secure-web/backend/api/serializers.py

"""
DRF Serializers for the API.

This module defines serializers for converting model instances
to JSON and validating input data.
"""

from rest_framework import serializers
from websites.models import Website, Scan, Issue, Metric, ScanStatus


class IssueSerializer(serializers.ModelSerializer):
    """Serializer for Issue model."""
    
    severity_display = serializers.CharField(source='get_severity_display', read_only=True)
    category_display = serializers.CharField(source='get_category_display', read_only=True)
    tool_display = serializers.CharField(source='get_tool_display', read_only=True)
    
    class Meta:
        model = Issue
        fields = [
            'id',
            'category',
            'category_display',
            'severity',
            'severity_display',
            'tool',
            'tool_display',
            'title',
            'description',
            'affected_url',
            'remediation',
            'created_at',
        ]
        read_only_fields = fields


class MetricSerializer(serializers.ModelSerializer):
    """Serializer for Metric model."""
    
    formatted_value = serializers.CharField(source='get_formatted_value', read_only=True)
    unit_display = serializers.CharField(source='get_unit_display', read_only=True)
    
    class Meta:
        model = Metric
        fields = [
            'id',
            'name',
            'display_name',
            'value',
            'unit',
            'unit_display',
            'formatted_value',
            'source',
            'score',
        ]
        read_only_fields = fields


class ScanListSerializer(serializers.ModelSerializer):
    """Serializer for Scan list views (minimal data)."""
    
    status_display = serializers.CharField(source='get_status_display', read_only=True)
    website_url = serializers.CharField(source='website.url', read_only=True)
    issues_count = serializers.SerializerMethodField()
    
    class Meta:
        model = Scan
        fields = [
            'id',
            'website_url',
            'status',
            'status_display',
            'created_at',
            'completed_at',
            'overall_score',
            'performance_score',
            'security_score',
            'issues_count',
        ]
        read_only_fields = fields
    
    def get_issues_count(self, obj):
        return obj.issues.count()


class ScanDetailSerializer(serializers.ModelSerializer):
    """Serializer for Scan detail views (full data)."""
    
    status_display = serializers.CharField(source='get_status_display', read_only=True)
    website_url = serializers.CharField(source='website.url', read_only=True)
    website_domain = serializers.CharField(source='website.domain', read_only=True)
    issues = IssueSerializer(many=True, read_only=True)
    metrics = MetricSerializer(many=True, read_only=True)
    issues_by_category = serializers.SerializerMethodField()
    issues_by_severity = serializers.SerializerMethodField()
    
    class Meta:
        model = Scan
        fields = [
            'id',
            'website_url',
            'website_domain',
            'status',
            'status_display',
            'created_at',
            'started_at',
            'completed_at',
            'overall_score',
            'performance_score',
            'accessibility_score',
            'seo_score',
            'best_practices_score',
            'security_score',
            'error_message',
            'issues',
            'metrics',
            'issues_by_category',
            'issues_by_severity',
        ]
        read_only_fields = fields
    
    def get_issues_by_category(self, obj):
        """Group issues by category."""
        from collections import defaultdict
        grouped = defaultdict(list)
        
        for issue in obj.issues.all():
            grouped[issue.category].append(IssueSerializer(issue).data)
        
        return dict(grouped)
    
    def get_issues_by_severity(self, obj):
        """Count issues by severity."""
        from django.db.models import Count
        
        counts = obj.issues.values('severity').annotate(count=Count('id'))
        return {item['severity']: item['count'] for item in counts}


class ScanCreateSerializer(serializers.Serializer):
    """Serializer for creating new scans."""
    
    url = serializers.URLField(
        required=True,
        help_text="The URL to scan (must be http or https)"
    )
    
    def validate_url(self, value):
        """Validate and normalize the URL."""
        from scanner.utils import validate_url
        
        is_valid, result = validate_url(value)
        
        if not is_valid:
            raise serializers.ValidationError(result)
        
        return result  # Return normalized URL
    
    def create(self, validated_data):
        """Create Website and Scan records."""
        from scanner.tasks import check_rate_limit, check_concurrent_scan_limit, run_scan_task
        
        url = validated_data['url']
        
        # Check rate limit
        rate_limit_error = check_rate_limit(url)
        if rate_limit_error:
            raise serializers.ValidationError({'url': rate_limit_error})
        
        # Check concurrent scan limit
        concurrent_error = check_concurrent_scan_limit()
        if concurrent_error:
            raise serializers.ValidationError({'non_field_errors': concurrent_error})
        
        # Get or create Website
        website, created = Website.objects.get_or_create(
            url=url,
            defaults={'domain': validated_data.get('domain', '')}
        )
        
        # Create Scan
        scan = Scan.objects.create(
            website=website,
            status=ScanStatus.PENDING
        )
        
        # Trigger Celery task
        task = run_scan_task.delay(str(scan.id))
        
        # Update scan with task ID
        scan.celery_task_id = task.id
        scan.save(update_fields=['celery_task_id'])
        
        return scan


class WebsiteSerializer(serializers.ModelSerializer):
    """Serializer for Website model."""
    
    scans_count = serializers.SerializerMethodField()
    latest_scan = serializers.SerializerMethodField()
    
    class Meta:
        model = Website
        fields = [
            'id',
            'url',
            'domain',
            'created_at',
            'last_scanned_at',
            'scans_count',
            'latest_scan',
        ]
        read_only_fields = fields
    
    def get_scans_count(self, obj):
        return obj.scans.count()
    
    def get_latest_scan(self, obj):
        latest = obj.scans.first()
        if latest:
            return ScanListSerializer(latest).data
        return None


class WebsiteDetailSerializer(WebsiteSerializer):
    """Detailed Website serializer with scan list."""
    
    scans = ScanListSerializer(many=True, read_only=True)
    
    class Meta(WebsiteSerializer.Meta):
        fields = WebsiteSerializer.Meta.fields + ['scans']


class HealthCheckSerializer(serializers.Serializer):
    """Serializer for health check response."""
    
    status = serializers.CharField()
    database = serializers.CharField()
    redis = serializers.CharField()
    celery = serializers.CharField()
    timestamp = serializers.DateTimeField()