tests/test_source_tracker.py - master - alsoalgo/CTF

tests/test_source_tracker.py
200 строк7 KB
alsoalgo
some improvements
14 дек 2025, 22:23
14 дек 2025, 22:2337d2ec0
"""
Тесты для SourceTracker
"""
import pytest
from src.components.source_tracker import SourceTracker, SourceLocation


class TestSourceLocation:
    """Тесты для SourceLocation"""
    
    def test_source_location_creation(self):
        """Тест создания SourceLocation"""
        location = SourceLocation(
            file_path="/path/to/file.pdf",
            file_name="file.pdf",
            page_number=1,
            coordinates={"x": 10, "y": 20, "width": 100, "height": 30},
            context="Some context",
            extraction_method="ocr",
            confidence=0.95
        )
        
        assert location.file_path == "/path/to/file.pdf"
        assert location.file_name == "file.pdf"
        assert location.page_number == 1
        assert location.coordinates == {"x": 10, "y": 20, "width": 100, "height": 30}
        assert location.context == "Some context"
        assert location.extraction_method == "ocr"
        assert location.confidence == 0.95
    
    def test_source_location_defaults(self):
        """Тест SourceLocation с значениями по умолчанию"""
        location = SourceLocation(
            file_path="/path/to/file.pdf",
            file_name="file.pdf"
        )
        
        assert location.page_number is None
        assert location.coordinates is None
        assert location.context is None
        assert location.extraction_method == "standard"
        assert location.confidence == 1.0
    
    def test_source_location_to_dict(self):
        """Тест преобразования SourceLocation в словарь"""
        location = SourceLocation(
            file_path="/path/to/file.pdf",
            file_name="file.pdf",
            page_number=1
        )
        
        data = location.to_dict()
        assert isinstance(data, dict)
        assert data['file_path'] == "/path/to/file.pdf"
        assert data['file_name'] == "file.pdf"
        assert data['page_number'] == 1
    
    def test_source_location_from_dict(self):
        """Тест создания SourceLocation из словаря"""
        data = {
            'file_path': "/path/to/file.pdf",
            'file_name': "file.pdf",
            'page_number': 1,
            'coordinates': None,
            'context': None,
            'extraction_method': "ocr",
            'confidence': 0.9
        }
        
        location = SourceLocation.from_dict(data)
        assert location.file_path == "/path/to/file.pdf"
        assert location.file_name == "file.pdf"
        assert location.page_number == 1
        assert location.extraction_method == "ocr"
        assert location.confidence == 0.9


class TestSourceTracker:
    """Тесты для SourceTracker"""
    
    def setup_method(self):
        """Настройка перед каждым тестом"""
        self.tracker = SourceTracker()
    
    def test_tracker_initialization(self):
        """Тест инициализации SourceTracker"""
        tracker = SourceTracker()
        assert tracker.sources == {}
    
    def test_add_source(self):
        """Тест добавления источника"""
        self.tracker.add_source(
            key="supplier",
            file_path="/path/to/invoice.pdf",
            page_number=1,
            coordinates={"x": 10, "y": 20},
            context="Supplier name",
            extraction_method="ocr",
            confidence=0.95
        )
        
        assert "supplier" in self.tracker.sources
        source = self.tracker.sources["supplier"]
        assert source.file_path == "/path/to/invoice.pdf"
        assert source.page_number == 1
        assert source.extraction_method == "ocr"
    
    def test_add_source_minimal(self):
        """Тест добавления источника с минимальными данными"""
        self.tracker.add_source(
            key="field1",
            file_path="/path/to/file.pdf"
        )
        
        assert "field1" in self.tracker.sources
        source = self.tracker.sources["field1"]
        assert source.file_name == "file.pdf"
        assert source.extraction_method == "standard"
        assert source.confidence == 1.0
    
    def test_get_source_existing(self):
        """Тест получения существующего источника"""
        self.tracker.add_source(
            key="test_key",
            file_path="/path/to/file.pdf"
        )
        
        source = self.tracker.get_source("test_key")
        assert source is not None
        assert source.file_path == "/path/to/file.pdf"
    
    def test_get_source_nonexistent(self):
        """Тест получения несуществующего источника"""
        source = self.tracker.get_source("nonexistent")
        assert source is None
    
    def test_get_all_sources(self):
        """Тест получения всех источников"""
        self.tracker.add_source("key1", "/path/to/file1.pdf")
        self.tracker.add_source("key2", "/path/to/file2.pdf")
        
        all_sources = self.tracker.get_all_sources()
        assert len(all_sources) == 2
        assert "key1" in all_sources
        assert "key2" in all_sources
        # Проверяем, что это копия, а не ссылка
        all_sources["key3"] = "test"
        assert "key3" not in self.tracker.sources
    
    def test_to_dict(self):
        """Тест преобразования SourceTracker в словарь"""
        self.tracker.add_source(
            key="test_key",
            file_path="/path/to/file.pdf",
            page_number=1
        )
        
        data = self.tracker.to_dict()
        assert isinstance(data, dict)
        assert "test_key" in data
        assert data["test_key"]["file_path"] == "/path/to/file.pdf"
        assert data["test_key"]["page_number"] == 1
    
    def test_from_dict(self):
        """Тест создания SourceTracker из словаря"""
        data = {
            "key1": {
                'file_path': "/path/to/file1.pdf",
                'file_name': "file1.pdf",
                'page_number': 1,
                'coordinates': None,
                'context': None,
                'extraction_method': "ocr",
                'confidence': 0.9
            },
            "key2": {
                'file_path': "/path/to/file2.pdf",
                'file_name': "file2.pdf",
                'page_number': None,
                'coordinates': None,
                'context': None,
                'extraction_method': "standard",
                'confidence': 1.0
            }
        }
        
        tracker = SourceTracker.from_dict(data)
        assert len(tracker.sources) == 2
        assert tracker.get_source("key1").file_path == "/path/to/file1.pdf"
        assert tracker.get_source("key2").file_path == "/path/to/file2.pdf"
    
    def test_multiple_sources(self):
        """Тест работы с несколькими источниками"""
        self.tracker.add_source("supplier", "/path/to/invoice.pdf", page_number=1)
        self.tracker.add_source("amount", "/path/to/invoice.pdf", page_number=1)
        self.tracker.add_source("date", "/path/to/contract.pdf", page_number=2)
        
        assert len(self.tracker.sources) == 3
        assert self.tracker.get_source("supplier").file_path == "/path/to/invoice.pdf"
        assert self.tracker.get_source("date").file_path == "/path/to/contract.pdf"