif %}
+      {% endwith %}
+      <p><strong>Missing M2M:</strong>
+        {% if item.missing_m2m %}
+          <ul>
+            {% for m in item.missing_m2m %}
+              <li>{{ m.model }}: {{ m.values|join:", " }}</li>
+            {% endfor %}
+          </ul>
+        {% else %}
+          None
+        {% endif %}
+      </p>
+      <div>
+        <label><input type="checkbox" name="selected" value="{{ item.index }}" checked> Include</label>
+      </div>
+    </div>
+  </div>
+{% endfor %}
+
+<div class="d-flex gap-2">
+  <button class="btn btn-success" hx-post="{% url 'sbas:import_llm_confirm' %}" hx-target="#import-result" hx-include="closest form">Import selected</button>
+  <button class="btn btn-secondary" hx-get="{% url 'sbas:import_llm_questions' %}" hx-target="#import-result">Cancel</button>
+</div>
+
+<div id="import-result"></div>
@@ -0,0 +1,21 @@
+<div class="alert alert-info">Imported {{ created }} questions.</div>
+{% if errors %}
+  <div class="alert alert-danger">
+    <h5>Errors</h5>
+    <ul>
+      {% for e in errors %}
+        <li>Item {{ e.index }}: {{ e.errors|join:', ' }}</li>
+      {% endfor %}
+    </ul>
+  </div>
+{% endif %}
+{% if actually_created %}
+  <div class="alert alert-secondary">
+    <h5>Created M2M records</h5>
+    <ul>
+      {% for k,v in actually_created.items %}
+        <li>{{ k }}: {{ v|join:', ' }}</li>
+      {% endfor %}
+    </ul>
+  </div>
+{% endif %}
@@ -94,6 +94,11 @@ urlpatterns = [
        views.import_llm_questions,
        name="import_llm_questions",
    ),
+    path(
+        "import_llm_confirm/",
+        views.import_llm_confirm,
+        name="import_llm_confirm",
+    ),
    #path(
    #    "exam/<int:pk>/scores/<int:cid>/<str:passcode>/",
    #    views.exam_scores_cid_user,
@@ -35,6 +35,7 @@ import re
 import logging

 logger = logging.getLogger(__name__)
+import uuid

 from loguru import logger

@@ -761,4 +762,130 @@ def import_llm_questions(request):
    for k, v in aggregated_actually_created.items():
        report["actually_created_m2m"][k] = v

-    return JsonResponse({"ok": True, "report": report})
+    # If this was a dry-run and the request came from HTMX, render a preview partial and store the parsed candidates in session
+    if dry_run and request.headers.get("HX-Request"):
+        session_key = f"llm_import_{uuid.uuid4().hex}"
+        # store minimal necessary data in session (the raw payloads)
+        request.session[session_key] = candidates
+        request.session.modified = True
+        preview_ctx = {
+            "report": report,
+            "items": report["per_item"],
+            "payloads": candidates,
+            "session_key": session_key,
+            "allow_create_m2m": allow_create_m2m,
+        }
+        return render(request, "sbas/partials/import_preview.html", preview_ctx)
+
+    # Non-HTMX callers get JSON
+    return JsonResponse({"ok": True, "report": report})
+
+
+@login_required
+@user_passes_test(lambda u: u.is_superuser)
+@require_http_methods(["POST"])
+def import_llm_confirm(request):
+    """Confirm import of previously-previewed candidates stored in session.
+
+    Expects POST with 'session_key' and 'selected' (comma-separated indices) and optional allow_create_m2m.
+    Returns an HTML fragment suitable for HTMX replacement summarising the import results.
+    """
+    session_key = request.POST.get("session_key")
+    # support both a comma-separated 'selected' string or multiple 'selected' values
+    selected_raw = request.POST.get("selected")
+    selected_list = request.POST.getlist("selected")
+    allow_create_m2m = request.POST.get("allow_create_m2m") in ("on", "true", "True", "1")
+    if selected_list:
+        # selected_list contains strings of indices
+        selected = ",".join(selected_list)
+    else:
+        selected = selected_raw
+    if not session_key or session_key not in request.session:
+        return render(request, "sbas/partials/import_error.html", {"errors": ["Session expired or invalid; please re-run preview."]}, status=400)
+
+    candidates = request.session.get(session_key, [])
+    indices = []
+    if selected:
+        for part in selected.split(','):
+            try:
+                indices.append(int(part))
+            except Exception:
+                continue
+
+    # default: import all if none specified
+    if not indices:
+        indices = list(range(len(candidates)))
+
+    # perform actual import for selected indices
+    from atlas.models import Finding, Structure, Condition, Presentation, Subspecialty
+
+    created = 0
+    errors = []
+    actually_created = defaultdict(list)
+    for idx in indices:
+        try:
+            payload = candidates[idx]
+        except Exception:
+            errors.append({"index": idx, "errors": ["Missing candidate"]})
+            continue
+
+        try:
+            with transaction.atomic():
+                q = Question()
+                q.stem = payload.get("stem", "").strip()
+                title = payload.get("title")
+                if title:
+                    q.stem = f"<strong>{title}</strong>\n" + q.stem
+                q.a_answer = payload.get("a_answer", "").strip()
+                q.a_feedback = payload.get("a_feedback", "")
+                q.b_answer = payload.get("b_answer", "").strip()
+                q.b_feedback = payload.get("b_feedback", "")
+                q.c_answer = payload.get("c_answer", "").strip()
+                q.c_feedback = payload.get("c_feedback", "")
+                q.d_answer = payload.get("d_answer", "").strip()
+                q.d_feedback = payload.get("d_feedback", "")
+                q.e_answer = payload.get("e_answer", "").strip()
+                q.e_feedback = payload.get("e_feedback", "")
+                q.feedback = payload.get("feedback", "")
+                q.best_answer = payload.get("best_answer")
+                cat_val = payload.get("category")
+                if isinstance(cat_val, str):
+                    cat_obj, _ = Category.objects.get_or_create(category=cat_val)
+                    q.category = cat_obj
+                q.save()
+
+                for key, model_cls in (("finding", Finding), ("structure", Structure), ("condition", Condition), ("presentation", Presentation), ("subspecialty", Subspecialty)):
+                    vals = payload.get(key) or []
+                    final_objs = []
+                    for v in vals:
+                        if isinstance(v, int):
+                            try:
+                                final_objs.append(model_cls.objects.get(pk=v))
+                            except model_cls.DoesNotExist:
+                                continue
+                        elif isinstance(v, str):
+                            qs = model_cls.objects.filter(name__iexact=v)
+                            if not qs.exists():
+                                qs = model_cls.objects.filter(name__icontains=v)
+                            if qs.exists():
+                                final_objs.extend(list(qs[:5]))
+                            else:
+                                if allow_create_m2m:
+                                    obj, created_flag = model_cls.objects.get_or_create(name=v)
+                                    final_objs.append(obj)
+                                    actually_created[model_cls.__name__].append(obj.pk)
+                    if final_objs:
+                        getattr(q, key).add(*[o.pk for o in final_objs])
+
+                created += 1
+        except Exception as e:
+            errors.append({"index": idx, "errors": [str(e)]})
+
+    # cleanup session
+    try:
+        del request.session[session_key]
+        request.session.modified = True
+    except Exception:
+        pass
+
+    return render(request, "sbas/partials/import_result.html", {"created": created, "errors": errors, "actually_created": dict(actually_created)})