Switched from geckodriver to chrome

This commit is contained in:
2022-09-18 19:26:55 +02:00
parent 7cf7422b46
commit db161e50c8
13 changed files with 135 additions and 61 deletions

View File

@@ -26,5 +26,4 @@ local_storage_path: /app/containerdata/files
debug_storage_path: /app/containerdata/debug/
default_download_path: /app/containerdata/tmp
remote_storage_path: /helbing_support/Files RM/Archiving
browser_profile_path: /app/containerdata/dependencies/7hlyfqxt.Auto News
blacklisted_href_domains: ["google.", "facebook."]
browser_profile_path: /user_data/news_fetch.profile

56
misc/youtube_batch.py Normal file
View File

@@ -0,0 +1,56 @@
import youtube_dl
from waybackpy import WaybackMachineSaveAPI # upload to archive.org
import time
urls = [
"https://www.youtube.com/watch?v=R4h_yiDIuQE",
"https://www.youtube.com/watch?v=-G8ZI1Jq8xA",
"https://www.youtube.com/watch?v=8eYBcASQIQI",
"https://www.thingiverse.com/thing:5463267",
"https://www.youtube.com/watch?v=cJoUSHJcV4E&t=0s",
"https://www.youtube.com/watch?v=UbBYZZBREBA&t=0s",
"https://www.youtube.com/watch?v=bQQn_vET4ys",
"https://www.youtube.com/watch?v=6FqNctiO06E",
"https://www.youtube.com/watch?v=ImnuJgj8XJo",
"https://www.youtube.com/watch?v=4QZQtSqaC34",
"https://www.youtube.com/watch?v=cW4qIjPMGkQ",
"https://www.youtube.com/watch?v=QWsUGpKfP8A",
"https://www.youtube.com/watch?v=a0PwEwLG9No",
"https://www.youtube.com/watch?v=Hd3lnWVIIpo",
"https://www.youtube.com/watch?v=JNtdAp-BdzI",
"https://en.wikipedia.org/wiki/Viktor_Schauberger",
"https://de.wikipedia.org/wiki/Viktor_Schauberger",
]
def post_download_hook(ret_code):
# print(ret_code)
if ret_code['status'] == 'finished':
file_loc = ret_code["filename"]
print(file_loc)
def save_video(url):
"""Saves video accoring to url and save path"""
ydl_opts = {
'format': 'best[height<=720]',
# 'outtmpl': f"{file_path}.%(ext)s", # basically the filename from the object, but with a custom extension depending on the download
'progress_hooks': [post_download_hook],
'updatetime': False
}
try:
with youtube_dl.YoutubeDL(ydl_opts) as ydl:
ydl.download([url])
# article file name is updated in self.post_download_hook
except Exception as e:
print(f"Youtube download crashed: {e}")
# for url in urls:
# save_video(url)
for url in urls:
user_agent = "Mozilla/5.0 (Windows NT 5.1; rv:40.0) Gecko/20100101 Firefox/40.0" # needed?
wayback = WaybackMachineSaveAPI(url, user_agent)
archive_url = wayback.save()
print(archive_url)
time.sleep(20)