tests/test_api.py

from __future__ import annotations

from pathlib import Path

import pytest
from fastapi.testclient import TestClient

from bl_ocr.api.main import create_app
from bl_ocr.application.service import InferenceRequest, InferenceResult, InferenceMetadata
from bl_ocr.core.config import Settings


def get_test_settings() -> Settings:
    return Settings(
        service_name="bl-ocr-backend-test",
        api_version="v1",
        mistral_api_key="fake-key",
        mistral_ocr_url="https://example.test/ocr",
        mistral_model="test-model",
        mistral_strict_schema=True,
        request_timeout_seconds=5.0,
        max_retries=1,
        max_upload_mb=1,
        schema_path=Path("schema.json"),
        example_path=Path("example.json"),
        cors_allow_origins=("*",),
    )


class MockInferenceService:
    def __init__(self, settings: Settings) -> None:
        self.settings = settings
        self.last_request = None

    def run(self, request: InferenceRequest) -> InferenceResult:
        self.last_request = request
        return InferenceResult(
            filename=request.filename,
            content_type=request.content_type,
            checksum_md5="d41d8cd98f00b204e9800998ecf8427e",
            annotation={"extracted": "yes"},
            schema_valid=True,
            schema_errors=[],
            validation_warnings=["warning1"] if request.validate else [],
            repairs=(
                [
                    {
                        "path": "products[0].pharmacist_price_ttc",
                        "original": 10.0,
                        "value": 6.0,
                        "reason": "derived from line_total_ttc / quantity",
                    }
                ]
                if request.repair
                else []
            ),
            raw_provider_response={"document_annotation": {"extracted": "yes"}},
            parse_error=None,
            metadata=InferenceMetadata(
                provider="mistral",
                model=self.settings.mistral_model,
                retry_count=0,
                critical_retry_reasons=[],
                retry_reasons=[],
                duration_ms=10,
            ),
        )


@pytest.fixture
def client_and_service() -> tuple[TestClient, MockInferenceService]:
    settings = get_test_settings()
    service = MockInferenceService(settings)
    app = create_app(settings=settings, inference_service=service)
    client = TestClient(app)
    return client, service


def test_health_endpoint(client_and_service):
    client, _ = client_and_service
    response = client.get("/health")
    assert response.status_code == 200
    data = response.json()
    assert data["status"] == "ok"
    assert data["service"] == "bl-ocr-backend-test"
    assert data["provider_configured"] is True


def test_schema_endpoint(client_and_service):
    client, _ = client_and_service
    response = client.get("/v1/schema")
    assert response.status_code == 200
    data = response.json()
    # Should load the schema from schema.json
    assert "$schema" in data or "type" in data


def test_ocr_upload_endpoint(client_and_service):
    client, service = client_and_service
    files = {"file": ("test.jpg", b"fake-image-bytes", "image/jpeg")}
    response = client.post("/v1/ocr?repair=true&validate=true", files=files)
    
    assert response.status_code == 200
    data = response.json()
    assert data["filename"] == "test.jpg"
    assert data["content_type"] == "image/jpeg"
    assert data["annotation"] == {"extracted": "yes"}
    assert data["schema_valid"] is True
    assert data["schema_errors"] == []
    assert data["validation_warnings"] == ["warning1"]
    assert data["repairs"] == [
        {
            "path": "products[0].pharmacist_price_ttc",
            "original": 10.0,
            "value": 6.0,
            "reason": "derived from line_total_ttc / quantity",
        }
    ]

    assert service.last_request is not None
    assert service.last_request.filename == "test.jpg"
    assert service.last_request.content_type == "image/jpeg"
    assert service.last_request.repair is True
    assert service.last_request.validate is True


def test_ocr_base64_endpoint(client_and_service):
    client, service = client_and_service
    # base64 representation of b"fake-image-bytes" is ZmFrZS1pbWFnZS1ieXRlcw==
    payload = {
        "image_base64": "ZmFrZS1pbWFnZS1ieXRlcw==",
        "filename": "test_b64.jpg",
        "content_type": "image/jpeg",
        "repair": True,
        "validate": False
    }
    response = client.post("/v1/ocr/base64", json=payload)
    
    assert response.status_code == 200
    data = response.json()
    assert data["filename"] == "test_b64.jpg"
    assert data["content_type"] == "image/jpeg"
    assert data["validation_warnings"] == []
    
    assert service.last_request is not None
    assert service.last_request.filename == "test_b64.jpg"
    assert service.last_request.image_bytes == b"fake-image-bytes"
    assert service.last_request.repair is True
    assert service.last_request.validate is False