|
@@ -510,6 +510,7 @@ def retrieve(url, args):
|
|
|
|
|
|
|
|
|
def audio_passes_filters(audio, args):
|
|
|
+ logging.debug(audio.end_date)
|
|
|
if not audio.valid:
|
|
|
return False
|
|
|
if args.max_len and audio.duration > args.max_len:
|
|
@@ -545,8 +546,10 @@ def main():
|
|
|
|
|
|
audio_by_source = OrderedDict()
|
|
|
for i, url in enumerate(sources):
|
|
|
- url_audios = retrieve(url, args)
|
|
|
+ url_audios = list(retrieve(url, args))
|
|
|
+ logging.debug("Found %d audios in %s", len(url_audios), url)
|
|
|
url_audios = [au for au in url_audios if audio_passes_filters(au, args)]
|
|
|
+ logging.debug("%d of those are passing filters", len(url_audios))
|
|
|
audio_by_source[url] = url_audios
|
|
|
if not url_audios:
|
|
|
weights[i] = 0
|
|
@@ -558,6 +561,8 @@ def main():
|
|
|
for source_url in sources:
|
|
|
audios += audio_by_source[source_url]
|
|
|
|
|
|
+ logging.debug("Found %d audios", len(audios))
|
|
|
+
|
|
|
# sort
|
|
|
if args.sort_by == "random":
|
|
|
random.shuffle(audios)
|