From be6c95c49772e62de33a27e7ff7b95af1254b991 Mon Sep 17 00:00:00 2001
From: Aditya Gupta <mail@adityagupta.dev>
Date: Fri, 3 Jul 2026 08:04:14 +0530
Subject: [PATCH] feat: Add resilience hardening for csv parser

---
 app/utils/csv_parser.py                   | 16 +++++++--
 tests/services/test_csv_parser.py         | 12 +++++++
 tests/services/test_llm_classification.py | 44 +++++++++++++++++++++++
 3 files changed, 69 insertions(+), 3 deletions(-)
 create mode 100644 tests/services/test_llm_classification.py

diff --git a/app/utils/csv_parser.py b/app/utils/csv_parser.py
index 02099c6..4e2fc19 100644
--- a/app/utils/csv_parser.py
+++ b/app/utils/csv_parser.py
@@ -75,14 +75,24 @@ def parse_and_validate_csv(content: bytes) -> List[Dict[str, Any]]:
             else None
         )
 
-        if not all([txn_id, date_str, amount_str, merchant]):
+        if not all([date_str, amount_str, merchant]):
             raise InvalidCSVException(
-                f"Row {line_num}: Empty values are not allowed in required fields"
+                f"Row {line_num}: Empty values are not allowed in required fields (date, amount, merchant)"
             )
 
+        # Auto-generate txn_id if missing or empty
+        if not txn_id or not txn_id.strip():
+            import uuid
+            txn_id = f"GEN_{uuid.uuid4().hex[:8].upper()}"
+        else:
+            txn_id = txn_id.strip()
+
         # Validate amount
+        cleaned_amount_str = (
+            amount_str.strip().replace("$", "").replace("₹", "").replace(",", "")
+        )
         try:
-            amount = float(amount_str.strip())
+            amount = float(cleaned_amount_str)
         except ValueError as e:
             raise InvalidCSVException(
                 f"Row {line_num}: Invalid amount '{amount_str}'. Must be a number."
diff --git a/tests/services/test_csv_parser.py b/tests/services/test_csv_parser.py
index a714b24..e0769de 100644
--- a/tests/services/test_csv_parser.py
+++ b/tests/services/test_csv_parser.py
@@ -52,3 +52,15 @@ def test_parse_invalid_date():
     with pytest.raises(InvalidCSVException) as exc_info:
         parse_and_validate_csv(invalid_csv.encode("utf-8"))
     assert "Invalid date format" in str(exc_info.value)
+
+
+def test_parse_amount_with_symbols():
+    csv_data = (
+        "txn_id,date,amount,merchant\n"
+        "TX1001,2026-06-25,$120.50,Weekly grocery shop\n"
+        'TX1002,2026-06-26,"₹1,500.00",Weekly grocery shop\n'
+    )
+    result = parse_and_validate_csv(csv_data.encode("utf-8"))
+    assert len(result) == 2
+    assert result[0]["amount"] == 120.50
+    assert result[1]["amount"] == 1500.00
diff --git a/tests/services/test_llm_classification.py b/tests/services/test_llm_classification.py
new file mode 100644
index 0000000..e550e46
--- /dev/null
+++ b/tests/services/test_llm_classification.py
@@ -0,0 +1,44 @@
+import pytest
+from app.worker import classify_transactions_batch
+from google import genai
+from unittest.mock import MagicMock, patch
+
+pytestmark = pytest.mark.asyncio
+
+
+async def test_classify_transactions_batch_retry_success():
+    # Mock GenAI client and interaction creation
+    mock_client = MagicMock(spec=genai.Client)
+    mock_interaction = MagicMock()
+    mock_interaction.output_text = '{"classifications": [{"txn_id": "TX1", "category": "Food"}, {"txn_id": "TX2", "category": "Shopping"}]}'
+    mock_client.interactions.create.return_value = mock_interaction
+
+    batch_txns = [
+        {"txn_id": "TX1", "merchant": "Swiggy", "amount": 150.0},
+        {"txn_id": "TX2", "merchant": "Amazon", "amount": 1200.0},
+    ]
+
+    results = await classify_transactions_batch(mock_client, batch_txns)
+    assert len(results) == 2
+    assert results["TX1"]["category"] == "Food"
+    assert results["TX1"]["failed"] is False
+    assert results["TX2"]["category"] == "Shopping"
+    assert results["TX2"]["failed"] is False
+
+
+async def test_classify_transactions_batch_retry_failure():
+    mock_client = MagicMock(spec=genai.Client)
+    # Simulate API call failure (e.g. invalid API key or connection error)
+    mock_client.interactions.create.side_effect = Exception("API Key Error")
+
+    batch_txns = [
+        {"txn_id": "TX1", "merchant": "Swiggy", "amount": 150.0},
+    ]
+
+    # Patch sleep to make retry test run instantly
+    with patch("asyncio.sleep", return_value=None):
+        results = await classify_transactions_batch(mock_client, batch_txns)
+
+    assert len(results) == 1
+    assert results["TX1"]["category"] == "Other"
+    assert results["TX1"]["failed"] is True