annotate kemonopartydownloader.py @ 88:4b8fd5492092

why does this work? committer: GitHub <noreply@github.com>
author Paper <37962225+mrpapersonic@users.noreply.github.com>
date Thu, 28 Jul 2022 20:09:42 -0400
parents bde647ac9554
children d2e0edd4a070
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
1 # example args.url: https://kemono.party/fanbox/user/5375435/post/2511461
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
2 # created by Paper in 2021
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
3 # please do not share without crediting me!
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
4 import argparse
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
5 import http.cookiejar
49
717feded576e Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 46
diff changeset
6 import os
717feded576e Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 46
diff changeset
7 import re
717feded576e Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 46
diff changeset
8 import requests # pip install requests
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
9 import time
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
10 import math
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
11 import zipfile
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
12 import urllib.parse
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
13 from urllib.error import HTTPError
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
14 from http.client import BadStatusLine
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
15
51
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
16
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
17 def download_folder_from_google_drive(link):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
18 session = requests.Session()
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
19 session.headers = {
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
20 'origin': 'https://drive.google.com',
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
21 'content-type': 'application/json',
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
22 }
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
23 key = "AIzaSyC1qbk75NzWBvSaDh6KnsjjA9pIrP4lYIE" # google anonymous key
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
24 takeoutjs = session.post(f"https://takeout-pa.clients6.google.com/v1/exports?key={key}", data='{{"items":[{{"id":"{0}"}}]}}'.format(link.split("?")[0].split("/")[-1])).json()
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
25 takeoutid = takeoutjs["exportJob"]["id"]
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
26 storagePath = None
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
27 while storagePath is None:
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
28 succeededjson = session.get("https://takeout-pa.clients6.google.com/v1/exports/{0}?key={1}".format(takeoutid, key)).json()
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
29 if succeededjson["exportJob"]["status"] == "SUCCEEDED":
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
30 storagePath = succeededjson["exportJob"]["archives"][0]["storagePath"]
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
31 time.sleep(1)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
32 size = 0
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
33 for path, dirs, files in os.walk("./{0}/Drive - {1}".format(output, sanitize(i["title"]))):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
34 for f in files:
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
35 fp = os.path.join(path, f)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
36 size += os.path.getsize(fp)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
37 if size >= int(succeededjson["exportJob"]["archives"][0]["sizeOfContents"]):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
38 print(" {0} already downloaded!".format(succeededjson["exportJob"]["archives"][0]["fileName"]))
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
39 return
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
40 response = session.get(storagePath, stream=True)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
41 amountdone = 0
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
42 with open(succeededjson["exportJob"]["archives"][0]["fileName"], "wb") as f:
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
43 for chunk in response.iter_content(1024):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
44 if chunk: # filter out keep-alive new chunks
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
45 f.write(chunk)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
46 amountdone += 1024
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
47 print(" downloading {0}: ".format(succeededjson["exportJob"]["archives"][0]["fileName"]) + " " + str(round((amountdone / int(succeededjson['exportJob']['archives'][0]['compressedSize'])) * 100, 2)) + "%\r", end="")
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
48 print(" downloaded {0}".format(succeededjson["exportJob"]["archives"][0]["fileName"]) + ": 100.00% ")
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
49 unzip(succeededjson["exportJob"]["archives"][0]["fileName"], "./{0}/Drive - {1}".format(output, sanitize(i["title"])))
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
50
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
51
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
52 def unzip(src_path, dst_dir, pwd=None):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
53 with zipfile.ZipFile(src_path) as zf:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
54 members = zf.namelist()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
55 for member in members:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
56 arch_info = zf.getinfo(member)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
57 arch_name = arch_info.filename.replace('/', os.path.sep)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
58 dst_path = os.path.join(dst_dir, arch_name)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
59 dst_path = os.path.normpath(dst_path)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
60 if not os.path.exists(dst_path):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
61 zf.extract(arch_info, dst_dir, pwd)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
62
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
63
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
64 def download_from_dropbox(link):
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
65 responsehead = req.head(link.split("?")[0] + "?dl=1", allow_redirects=True)
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
66 if responsehead.status_code == 404:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
67 print(" dropbox link not available!")
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
68 return
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
69 if not os.path.exists(output + "/Dropbox - " + sanitize(i["title"])):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
70 os.makedirs(output + "/Dropbox - " + sanitize(i["title"]))
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
71 filename = output + "/Dropbox - " + sanitize(i["title"]) + "/" + sanitize(responsehead.headers["Content-Disposition"].split("'")[-1])
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
72 if os.path.exists(urllib.parse.unquote(os.path.splitext(filename)[0])) and os.path.isdir(urllib.parse.unquote(os.path.splitext(filename)[0])):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
73 print(" file(s) already downloaded!")
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
74 return
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
75 if os.path.exists(filename):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
76 filesize = os.stat(filename).st_size
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
77 else:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
78 filesize = 0
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
79 serverfilesize = int(responsehead.headers["Content-Length"])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
80 if filesize < serverfilesize:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
81 with req.get(link.split("?")[0] + "?dl=1", stream=True, headers={"Range": f"bytes={filesize}-"}) as r:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
82 r.raise_for_status()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
83 with open(filename, "ab") as f:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
84 for chunk in r.iter_content(chunk_size=4096):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
85 f.write(chunk)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
86 filesize += 4096
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
87 print(" file {0} downloading: ".format(urllib.parse.unquote(responsehead.headers["Content-Disposition"].split("'")[-1])) + str(round((filesize / serverfilesize) * 100)) + "%\r", end="")
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
88 print(" {0} successfully downloaded!".format(urllib.parse.unquote(responsehead.headers["Content-Disposition"].split("'")[-1])))
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
89 if responsehead.headers["Content-Disposition"].split("'")[-1].endswith(".zip"):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
90 unzip(filename, urllib.parse.unquote(os.path.splitext(filename)[0]))
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
91 os.remove(filename)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
92
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
93
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
94 def download_file_from_google_drive(id, dir=""): # https://stackoverflow.com/questions/25010369/wget-curl-large-file-from-google-drive/39225039
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
95 def get_confirm_token(response):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
96 for key, value in response.cookies.items():
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
97 if key.startswith('download_warning'):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
98 return value
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
99
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
100 return None
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
101
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
102 def save_response_content(response):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
103 amountdone = 0
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
104 CHUNK_SIZE = 4096
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
105 if not os.path.exists(output + "/Drive - " + sanitize(i["title"]) + "/" + dir):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
106 os.makedirs(output + "/Drive - " + sanitize(i["title"]) + "/" + dir)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
107 destination = output + "/Drive - " + sanitize(i["title"]) + "/" + dir + "/" + sanitize(response.headers["Content-Disposition"].split("'")[-1])
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
108 if os.path.exists(destination):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
109 filesize = os.stat(destination).st_size
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
110 else:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
111 filesize = 0
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
112
55
4e5000c9b48f Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 54
diff changeset
113 if os.path.exists(destination) and filesize == int(response.headers["Content-Range"].partition('/')[-1]):
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
114 print(" " + os.path.basename(destination) + " already downloaded!")
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
115 return
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
116
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
117 with open(destination, "wb") as f:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
118 for chunk in response.iter_content(CHUNK_SIZE):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
119 if chunk: # filter out keep-alive new chunks
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
120 f.write(chunk)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
121 amountdone += CHUNK_SIZE
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
122 print(" downloading {0}: ".format(os.path.basename(destination)) + " " + str(round((amountdone / int(response.headers["Content-Range"].partition('/')[-1])) * 100, 2)) + "%\r", end="")
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
123 print(" downloaded {0}".format(os.path.basename(destination)) + ": 100.00% ")
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
124
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
125 URL = "https://docs.google.com/uc?export=download"
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
126
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
127 session = requests.Session()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
128
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
129 headers = {
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
130 "Range": "bytes=0-",
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
131 }
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
132
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
133 session.proxies = req.proxies
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
134
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
135 response = session.get(URL, headers=headers, params={'id': id}, stream=True)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
136
55
4e5000c9b48f Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 54
diff changeset
137 while response.status_code == 403:
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
138 time.sleep(30)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
139 response = session.get(URL, headers=headers, params={'id': id}, stream=True)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
140
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
141 if response.status_code == 404:
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
142 return # bypass when root folder has no files
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
143
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
144 token = get_confirm_token(response)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
145
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
146 if token:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
147 params = {'id': id, 'confirm': token}
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
148 response = session.get(URL, headers=headers, params=params, stream=True)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
149
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
150 save_response_content(response)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
151
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
152
49
717feded576e Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 46
diff changeset
153 def sanitize(filename):
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
154 return re.sub(r"[\/:*?\"<>|]", "_", filename).strip()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
155
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
156
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
157 def find_urls(s):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
158 urllist = []
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
159 for findall in re.findall(r"""http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\(\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+""", s):
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
160 urllist.append(findall.split("<")[0].split(">")[-1])
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
161 return urllist
51
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
162
49
717feded576e Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 46
diff changeset
163
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
164 def downloadfile(i, x, count):
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
165 filename = "{4}/{0}_{1}p_{2}_{3}".format(i["id"], count, sanitize(i["title"]), os.path.basename(x["path"]), output)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
166 amountdone = 0
51
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
167 if os.path.exists(filename):
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
168 filesize = os.stat(filename).st_size
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
169 else:
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
170 filesize = 0
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
171 serverhead = req.head("https://data.kemono.party" + x['path'])
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
172 for i in range(500):
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
173 serverfilesize = int(serverhead.headers["Content-Length"])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
174 if filesize < serverfilesize:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
175 with req.get(f"https://data.kemono.party{x['path']}", stream=True, headers={"Range": f"bytes={filesize}-"}) as r:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
176 r.raise_for_status()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
177 with open(filename, "ab") as f:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
178 for chunk in r.iter_content(chunk_size=4096):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
179 f.write(chunk)
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
180 amountdone += len(chunk)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
181 print(" downloading image " + str(count) + ": " + str(round(((filesize + amountdone) / serverfilesize) * 100, 2)) + "%\r", end="")
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
182 print(" downloaded image " + str(count) + ": 100.00% ")
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
183 return
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
184 else:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
185 print(" image " + str(count) + " already downloaded!")
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
186 return
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
187 time.sleep(10)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
188 print(" download timed out!")
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
189 return
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
190
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
191
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
192 def parse_json(i, count):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
193 seen = set()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
194 unique_urls = []
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
195 for url in find_urls(i["content"]):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
196 if url.startswith("https://drive.google.com/drive/folders"):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
197 if url.split("/")[-1].split("?")[0] not in seen:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
198 unique_urls.append(url)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
199 seen.add(url.split("/")[-1].split("?")[0])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
200 elif url.startswith("https://drive.google.com/open?id="):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
201 if url.split("?id=")[-1] not in seen:
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
202 unique_urls.append(req.head(url).headers["Location"])
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
203 seen.add(url.split("/")[-1].split("?")[0])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
204 elif url.startswith("https://drive.google.com/file/"):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
205 if url.split("?")[0].split("/")[-2] not in seen:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
206 unique_urls.append(url)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
207 seen.add(url.split("?")[0].split("/")[-2])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
208 elif url.startswith("https://www.dropbox.com"):
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
209 print(" Dropbox link found! attempting to download its files...")
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
210 download_from_dropbox(url)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
211 else: # TODO: add MEGA, or some sort of other file hosting website(s). gdrive and dropbox seem like the most popular ones atm
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
212 pass
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
213 for url in unique_urls:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
214 if url.startswith("https://drive.google.com/drive/folders/"):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
215 # Google Drive folder downloading
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
216 print(" Google Drive link found! attempting to download its files...")
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
217 download_folder_from_google_drive(url)
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
218 elif url.startswith("https://drive.google.com/file/"):
55
4e5000c9b48f Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 54
diff changeset
219 print(" Google Drive link found! attempting to download its files...")
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
220 download_file_from_google_drive(url.split("?")[0].split("/")[-2])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
221 for x in i["attachments"]:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
222 count += 1
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
223 while not os.path.exists("{4}/{0}_{1}p_{2}_{3}".format(int(i["id"]) - 1, count, sanitize(i["title"]), os.path.basename(x["path"]), output)):
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
224 try:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
225 downloadfile(i, x, count)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
226 break
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
227 except HTTPError:
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
228 while 1:
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
229 time.sleep(10)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
230 downloadfile(i, x, count)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
231 except BadStatusLine: # DDoS-GUARD
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
232 while 1:
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
233 time.sleep(10)
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
234 downloadfile(i, x, count)
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
235 except Exception as e:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
236 print(e)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
237 time.sleep(10)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
238
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
239
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
240 def get_amount_of_posts(s, u):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
241 amount = 0
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
242 while 1:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
243 data = req.get("https://kemono.party/api/{0}/user/{1}?o={2}".format(s, u, amount)).json()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
244 if len(data) < 25:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
245 return math.ceil((amount + 1) / 25)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
246 amount += 25
51
6eb6c657c7eb Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 49
diff changeset
247
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
248
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
249 parser = argparse.ArgumentParser(description="Downloads files from kemono.party")
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
250 parser.add_argument("-u", "--url", help="user URL", metavar='<url>', required=True)
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
251 parser.add_argument("-c", "--cookies", help="", metavar='<cookies>', required=True) # required because of DDoS-GUARD
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
252 parser.add_argument("-p", "--proxy", help="proxy\n supported types: http, https, socks5 (requires pysocks)", metavar='<proxy>') # SOCKS proxy support is through PySocks - pip install pysocks
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
253 parser.add_argument("-o", "--output", help="output folder, defaults to user ID", metavar='<output>')
56
bde647ac9554 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 55
diff changeset
254 parser.add_argument("--test-download-services", dest="testdownloadservices", action="store_true", help="test download services")
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
255 args = parser.parse_args()
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
256
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
257 req = requests.Session()
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
258
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
259 if args.proxy:
53
ae64a0c8831b Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 51
diff changeset
260 req.proxies = {}
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
261 if args.proxy[:6] == "socks5":
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
262 httpproxy = args.proxy
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
263 httpsproxy = args.proxy
53
ae64a0c8831b Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 51
diff changeset
264 elif args.proxy[:5] == "https":
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
265 httpsproxy = args.proxy
53
ae64a0c8831b Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 51
diff changeset
266 elif args.proxy[:4] == "http":
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
267 httpproxy = args.proxy
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
268 else:
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
269 print("unknown proxy format! defaulting to HTTP...")
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
270 httpproxy = args.proxy
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
271 if httpproxy:
53
ae64a0c8831b Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 51
diff changeset
272 req.proxies["http"] = httpproxy
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
273 if httpsproxy:
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
274 req.proxies["https"] = httpsproxy
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
275
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
276 cj = http.cookiejar.MozillaCookieJar(args.cookies)
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
277 cj.load(ignore_expires=True)
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
278 req.cookies = cj
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
279
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
280 try:
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
281 int(args.url)
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
282 print("do not input user IDs in --url! use a link instead")
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
283 exit()
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
284 except Exception:
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
285 pass
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
286
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
287 if args.url.split("/")[-2] == "post":
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
288 service = args.url.split("/")[-5]
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
289 user = args.url.split("/")[-3]
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
290 post = args.url.split("/")[-1]
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
291 elif args.url.split("/")[-2] == "user":
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
292 service = args.url.split("/")[-3]
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
293 user = args.url.split("/")[-1]
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
294
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
295 if not args.output:
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
296 output = user
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
297 else:
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
298 output = args.output
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
299
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
300 if not os.path.isdir(output):
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
301 if os.path.exists(output):
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
302 os.remove(output)
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
303 os.makedirs(output)
46
522ad91a230e Add kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents:
diff changeset
304
54
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
305 try:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
306 post
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
307 pages = 1
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
308 except Exception:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
309 pages = get_amount_of_posts(service, user)
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
310 for page in range(pages):
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
311 try:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
312 post
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
313 userdata = req.get("https://kemono.party/api/{0}/user/{1}/post/{2}".format(service, user, post)).json()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
314 except Exception:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
315 userdata = req.get("https://kemono.party/api/{0}/user/{1}?o={2}".format(service, user, (page * 25))).json()
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
316 for i in userdata:
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
317 print(i["id"])
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
318 count = 0
5a5d47a795c6 Update kemonopartydownloader.py
Paper <37962225+mrpapersonic@users.noreply.github.com>
parents: 53
diff changeset
319 parse_json(i, count)