"""Registry API client for CmdForge.

Handles all HTTP communication with the registry server.
"""

import hashlib
import json
import time
from dataclasses import dataclass, field
from datetime import datetime, timedelta
from pathlib import Path
from typing import Optional, List, Dict, Any
from urllib.parse import urljoin, urlencode

import requests

from .config import (
    load_config,
    get_registry_url,
    get_registry_token,
    get_client_id,
    CONFIG_DIR
)


# Local cache directory
CACHE_DIR = CONFIG_DIR / "registry"
INDEX_CACHE_FILE = CACHE_DIR / "index.json"
INDEX_CACHE_MAX_AGE = timedelta(hours=24)


@dataclass
class RegistryError(Exception):
    """Base exception for registry errors."""
    code: str
    message: str
    details: Optional[Dict] = None
    http_status: int = 0

    def __str__(self):
        return f"{self.code}: {self.message}"


@dataclass
class RateLimitError(RegistryError):
    """Raised when rate limited by the registry."""
    retry_after: int = 60

    def __init__(self, retry_after: int = 60):
        super().__init__(
            code="RATE_LIMITED",
            message=f"Rate limited. Retry after {retry_after} seconds.",
            http_status=429
        )
        self.retry_after = retry_after


@dataclass
class PaginatedResponse:
    """Paginated API response."""
    data: List[Dict]
    page: int = 1
    per_page: int = 20
    total: int = 0
    total_pages: int = 0


@dataclass
class ToolInfo:
    """Tool information from the registry."""
    owner: str
    name: str
    version: str
    description: str = ""
    category: str = ""
    tags: List[str] = field(default_factory=list)
    downloads: int = 0
    deprecated: bool = False
    deprecated_message: str = ""
    replacement: str = ""
    published_at: str = ""
    readme: str = ""

    @property
    def full_name(self) -> str:
        return f"{self.owner}/{self.name}"

    @classmethod
    def from_dict(cls, data: dict) -> "ToolInfo":
        return cls(
            owner=data.get("owner", ""),
            name=data.get("name", ""),
            version=data.get("version", ""),
            description=data.get("description", ""),
            category=data.get("category", ""),
            tags=data.get("tags", []),
            downloads=data.get("downloads", 0),
            deprecated=data.get("deprecated", False),
            deprecated_message=data.get("deprecated_message", ""),
            replacement=data.get("replacement", ""),
            published_at=data.get("published_at", ""),
            readme=data.get("readme", "")
        )


@dataclass
class DownloadResult:
    """Result of downloading a tool."""
    owner: str
    name: str
    resolved_version: str
    config_yaml: str
    readme: str = ""


class RegistryClient:
    """Client for interacting with the CmdForge registry API."""

    def __init__(
        self,
        base_url: Optional[str] = None,
        token: Optional[str] = None,
        timeout: int = 30,
        max_retries: int = 3
    ):
        """
        Initialize the registry client.

        Args:
            base_url: Registry API base URL (default: from config)
            token: Auth token for authenticated requests (default: from config)
            timeout: Request timeout in seconds
            max_retries: Maximum number of retries for failed requests
        """
        self.base_url = base_url or get_registry_url()
        self.token = token or get_registry_token()
        self.timeout = timeout
        self.max_retries = max_retries
        self.client_id = get_client_id()

        # Session for connection pooling
        self._session = requests.Session()
        self._session.headers.update({
            "User-Agent": "CmdForge-CLI/1.0",
            "X-CmdForge-Client": "cli/1.0.0",
            "Accept": "application/json"
        })

        # Add client ID header
        if self.client_id:
            self._session.headers["X-Client-ID"] = self.client_id

    def _url(self, path: str) -> str:
        """Build full URL from path."""
        # Ensure base_url ends without /api/v1 duplication
        base = self.base_url.rstrip("/")
        if not path.startswith("/"):
            path = "/" + path
        return base + path

    def _auth_headers(self) -> Dict[str, str]:
        """Get authentication headers if token is available."""
        if self.token:
            return {"Authorization": f"Bearer {self.token}"}
        return {}

    def _request(
        self,
        method: str,
        path: str,
        params: Optional[Dict] = None,
        json_data: Optional[Dict] = None,
        require_auth: bool = False,
        etag: Optional[str] = None
    ) -> requests.Response:
        """
        Make an HTTP request with retry logic.

        Args:
            method: HTTP method
            path: API path
            params: Query parameters
            json_data: JSON body data
            require_auth: Whether auth is required
            etag: ETag for conditional requests

        Returns:
            Response object

        Raises:
            RegistryError: On API errors
            RateLimitError: When rate limited
        """
        url = self._url(path)
        headers = {}

        if require_auth:
            if not self.token:
                raise RegistryError(
                    code="UNAUTHORIZED",
                    message="Authentication required. Set registry token with 'cmdforge config set-token'",
                    http_status=401
                )
            headers.update(self._auth_headers())

        if etag:
            headers["If-None-Match"] = etag

        last_error = None
        for attempt in range(self.max_retries):
            try:
                response = self._session.request(
                    method=method,
                    url=url,
                    params=params,
                    json=json_data,
                    headers=headers,
                    timeout=self.timeout
                )

                # Handle rate limiting
                if response.status_code == 429:
                    retry_after = int(response.headers.get("Retry-After", 60))
                    if attempt < self.max_retries - 1:
                        time.sleep(min(retry_after, 30))  # Cap wait at 30s per attempt
                        continue
                    raise RateLimitError(retry_after=retry_after)

                # Handle server errors with retry
                if response.status_code >= 500:
                    if attempt < self.max_retries - 1:
                        time.sleep(2 ** attempt)  # Exponential backoff
                        continue

                return response

            except requests.exceptions.Timeout:
                last_error = RegistryError(
                    code="TIMEOUT",
                    message="Request timed out"
                )
                if attempt < self.max_retries - 1:
                    time.sleep(2 ** attempt)
                    continue

            except requests.exceptions.ConnectionError:
                last_error = RegistryError(
                    code="CONNECTION_ERROR",
                    message="Could not connect to registry"
                )
                if attempt < self.max_retries - 1:
                    time.sleep(2 ** attempt)
                    continue

        raise last_error or RegistryError(
            code="REQUEST_FAILED",
            message="Request failed after retries"
        )

    def _handle_error_response(self, response: requests.Response) -> None:
        """Parse and raise appropriate error from error response."""
        try:
            data = response.json()
            error = data.get("error", {})
            raise RegistryError(
                code=error.get("code", "UNKNOWN_ERROR"),
                message=error.get("message", "Unknown error"),
                details=error.get("details"),
                http_status=response.status_code
            )
        except (json.JSONDecodeError, KeyError):
            raise RegistryError(
                code="UNKNOWN_ERROR",
                message=f"HTTP {response.status_code}: {response.text[:200]}",
                http_status=response.status_code
            )

    # -------------------------------------------------------------------------
    # Public API Methods
    # -------------------------------------------------------------------------

    def list_tools(
        self,
        category: Optional[str] = None,
        page: int = 1,
        per_page: int = 20,
        sort: str = "downloads",
        order: str = "desc"
    ) -> PaginatedResponse:
        """
        List tools from the registry.

        Args:
            category: Filter by category
            page: Page number (1-indexed)
            per_page: Items per page (max 100)
            sort: Sort field (downloads, published_at, name)
            order: Sort order (asc, desc)

        Returns:
            PaginatedResponse with tool data
        """
        params = {
            "page": page,
            "per_page": min(per_page, 100),
            "sort": sort,
            "order": order
        }
        if category:
            params["category"] = category

        response = self._request("GET", "/tools", params=params)

        if response.status_code != 200:
            self._handle_error_response(response)

        data = response.json()
        meta = data.get("meta", {})

        return PaginatedResponse(
            data=data.get("data", []),
            page=meta.get("page", page),
            per_page=meta.get("per_page", per_page),
            total=meta.get("total", 0),
            total_pages=meta.get("total_pages", 0)
        )

    def search_tools(
        self,
        query: str,
        category: Optional[str] = None,
        page: int = 1,
        per_page: int = 20,
        sort: str = "relevance"
    ) -> PaginatedResponse:
        """
        Search for tools in the registry.

        Args:
            query: Search query
            category: Filter by category
            page: Page number
            per_page: Items per page
            sort: Sort field (relevance, downloads, published_at)

        Returns:
            PaginatedResponse with matching tools
        """
        params = {
            "q": query,
            "page": page,
            "per_page": min(per_page, 100),
            "sort": sort
        }
        if category:
            params["category"] = category

        response = self._request("GET", "/tools/search", params=params)

        if response.status_code != 200:
            self._handle_error_response(response)

        data = response.json()
        meta = data.get("meta", {})

        return PaginatedResponse(
            data=data.get("data", []),
            page=meta.get("page", page),
            per_page=meta.get("per_page", per_page),
            total=meta.get("total", 0),
            total_pages=meta.get("total_pages", 0)
        )

    def get_tool(self, owner: str, name: str) -> ToolInfo:
        """
        Get detailed information about a tool.

        Args:
            owner: Tool owner (namespace)
            name: Tool name

        Returns:
            ToolInfo object
        """
        response = self._request("GET", f"/tools/{owner}/{name}")

        if response.status_code == 404:
            raise RegistryError(
                code="TOOL_NOT_FOUND",
                message=f"Tool '{owner}/{name}' not found",
                http_status=404
            )

        if response.status_code != 200:
            self._handle_error_response(response)

        data = response.json().get("data", {})
        return ToolInfo.from_dict(data)

    def get_tool_versions(self, owner: str, name: str) -> List[str]:
        """
        Get all versions of a tool.

        Args:
            owner: Tool owner
            name: Tool name

        Returns:
            List of version strings (sorted newest first)
        """
        response = self._request("GET", f"/tools/{owner}/{name}/versions")

        if response.status_code == 404:
            raise RegistryError(
                code="TOOL_NOT_FOUND",
                message=f"Tool '{owner}/{name}' not found",
                http_status=404
            )

        if response.status_code != 200:
            self._handle_error_response(response)

        data = response.json()
        return data.get("data", {}).get("versions", [])

    def download_tool(
        self,
        owner: str,
        name: str,
        version: Optional[str] = None,
        install: bool = True
    ) -> DownloadResult:
        """
        Download a tool's configuration.

        Args:
            owner: Tool owner
            name: Tool name
            version: Version or constraint (default: latest)
            install: Whether to count as install for stats

        Returns:
            DownloadResult with config YAML
        """
        params = {"install": str(install).lower()}
        if version:
            params["version"] = version

        response = self._request(
            "GET",
            f"/tools/{owner}/{name}/download",
            params=params
        )

        if response.status_code == 404:
            error_data = {}
            try:
                error_data = response.json().get("error", {})
            except json.JSONDecodeError:
                pass

            code = error_data.get("code", "TOOL_NOT_FOUND")
            message = error_data.get("message", f"Tool '{owner}/{name}' not found")

            raise RegistryError(
                code=code,
                message=message,
                details=error_data.get("details"),
                http_status=404
            )

        if response.status_code != 200:
            self._handle_error_response(response)

        data = response.json().get("data", {})

        return DownloadResult(
            owner=data.get("owner", owner),
            name=data.get("name", name),
            resolved_version=data.get("resolved_version", ""),
            config_yaml=data.get("config", ""),
            readme=data.get("readme", "")
        )

    def get_categories(self) -> List[Dict[str, Any]]:
        """
        Get list of tool categories.

        Returns:
            List of category dicts with name, description, icon
        """
        response = self._request("GET", "/categories")

        if response.status_code != 200:
            self._handle_error_response(response)

        return response.json().get("data", [])

    def publish_tool(
        self,
        config_yaml: str,
        readme: str = "",
        dry_run: bool = False
    ) -> Dict[str, Any]:
        """
        Publish a tool to the registry.

        Args:
            config_yaml: Tool configuration YAML content
            readme: README.md content
            dry_run: If True, validate without publishing

        Returns:
            Dict with PR URL or validation results
        """
        payload = {
            "config": config_yaml,
            "readme": readme,
            "dry_run": dry_run
        }

        response = self._request(
            "POST",
            "/tools",
            json_data=payload,
            require_auth=True
        )

        if response.status_code == 409:
            # Version already exists
            self._handle_error_response(response)

        if response.status_code not in (200, 201):
            self._handle_error_response(response)

        return response.json().get("data", {})

    def get_my_tools(self) -> List[ToolInfo]:
        """
        Get tools published by the authenticated user.

        Returns:
            List of ToolInfo objects
        """
        response = self._request("GET", "/me/tools", require_auth=True)

        if response.status_code != 200:
            self._handle_error_response(response)

        tools = response.json().get("data", [])
        return [ToolInfo.from_dict(t) for t in tools]

    def get_popular_tools(self, limit: int = 10) -> List[ToolInfo]:
        """
        Get most popular tools.

        Args:
            limit: Maximum number of tools to return

        Returns:
            List of ToolInfo objects
        """
        response = self._request(
            "GET",
            "/stats/popular",
            params={"limit": limit}
        )

        if response.status_code != 200:
            self._handle_error_response(response)

        tools = response.json().get("data", [])
        return [ToolInfo.from_dict(t) for t in tools]

    # -------------------------------------------------------------------------
    # Index Caching
    # -------------------------------------------------------------------------

    def get_index(self, force_refresh: bool = False) -> Dict[str, Any]:
        """
        Get the full tool index, using cache when possible.

        Args:
            force_refresh: Force refresh from server

        Returns:
            Index dict with tools list
        """
        # Check cache first
        if not force_refresh:
            cached = self._load_cached_index()
            if cached:
                return cached

        # Fetch from server
        etag = self._get_cached_etag()
        response = self._request("GET", "/index.json", etag=etag)

        if response.status_code == 304:
            # Not modified, use cache
            cached = self._load_cached_index()
            if cached:
                return cached

        if response.status_code != 200:
            # Try to use stale cache on error
            cached = self._load_cached_index(ignore_age=True)
            if cached:
                return cached
            self._handle_error_response(response)

        data = response.json()

        # Cache the response
        new_etag = response.headers.get("ETag")
        self._save_cached_index(data, new_etag)

        return data

    def _load_cached_index(self, ignore_age: bool = False) -> Optional[Dict]:
        """Load cached index if valid."""
        if not INDEX_CACHE_FILE.exists():
            return None

        try:
            cache_data = json.loads(INDEX_CACHE_FILE.read_text())

            # Check age
            if not ignore_age:
                cached_at = datetime.fromisoformat(cache_data.get("_cached_at", ""))
                if datetime.now() - cached_at > INDEX_CACHE_MAX_AGE:
                    return None

            # Verify checksum
            if not self._verify_index_checksum(cache_data):
                return None

            return cache_data

        except (json.JSONDecodeError, KeyError, ValueError):
            return None

    def _save_cached_index(self, data: Dict, etag: Optional[str] = None) -> None:
        """Save index to cache."""
        CACHE_DIR.mkdir(parents=True, exist_ok=True)

        data["_cached_at"] = datetime.now().isoformat()
        if etag:
            data["_etag"] = etag

        INDEX_CACHE_FILE.write_text(json.dumps(data, indent=2))

    def _get_cached_etag(self) -> Optional[str]:
        """Get ETag from cached index."""
        if not INDEX_CACHE_FILE.exists():
            return None

        try:
            cache_data = json.loads(INDEX_CACHE_FILE.read_text())
            return cache_data.get("_etag")
        except (json.JSONDecodeError, KeyError):
            return None

    def _verify_index_checksum(self, data: Dict) -> bool:
        """Verify cached index integrity."""
        checksum = data.get("checksum", "")
        if not checksum:
            return True  # No checksum to verify

        # Compute checksum of tools list
        tools = data.get("tools", [])
        content = json.dumps(tools, sort_keys=True)
        computed = "sha256:" + hashlib.sha256(content.encode()).hexdigest()

        return computed == checksum

    def clear_cache(self) -> None:
        """Clear the local index cache."""
        if INDEX_CACHE_FILE.exists():
            INDEX_CACHE_FILE.unlink()


# -------------------------------------------------------------------------
# Convenience functions
# -------------------------------------------------------------------------

def get_client() -> RegistryClient:
    """Get a configured registry client instance."""
    return RegistryClient()


def search(query: str, **kwargs) -> PaginatedResponse:
    """Search the registry for tools."""
    return get_client().search_tools(query, **kwargs)


def install_tool(tool_spec: str, version: Optional[str] = None) -> DownloadResult:
    """
    Download a tool for installation.

    Args:
        tool_spec: Tool specification (owner/name or just name)
        version: Version constraint

    Returns:
        DownloadResult with config YAML
    """
    client = get_client()

    # Parse tool spec
    if "/" in tool_spec:
        owner, name = tool_spec.split("/", 1)
    else:
        # Shorthand - try official namespace first
        owner = "official"
        name = tool_spec

    try:
        return client.download_tool(owner, name, version=version, install=True)
    except RegistryError as e:
        if e.code == "TOOL_NOT_FOUND" and owner == "official":
            # Fall back to searching for most popular tool with this name
            results = client.search_tools(name, per_page=1)
            if results.data:
                first = results.data[0]
                return client.download_tool(
                    first["owner"],
                    first["name"],
                    version=version,
                    install=True
                )
        raise