draft: moving page interface to downloader module
All checks were successful
ci/woodpecker/push/woodpecker Pipeline was successful
All checks were successful
ci/woodpecker/push/woodpecker Pipeline was successful
This commit is contained in:
parent
c683394228
commit
cd2e7d7173
@ -1,89 +1,24 @@
|
|||||||
import random
|
import random
|
||||||
from typing import Set, Type, Dict, List
|
|
||||||
from pathlib import Path
|
|
||||||
import re
|
import re
|
||||||
|
from pathlib import Path
|
||||||
|
from typing import Dict, List, Set, Type
|
||||||
|
|
||||||
from .utils import cli_function
|
from .. import console
|
||||||
from .options.first_config import initial_config
|
from ..download import Downloader
|
||||||
|
from ..download.results import GoToResults, Option, PageResults, Results
|
||||||
from ..utils import output, BColors
|
from ..objects import Album, Artist, DatabaseObject, Song
|
||||||
from ..utils.config import write_config, main_settings
|
from ..pages import Page
|
||||||
from ..utils.shared import URL_PATTERN
|
from ..utils import BColors, output
|
||||||
from ..utils.string_processing import fit_to_file_system
|
from ..utils.config import main_settings, write_config
|
||||||
from ..utils.support_classes.query import Query
|
from ..utils.enums.colors import BColors
|
||||||
from ..utils.support_classes.download_result import DownloadResult
|
|
||||||
from ..utils.exception import MKInvalidInputException
|
from ..utils.exception import MKInvalidInputException
|
||||||
from ..utils.exception.download import UrlNotFoundException
|
from ..utils.exception.download import UrlNotFoundException
|
||||||
from ..utils.enums.colors import BColors
|
from ..utils.shared import URL_PATTERN
|
||||||
from .. import console
|
from ..utils.string_processing import fit_to_file_system
|
||||||
|
from ..utils.support_classes.download_result import DownloadResult
|
||||||
from ..download.results import Results, Option, PageResults, GoToResults
|
from ..utils.support_classes.query import Query
|
||||||
from ..download.page_attributes import Pages
|
from .options.first_config import initial_config
|
||||||
from ..pages import Page
|
from .utils import cli_function
|
||||||
from ..objects import Song, Album, Artist, DatabaseObject
|
|
||||||
|
|
||||||
"""
|
|
||||||
This is the implementation of the Shell
|
|
||||||
|
|
||||||
# Behaviour
|
|
||||||
|
|
||||||
## Searching
|
|
||||||
|
|
||||||
```mkshell
|
|
||||||
> s: {querry or url}
|
|
||||||
|
|
||||||
# examples
|
|
||||||
> s: https://musify.club/release/some-random-release-183028492
|
|
||||||
> s: r: #a an Artist #r some random Release
|
|
||||||
```
|
|
||||||
|
|
||||||
Searches for an url, or an query
|
|
||||||
|
|
||||||
### Query Syntax
|
|
||||||
|
|
||||||
```
|
|
||||||
#a {artist} #r {release} #t {track}
|
|
||||||
```
|
|
||||||
|
|
||||||
You can escape stuff like `#` doing this: `\#`
|
|
||||||
|
|
||||||
## Downloading
|
|
||||||
|
|
||||||
To download something, you either need a direct link, or you need to have already searched for options
|
|
||||||
|
|
||||||
```mkshell
|
|
||||||
> d: {option ids or direct url}
|
|
||||||
|
|
||||||
# examples
|
|
||||||
> d: 0, 3, 4
|
|
||||||
> d: 1
|
|
||||||
> d: https://musify.club/release/some-random-release-183028492
|
|
||||||
```
|
|
||||||
|
|
||||||
## Misc
|
|
||||||
|
|
||||||
### Exit
|
|
||||||
|
|
||||||
```mkshell
|
|
||||||
> q
|
|
||||||
> quit
|
|
||||||
> exit
|
|
||||||
> abort
|
|
||||||
```
|
|
||||||
|
|
||||||
### Current Options
|
|
||||||
|
|
||||||
```mkshell
|
|
||||||
> .
|
|
||||||
```
|
|
||||||
|
|
||||||
### Previous Options
|
|
||||||
|
|
||||||
```
|
|
||||||
> ..
|
|
||||||
```
|
|
||||||
|
|
||||||
"""
|
|
||||||
|
|
||||||
EXIT_COMMANDS = {"q", "quit", "exit", "abort"}
|
EXIT_COMMANDS = {"q", "quit", "exit", "abort"}
|
||||||
ALPHABET = "abcdefghijklmnopqrstuvwxyz"
|
ALPHABET = "abcdefghijklmnopqrstuvwxyz"
|
||||||
@ -143,7 +78,7 @@ def help_message():
|
|||||||
print()
|
print()
|
||||||
|
|
||||||
|
|
||||||
class Downloader:
|
class CliDownloader:
|
||||||
def __init__(
|
def __init__(
|
||||||
self,
|
self,
|
||||||
exclude_pages: Set[Type[Page]] = None,
|
exclude_pages: Set[Type[Page]] = None,
|
||||||
@ -153,7 +88,7 @@ class Downloader:
|
|||||||
genre: str = None,
|
genre: str = None,
|
||||||
process_metadata_anyway: bool = False,
|
process_metadata_anyway: bool = False,
|
||||||
) -> None:
|
) -> None:
|
||||||
self.pages: Pages = Pages(exclude_pages=exclude_pages, exclude_shady=exclude_shady)
|
self.pages: Downloader = Downloader(exclude_pages=exclude_pages, exclude_shady=exclude_shady)
|
||||||
|
|
||||||
self.page_dict: Dict[str, Type[Page]] = dict()
|
self.page_dict: Dict[str, Type[Page]] = dict()
|
||||||
|
|
||||||
@ -446,7 +381,7 @@ def download(
|
|||||||
else:
|
else:
|
||||||
print(f"{BColors.FAIL.value}Something went wrong configuring.{BColors.ENDC.value}")
|
print(f"{BColors.FAIL.value}Something went wrong configuring.{BColors.ENDC.value}")
|
||||||
|
|
||||||
shell = Downloader(genre=genre, process_metadata_anyway=process_metadata_anyway)
|
shell = CliDownloader(genre=genre, process_metadata_anyway=process_metadata_anyway)
|
||||||
|
|
||||||
if command_list is not None:
|
if command_list is not None:
|
||||||
for command in command_list:
|
for command in command_list:
|
||||||
|
@ -1,8 +1,36 @@
|
|||||||
from dataclasses import dataclass, field
|
from __future__ import annotations
|
||||||
from typing import Set
|
|
||||||
|
|
||||||
from ..utils.config import main_settings
|
import logging
|
||||||
|
import random
|
||||||
|
import re
|
||||||
|
from collections import defaultdict
|
||||||
|
from copy import copy
|
||||||
|
from dataclasses import dataclass, field
|
||||||
|
from pathlib import Path
|
||||||
|
from string import Formatter
|
||||||
|
from typing import (TYPE_CHECKING, Dict, List, Optional, Set, Tuple, Type,
|
||||||
|
TypedDict, Union)
|
||||||
|
|
||||||
|
import requests
|
||||||
|
from bs4 import BeautifulSoup
|
||||||
|
|
||||||
|
from ..audio import correct_codec, write_metadata_to_target
|
||||||
|
from ..connection import Connection
|
||||||
|
from ..objects import Album, Artist, Collection
|
||||||
|
from ..objects import DatabaseObject as DataObject
|
||||||
|
from ..objects import Label, Options, Song, Source, Target
|
||||||
|
from ..utils import BColors, output, trace
|
||||||
|
from ..utils.config import main_settings, youtube_settings
|
||||||
|
from ..utils.enums import ALL_SOURCE_TYPES, SourceType
|
||||||
from ..utils.enums.album import AlbumType
|
from ..utils.enums.album import AlbumType
|
||||||
|
from ..utils.exception import MKMissingNameException
|
||||||
|
from ..utils.exception.download import UrlNotFoundException
|
||||||
|
from ..utils.path_manager import LOCATIONS
|
||||||
|
from ..utils.shared import DEBUG_PAGES
|
||||||
|
from ..utils.string_processing import fit_to_file_system
|
||||||
|
from ..utils.support_classes.download_result import DownloadResult
|
||||||
|
from ..utils.support_classes.query import Query
|
||||||
|
from .results import SearchResults
|
||||||
|
|
||||||
|
|
||||||
@dataclass
|
@dataclass
|
||||||
@ -19,3 +47,401 @@ class DownloadOptions:
|
|||||||
download_again_if_found: bool = False
|
download_again_if_found: bool = False
|
||||||
process_audio_if_found: bool = False
|
process_audio_if_found: bool = False
|
||||||
process_metadata_if_found: bool = True
|
process_metadata_if_found: bool = True
|
||||||
|
|
||||||
|
|
||||||
|
fetch_map = {
|
||||||
|
Song: "fetch_song",
|
||||||
|
Album: "fetch_album",
|
||||||
|
Artist: "fetch_artist",
|
||||||
|
Label: "fetch_label",
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
class Downloader:
|
||||||
|
def __init__(
|
||||||
|
self,
|
||||||
|
auto_register_pages: bool = True,
|
||||||
|
download_options: DownloadOptions = None,
|
||||||
|
fetch_options: FetchOptions = None,
|
||||||
|
**kwargs
|
||||||
|
):
|
||||||
|
self.LOGGER = logging.getLogger("download")
|
||||||
|
|
||||||
|
self.download_options: DownloadOptions = download_options or DownloadOptions()
|
||||||
|
self.fetch_options: FetchOptions = fetch_options or FetchOptions()
|
||||||
|
|
||||||
|
self._registered_pages: Dict[Type[Page], Set[Page]] = defaultdict(set)
|
||||||
|
if auto_register_pages:
|
||||||
|
self.scan_for_pages(**kwargs)
|
||||||
|
|
||||||
|
def register_page(self, page_type: Type[Page], **kwargs):
|
||||||
|
if page_type in _registered_pages:
|
||||||
|
return
|
||||||
|
|
||||||
|
self._registered_pages[page_type].add(page_type(
|
||||||
|
download_options=self.download_options,
|
||||||
|
fetch_options=self.fetch_options,
|
||||||
|
**kwargs
|
||||||
|
))
|
||||||
|
|
||||||
|
def deregister_page(self, page_type: Type[Page]):
|
||||||
|
if page_type not in _registered_pages:
|
||||||
|
return
|
||||||
|
|
||||||
|
for p in self._registered_pages[page_type]:
|
||||||
|
p.__del__()
|
||||||
|
del self._registered_pages[page_type]
|
||||||
|
|
||||||
|
def scan_for_pages(self, **kwargs):
|
||||||
|
# assuming the wanted pages are the leaf classes of the interface
|
||||||
|
leaf_classes = []
|
||||||
|
|
||||||
|
class_list = [Page]
|
||||||
|
while len(class_list):
|
||||||
|
_class = class_list.pop()
|
||||||
|
class_subclasses = _class.__subclasses__()
|
||||||
|
|
||||||
|
if len(class_subclasses) == 0:
|
||||||
|
if _class.REGISTER:
|
||||||
|
leaf_classes.append(_class)
|
||||||
|
else:
|
||||||
|
class_list.extend(class_subclasses)
|
||||||
|
|
||||||
|
for leaf_class in leaf_classes:
|
||||||
|
self.register_page(leaf_class, **kwargs)
|
||||||
|
|
||||||
|
def get_pages(self, *page_types: List[Type[Page]]) -> Generator[Page, None, None]:
|
||||||
|
if len(page_types) == 0:
|
||||||
|
page_types = _registered_pages.keys()
|
||||||
|
|
||||||
|
for page_type in page_types:
|
||||||
|
yield from self._registered_pages[page_type]
|
||||||
|
|
||||||
|
def search(self, query: Query) -> SearchResults:
|
||||||
|
result = SearchResults()
|
||||||
|
|
||||||
|
for page in self.get_pages():
|
||||||
|
result.add(
|
||||||
|
page=type(page),
|
||||||
|
search_result=page.search(query=query)
|
||||||
|
)
|
||||||
|
|
||||||
|
return result
|
||||||
|
|
||||||
|
def fetch_details(self, data_object: DataObject, stop_at_level: int = 1, **kwargs) -> DataObject:
|
||||||
|
source: Source
|
||||||
|
for source in data_object.source_collection.get_sources(source_type_sorting={
|
||||||
|
"only_with_page": True,
|
||||||
|
}):
|
||||||
|
new_data_object = self.fetch_from_source(source=source, stop_at_level=stop_at_level)
|
||||||
|
if new_data_object is not None:
|
||||||
|
data_object.merge(new_data_object)
|
||||||
|
|
||||||
|
return data_object
|
||||||
|
|
||||||
|
def fetch_from_source(self, source: Source, **kwargs) -> Optional[DataObject]:
|
||||||
|
if not source.has_page:
|
||||||
|
return None
|
||||||
|
|
||||||
|
source_type = source.page.get_source_type(source=source)
|
||||||
|
if source_type is None:
|
||||||
|
self.LOGGER.debug(f"Could not determine source type for {source}.")
|
||||||
|
return None
|
||||||
|
|
||||||
|
func = getattr(source.page, fetch_map[source_type])
|
||||||
|
|
||||||
|
# fetching the data object and marking it as fetched
|
||||||
|
data_object: DataObject = func(source=source, **kwargs)
|
||||||
|
data_object.mark_as_fetched(source.hash_url)
|
||||||
|
return data_object
|
||||||
|
|
||||||
|
def fetch_from_url(self, url: str) -> Optional[DataObject]:
|
||||||
|
source = Source.match_url(url, ALL_SOURCE_TYPES.MANUAL)
|
||||||
|
if source is None:
|
||||||
|
return None
|
||||||
|
|
||||||
|
return self.fetch_from_source(source=source)
|
||||||
|
|
||||||
|
def _skip_object(self, data_object: DataObject) -> bool:
|
||||||
|
if isinstance(data_object, Album):
|
||||||
|
if not self.download_options.download_all and data_object.album_type in self.download_options.album_type_blacklist:
|
||||||
|
return True
|
||||||
|
|
||||||
|
return False
|
||||||
|
|
||||||
|
def download(self, data_object: DataObject, genre: str, **kwargs) -> DownloadResult:
|
||||||
|
# fetch the given object
|
||||||
|
self.fetch_details(data_object)
|
||||||
|
output(f"\nDownloading {data_object.option_string}...", color=BColors.BOLD)
|
||||||
|
|
||||||
|
# fetching all parent objects (e.g. if you only download a song)
|
||||||
|
if not kwargs.get("fetched_upwards", False):
|
||||||
|
to_fetch: List[DataObject] = [data_object]
|
||||||
|
|
||||||
|
while len(to_fetch) > 0:
|
||||||
|
new_to_fetch = []
|
||||||
|
for d in to_fetch:
|
||||||
|
if self._skip_object(d):
|
||||||
|
continue
|
||||||
|
|
||||||
|
self.fetch_details(d)
|
||||||
|
|
||||||
|
for c in d.get_parent_collections():
|
||||||
|
new_to_fetch.extend(c)
|
||||||
|
|
||||||
|
to_fetch = new_to_fetch
|
||||||
|
|
||||||
|
kwargs["fetched_upwards"] = True
|
||||||
|
|
||||||
|
# download all children
|
||||||
|
download_result: DownloadResult = DownloadResult()
|
||||||
|
for c in data_object.get_child_collections():
|
||||||
|
for d in c:
|
||||||
|
if self._skip_object(d):
|
||||||
|
continue
|
||||||
|
|
||||||
|
download_result.merge(self.download(d, genre, **kwargs))
|
||||||
|
|
||||||
|
# actually download if the object is a song
|
||||||
|
if isinstance(data_object, Song):
|
||||||
|
"""
|
||||||
|
TODO
|
||||||
|
add the traced artist and album to the naming.
|
||||||
|
I am able to do that, because duplicate values are removed later on.
|
||||||
|
"""
|
||||||
|
|
||||||
|
self._download_song(data_object, naming={
|
||||||
|
"genre": [genre],
|
||||||
|
"audio_format": [main_settings["audio_format"]],
|
||||||
|
})
|
||||||
|
|
||||||
|
return download_result
|
||||||
|
|
||||||
|
def _extract_fields_from_template(self, path_template: str) -> Set[str]:
|
||||||
|
return set(re.findall(r"{([^}]+)}", path_template))
|
||||||
|
|
||||||
|
def _parse_path_template(self, path_template: str, naming: Dict[str, List[str]]) -> str:
|
||||||
|
field_names: Set[str] = self._extract_fields_from_template(path_template)
|
||||||
|
|
||||||
|
for field in field_names:
|
||||||
|
if len(naming[field]) == 0:
|
||||||
|
raise MKMissingNameException(f"Missing field for {field}.")
|
||||||
|
|
||||||
|
path_template = path_template.replace(f"{{{field}}}", naming[field][0])
|
||||||
|
|
||||||
|
return path_template
|
||||||
|
|
||||||
|
def _download_song(self, song: Song, naming: dict) -> DownloadOptions:
|
||||||
|
"""
|
||||||
|
TODO
|
||||||
|
Search the song in the file system.
|
||||||
|
"""
|
||||||
|
r = DownloadResult(total=1)
|
||||||
|
|
||||||
|
# pre process the data recursively
|
||||||
|
song.compile()
|
||||||
|
|
||||||
|
# manage the naming
|
||||||
|
naming: Dict[str, List[str]] = defaultdict(list, naming)
|
||||||
|
naming["song"].append(song.title_value)
|
||||||
|
naming["isrc"].append(song.isrc)
|
||||||
|
naming["album"].extend(a.title_value for a in song.album_collection)
|
||||||
|
naming["album_type"].extend(a.album_type.value for a in song.album_collection)
|
||||||
|
naming["artist"].extend(a.name for a in song.artist_collection)
|
||||||
|
naming["artist"].extend(a.name for a in song.feature_artist_collection)
|
||||||
|
for a in song.album_collection:
|
||||||
|
naming["label"].extend([l.title_value for l in a.label_collection])
|
||||||
|
# removing duplicates from the naming, and process the strings
|
||||||
|
for key, value in naming.items():
|
||||||
|
# https://stackoverflow.com/a/17016257
|
||||||
|
naming[key] = list(dict.fromkeys(value))
|
||||||
|
song.genre = naming["genre"][0]
|
||||||
|
|
||||||
|
# manage the targets
|
||||||
|
tmp: Target = Target.temp(file_extension=main_settings["audio_format"])
|
||||||
|
|
||||||
|
song.target_collection.append(Target(
|
||||||
|
relative_to_music_dir=True,
|
||||||
|
file_path=Path(
|
||||||
|
self._parse_path_template(main_settings["download_path"], naming=naming),
|
||||||
|
self._parse_path_template(main_settings["download_file"], naming=naming),
|
||||||
|
)
|
||||||
|
))
|
||||||
|
for target in song.target_collection:
|
||||||
|
if target.exists:
|
||||||
|
output(f'{target.file_path} {BColors.OKGREEN.value}[already exists]', color=BColors.GREY)
|
||||||
|
r.found_on_disk += 1
|
||||||
|
|
||||||
|
if not self.download_options.download_again_if_found:
|
||||||
|
target.copy_content(tmp)
|
||||||
|
else:
|
||||||
|
target.create_path()
|
||||||
|
output(f'{target.file_path}', color=BColors.GREY)
|
||||||
|
|
||||||
|
# this streams from every available source until something succeeds, setting the skip intervals to the values of the according source
|
||||||
|
used_source: Optional[Source] = None
|
||||||
|
skip_intervals: List[Tuple[float, float]] = []
|
||||||
|
for source in song.source_collection.get_sources(source_type_sorting={
|
||||||
|
"only_with_page": True,
|
||||||
|
"sort_key": lambda page: page.download_priority,
|
||||||
|
"reverse": True,
|
||||||
|
}):
|
||||||
|
if tmp.exists:
|
||||||
|
break
|
||||||
|
|
||||||
|
used_source = source
|
||||||
|
streaming_results = source.page.download_song_to_target(source=source, target=tmp, desc="download")
|
||||||
|
skip_intervals = source.page.get_skip_intervals(song=song, source=source)
|
||||||
|
|
||||||
|
# if something has been downloaded but it somehow failed, delete the file
|
||||||
|
if streaming_results.is_fatal_error and tmp.exists:
|
||||||
|
tmp.delete()
|
||||||
|
|
||||||
|
# if everything went right, the file should exist now
|
||||||
|
if not tmp.exists:
|
||||||
|
if used_source is None:
|
||||||
|
r.error_message = f"No source found for {song.option_string}."
|
||||||
|
else:
|
||||||
|
r.error_message = f"Something went wrong downloading {song.option_string}."
|
||||||
|
return r
|
||||||
|
|
||||||
|
# post process the audio
|
||||||
|
found_on_disk = used_source is None
|
||||||
|
if not found_on_disk or self.download_options.process_audio_if_found:
|
||||||
|
correct_codec(target=tmp, skip_intervals=skip_intervals)
|
||||||
|
r.sponsor_segments = len(skip_intervals)
|
||||||
|
|
||||||
|
if used_source is not None:
|
||||||
|
used_source.page.post_process_hook(song=song, temp_target=tmp)
|
||||||
|
|
||||||
|
if not found_on_disk or self.download_options.process_metadata_if_found:
|
||||||
|
write_metadata_to_target(metadata=song.metadata, target=tmp, song=song)
|
||||||
|
|
||||||
|
# copy the tmp target to the final locations
|
||||||
|
for target in song.target_collection:
|
||||||
|
tmp.copy_content(target)
|
||||||
|
|
||||||
|
tmp.delete()
|
||||||
|
return r
|
||||||
|
|
||||||
|
def fetch_url(self, url: str, **kwargs) -> DataObject:
|
||||||
|
source = Source.match_url(url, ALL_SOURCE_TYPES.MANUAL)
|
||||||
|
|
||||||
|
if source is None or source.page is None:
|
||||||
|
raise UrlNotFoundException(url=url)
|
||||||
|
|
||||||
|
return source.page.fetch_object_from_source(source=source, **kwargs)
|
||||||
|
|
||||||
|
|
||||||
|
class Page:
|
||||||
|
REGISTER = True
|
||||||
|
SOURCE_TYPE: SourceType
|
||||||
|
LOGGER: logging.Logger
|
||||||
|
|
||||||
|
def __new__(cls, *args, **kwargs):
|
||||||
|
cls.LOGGER = logging.getLogger(cls.__name__)
|
||||||
|
return super().__new__(cls)
|
||||||
|
|
||||||
|
def __init__(self, download_options: DownloadOptions = None, fetch_options: FetchOptions = None, **kwargs):
|
||||||
|
self.SOURCE_TYPE.register_page(self)
|
||||||
|
|
||||||
|
self.download_options: DownloadOptions = download_options or DownloadOptions()
|
||||||
|
self.fetch_options: FetchOptions = fetch_options or FetchOptions()
|
||||||
|
|
||||||
|
def __del__(self):
|
||||||
|
self.SOURCE_TYPE.deregister_page()
|
||||||
|
|
||||||
|
def _search_regex(self, pattern, string, default=None, fatal=True, flags=0, group=None):
|
||||||
|
"""
|
||||||
|
Perform a regex search on the given string, using a single or a list of
|
||||||
|
patterns returning the first matching group.
|
||||||
|
In case of failure return a default value or raise a WARNING or a
|
||||||
|
RegexNotFoundError, depending on fatal, specifying the field name.
|
||||||
|
"""
|
||||||
|
|
||||||
|
if isinstance(pattern, str):
|
||||||
|
mobj = re.search(pattern, string, flags)
|
||||||
|
else:
|
||||||
|
for p in pattern:
|
||||||
|
mobj = re.search(p, string, flags)
|
||||||
|
if mobj:
|
||||||
|
break
|
||||||
|
|
||||||
|
if mobj:
|
||||||
|
if group is None:
|
||||||
|
# return the first matching group
|
||||||
|
return next(g for g in mobj.groups() if g is not None)
|
||||||
|
elif isinstance(group, (list, tuple)):
|
||||||
|
return tuple(mobj.group(g) for g in group)
|
||||||
|
else:
|
||||||
|
return mobj.group(group)
|
||||||
|
|
||||||
|
return default
|
||||||
|
|
||||||
|
def get_source_type(self, source: Source) -> Optional[Type[DataObject]]:
|
||||||
|
return None
|
||||||
|
|
||||||
|
def get_soup_from_response(self, r: requests.Response) -> BeautifulSoup:
|
||||||
|
return BeautifulSoup(r.content, "html.parser")
|
||||||
|
|
||||||
|
# to search stuff
|
||||||
|
def search(self, query: Query) -> List[DataObject]:
|
||||||
|
music_object = query.music_object
|
||||||
|
|
||||||
|
search_functions = {
|
||||||
|
Song: self.song_search,
|
||||||
|
Album: self.album_search,
|
||||||
|
Artist: self.artist_search,
|
||||||
|
Label: self.label_search
|
||||||
|
}
|
||||||
|
|
||||||
|
if type(music_object) in search_functions:
|
||||||
|
r = search_functions[type(music_object)](music_object)
|
||||||
|
if r is not None and len(r) > 0:
|
||||||
|
return r
|
||||||
|
|
||||||
|
r = []
|
||||||
|
for default_query in query.default_search:
|
||||||
|
for single_option in self.general_search(default_query):
|
||||||
|
r.append(single_option)
|
||||||
|
|
||||||
|
return r
|
||||||
|
|
||||||
|
def general_search(self, search_query: str) -> List[DataObject]:
|
||||||
|
return []
|
||||||
|
|
||||||
|
def label_search(self, label: Label) -> List[Label]:
|
||||||
|
return []
|
||||||
|
|
||||||
|
def artist_search(self, artist: Artist) -> List[Artist]:
|
||||||
|
return []
|
||||||
|
|
||||||
|
def album_search(self, album: Album) -> List[Album]:
|
||||||
|
return []
|
||||||
|
|
||||||
|
def song_search(self, song: Song) -> List[Song]:
|
||||||
|
return []
|
||||||
|
|
||||||
|
# to fetch stuff
|
||||||
|
def fetch_song(self, source: Source, stop_at_level: int = 1) -> Song:
|
||||||
|
return Song()
|
||||||
|
|
||||||
|
def fetch_album(self, source: Source, stop_at_level: int = 1) -> Album:
|
||||||
|
return Album()
|
||||||
|
|
||||||
|
def fetch_artist(self, source: Source, stop_at_level: int = 1) -> Artist:
|
||||||
|
return Artist()
|
||||||
|
|
||||||
|
def fetch_label(self, source: Source, stop_at_level: int = 1) -> Label:
|
||||||
|
return Label()
|
||||||
|
|
||||||
|
# to download stuff
|
||||||
|
def get_skip_intervals(self, song: Song, source: Source) -> List[Tuple[float, float]]:
|
||||||
|
return []
|
||||||
|
|
||||||
|
def post_process_hook(self, song: Song, temp_target: Target, **kwargs):
|
||||||
|
pass
|
||||||
|
|
||||||
|
def download_song_to_target(self, source: Source, target: Target, desc: str = None) -> DownloadResult:
|
||||||
|
return DownloadResult()
|
||||||
|
|
||||||
|
Loading…
Reference in New Issue
Block a user