1
0
mirror of https://git.sr.ht/~cadence/NewLeaf synced 2024-09-28 22:17:29 +00:00

Move extract_yt_initial_data to function

This commit is contained in:
Cadence Ember 2020-08-11 23:33:14 +12:00
parent 080b00bc0c
commit 577cdd8a24
No known key found for this signature in database
GPG Key ID: 128B99B1B74A6412

View File

@ -26,6 +26,16 @@ def length_text_to_seconds(text):
s = text.split(":")
return sum([int(x) * 60**(len(s)-i-1) for i, x in enumerate(s)])
r_yt_intial_data = re.compile(r"""^\s*window\["ytInitialData"\] = (\{.*\});\n?$""", re.M)
def extract_yt_initial_data(content):
m_yt_initial_data = re.search(r_yt_intial_data, content)
if m_yt_initial_data:
yt_initial_data = json.loads(m_yt_initial_data.group(1))
return yt_initial_data
else:
raise Exception("Could not match ytInitialData in content")
class Second(object):
def __init__(self):
self.video_cache = TTLCache(maxsize=50, ttl=300)
@ -166,12 +176,10 @@ class Second(object):
if len(possible_files) == 1:
filename = possible_files[0]
with open(filename) as file:
r_yt_intial_data = re.compile(r"""^\s*window\["ytInitialData"\] = (\{.*\});\n?$""")
r_yt_player_config = re.compile(r"""^\s*[^"]+"cfg"[^"]+ytplayer\.config = (\{.*\});ytplayer\.web_player_context_config = {".""")
for line in file:
m_yt_initial_data = re.search(r_yt_intial_data, line)
if m_yt_initial_data:
yt_initial_data = json.loads(m_yt_initial_data.group(1))
content = file.read()
yt_initial_data = extract_yt_initial_data(content)
views = yt_initial_data["contents"]["twoColumnWatchNextResults"]["results"]["results"]["contents"][0]\
["videoPrimaryInfoRenderer"]["viewCount"]["videoViewCountRenderer"]
result["second__viewCountText"] = views["viewCount"]["simpleText"]