"""
Template Sources Configuration for Legal Templates RAG.

Defines all source repositories and their license metadata for the
bp_legal_templates collection. Sources are organized by license type
for proper attribution compliance.

License Types:
- PUBLIC_DOMAIN: German official works (§5 UrhG) - no attribution needed
- CC0: Public Domain Dedication - no attribution needed (recommended)
- UNLICENSE: Public Domain equivalent - no attribution needed
- MIT: Attribution required on redistribution
- CC_BY_4: Attribution + change notices required
- REUSE_NOTICE: May quote with source, no distortion allowed
"""

from dataclasses import dataclass, field
from enum import Enum
from typing import List, Optional


class LicenseType(Enum):
    """License types for template sources with compliance requirements."""
    PUBLIC_DOMAIN = "public_domain"  # §5 UrhG amtliche Werke
    CC0 = "cc0"                      # CC0 1.0 Universal
    UNLICENSE = "unlicense"          # Unlicense (public domain)
    MIT = "mit"                      # MIT License
    CC_BY_4 = "cc_by_4"              # CC BY 4.0 International
    REUSE_NOTICE = "reuse_notice"    # EU reuse notice (source required)


@dataclass
class LicenseInfo:
    """Detailed license information for compliance."""
    id: LicenseType
    name: str
    url: str
    attribution_required: bool
    share_alike: bool = False
    no_derivatives: bool = False
    commercial_use: bool = True
    training_allowed: bool = True
    output_allowed: bool = True
    modification_allowed: bool = True
    distortion_prohibited: bool = False
    attribution_template: Optional[str] = None

    def get_attribution_text(self, source_name: str, source_url: str) -> str:
        """Generate attribution text for this license type."""
        if not self.attribution_required:
            return ""
        if self.attribution_template:
            return self.attribution_template.format(
                source_name=source_name,
                source_url=source_url,
                license_name=self.name,
                license_url=self.url
            )
        return f"Source: {source_name} ({self.name})"


# License definitions with full compliance info
LICENSES: dict[LicenseType, LicenseInfo] = {
    LicenseType.PUBLIC_DOMAIN: LicenseInfo(
        id=LicenseType.PUBLIC_DOMAIN,
        name="Public Domain (§5 UrhG)",
        url="https://www.gesetze-im-internet.de/urhg/__5.html",
        attribution_required=False,
        training_allowed=True,
        output_allowed=True,
        modification_allowed=True,
    ),
    LicenseType.CC0: LicenseInfo(
        id=LicenseType.CC0,
        name="CC0 1.0 Universal",
        url="https://creativecommons.org/publicdomain/zero/1.0/",
        attribution_required=False,  # Not required but recommended
        training_allowed=True,
        output_allowed=True,
        modification_allowed=True,
        attribution_template="[{source_name}]({source_url}) - CC0 1.0",
    ),
    LicenseType.UNLICENSE: LicenseInfo(
        id=LicenseType.UNLICENSE,
        name="Unlicense",
        url="https://unlicense.org/",
        attribution_required=False,
        training_allowed=True,
        output_allowed=True,
        modification_allowed=True,
    ),
    LicenseType.MIT: LicenseInfo(
        id=LicenseType.MIT,
        name="MIT License",
        url="https://opensource.org/licenses/MIT",
        attribution_required=True,
        training_allowed=True,
        output_allowed=True,
        modification_allowed=True,
        attribution_template="Based on [{source_name}]({source_url}) - MIT License",
    ),
    LicenseType.CC_BY_4: LicenseInfo(
        id=LicenseType.CC_BY_4,
        name="CC BY 4.0 International",
        url="https://creativecommons.org/licenses/by/4.0/",
        attribution_required=True,
        training_allowed=False,  # CC BY 4.0 may restrict training
        output_allowed=True,
        modification_allowed=True,
        attribution_template=(
            "Adapted from [{source_name}]({source_url}), "
            "licensed under [CC BY 4.0]({license_url}). Changes were made."
        ),
    ),
    LicenseType.REUSE_NOTICE: LicenseInfo(
        id=LicenseType.REUSE_NOTICE,
        name="EU Reuse Notice",
        url="https://commission.europa.eu/legal-notice_en",
        attribution_required=True,
        training_allowed=False,
        output_allowed=True,
        modification_allowed=False,
        distortion_prohibited=True,
        attribution_template="Source: {source_name} ({source_url})",
    ),
}


@dataclass
class SourceConfig:
    """Configuration for a template source repository."""
    name: str
    license_type: LicenseType
    template_types: List[str]
    languages: List[str]
    jurisdiction: str
    description: str
    repo_url: Optional[str] = None
    web_url: Optional[str] = None
    file_patterns: List[str] = field(default_factory=lambda: ["*.md", "*.txt", "*.html"])
    exclude_patterns: List[str] = field(default_factory=list)
    priority: int = 1  # 1 = highest priority (CC0), 5 = lowest (REUSE_NOTICE)
    enabled: bool = True

    @property
    def license_info(self) -> LicenseInfo:
        """Get the full license information for this source."""
        return LICENSES[self.license_type]

    def get_source_url(self) -> str:
        """Get the primary URL for this source."""
        return self.repo_url or self.web_url or ""


# =============================================================================
# Phase 1: CC0-Quellen (Höchste Priorität - keine Attribution nötig)
# =============================================================================

TEMPLATE_SOURCES: List[SourceConfig] = [
    # GitHub Site Policy (CC0)
    SourceConfig(
        name="github-site-policy",
        repo_url="https://github.com/github/site-policy",
        license_type=LicenseType.CC0,
        template_types=["terms_of_service", "privacy_policy", "community_guidelines", "acceptable_use"],
        languages=["en"],
        jurisdiction="US",
        description="GitHub's site policies including Terms of Service, Privacy Policy, and Community Guidelines. High-quality, well-structured legal templates.",
        file_patterns=["Policies/*.md", "*.md"],
        exclude_patterns=["README.md", "CONTRIBUTING.md", "LICENSE.md", "archived/*"],
        priority=1,
    ),

    # opr.vc DSGVO Muster (CC0)
    SourceConfig(
        name="opr-vc",
        repo_url="https://github.com/oprvc/oprvc.github.io",
        web_url="https://opr.vc/",
        license_type=LicenseType.CC0,
        template_types=["privacy_policy", "impressum"],
        languages=["de"],
        jurisdiction="DE",
        description="Open Privacy Resource - DSGVO-konforme Mustertexte für Datenschutzerklärungen und Impressum. Speziell für deutsche Websites.",
        file_patterns=["*.md", "*.html", "_posts/*.md"],
        priority=1,
    ),

    # Open Gov Foundation (CC0)
    SourceConfig(
        name="opengovfoundation-site-policy",
        repo_url="https://github.com/opengovfoundation/site-policy",
        license_type=LicenseType.CC0,
        template_types=["terms_of_service", "privacy_policy", "copyright_policy"],
        languages=["en"],
        jurisdiction="US",
        description="OpenGov Foundation's site policies. Clean, reusable templates for open government projects.",
        file_patterns=["*.md"],
        priority=1,
    ),

    # Creative Commons Legal Tools Data (CC0)
    SourceConfig(
        name="cc-legal-tools-data",
        repo_url="https://github.com/creativecommons/cc-legal-tools-data",
        license_type=LicenseType.CC0,
        template_types=["license_text"],
        languages=["de", "en"],
        jurisdiction="INTL",
        description="Creative Commons license texts in multiple languages. Useful as reference for license templates.",
        file_patterns=["legalcode/**/legalcode.de.html", "legalcode/**/legalcode.en.html"],
        priority=1,
    ),

    # =============================================================================
    # Phase 2: MIT-Quellen (Attribution bei Weitergabe)
    # =============================================================================

    # Webflorist Privacy Policy Text (MIT)
    SourceConfig(
        name="webflorist-privacy-policy",
        repo_url="https://github.com/webflorist/privacy-policy-text",
        license_type=LicenseType.MIT,
        template_types=["privacy_policy"],
        languages=["de", "en"],
        jurisdiction="EU",
        description="Modular GDPR-compliant privacy policy texts in JSON/PHP format. Highly customizable with variable sections.",
        file_patterns=["src/**/*.json", "src/**/*.php", "*.md"],
        priority=2,
    ),

    # Tempest Privacy Policy Generator (MIT)
    SourceConfig(
        name="tempest-privacy-policy",
        repo_url="https://github.com/Tempest-Solutions-Company/privacy-policy-generator",
        license_type=LicenseType.MIT,
        template_types=["privacy_policy"],
        languages=["en"],
        jurisdiction="INTL",
        description="Privacy policy generator with templates for various use cases.",
        file_patterns=["templates/*.md", "src/**/*.txt", "*.md"],
        priority=2,
    ),

    # Tempest Terms of Service Generator (MIT)
    SourceConfig(
        name="tempest-terms-of-service",
        repo_url="https://github.com/Tempest-Solutions-Company/terms-of-service-generator",
        license_type=LicenseType.MIT,
        template_types=["terms_of_service", "dpa"],
        languages=["en"],
        jurisdiction="INTL",
        description="Terms of Service and DPA clause generator templates.",
        file_patterns=["templates/*.md", "src/**/*.txt", "*.md"],
        priority=2,
    ),

    # Tempest Cookie Banner (MIT)
    SourceConfig(
        name="tempest-cookie-banner",
        repo_url="https://github.com/Tempest-Solutions-Company/cookie-banner-consent-solution",
        license_type=LicenseType.MIT,
        template_types=["cookie_banner", "cookie_policy"],
        languages=["en"],
        jurisdiction="EU",
        description="Cookie consent banner texts and templates for GDPR/ePrivacy compliance.",
        file_patterns=["templates/*.md", "src/**/*.txt", "*.md", "locales/*.json"],
        priority=2,
    ),

    # =============================================================================
    # Phase 3: CC BY 4.0 (Attribution + Änderungskennzeichnung)
    # =============================================================================

    # Common Paper Standards (CC BY 4.0)
    SourceConfig(
        name="common-paper-standards",
        repo_url="https://github.com/CommonPaper/SLA",
        web_url="https://commonpaper.com/standards/",
        license_type=LicenseType.CC_BY_4,
        template_types=["sla", "cloud_service_agreement", "terms_of_service", "nda", "dpa"],
        languages=["en"],
        jurisdiction="US",
        description="Common Paper's standardized B2B SaaS contract templates. Industry-standard agreements for cloud services.",
        file_patterns=["*.md", "versions/**/*.md"],
        priority=3,
    ),

    # Datennutzungsklauseln Muster (CC BY 4.0)
    SourceConfig(
        name="datennutzungsklauseln-muster",
        repo_url="https://gitlab.opencode.de/wernerth/datennutzungsklauseln-muster",
        license_type=LicenseType.CC_BY_4,
        template_types=["data_usage_clause", "dpa"],
        languages=["de"],
        jurisdiction="DE",
        description="B2B Datennutzungsklauseln für Verträge. Speziell für deutsche Unternehmen.",
        file_patterns=["*.md", "klauseln/*.md"],
        priority=3,
    ),

    # =============================================================================
    # Phase 4: Amtliche Werke (§5 UrhG - urheberrechtsfrei, Referenz)
    # =============================================================================

    # Bundestag Gesetze (Unlicense)
    SourceConfig(
        name="bundestag-gesetze",
        repo_url="https://github.com/bundestag/gesetze",
        license_type=LicenseType.UNLICENSE,
        template_types=["law_reference"],
        languages=["de"],
        jurisdiction="DE",
        description="Deutsche Bundesgesetze im Markdown-Format. Referenz für DDG, TDDDG, EGBGB Muster.",
        file_patterns=["d/ddg/*.md", "t/tdddg/*.md", "e/egbgb/*.md", "b/bgb/*.md"],
        priority=4,
    ),

    # Gesetze im Internet (Public Domain via §5 UrhG)
    SourceConfig(
        name="gesetze-im-internet",
        web_url="https://www.gesetze-im-internet.de/",
        license_type=LicenseType.PUBLIC_DOMAIN,
        template_types=["law_reference", "widerruf", "impressum"],
        languages=["de"],
        jurisdiction="DE",
        description="Amtliche Gesetzestexte. DDG §5 (Impressum), TDDDG §25, EGBGB Muster-Widerrufsformular.",
        file_patterns=[],  # Web scraping required
        enabled=False,  # Requires custom web crawler
        priority=4,
    ),

    # EUR-Lex (Public Domain + Reuse Notice)
    SourceConfig(
        name="eur-lex",
        web_url="https://eur-lex.europa.eu/",
        license_type=LicenseType.PUBLIC_DOMAIN,
        template_types=["scc", "law_reference"],
        languages=["de", "en"],
        jurisdiction="EU",
        description="EU-Recht: DSGVO Artikel, DSA, SCC (Durchführungsbeschluss 2021/914).",
        file_patterns=[],  # Web scraping required
        enabled=False,  # Requires custom web crawler
        priority=4,
    ),

    # =============================================================================
    # Phase 5: Reuse-Notices (Guidance als Referenz)
    # =============================================================================

    # EDPB Guidelines (Reuse Notice)
    SourceConfig(
        name="edpb-guidelines",
        web_url="https://www.edpb.europa.eu/",
        license_type=LicenseType.REUSE_NOTICE,
        template_types=["guidance"],
        languages=["de", "en"],
        jurisdiction="EU",
        description="EDPB Datenschutz-Guidelines und FAQs. Als Referenz verwendbar, keine Sinnentstellung erlaubt.",
        file_patterns=[],  # Web scraping required
        enabled=False,  # Requires custom web crawler
        priority=5,
    ),

    # EDPS Resources (Reuse Notice)
    SourceConfig(
        name="edps-resources",
        web_url="https://www.edps.europa.eu/",
        license_type=LicenseType.REUSE_NOTICE,
        template_types=["guidance"],
        languages=["de", "en"],
        jurisdiction="EU",
        description="EDPS Datenschutz-Ressourcen und FAQs. Als Referenz verwendbar.",
        file_patterns=[],  # Web scraping required
        enabled=False,  # Requires custom web crawler
        priority=5,
    ),

    # EU Commission Policies (CC BY 4.0)
    SourceConfig(
        name="eu-commission-policies",
        web_url="https://commission.europa.eu/",
        license_type=LicenseType.CC_BY_4,
        template_types=["guidance", "policy"],
        languages=["de", "en"],
        jurisdiction="EU",
        description="EU-Kommission Policy-Dokumente. CC BY 4.0 lizenziert.",
        file_patterns=[],  # Web scraping required
        enabled=False,  # Requires custom web crawler
        priority=5,
    ),
]


def get_enabled_sources() -> List[SourceConfig]:
    """Get all enabled template sources."""
    return [s for s in TEMPLATE_SOURCES if s.enabled]


def get_sources_by_priority(max_priority: int = 5) -> List[SourceConfig]:
    """Get sources filtered by priority level (lower = higher priority)."""
    return sorted(
        [s for s in get_enabled_sources() if s.priority <= max_priority],
        key=lambda s: s.priority
    )


def get_sources_by_license(license_type: LicenseType) -> List[SourceConfig]:
    """Get sources filtered by license type."""
    return [s for s in get_enabled_sources() if s.license_type == license_type]


def get_sources_by_template_type(template_type: str) -> List[SourceConfig]:
    """Get sources that provide a specific template type."""
    return [s for s in get_enabled_sources() if template_type in s.template_types]


def get_sources_by_language(language: str) -> List[SourceConfig]:
    """Get sources that provide content in a specific language."""
    return [s for s in get_enabled_sources() if language in s.languages]


def get_sources_by_jurisdiction(jurisdiction: str) -> List[SourceConfig]:
    """Get sources for a specific jurisdiction."""
    return [s for s in get_enabled_sources() if s.jurisdiction == jurisdiction]


# Template type definitions for documentation
TEMPLATE_TYPES = {
    "privacy_policy": "Datenschutzerklärung / Privacy Policy",
    "terms_of_service": "Nutzungsbedingungen / Terms of Service",
    "agb": "Allgemeine Geschäftsbedingungen",
    "cookie_banner": "Cookie-Banner Text",
    "cookie_policy": "Cookie-Richtlinie / Cookie Policy",
    "impressum": "Impressum / Legal Notice",
    "widerruf": "Widerrufsbelehrung / Cancellation Policy",
    "dpa": "Auftragsverarbeitungsvertrag / Data Processing Agreement",
    "sla": "Service Level Agreement",
    "nda": "Geheimhaltungsvereinbarung / Non-Disclosure Agreement",
    "cloud_service_agreement": "Cloud-Dienstleistungsvertrag",
    "data_usage_clause": "Datennutzungsklausel",
    "acceptable_use": "Acceptable Use Policy",
    "community_guidelines": "Community-Richtlinien",
    "copyright_policy": "Urheberrechtsrichtlinie",
    "license_text": "Lizenztext",
    "law_reference": "Gesetzesreferenz (nicht als Vorlage)",
    "guidance": "Behördliche Guidance (nur Referenz)",
    "policy": "Policy-Dokument",
}


# Jurisdiction definitions
JURISDICTIONS = {
    "DE": "Deutschland",
    "AT": "Österreich",
    "CH": "Schweiz",
    "EU": "Europäische Union",
    "US": "United States",
    "INTL": "International",
}