Medios-Macina/ProviderCore/registry.py

"""Provider registry.

Concrete provider implementations live in the `Provider/` package.
This module is the single source of truth for provider discovery.
"""

from __future__ import annotations

from typing import Any, Dict, Optional, Sequence, Type
import sys
from urllib.parse import urlparse

from SYS.logger import log

from ProviderCore.base import Provider, SearchProvider, FileProvider, SearchResult
from Provider.alldebrid import AllDebrid
from Provider.bandcamp import Bandcamp
from Provider.libgen import Libgen
from Provider.matrix import Matrix
from Provider.openlibrary import OpenLibrary
from Provider.soulseek import Soulseek, download_soulseek_file
from Provider.telegram import Telegram
from Provider.youtube import YouTube
from Provider.fileio import FileIO
from Provider.zeroxzero import ZeroXZero
from Provider.loc import LOC
from Provider.internetarchive import InternetArchive
from Provider.podcastindex import PodcastIndex
from Provider.HIFI import HIFI

_PROVIDERS: Dict[str,
                 Type[Provider]] = {
                     # Search-capable providers
                     "alldebrid": AllDebrid,
                     "libgen": Libgen,
                     "openlibrary": OpenLibrary,
                     "internetarchive": InternetArchive,
                    "hifi": HIFI,
                     "soulseek": Soulseek,
                     "bandcamp": Bandcamp,
                     "youtube": YouTube,
                     "telegram": Telegram,
                     "loc": LOC,
                     "podcastindex": PodcastIndex,
                     # Upload-capable providers
                     "0x0": ZeroXZero,
                     "file.io": FileIO,
                     "matrix": Matrix,
                 }


def is_known_provider_name(name: str) -> bool:
    """Return True if `name` matches a registered provider key.

    This is intentionally cheap (no imports/instantiation) so callers can
    probe UI strings (table names, store names, etc.) without triggering
    noisy 'Unknown provider' logs.
    """

    return (name or "").strip().lower() in _PROVIDERS


def _supports_search(provider: Provider) -> bool:
    return provider.__class__.search is not Provider.search


def _supports_upload(provider: Provider) -> bool:
    return provider.__class__.upload is not Provider.upload


def get_provider(name: str,
                 config: Optional[Dict[str,
                                       Any]] = None) -> Optional[Provider]:
    """Get a provider by name (unified registry)."""

    provider_class = _PROVIDERS.get((name or "").lower())
    if provider_class is None:
        log(f"[provider] Unknown provider: {name}", file=sys.stderr)
        return None

    try:
        provider = provider_class(config)
        if not provider.validate():
            log(f"[provider] Provider '{name}' is not available", file=sys.stderr)
            return None
        return provider
    except Exception as exc:
        log(f"[provider] Error initializing '{name}': {exc}", file=sys.stderr)
        return None


def list_providers(config: Optional[Dict[str, Any]] = None) -> Dict[str, bool]:
    """List all providers and their availability."""

    availability: Dict[str,
                       bool] = {}
    for name, provider_class in _PROVIDERS.items():
        try:
            provider = provider_class(config)
            availability[name] = provider.validate()
        except Exception:
            availability[name] = False
    return availability


def get_search_provider(name: str,
                        config: Optional[Dict[str,
                                              Any]] = None) -> Optional[SearchProvider]:
    """Get a search-capable provider by name (compat API)."""

    provider = get_provider(name, config)
    if provider is None:
        return None
    if not _supports_search(provider):
        log(f"[provider] Provider '{name}' does not support search", file=sys.stderr)
        return None
    return provider  # type: ignore[return-value]


def list_search_providers(config: Optional[Dict[str, Any]] = None) -> Dict[str, bool]:
    """List all search providers and their availability."""

    availability: Dict[str,
                       bool] = {}
    for name, provider_class in _PROVIDERS.items():
        try:
            provider = provider_class(config)
            availability[name] = bool(
                provider.validate() and _supports_search(provider)
            )
        except Exception:
            availability[name] = False
    return availability


def get_file_provider(name: str,
                      config: Optional[Dict[str,
                                            Any]] = None) -> Optional[FileProvider]:
    """Get an upload-capable provider by name (compat API)."""

    provider = get_provider(name, config)
    if provider is None:
        return None
    if not _supports_upload(provider):
        log(f"[provider] Provider '{name}' does not support upload", file=sys.stderr)
        return None
    return provider  # type: ignore[return-value]


def list_file_providers(config: Optional[Dict[str, Any]] = None) -> Dict[str, bool]:
    """List all file providers and their availability."""

    availability: Dict[str,
                       bool] = {}
    for name, provider_class in _PROVIDERS.items():
        try:
            provider = provider_class(config)
            availability[name] = bool(
                provider.validate() and _supports_upload(provider)
            )
        except Exception:
            availability[name] = False
    return availability


def match_provider_name_for_url(url: str) -> Optional[str]:
    """Return a registered provider name that claims the URL's domain.

    Providers can declare domains via a class attribute `URL_DOMAINS` (sequence of strings).
    This matcher is intentionally cheap (no provider instantiation, no network).
    """

    try:
        parsed = urlparse(str(url))
        host = (parsed.hostname or "").strip().lower()
        path = (parsed.path or "").strip()
    except Exception:
        host = ""
        path = ""

    if not host:
        return None

    # Prefer Internet Archive for archive.org links unless the URL clearly refers
    # to a borrow/loan flow (handled by OpenLibrary provider).
    #
    # This keeps direct downloads and item pages routed to `internetarchive`, while
    # preserving OpenLibrary's scripted borrow pipeline for loan/reader URLs.
    if host == "openlibrary.org" or host.endswith(".openlibrary.org"):
        return "openlibrary" if "openlibrary" in _PROVIDERS else None

    if host == "archive.org" or host.endswith(".archive.org"):
        low_path = str(path or "").lower()
        is_borrowish = (
            low_path.startswith("/borrow/") or low_path.startswith("/stream/")
            or low_path.startswith("/services/loans/") or "/services/loans/" in low_path
        )
        if is_borrowish:
            return "openlibrary" if "openlibrary" in _PROVIDERS else None
        return "internetarchive" if "internetarchive" in _PROVIDERS else None

    for name, provider_class in _PROVIDERS.items():
        domains = getattr(provider_class, "URL_DOMAINS", None)
        if not isinstance(domains, (list, tuple)):
            continue
        for d in domains:
            dom = str(d or "").strip().lower()
            if not dom:
                continue
            if host == dom or host.endswith("." + dom):
                return name

    return None


def get_provider_for_url(url: str,
                         config: Optional[Dict[str,
                                               Any]] = None) -> Optional[Provider]:
    """Instantiate and return the matching provider for a URL, if any."""

    name = match_provider_name_for_url(url)
    if not name:
        return None
    return get_provider(name, config)


__all__ = [
    "SearchResult",
    "Provider",
    "SearchProvider",
    "FileProvider",
    "get_provider",
    "list_providers",
    "get_search_provider",
    "list_search_providers",
    "get_file_provider",
    "list_file_providers",
    "match_provider_name_for_url",
    "get_provider_for_url",
    "download_soulseek_file",
]