refactor 2

This commit is contained in:
Шурупов Илья Викторович 2026-01-04 23:40:41 +03:00
parent 72af96eada
commit 39c88564ba
7 changed files with 426 additions and 503 deletions

View file

@ -1,9 +1,10 @@
from datetime import datetime
from src.Library import *
from typing import Any, Iterable
from pathlib import Path
import json
from src.Library import *
from datetime import datetime
from pathlib import Path
def get_artists_str(artists_pack):
@ -11,129 +12,117 @@ def get_artists_str(artists_pack):
artists = " ".join(track_artists)
return artists
class Parser:
def __init__(self):
# state: Spotify track ID -> Track object
self.track_map: dict[str, Track] = {}
self.artist_map: dict[str, Artist] = {}
self.album_map: dict[str, Album] = {}
self.library = Library()
# -------------------------
# Low-level helpers
# -------------------------
@staticmethod
def _parse_spotify_datetime(value: str | None) -> datetime | None:
if not value:
return None
return datetime.fromisoformat(value.replace("Z", "+00:00"))
def parse(self, playlists: Path, tracks: Path, top_tracks: Path, top_artists: Path) -> Library:
self.library = Library()
@staticmethod
def _parse_spotify_artist(a: dict[str, Any]) -> SpotifyArtist:
return SpotifyArtist(
id=a["id"],
name=a["name"],
uri=a["uri"],
url=a["external_urls"]["spotify"],
)
self.parse_tracks(json.loads(tracks.read_text()))
self.parse_playlists(json.loads(playlists.read_text()))
@staticmethod
def _parse_spotify_album(a: dict[str, Any]) -> SpotifyAlbum:
return SpotifyAlbum(
id=a["id"],
name=a["name"],
release_date=a.get("release_date"),
total_tracks=a.get("total_tracks", 0),
uri=a["uri"],
url=a["external_urls"]["spotify"],
)
self.parse_top_artists(json.loads(top_artists.read_text()))
self.parse_top_tracks(json.loads(top_tracks.read_text()))
def _parse_spotify_data(self, track_json: dict[str, Any]) -> SpotifyData:
return SpotifyData(
id=track_json["id"],
uri=track_json["uri"],
url=track_json["external_urls"]["spotify"],
href=track_json["href"],
popularity=track_json.get("popularity"),
is_playable=track_json.get("is_playable"),
is_local=track_json.get("is_local"),
external_ids=track_json.get("external_ids", {}),
artists=[self._parse_spotify_artist(a) for a in track_json.get("artists", [])],
album=self._parse_spotify_album(track_json["album"])
if track_json.get("album")
else None,
)
self.add_tracks_to_albums_and_artists()
# -------------------------
# Track parsers
# -------------------------
def parse_track_item(self, item: dict[str, Any]) -> Track:
"""
Parses a single track item and returns Track object.
Reuses existing Track from track_map if already parsed.
"""
t = item["track"]
track_id = t["id"]
return self.library
def add_tracks_to_albums_and_artists(self):
for track_id, track in self.track_map.items():
for artist in track.artists:
if track not in artist.tracks:
artist.tracks.append(track)
if track not in track.album.tracks:
track.album.tracks.append(track)
def parse_playlists(self, data):
for name, tracks in data.items():
playlist = Playlist()
playlist.title = name
for track in tracks:
track_data = track.get("track")
if track_data:
track_id = track_data.get("id")
playlist.tracks.append(self.parse_track(track_data, track.get("added_at")))
self.library.playlists.append(playlist)
def parse_top_tracks(self, data):
for track in data:
if track.get("id") in self.track_map:
self.library.top_tracks.append(self.track_map.get(track.get("id")))
def parse_top_artists(self, data):
for artist in data:
if artist.get("id") in self.artist_map:
self.library.top_artists.append(self.artist_map.get(artist.get("id")))
def parse_tracks(self, data):
for track in data:
self.parse_track(track.get("track"), track.get("added_at"))
for track in data:
self.library.liked_tracks.append(self.track_map.get(track.get("track").get("id")))
def parse_track(self, data: dict[str, Any], added_time : datetime) -> Track:
track_id = data.get("id")
# return existing track if already parsed
if track_id in self.track_map:
return self.track_map[track_id]
track = Track(
title=t["name"],
artists=[a["name"] for a in t.get("artists", [])],
album=t["album"]["name"] if t.get("album") else None,
duration_ms=t.get("duration_ms"),
added_at=self._parse_spotify_datetime(item.get("added_at")),
spotify=self._parse_spotify_data(t),
id=self._parse_spotify_data(t).id
)
track = Track()
self.track_map[track_id] = track
track.id = track_id
track.title = data.get("name")
track.artists = [self.parse_artist(artist) for artist in data.get("artists")]
track.album = self.parse_album(data.get("album"))
track.duration_ms = data.get("duration_ms")
self.library.tracks.append(track)
return track
def parse_track_items(self, items: Iterable[dict[str, Any]]) -> list[Track]:
return [self.parse_track_item(i) for i in items if i.get("track")]
def parse_album(self, data):
album_id = data.get("id")
# -------------------------
# Playlist parsers
# -------------------------
def parse_playlist_content(self, playlist_name: str, items: list[dict[str, Any]]) -> Playlist:
tracks = self.parse_track_items(items)
return Playlist(name=playlist_name, tracks=tracks)
if album_id in self.album_map:
return self.album_map[album_id]
def parse_all_playlist_content(self, data: dict[str, list[dict[str, Any]]]) -> list[Playlist]:
playlists: list[Playlist] = []
for name, items in data.items():
playlists.append(self.parse_playlist_content(name, items))
return playlists
album = Album()
def parse_all_tracks(self, items: dict[str, list[dict[str, Any]]]) -> list[Track]:
return [self.parse_track_item(i) for i in items]
self.album_map[album_id] = album
# -------------------------
# Library builder
# -------------------------
def build_library_from_playlist_content(
self, playlist_content_json: dict[str, list[dict[str, Any]]],
tracks_content_json: dict[str, list[dict[str, Any]]]
) -> Library:
library = Library()
album.id = data.get("id")
album.title = data.get("name")
album.artists = [self.parse_artist(artist) for artist in data.get("artists")]
self.parse_all_tracks(tracks_content_json)
self.library.albums.append(album)
library.tracklists.append(TrackList(tracks=list((self.track_map.values()))))
return album
library.playlists = self.parse_all_playlist_content(playlist_content_json)
library.tracks.extend(self.track_map.values())
def parse_artist(self, data: dict[str, Any]) -> Artist:
artist_id = data.get("id")
return library
if artist_id in self.artist_map:
return self.artist_map[artist_id]
# -------------------------
# Convenience method
# -------------------------
def parse(
self,
playlist_content_file: Path,
tracks_file: Path,
) -> Library:
playlist_content = json.loads(playlist_content_file.read_text())
tracks_content = json.loads(tracks_file.read_text())
return self.build_library_from_playlist_content(playlist_content, tracks_content)
artist = Artist()
self.artist_map[artist_id] = artist
artist.id = data.get("id")
artist.title = data.get("name")
self.library.artists.append(artist)
return artist