def separate_into_parts(self, input_path: str, output_path: Path):
# Check if we downloaded a webpage instead of the actual model file
file_exists = self.model_file_path.is_file()
mime = None
if file_exists:
mime = magic.from_file(str(self.model_file_path), mime=True)
download_and_verify(MODEL_URL,
self.model_dir,
self.model_file_path,
force=(file_exists and mime == 'text/html'))
parts = {
'vocals': True,
'drums': True,
'bass': True,
'other': True
}
estimates = self.get_estimates(input_path, parts)
# Export all source MP3s in parallel
pool = Pool()
tasks = []
output_path = Path(output_path)
for name, estimate in estimates.items():
filename = f'{name}.mp3'
print(f'Exporting {name} MP3...')
task = pool.apply_async(self.audio_adapter.save, (output_path / filename, estimate, self.sample_rate, 'mp3', self.bitrate))
tasks.append(task)
pool.close()
pool.join()
def separate_into_parts(self, input_path: str, output_path: str):
"""Creates a dynamic mix
:param input_path: Input path
:param output_path: Output path
"""
input_path = Path(input_path)
output_path = Path(output_path)
model = self.get_model()
raw_sources = self.apply_model(model, input_path)
# Export all source MP3s in parallel
pool = Pool()
tasks = []
for source, name in zip(raw_sources,
['drums', 'bass', 'other', 'vocals']):
source = source.cpu().transpose(0, 1).numpy()
filename = f'{name}.mp3'
print(f'Exporting {name} MP3...')
task = pool.apply_async(self.audio_adapter.save,
(output_path / filename, source,
self.sample_rate, 'mp3', self.bitrate))
tasks.append(task)
try:
pool.close()
pool.join()
except SoftTimeLimitExceeded as e:
pool.terminate()
raise e
def separate_into_parts(self, input_path: str, output_path: str):
"""Creates a dynamic mix
:param input_path: Input path
:param output_path: Output path
"""
input_path = Path(input_path)
output_path = Path(output_path)
self.download_and_verify()
raw_sources = self.apply_model(input_path)
# Export all source MP3s in parallel
pool = Pool()
tasks = []
for source, name in zip(raw_sources,
['drums', 'bass', 'other', 'vocals']):
source = (source * 2**15).clamp_(-2**15, 2**15 - 1).short()
source = source.cpu().transpose(0, 1).numpy()
filename = f'{name}.mp3'
print(f'Exporting {name} MP3...')
task = pool.apply_async(encode_mp3,
(source, str(output_path / filename),
self.bitrate, self.verbose))
tasks.append(task)
try:
pool.close()
pool.join()
except SoftTimeLimitExceeded as e:
pool.terminate()
raise e
def separate_into_parts(self, input_path: str, output_path: Path):
self.download_and_verify()
ctx = get_extension_context(self.context)
nn.set_default_context(ctx)
nn.set_auto_forward(True)
audio, _ = self.audio_adapter.load(input_path,
sample_rate=self.sample_rate)
if audio.shape[1] > 2:
warnings.warn('Channel count > 2! '
'Only the first two channels will be processed!')
audio = audio[:, :2]
if audio.shape[1] == 1:
print('received mono file, so duplicate channels')
audio = np.repeat(audio, 2, axis=1)
print('Separating...')
estimates = separate(audio,
model_path=str(self.model_file_path),
niter=self.iterations,
alpha=self.alpha,
softmask=self.softmask,
residual_model=self.residual_model)
output_path = Path(output_path)
# Export all source MP3s in parallel
pool = Pool()
tasks = []
for name, estimate in estimates.items():
filename = f'{name}.mp3'
print(f'Exporting {name} MP3...')
task = pool.apply_async(self.audio_adapter.save, (os.path.join(
output_path,
filename), estimate, self.sample_rate, 'mp3', self.bitrate))
tasks.append(task)
pool.close()
pool.join()
def separate_into_parts(self, input_path: str, output_path: Path):
self.download_and_verify()
estimates = self.get_estimates(input_path)
# Export all source MP3s in parallel
pool = Pool()
tasks = []
output_path = Path(output_path)
for name, estimate in estimates.items():
filename = f'{name}.mp3'
print(f'Exporting {name} MP3...')
task = pool.apply_async(self.audio_adapter.save, (os.path.join(
output_path,
filename), estimate, self.sample_rate, 'mp3', self.bitrate))
tasks.append(task)
pool.close()
pool.join()
class RenderAndSave():
def __init__(self, async_mode=True):
"""
:param async_mode: False - используется для теста
"""
self.async_mode = async_mode
@staticmethod
def get_available_sites(news_portal=None, news_dep=None) -> list:
"""
Получение списка доступных сайтов
:param news_portal: новостной портал, по активным тематическим разделам которого будет выполнен поиск, если
не задан - поиск ведется по всем активным
:param news_dep: новостной раздел портала
:return: query
"""
if news_portal:
return [
Site.objects.filter(is_active=True,
news_portal=news_portal,
news_department=news_dep)
]
return [site for site in Site.objects.filter(is_active=True)]
@staticmethod
def _save_postgr(results):
"""
Сохранение в Postgresql результатов парсинга
:param results: список с корежами, содержащими словарь, описывающий результат парсинга отдельной статьи
:return:
"""
for part in results:
for result in part:
Article.objects.get_or_create(
link=result['news_link'],
has_prices=get_has_prices(result['main_text']),
has_percents=get_has_percents(result['main_text']),
frequent_words=get_frequent_words(result['main_text']),
content=result)
def _async_worker(self, sites_list) -> tuple:
"""
Запуск парсера в асинхронном режиме.
:param sites_list: список сайтов для анализа
:return: спискок кортежей с резульатами парсинга
"""
self.process_pool = Pool(processes=settings.PROCESS_AMOUNT)
results = [
self.process_pool.apply_async(
settings.AVAILABLE_RENDERS[site.news_portal],
args=(site.target_url, )) for site in sites_list
]
clean_data = [i.get() for i in results]
self.process_pool.close()
self.process_pool.join()
return clean_data
def _sync_worker(self, site) -> list:
"""
Запуск парсера в синхронном режиме.
:param sites_list: сайт для анализа
:return: спискок кортежей с резульатами парсинга
"""
try:
site = site[0]
return [
settings.AVAILABLE_RENDERS[site.news_portal].__call__(
site.target_url)
]
except Exception as err:
logger.error(err)
def run_parser(self) -> list:
"""
Запуск парсера.
:return:
"""
sites_list = self.get_available_sites()
if self.async_mode:
data_from_site = self._async_worker(sites_list)
else:
data_from_site = self._sync_worker(sites_list)
if not data_from_site:
logger.info("Ошибка рабзора сайта")
raise Exception("Ошибка рабзора сайта")
self._save_postgr(data_from_site)
return [i.target_url for i in sites_list]
def crawl_keywords(what):
print(what)
p = Pool(3)
p.apply_async(crawl)
p.close()
p.join()