"""
Tests for Grid Detection Service v4

Tests cover:
- mm coordinate conversion
- Deskew calculation
- Column detection with 1mm margin
- Data class functionality

Lizenz: Apache 2.0 (kommerziell nutzbar)
"""

import pytest
import math
from typing import List

# Import the service under test
import sys
sys.path.insert(0, '/app')

from services.grid_detection_service import (
    GridDetectionService,
    OCRRegion,
    GridCell,
    CellStatus,
    ColumnType,
    A4_WIDTH_MM,
    A4_HEIGHT_MM,
    COLUMN_MARGIN_MM,
    COLUMN_MARGIN_PCT
)


class TestOCRRegionMMConversion:
    """Test mm coordinate conversion for OCR regions."""

    def test_x_mm_conversion(self):
        """Test X coordinate conversion from percent to mm."""
        # 50% of A4 width = 105mm
        region = OCRRegion(text="test", confidence=0.9, x=50.0, y=0.0, width=10.0, height=5.0)
        assert region.x_mm == 105.0

    def test_y_mm_conversion(self):
        """Test Y coordinate conversion from percent to mm."""
        # 33.33% of A4 height = 99mm (approx)
        region = OCRRegion(text="test", confidence=0.9, x=0.0, y=33.33, width=10.0, height=5.0)
        assert abs(region.y_mm - 99.0) < 0.5

    def test_width_mm_conversion(self):
        """Test width conversion from percent to mm."""
        # 10% of A4 width = 21mm
        region = OCRRegion(text="test", confidence=0.9, x=0.0, y=0.0, width=10.0, height=5.0)
        assert region.width_mm == 21.0

    def test_height_mm_conversion(self):
        """Test height conversion from percent to mm."""
        # 5% of A4 height = 14.85mm
        region = OCRRegion(text="test", confidence=0.9, x=0.0, y=0.0, width=10.0, height=5.0)
        assert abs(region.height_mm - 14.85) < 0.01

    def test_center_coordinates(self):
        """Test center coordinate calculation."""
        region = OCRRegion(text="test", confidence=0.9, x=10.0, y=20.0, width=20.0, height=10.0)
        assert region.center_x == 20.0
        assert region.center_y == 25.0

    def test_right_bottom_edges(self):
        """Test right and bottom edge calculation."""
        region = OCRRegion(text="test", confidence=0.9, x=10.0, y=20.0, width=30.0, height=15.0)
        assert region.right == 40.0
        assert region.bottom == 35.0


class TestGridCellMMConversion:
    """Test mm coordinate conversion for grid cells."""

    def test_cell_to_dict_includes_mm(self):
        """Test that to_dict includes mm coordinates."""
        cell = GridCell(row=0, col=0, x=10.0, y=20.0, width=30.0, height=5.0, text="hello")
        result = cell.to_dict()

        assert "x_mm" in result
        assert "y_mm" in result
        assert "width_mm" in result
        assert "height_mm" in result

        # 10% of 210mm = 21mm
        assert result["x_mm"] == 21.0
        # 20% of 297mm = 59.4mm
        assert result["y_mm"] == 59.4

    def test_cell_mm_coordinates(self):
        """Test direct mm property access."""
        cell = GridCell(row=0, col=0, x=50.0, y=50.0, width=20.0, height=3.0)

        assert cell.x_mm == 105.0  # 50% of 210mm
        assert cell.y_mm == 148.5  # 50% of 297mm
        assert cell.width_mm == 42.0  # 20% of 210mm
        assert abs(cell.height_mm - 8.91) < 0.01  # 3% of 297mm

    def test_cell_to_dict_includes_all_fields(self):
        """Test that to_dict includes all expected fields."""
        cell = GridCell(
            row=1, col=2, x=10.0, y=20.0, width=30.0, height=5.0,
            text="test", confidence=0.95, status=CellStatus.RECOGNIZED,
            column_type=ColumnType.ENGLISH, logical_row=0, logical_col=0,
            is_continuation=False
        )
        result = cell.to_dict()

        assert result["row"] == 1
        assert result["col"] == 2
        assert result["text"] == "test"
        assert result["confidence"] == 0.95
        assert result["status"] == "recognized"
        assert result["column_type"] == "english"
        assert result["logical_row"] == 0
        assert result["logical_col"] == 0
        assert result["is_continuation"] == False


class TestA4Constants:
    """Test A4 dimension constants."""

    def test_a4_width_mm(self):
        """Verify A4 width is 210mm."""
        assert A4_WIDTH_MM == 210.0

    def test_a4_height_mm(self):
        """Verify A4 height is 297mm."""
        assert A4_HEIGHT_MM == 297.0

    def test_column_margin_mm(self):
        """Verify column margin is 1mm."""
        assert COLUMN_MARGIN_MM == 1.0

    def test_column_margin_percent(self):
        """Verify column margin percentage calculation."""
        expected = (1.0 / 210.0) * 100
        assert abs(COLUMN_MARGIN_PCT - expected) < 0.001


class TestGridDetectionServiceInit:
    """Test GridDetectionService initialization."""

    def test_init_with_defaults(self):
        """Test service initializes with default parameters."""
        service = GridDetectionService()
        assert service.y_tolerance_pct == 1.5
        assert service.padding_pct == 0.3
        assert service.column_margin_mm == COLUMN_MARGIN_MM

    def test_init_with_custom_params(self):
        """Test service initializes with custom parameters."""
        service = GridDetectionService(
            y_tolerance_pct=2.0,
            padding_pct=0.5,
            column_margin_mm=2.0
        )
        assert service.y_tolerance_pct == 2.0
        assert service.padding_pct == 0.5
        assert service.column_margin_mm == 2.0


class TestDeskewCalculation:
    """Test deskew angle calculation."""

    def test_calculate_deskew_no_regions(self):
        """Test deskew returns 0 for empty regions."""
        service = GridDetectionService()
        angle = service.calculate_deskew_angle([])
        assert angle == 0.0

    def test_calculate_deskew_few_regions(self):
        """Test deskew returns 0 for too few regions."""
        service = GridDetectionService()
        regions = [
            OCRRegion(text="a", confidence=0.9, x=10.0, y=10.0, width=5.0, height=2.0),
        ]
        angle = service.calculate_deskew_angle(regions)
        assert angle == 0.0

    def test_calculate_deskew_perfectly_aligned(self):
        """Test deskew returns near-zero for perfectly aligned text."""
        service = GridDetectionService()
        # Perfectly vertical alignment at x=10%
        regions = [
            OCRRegion(text="a", confidence=0.9, x=10.0, y=10.0, width=5.0, height=2.0),
            OCRRegion(text="b", confidence=0.9, x=10.0, y=20.0, width=5.0, height=2.0),
            OCRRegion(text="c", confidence=0.9, x=10.0, y=30.0, width=5.0, height=2.0),
            OCRRegion(text="d", confidence=0.9, x=10.0, y=40.0, width=5.0, height=2.0),
            OCRRegion(text="e", confidence=0.9, x=10.0, y=50.0, width=5.0, height=2.0),
        ]
        angle = service.calculate_deskew_angle(regions)
        assert abs(angle) < 0.5  # Should be very close to 0

    def test_calculate_deskew_tilted_right(self):
        """Test deskew detects right tilt."""
        service = GridDetectionService()
        # Text tilts right as we go down (x increases with y)
        regions = [
            OCRRegion(text="a", confidence=0.9, x=10.0, y=10.0, width=5.0, height=2.0),
            OCRRegion(text="b", confidence=0.9, x=11.0, y=20.0, width=5.0, height=2.0),
            OCRRegion(text="c", confidence=0.9, x=12.0, y=30.0, width=5.0, height=2.0),
            OCRRegion(text="d", confidence=0.9, x=13.0, y=40.0, width=5.0, height=2.0),
            OCRRegion(text="e", confidence=0.9, x=14.0, y=50.0, width=5.0, height=2.0),
        ]
        angle = service.calculate_deskew_angle(regions)
        assert angle > 0  # Positive angle for right tilt

    def test_calculate_deskew_max_angle(self):
        """Test deskew is clamped to max 5 degrees."""
        service = GridDetectionService()
        # Extreme tilt
        regions = [
            OCRRegion(text="a", confidence=0.9, x=5.0, y=10.0, width=5.0, height=2.0),
            OCRRegion(text="b", confidence=0.9, x=15.0, y=20.0, width=5.0, height=2.0),
            OCRRegion(text="c", confidence=0.9, x=25.0, y=30.0, width=5.0, height=2.0),
            OCRRegion(text="d", confidence=0.9, x=35.0, y=40.0, width=5.0, height=2.0),
            OCRRegion(text="e", confidence=0.9, x=45.0, y=50.0, width=5.0, height=2.0),
        ]
        angle = service.calculate_deskew_angle(regions)
        assert abs(angle) <= 5.0  # Clamped to ±5°


class TestDeskewApplication:
    """Test deskew coordinate transformation."""

    def test_apply_deskew_zero_angle(self):
        """Test no transformation for zero angle."""
        service = GridDetectionService()
        regions = [
            OCRRegion(text="a", confidence=0.9, x=10.0, y=20.0, width=5.0, height=2.0),
        ]
        result = service.apply_deskew_to_regions(regions, 0.0)

        assert len(result) == 1
        assert result[0].x == 10.0
        assert result[0].y == 20.0

    def test_apply_deskew_preserves_text(self):
        """Test deskew preserves text and confidence."""
        service = GridDetectionService()
        regions = [
            OCRRegion(text="hello", confidence=0.95, x=10.0, y=20.0, width=5.0, height=2.0),
        ]
        result = service.apply_deskew_to_regions(regions, 2.0)

        assert result[0].text == "hello"
        assert result[0].confidence == 0.95


class TestCellStatus:
    """Test cell status classification."""

    def test_cell_status_empty(self):
        """Test empty cell status."""
        cell = GridCell(row=0, col=0, x=0, y=0, width=10, height=5, text="")
        assert cell.status == CellStatus.EMPTY

    def test_cell_status_recognized(self):
        """Test recognized cell status."""
        cell = GridCell(
            row=0, col=0, x=0, y=0, width=10, height=5,
            text="hello", confidence=0.9, status=CellStatus.RECOGNIZED
        )
        assert cell.status == CellStatus.RECOGNIZED

    def test_cell_status_problematic(self):
        """Test problematic cell (low confidence)."""
        cell = GridCell(
            row=0, col=0, x=0, y=0, width=10, height=5,
            text="hello", confidence=0.3, status=CellStatus.PROBLEMATIC
        )
        assert cell.status == CellStatus.PROBLEMATIC


class TestColumnType:
    """Test column type enum."""

    def test_column_type_values(self):
        """Test column type enum values."""
        assert ColumnType.ENGLISH.value == "english"
        assert ColumnType.GERMAN.value == "german"
        assert ColumnType.EXAMPLE.value == "example"
        assert ColumnType.UNKNOWN.value == "unknown"


class TestDetectGrid:
    """Test grid detection functionality."""

    def test_detect_grid_empty_regions(self):
        """Test grid detection with empty regions."""
        service = GridDetectionService()
        result = service.detect_grid([])

        assert result.rows == 0
        assert result.columns == 0
        assert len(result.cells) == 0

    def test_detect_grid_single_word(self):
        """Test grid detection with single word."""
        service = GridDetectionService()
        regions = [
            OCRRegion(text="house", confidence=0.9, x=10.0, y=10.0, width=10.0, height=2.0),
        ]
        result = service.detect_grid(regions)

        assert result.rows >= 1
        assert result.columns >= 1

    def test_detect_grid_result_has_page_dimensions(self):
        """Test that result includes page dimensions."""
        service = GridDetectionService()
        regions = [
            OCRRegion(text="house", confidence=0.9, x=10.0, y=10.0, width=10.0, height=2.0),
        ]
        result = service.detect_grid(regions)
        result_dict = result.to_dict()

        assert "page_dimensions" in result_dict
        assert result_dict["page_dimensions"]["width_mm"] == 210.0
        assert result_dict["page_dimensions"]["height_mm"] == 297.0
        assert result_dict["page_dimensions"]["format"] == "A4"

    def test_detect_grid_result_has_stats(self):
        """Test that result includes stats."""
        service = GridDetectionService()
        regions = [
            OCRRegion(text="house", confidence=0.9, x=10.0, y=10.0, width=10.0, height=2.0),
            OCRRegion(text="Haus", confidence=0.8, x=50.0, y=10.0, width=8.0, height=2.0),
        ]
        result = service.detect_grid(regions)
        result_dict = result.to_dict()

        assert "stats" in result_dict
        assert "recognized" in result_dict["stats"]
        assert "coverage" in result_dict["stats"]


class TestIntegration:
    """Integration tests for full analysis pipeline."""

    def test_full_vocabulary_table_analysis(self):
        """Test analysis of a typical vocabulary table."""
        service = GridDetectionService()

        # Simulate a vocabulary table with 3 columns
        regions = [
            # Row 1
            OCRRegion(text="house", confidence=0.95, x=10.0, y=15.0, width=12.0, height=2.5),
            OCRRegion(text="Haus", confidence=0.92, x=45.0, y=15.0, width=8.0, height=2.5),
            OCRRegion(text="This is a house.", confidence=0.88, x=70.0, y=15.0, width=25.0, height=2.5),
            # Row 2
            OCRRegion(text="car", confidence=0.94, x=10.0, y=22.0, width=8.0, height=2.5),
            OCRRegion(text="Auto", confidence=0.91, x=45.0, y=22.0, width=9.0, height=2.5),
            OCRRegion(text="I drive a car.", confidence=0.85, x=70.0, y=22.0, width=22.0, height=2.5),
            # Row 3
            OCRRegion(text="tree", confidence=0.96, x=10.0, y=29.0, width=9.0, height=2.5),
            OCRRegion(text="Baum", confidence=0.93, x=45.0, y=29.0, width=10.0, height=2.5),
            OCRRegion(text="The tree is tall.", confidence=0.87, x=70.0, y=29.0, width=24.0, height=2.5),
        ]

        result = service.detect_grid(regions)
        result_dict = result.to_dict()

        # Verify structure
        assert "cells" in result_dict
        assert "page_dimensions" in result_dict
        assert "stats" in result_dict

        # Verify page dimensions
        assert result_dict["page_dimensions"]["format"] == "A4"

        # Verify cells have mm coordinates
        if len(result_dict["cells"]) > 0 and len(result_dict["cells"][0]) > 0:
            cell = result_dict["cells"][0][0]
            assert "x_mm" in cell
            assert "y_mm" in cell
            assert "width_mm" in cell
            assert "height_mm" in cell


if __name__ == "__main__":
    pytest.main([__file__, "-v"])