feat: add graph_description file, send failure notification email, remove redundant error handling

Author: Nigma-Ks

.gitignore

@@ -225,3 +225,6 @@ backend/resources/dataset.json
 
 # Pre-commit
 .pre-commit-config.local
+
+# Memory benchmark scripts, inputs and results
+mem_benchmark/

backend/utils/results_sender.py

@@ -246,6 +246,47 @@ def send_experiment_results_by_email(
             os.remove(zip_path)
 
 
+def send_task_failed_email(
+    to_email: str,
+    experiment_name: str,
+    model_file_name: str,
+):
+    load_dotenv()
+
+    from_email = require_env("EMAIL")
+    from_password = require_env("APP_PASSWORD")
+
+    model_name = (
+        model_file_name[:-5] if model_file_name.endswith(".onnx") else model_file_name
+    )
+
+    msg = EmailMessage()
+    msg["From"] = from_email
+    msg["To"] = to_email
+    msg["Subject"] = f"Experiment '{experiment_name}' failed — please retry"
+
+    body = (
+        f"\n\n"
+        f"An error occurred while running the experiment '{experiment_name}' "
+        f"with model '{model_name}'.\n\n"
+        f"Please retry your experiment. If the problem persists, contact support.\n"
+    )
+
+    msg.set_content(body)
+
+    context = ssl.create_default_context()
+
+    try:
+        with smtplib.SMTP("smtp.gmail.com", 587, timeout=SMTP_TIMEOUT) as server:
+            server.ehlo()
+            server.starttls(context=context)
+            server.ehlo()
+            server.login(from_email, from_password)
+            server.send_message(msg)
+    except (smtplib.SMTPException, TimeoutError, OSError):
+        logger.exception("Failed to send failure notification email")
+
+
 def send_publish_results_by_email(
     to_email: str,
     metrics: RunstratMetrics,

backend/utils/task.py

@@ -30,7 +30,10 @@
 )
 from backend.storage.minio_client import download_file, upload_file
 from backend.utils.docker_runner import Compstrat, run_pipeline
-from backend.utils.results_sender import send_experiment_results_by_email
+from backend.utils.results_sender import (
+    send_experiment_results_by_email,
+    send_task_failed_email,
+)
 from backend.utils.token_store import mark_task_completed
 
 celery_app = Celery("tasks", broker=REDIS_URL, backend=REDIS_URL)
@@ -86,98 +89,63 @@ def process_and_cleanup_task(
     metrics_by_lang: dict[str, RunstratMetrics] = {}
 
     try:
-        model_object_key = None
-        try:
-            model_object_key = upload_file(model_path, f"models/{task_uid}/model.onnx")
-        except Exception:
-            logger.exception("MinIO model upload failed for task %s", task_uid)
+        model_object_key = upload_file(model_path, f"models/{task_uid}/model.onnx")
 
         for lang in langs_to_run:
-            try:
-                csv_src = LANGUAGE_CSVS[lang]
-                dest = get_thread_filepath(task_uid, LAUNCH_INFO_FILE)
-                os.makedirs(os.path.dirname(dest), exist_ok=True)
-                shutil.copy2(csv_src, dest)
-
-                methods_launched = _count_methods_in_csv(csv_src)
-                suffix = f"_{lang}"
-                run_pipeline(task_uid, out_suffix=suffix)
-
-                ai_csv = _ai_csv_path(task_uid, suffix)
-                metrics = compute_metrics(ai_csv)
-                metrics_by_lang[lang] = metrics
-
-                results_object_key = None
-                try:
-                    results_object_key = upload_file(
-                        ai_csv, f"results/{task_uid}/{lang}/AI.csv"
-                    )
-                except Exception:
-                    logger.exception(
-                        "MinIO results upload failed for task %s lang %s",
-                        task_uid,
-                        lang,
-                    )
-
-                try:
-                    save_experiment(
-                        experiment_name=experiment,
-                        model_name=model_name,
-                        email=email,
-                        metrics=metrics,
-                        methods_launched=methods_launched,
-                        language=lang,
-                        model_object_key=model_object_key,
-                        results_object_key=results_object_key,
-                    )
-                except Exception:
-                    logger.exception(
-                        "DB save failed for task %s lang %s", task_uid, lang
-                    )
-            except Exception:
-                logger.exception("Pipeline failed for task %s lang %s", task_uid, lang)
-
-        if language == "all" and len(metrics_by_lang) >= 1:
-            all_ai_csvs = [
-                _ai_csv_path(task_uid, f"_{lang}")
-                for lang in langs_to_run
-                if lang in metrics_by_lang
-            ]
-            try:
-                combined = combine_metrics(all_ai_csvs)
-                total_launched = sum(
-                    _count_methods_in_csv(LANGUAGE_CSVS[lang])
-                    for lang in langs_to_run
-                    if lang in metrics_by_lang
-                )
-
-                merged_csv_path = os.path.join(
-                    get_thread_filepath(task_uid, RESULTS_DIR), "ai_all.csv"
-                )
-                merge_csvs(all_ai_csvs, merged_csv_path)
-
-                all_results_key = None
-                try:
-                    all_results_key = upload_file(
-                        merged_csv_path, f"results/{task_uid}/all/AI.csv"
-                    )
-                except Exception:
-                    logger.exception(
-                        "MinIO all-results upload failed for task %s", task_uid
-                    )
-
-                save_experiment(
-                    experiment_name=experiment,
-                    model_name=model_name,
-                    email=email,
-                    metrics=combined,
-                    methods_launched=total_launched,
-                    language="all",
-                    model_object_key=model_object_key,
-                    results_object_key=all_results_key,
-                )
-            except Exception:
-                logger.exception("Aggregated DB save failed for task %s", task_uid)
+            csv_src = LANGUAGE_CSVS[lang]
+            dest = get_thread_filepath(task_uid, LAUNCH_INFO_FILE)
+            os.makedirs(os.path.dirname(dest), exist_ok=True)
+            shutil.copy2(csv_src, dest)
+
+            methods_launched = _count_methods_in_csv(csv_src)
+            suffix = f"_{lang}"
+            run_pipeline(task_uid, out_suffix=suffix)
+
+            ai_csv = _ai_csv_path(task_uid, suffix)
+            metrics = compute_metrics(ai_csv)
+            metrics_by_lang[lang] = metrics
+
+            results_object_key = upload_file(
+                ai_csv, f"results/{task_uid}/{lang}/AI.csv"
+            )
+
+            save_experiment(
+                experiment_name=experiment,
+                model_name=model_name,
+                email=email,
+                metrics=metrics,
+                methods_launched=methods_launched,
+                language=lang,
+                model_object_key=model_object_key,
+                results_object_key=results_object_key,
+            )
+
+        if language == "all":
+            all_ai_csvs = [_ai_csv_path(task_uid, f"_{lang}") for lang in langs_to_run]
+            combined = combine_metrics(all_ai_csvs)
+            total_launched = sum(
+                _count_methods_in_csv(LANGUAGE_CSVS[lang]) for lang in langs_to_run
+            )
+
+            merged_csv_path = os.path.join(
+                get_thread_filepath(task_uid, RESULTS_DIR), "ai_all.csv"
+            )
+            merge_csvs(all_ai_csvs, merged_csv_path)
+
+            all_results_key = upload_file(
+                merged_csv_path, f"results/{task_uid}/all/AI.csv"
+            )
+
+            save_experiment(
+                experiment_name=experiment,
+                model_name=model_name,
+                email=email,
+                metrics=combined,
+                methods_launched=total_launched,
+                language="all",
+                model_object_key=model_object_key,
+                results_object_key=all_results_key,
+            )
 
         send_experiment_results_by_email(
             email,
@@ -188,6 +156,12 @@ def process_and_cleanup_task(
         )
     except Exception:
         logger.exception("Task %s failed", task_uid)
+        try:
+            send_task_failed_email(email, experiment, filename)
+        except Exception:
+            logger.exception(
+                "Failed to send failure notification for task %s", task_uid
+            )
 
     finally:
         try:

frontend/public/graph_description.pdf

@@ -224,7 +224,7 @@ const ModelInterfacePage: React.FC = () => {
         </section>
 
         {/* Spec link */}
-        <section>
+        <section className="mb-10">
           <Title level={4}>Specification</Title>
           <Paragraph style={{ fontSize: 15, lineHeight: 1.9, color: '#374151' }}>
             The authoritative interface definition (inputs, outputs, and graph
@@ -236,6 +236,21 @@ const ModelInterfacePage: React.FC = () => {
             </Button>
           </a>
         </section>
+
+        {/* Graph description */}
+        <section>
+          <Title level={4}>Graph Structure</Title>
+          <Paragraph style={{ fontSize: 15, lineHeight: 1.9, color: '#374151' }}>
+            A detailed description of the graph representation of the symbolic machine
+            configuration — vertex types, edge semantics, and attributes — is available
+            in the reference document:
+          </Paragraph>
+          <a href="/graph_description.pdf" target="_blank" rel="noreferrer">
+            <Button icon={<LinkOutlined />} type="default">
+              Graph Representation of Symbolic Execution (PDF)
+            </Button>
+          </a>
+        </section>
       </div>
     </div>
   );