feat: Google Vision primary OCR with Auth0 WIF and monthly usage cap (#127) #128

Merged
egullickson merged 8 commits from issue-127-google-vision-primary-ocr into main 2026-02-11 01:46:21 +00:00
Showing only changes of commit e6dd7492a1 - Show all commits

View File

@@ -355,7 +355,7 @@ class TestCloudEngine:
from app.engines.cloud_engine import CloudEngine
engine = CloudEngine(key_path="/nonexistent/key.json")
with pytest.raises(EngineUnavailableError, match="key not found"):
with pytest.raises(EngineUnavailableError, match="credential config not found"):
engine._get_client()
@patch("os.path.isfile", return_value=True)
@@ -414,6 +414,16 @@ class TestCloudEngine:
class TestHybridEngine:
"""Tests for HybridEngine with monthly Vision API cap."""
def _mock_redis(self, current_count: int = 0) -> MagicMock:
"""Create a mock Redis instance with a configurable counter value."""
mock_r = MagicMock()
mock_r.get.return_value = str(current_count) if current_count else None
mock_pipe = MagicMock()
mock_r.pipeline.return_value = mock_pipe
return mock_r
def test_name_with_fallback(self) -> None:
from app.engines.hybrid_engine import HybridEngine
@@ -432,13 +442,15 @@ class TestHybridEngine:
engine = HybridEngine(primary=primary)
assert engine.name == "hybrid(paddleocr+none)"
# --- Local-primary path (original confidence-based fallback) ---
def test_high_confidence_skips_fallback(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "cloud"
fallback.name = "tesseract"
primary.recognize.return_value = _make_result("VIN123", 0.95, "paddleocr")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
@@ -447,22 +459,6 @@ class TestHybridEngine:
assert result.engine_name == "paddleocr"
fallback.recognize.assert_not_called()
def test_low_confidence_triggers_fallback(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "google_vision"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.return_value = _make_result("VIN456", 0.92, "google_vision")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN456"
assert result.engine_name == "google_vision"
fallback.recognize.assert_called_once()
def test_low_confidence_no_fallback_returns_primary(self) -> None:
from app.engines.hybrid_engine import HybridEngine
@@ -474,6 +470,57 @@ class TestHybridEngine:
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
def test_exact_threshold_skips_fallback(self) -> None:
"""When confidence == threshold, no fallback needed (>= check)."""
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "tesseract"
primary.recognize.return_value = _make_result("VIN", 0.6, "paddleocr")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
result = engine.recognize(b"img", OcrConfig())
assert result.engine_name == "paddleocr"
fallback.recognize.assert_not_called()
# --- Local-primary with cloud fallback (subject to monthly cap) ---
def test_low_confidence_triggers_cloud_fallback(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "google_vision"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.return_value = _make_result("VIN456", 0.92, "google_vision")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
engine._redis = self._mock_redis(current_count=0)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN456"
assert result.engine_name == "google_vision"
def test_cloud_fallback_skipped_when_limit_reached(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "google_vision"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=100
)
engine._redis = self._mock_redis(current_count=100)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
assert result.engine_name == "paddleocr"
fallback.recognize.assert_not_called()
def test_fallback_lower_confidence_returns_primary(self) -> None:
from app.engines.hybrid_engine import HybridEngine
@@ -485,10 +532,11 @@ class TestHybridEngine:
fallback.recognize.return_value = _make_result("VIN456", 0.3, "google_vision")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
engine._redis = self._mock_redis(current_count=0)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
def test_fallback_engine_error_returns_primary(self) -> None:
def test_cloud_fallback_error_returns_primary(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
@@ -499,25 +547,14 @@ class TestHybridEngine:
fallback.recognize.side_effect = EngineUnavailableError("key missing")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
def test_fallback_unexpected_error_returns_primary(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "google_vision"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.side_effect = RuntimeError("network error")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
engine._redis = self._mock_redis(current_count=0)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
@patch("app.engines.hybrid_engine.time")
def test_fallback_timeout_returns_primary(self, mock_time: MagicMock) -> None:
def test_cloud_fallback_timeout_returns_primary(
self, mock_time: MagicMock
) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
@@ -525,28 +562,211 @@ class TestHybridEngine:
primary.name = "paddleocr"
fallback.name = "google_vision"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.return_value = _make_result("VIN456", 0.92, "google_vision")
# Simulate 6-second delay (exceeds 5s limit)
fallback.recognize.return_value = _make_result(
"VIN456", 0.92, "google_vision"
)
mock_time.monotonic.side_effect = [0.0, 6.0]
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
engine._redis = self._mock_redis(current_count=0)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123" # timeout -> use primary
assert result.text == "VIN123"
def test_exact_threshold_skips_fallback(self) -> None:
"""When confidence == threshold, no fallback needed (>= check)."""
# --- Cloud-primary path ---
def test_cloud_primary_returns_result_when_under_limit(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
fallback.name = "paddleocr"
primary.recognize.return_value = _make_result(
"VIN789", 0.95, "google_vision"
)
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=1000
)
engine._redis = self._mock_redis(current_count=500)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN789"
assert result.engine_name == "google_vision"
fallback.recognize.assert_not_called()
def test_cloud_primary_falls_back_when_limit_reached(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
fallback.name = "paddleocr"
fallback.recognize.return_value = _make_result(
"VIN_LOCAL", 0.75, "paddleocr"
)
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=1000
)
engine._redis = self._mock_redis(current_count=1000)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN_LOCAL"
assert result.engine_name == "paddleocr"
primary.recognize.assert_not_called()
def test_cloud_primary_no_fallback_raises_when_limit_reached(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
engine = HybridEngine(
primary=primary, fallback=None, threshold=0.6, monthly_limit=1000
)
engine._redis = self._mock_redis(current_count=1000)
with pytest.raises(EngineProcessingError, match="no fallback"):
engine.recognize(b"img", OcrConfig())
def test_cloud_primary_error_falls_back(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
fallback.name = "paddleocr"
primary.recognize.side_effect = EngineUnavailableError("API down")
fallback.recognize.return_value = _make_result(
"VIN_LOCAL", 0.75, "paddleocr"
)
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=1000
)
engine._redis = self._mock_redis(current_count=500)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN_LOCAL"
assert result.engine_name == "paddleocr"
# --- Redis counter behavior ---
def test_counter_increments_on_successful_cloud_call(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
fallback.name = "paddleocr"
primary.recognize.return_value = _make_result(
"VIN789", 0.95, "google_vision"
)
mock_r = self._mock_redis(current_count=10)
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=1000
)
engine._redis = mock_r
engine.recognize(b"img", OcrConfig())
mock_r.pipeline.assert_called_once()
pipe = mock_r.pipeline.return_value
pipe.incr.assert_called_once()
pipe.expire.assert_called_once()
pipe.execute.assert_called_once()
def test_counter_not_incremented_when_limit_reached(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
fallback.name = "paddleocr"
fallback.recognize.return_value = _make_result(
"VIN_LOCAL", 0.75, "paddleocr"
)
mock_r = self._mock_redis(current_count=1000)
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=1000
)
engine._redis = mock_r
engine.recognize(b"img", OcrConfig())
mock_r.pipeline.assert_not_called()
def test_redis_failure_assumes_limit_not_reached(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "google_vision"
fallback.name = "paddleocr"
primary.recognize.return_value = _make_result(
"VIN789", 0.95, "google_vision"
)
mock_r = MagicMock()
mock_r.get.side_effect = Exception("Redis connection refused")
mock_pipe = MagicMock()
mock_r.pipeline.return_value = mock_pipe
engine = HybridEngine(
primary=primary, fallback=fallback, threshold=0.6, monthly_limit=1000
)
engine._redis = mock_r
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN789"
# --- Non-cloud fallback path (no cap needed) ---
def test_non_cloud_fallback_not_subject_to_cap(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "cloud"
primary.recognize.return_value = _make_result("VIN", 0.6, "paddleocr")
fallback.name = "tesseract"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.return_value = _make_result(
"VIN456", 0.92, "tesseract"
)
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
result = engine.recognize(b"img", OcrConfig())
assert result.engine_name == "paddleocr"
fallback.recognize.assert_not_called()
assert result.text == "VIN456"
assert result.engine_name == "tesseract"
@patch("app.engines.hybrid_engine.time")
def test_non_cloud_fallback_timeout_returns_primary(
self, mock_time: MagicMock
) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "tesseract"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.return_value = _make_result(
"VIN456", 0.92, "tesseract"
)
mock_time.monotonic.side_effect = [0.0, 6.0]
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
def test_non_cloud_fallback_error_returns_primary(self) -> None:
from app.engines.hybrid_engine import HybridEngine
primary = MagicMock(spec=OcrEngine)
fallback = MagicMock(spec=OcrEngine)
primary.name = "paddleocr"
fallback.name = "tesseract"
primary.recognize.return_value = _make_result("VIN123", 0.3, "paddleocr")
fallback.recognize.side_effect = RuntimeError("crash")
engine = HybridEngine(primary=primary, fallback=fallback, threshold=0.6)
result = engine.recognize(b"img", OcrConfig())
assert result.text == "VIN123"
# ---------------------------------------------------------------------------
@@ -599,6 +819,7 @@ class TestEngineFactory:
mock_settings.ocr_primary_engine = "paddleocr"
mock_settings.ocr_fallback_engine = "google_vision"
mock_settings.ocr_fallback_threshold = 0.7
mock_settings.vision_monthly_limit = 1000
mock_primary = MagicMock(spec=OcrEngine)
mock_fallback = MagicMock(spec=OcrEngine)
mock_create.side_effect = [mock_primary, mock_fallback]