feat: removed children from collections after merging

music_kraken/__init__.py

@@ -46,7 +46,7 @@ init_logging()
 from . import cli
 
 if DEBUG:
-    sys.setrecursionlimit(500)
+    sys.setrecursionlimit(300)
 
 
 if main_settings['modify_gc']:

music_kraken/objects/collection.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from collections import defaultdict
-from typing import TypeVar, Generic, Dict, Optional, Iterable, List, Iterator, Tuple
+from typing import TypeVar, Generic, Dict, Optional, Iterable, List, Iterator, Tuple, Generator
 from .parents import OuterProxy
 
 T = TypeVar('T', bound=OuterProxy)
@@ -35,39 +35,24 @@ class Collection(Generic[T]):
         # Key: collection attribute (str) of appended element
         # Value: main collection to sync to
         self.contain_given_in_attribute: Dict[str, Collection] = contain_given_in_attribute or {}
-        self.contain_attribute_in_given: Dict[str, Collection] = contain_attribute_in_given or {}
         self.append_object_to_attribute: Dict[str, T] = append_object_to_attribute or {}
 
-        self.contain_self_on_append: List[str] = []
-
         self._indexed_values = defaultdict(set)
         self._indexed_to_objects = defaultdict(list)
 
         self.extend(data)
 
     def _map_element(self, __object: T, from_map: bool = False):
-        if __object.id in self._contains_ids:
-            return
-
+        __object._inner._mapped_in_collection.add(self)
         self._contains_ids.add(__object.id)
 
         for name, value in __object.indexing_values:
-            if value is None:
+            if value is None or value == __object._inner._default_values.get(name):
                 continue
 
             self._indexed_values[name].add(value)
             self._indexed_to_objects[value].append(__object)
 
-        if not from_map:
-            for attribute, new_object in self.contain_given_in_attribute.items():
-                __object.__getattribute__(attribute).contain_collection_inside(new_object)
-
-            for attribute, new_object in self.contain_attribute_in_given.items():
-                new_object.contain_collection_inside(__object.__getattribute__(attribute))
-
-            for attribute, new_object in self.append_object_to_attribute.items():
-                __object.__getattribute__(attribute).append(new_object)
-
     def _unmap_element(self, __object: T):
         if __object.id in self._contains_ids:
             self._contains_ids.remove(__object.id)
@@ -186,15 +171,8 @@ class Collection(Generic[T]):
     def contains(self, __object: T) -> bool:
         return len(self._contained_in_sub(__object)) > 0
 
-    def _append(self, __object: T, from_map: bool = False):
-        print(self, __object)
-        self._map_element(__object, from_map=from_map)
-        self._data.append(__object)
-
     def _find_object_in_self(self, __object: T) -> Optional[T]:
         for name, value in __object.indexing_values:
-            if value is None or value == __object._default_factories.get(name, lambda: None)():
-                continue
             if value in self._indexed_values[name]:
                 return self._indexed_to_objects[value][0]
 
@@ -211,6 +189,7 @@ class Collection(Generic[T]):
         if no_sibling:
             return self, None
 
+        """
         # find in siblings and all children of siblings
         for parent in self.parents:
             for sibling in parent.children:
@@ -220,6 +199,7 @@ class Collection(Generic[T]):
                 o, other_object = sibling._find_object(__object, no_sibling=True)
                 if other_object is not None:
                     return o, other_object
+        """
 
         return self, None
 
@@ -235,7 +215,7 @@ class Collection(Generic[T]):
         :return:
         """
 
-        if __object is None or __object.id in self._contains_ids:
+        if __object is None:
             return
 
         append_to, existing_object = self._find_object(__object)
@@ -243,54 +223,47 @@ class Collection(Generic[T]):
         if existing_object is None:
             # append
             append_to._data.append(__object)
-            append_to._map_element(__object, from_map=from_map)
+            append_to._map_element(__object)
+
+            # only modify collections if the object actually has been appended
+            for collection_attribute, child_collection in self.contain_given_in_attribute.items():
+                __object.__getattribute__(collection_attribute).contain_collection_inside(child_collection, __object)
+
+            for attribute, new_object in self.append_object_to_attribute.items():
+                __object.__getattribute__(attribute).append(new_object)
         else:
-            # merge
+            # merge only if the two objects are not the same
+            if existing_object.id == __object.id:
+                return
+
             append_to._unmap_element(existing_object)
             existing_object.merge(__object)
-            append_to._map_element(existing_object, from_map=from_map)
+            append_to._map_element(existing_object)
 
 
-    def extend(self, __iterable: Optional[Iterable[T]], from_map: bool = False):
+    def extend(self, __iterable: Optional[Generator[T, None, None]]):
         if __iterable is None:
             return
 
         for __object in __iterable:
-            self.append(__object, from_map=from_map)
+            self.append(__object)
 
-    def sync_with_other_collection(self, equal_collection: Collection):
-        """
-        If two collections always need to have the same values, this can be used.
-
-        Internally:
-        1. import the data from other to self
-            - _data
-            - contained_collections
-        2. replace all refs from the other object, with refs from this object
-        """
-        if equal_collection is self:
-            return
-
-        # don't add the elements from the subelements from the other collection.
-        # this will be done in the next step.
-        self.extend(equal_collection._data)
-        # add all submodules
-        for equal_sub_collection in equal_collection.children:
-            self.contain_collection_inside(equal_sub_collection)
-
-    def contain_collection_inside(self, sub_collection: Collection):
+    def contain_collection_inside(self, sub_collection: Collection, _object: T):
         """
         This collection will ALWAYS contain everything from the passed in collection
         """
         if self is sub_collection or sub_collection in self.children:
             return
 
+        _object._inner._is_collection_child[self] = sub_collection
+        _object._inner._is_collection_parent[sub_collection] = self
+
         self.children.append(sub_collection)
         sub_collection.parents.append(self)
 
     @property
     def data(self) -> List[T]:
-        return list(i for i in self.__iter__())
+        return list(self.__iter__())
 
     def __len__(self) -> int:
         return len(self._data) + sum(len(collection) for collection in self.children)
@@ -312,7 +285,7 @@ class Collection(Generic[T]):
             yield from c.__iter__(finished_ids=finished_ids)
 
     def __merge__(self, __other: Collection, override: bool = False):
-        self.extend(__other._data, from_map=True)
+        self.extend(__other.__iter__())
 
     def __getitem__(self, item: int):
         if item < len(self._data):

music_kraken/objects/parents.py

@@ -29,12 +29,16 @@ class InnerData:
     _refers_to_instances: set = None
 
     def __init__(self, object_type, **kwargs):
-        self._refers_to_instances =set()
+        self._refers_to_instances = set()
+
+        # collection : collection that is a collection of self
+        self._is_collection_child: Dict[Collection, Collection] = {}
+        self._is_collection_parent: Dict[Collection, Collection] = {}
         
         # initialize the default values
-        self.__default_values = {}
+        self._default_values = {}
         for name, factory in object_type._default_factories.items():
-            self.__default_values[name] = factory()
+            self._default_values[name] = factory()
 
         for key, value in kwargs.items():
             self.__setattr__(key, value)
@@ -48,7 +52,7 @@ class InnerData:
 
         for key, value in __other.__dict__.copy().items():
             # 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)):
+            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
 
@@ -183,7 +187,7 @@ class OuterProxy:
         if __other is None:
             return
 
-        object_trace(f"merging {type(self).__name__} [{self.title_string}] with {type(__other).__name__} [{__other.title_string}]")
+        object_trace(f"merging {type(self).__name__} [{self.title_string} | {self.id}] with {type(__other).__name__} [{__other.title_string} | {__other.id}]")
 
         a = self
         b = __other
@@ -196,11 +200,20 @@ class OuterProxy:
             a, b = b, a
 
         a._inner.__merge__(b._inner, override=override)
+        for collection, child_collection in b._inner._is_collection_child.items():
+            collection.children.remove(child_collection)
+            
+        for collection, parent_collection in b._inner._is_collection_parent.items():
+            collection.parents.remove(parent_collection)
+
         a._inner._refers_to_instances.update(b._inner._refers_to_instances)
 
         for instance in b._inner._refers_to_instances:
             instance._inner = a._inner
 
+    def __merge__(self, __other: Optional[OuterProxy], override: bool = False):
+        self.merge(__other, override)
+
     def mark_as_fetched(self, *url_hash_list: List[str]):
         for url_hash in url_hash_list:
             self._fetched_from[url_hash] = {

music_kraken/objects/song.py

@@ -49,6 +49,7 @@ class Song(Base):
     source_collection: SourceCollection
     target_collection: Collection[Target]
     lyrics_collection: Collection[Lyrics]
+
     main_artist_collection: Collection[Artist]
     feature_artist_collection: Collection[Artist]
     album_collection: Collection[Album]
@@ -241,13 +242,17 @@ class Album(Base):
     UPWARDS_COLLECTION_STRING_ATTRIBUTES = ("artist_collection", "label_collection")
 
     def __init_collections__(self):
-        self.song_collection.contain_attribute_in_given = {
-            "main_artist_collection": self.artist_collection
-        }
         self.song_collection.append_object_to_attribute = {
             "album_collection": self
         }
 
+        self.artist_collection.append_object_to_attribute = {
+            "main_album_collection": self
+        }
+        self.artist_collection.contain_given_in_attribute = {
+            "label_collection": self.label_collection
+        }
+
     def _add_other_db_objects(self, object_type: Type[OuterProxy], object_list: List[OuterProxy]):
         if object_type is Song:
             self.song_collection.extend(object_list)
@@ -642,6 +647,15 @@ class Label(Base):
                          contact_list=contact_list, album_list=album_list, current_artist_list=current_artist_list,
                          **kwargs)
 
+    def __init_collections__(self):
+        self.album_collection.append_object_to_attribute = {
+            "label_collection": self
+        }
+
+        self.current_artist_collection.append_object_to_attribute = {
+            "label_collection": self
+        }
+
     @property
     def indexing_values(self) -> List[Tuple[str, object]]:
         return [

music_kraken/pages/musify.py

@@ -1,7 +1,7 @@
 from collections import defaultdict
 from dataclasses import dataclass
 from enum import Enum
-from typing import List, Optional, Type, Union
+from typing import List, Optional, Type, Union, Generator
 from urllib.parse import urlparse
 
 import pycountry
@@ -1056,7 +1056,7 @@ class Musify(Page):
             date=date
         )
 
-    def _get_discography(self, url: MusifyUrl, artist_name: str = None, stop_at_level: int = 1) -> List[Album]:
+    def _get_discography(self, url: MusifyUrl, artist_name: str = None, stop_at_level: int = 1) -> Generator[Album, None, None]:
         """
         POST https://musify.club/artist/filteralbums
         ArtistID: 280348
@@ -1077,18 +1077,8 @@ class Musify(Page):
             return []
         soup: BeautifulSoup = BeautifulSoup(r.content, features="html.parser")
 
-        discography: List[Album] = []
         for card_soup in soup.find_all("div", {"class": "card"}):
-            new_album: Album = self._parse_album_card(card_soup, artist_name)
-            album_source: Source
-            
-            if stop_at_level > 1:
-                for album_source in new_album.source_collection.get_sources_from_page(self.SOURCE_TYPE):
-                    new_album.merge(self.fetch_album(album_source, stop_at_level=stop_at_level-1))
-            
-            discography.append(new_album)
-
-        return discography
+            yield self._parse_album_card(card_soup, artist_name)
 
     def fetch_artist(self, source: Source, stop_at_level: int = 1) -> Artist:
         """
@@ -1110,8 +1100,7 @@ class Musify(Page):
 
         artist = self._get_artist_attributes(url)
 
-        discography: List[Album] = self._get_discography(url, artist.name)
-        artist.main_album_collection.extend(discography)
+        artist.main_album_collection.extend(self._get_discography(url, artist.name))
         
         return artist
 

music_kraken/utils/shared.py

@@ -15,7 +15,7 @@ __stage__ = os.getenv("STAGE", "prod")
 DEBUG = (__stage__ == "dev") and True
 DEBUG_LOGGING = DEBUG and True
 DEBUG_TRACE = DEBUG and True
-DEBUG_OBJECT_TRACE = DEBUG and False
+DEBUG_OBJECT_TRACE = DEBUG and True
 DEBUG_YOUTUBE_INITIALIZING = DEBUG and False
 DEBUG_PAGES = DEBUG and False
 DEBUG_DUMP = DEBUG and True