from __future__ import annotations import random from collections import defaultdict from functools import lru_cache from typing import Optional, Dict, Tuple, List, Type, Generic, Any, TypeVar, Set from pathlib import Path import inspect from .metadata import Metadata from ..utils import get_unix_time, object_trace, generate_id from ..utils.config import logging_settings, main_settings from ..utils.shared import HIGHEST_ID, DEBUG_PRINT_ID from ..utils.hacking import MetaClass LOGGER = logging_settings["object_logger"] P = TypeVar("P", bound="OuterProxy") class InnerData: """ This is the core class, which is used for every Data class. The attributes are set, and can be merged. The concept is, that the outer class proxies this class. If the data in the wrapper class has to be merged, then this class is just replaced and garbage collected. """ _refers_to_instances: set = None _is_in_collection: set = None _has_data: bool = False """ Attribute versions keep track, of if the attribute has been changed. """ def __init__(self, object_type, **kwargs): self._refers_to_instances = set() self._is_in_collection = set() self._fetched_from: dict = {} # initialize the default values self._default_values = {} for name, factory in object_type._default_factories.items(): self._default_values[name] = factory() for key, value in kwargs.items(): if hasattr(value, "__is_collection__"): value._collection_for[self] = key self.__setattr__(key, value) if self._has_data: continue def __setattr__(self, key: str, value): if self._has_data or not hasattr(self, "_default_values"): return super().__setattr__(key, value) super().__setattr__("_has_data", not (key in self._default_values and self._default_values[key] == value)) return super().__setattr__(key, value) def __hash__(self): return self.id def __merge__(self, __other: InnerData, **kwargs): """ :param __other: :return: """ self._fetched_from.update(__other._fetched_from) self._is_in_collection.update(__other._is_in_collection) for key, value in __other.__dict__.copy().items(): if key.startswith("_"): continue if hasattr(value, "__is_collection__") and key in self.__dict__: self.__getattribute__(key).__merge__(value, **kwargs) continue # just set the other value if self doesn't already have it if key not in self.__dict__ or (key in self.__dict__ and self.__dict__[key] == self._default_values.get(key)): self.__setattr__(key, value) continue # if the object of value implemented __merge__, it merges existing = self.__getattribute__(key) if hasattr(existing, "__merge__"): existing.__merge__(value, **kwargs) class OuterProxy: """ Wraps the inner data, and provides apis, to naturally access those values. """ _default_factories: dict = {} _outer_attribute: Set[str] = {"options", "metadata", "indexing_values", "option_string"} DOWNWARDS_COLLECTION_STRING_ATTRIBUTES = tuple() UPWARDS_COLLECTION_STRING_ATTRIBUTES = tuple() def __init__(self, _id: int = None, dynamic: bool = False, **kwargs): _automatic_id: bool = False if _id is None: """ generates a random integer id the range is defined in the config """ _id = generate_id() _automatic_id = True kwargs["automatic_id"] = _automatic_id kwargs["id"] = _id kwargs["dynamic"] = dynamic for name, factory in type(self)._default_factories.items(): if kwargs.get(name, None) is None: kwargs[name] = factory() collection_data: Dict[str, list] = {} for name, value in kwargs.copy().items(): if isinstance(value, list) and name.endswith("_list"): collection_name = name.replace("_list", "_collection") collection_data[collection_name] = value del kwargs[name] self._inner: InnerData = InnerData(type(self), **kwargs) self._inner._refers_to_instances.add(self) object_trace(f"creating {type(self).__name__} [{self.option_string}]") self.__init_collections__() for name, data_list in collection_data.items(): collection = self._inner.__getattribute__(name) collection.extend(data_list) self._inner.__setattr__(name, collection) def __init_collections__(self): pass def __getattribute__(self, __name: str) -> Any: """ Returns the attribute of _inner if the attribute exists, else it returns the attribute of self. That the _inner gets checked first is essential for the type hints. :param __name: :return: """ if __name.startswith("_") or __name in self._outer_attribute or __name.isupper(): return object.__getattribute__(self, __name) _inner: InnerData = super().__getattribute__("_inner") try: return _inner.__getattribute__(__name) except AttributeError: return super().__getattribute__(__name) def __setattr__(self, __name, __value): if not __name.startswith("_") and hasattr(self, "_inner"): _inner: InnerData = super().__getattribute__("_inner") return _inner.__setattr__(__name, __value) return super().__setattr__(__name, __value) def _add_other_db_objects(self, object_type: Type[OuterProxy], object_list: List[OuterProxy]): pass def add_list_of_other_objects(self, object_list: List[OuterProxy]): d: Dict[Type[OuterProxy], List[OuterProxy]] = defaultdict(list) for db_object in object_list: d[type(db_object)].append(db_object) for key, value in d.items(): self._add_other_db_objects(key, value) def __hash__(self): return id(self) def __eq__(self, other: Any): return self.__hash__() == other.__hash__() def merge(self, __other: Optional[OuterProxy], **kwargs): """ 1. merges the data of __other in self 2. replaces the data of __other with the data of self :param __other: :return: """ if __other is None: return a = self b = __other if a.id == b.id: return # switch instances if more efficient if len(b._inner._refers_to_instances) > len(a._inner._refers_to_instances): a, b = b, a object_trace(f"merging {a.option_string} | {b.option_string}") old_inner = b._inner for instance in b._inner._refers_to_instances.copy(): instance._inner = a._inner a._inner._refers_to_instances.add(instance) a._inner.__merge__(old_inner, **kwargs) del old_inner def __merge__(self, __other: Optional[OuterProxy], **kwargs): self.merge(__other, **kwargs) def mark_as_fetched(self, *url_hash_list: List[str]): for url_hash in url_hash_list: self._inner._fetched_from[url_hash] = { "time": get_unix_time(), "url": url_hash, } def already_fetched_from(self, url_hash: str) -> bool: res = self._inner._fetched_from.get(url_hash, None) if res is None: return False return get_unix_time() - res["time"] < main_settings["refresh_after"] @property def metadata(self) -> Metadata: """ This is an interface. :return: """ return Metadata() @property def options(self) -> List[P]: r = [] for collection_string_attribute in self.UPWARDS_COLLECTION_STRING_ATTRIBUTES: r.extend(self.__getattribute__(collection_string_attribute)) r.append(self) for collection_string_attribute in self.DOWNWARDS_COLLECTION_STRING_ATTRIBUTES: r.extend(self.__getattribute__(collection_string_attribute)) return r @property def option_string(self) -> str: return self.title_string INDEX_DEPENDS_ON: List[str] = [] @property def indexing_values(self) -> List[Tuple[str, object]]: """ This is an interface. It is supposed to return a map of the name and values for all important attributes. This helps in comparing classes for equal data (e.g. being the same song but different attributes) TODO Rewrite this approach into a approach, that is centered around statistics, and not binaries. Instead of: one of this matches, it is the same This: If enough attributes are similar enough, they are the same Returns: List[Tuple[str, object]]: the first element in the tuple is the name of the attribute, the second the value. """ return [] @property @lru_cache() def all_collections(self): r = [] for key in self._default_factories: val = self._inner.__getattribute__(key) if hasattr(val, "__is_collection__"): r.append(val) return r TITEL = "id" @property def title_string(self) -> str: return str(self.__getattribute__(self.TITEL)) + (f" {self.id}" if DEBUG_PRINT_ID else "") def __repr__(self): return f"{type(self).__name__}({self.title_string})"