





от: cloud 2024-07-17 14:34:52
python скачивание архива изображений вк
[507]
python
https://github.com/cl0oud/publ ...
python save archive image vk.com
[code]import time
import os
#python -m pip install requests
import requests
import sys
sys.setrecursionlimit(1000000000)
def is_url_image(url, j = 0):
#time.sleep(1)
try:
#image_formats = ("image/png", "image/jpeg", "image/jpg")
#r = requests.head(url)
#if r.headers["content-type"] in image_formats:
# return True
#return False
r = requests.get(url)
if r.status_code == 200:
return True
return False
except:
if j < 3:
j += 1
is_url_image(url, j)
else:
return False
def fileSave(url, fileTest, j = 0):
print(url + ' - ' + fileTest)
if j < 3:
if not os.path.exists(fileTest):
if is_url_image(url):
response = requests.get(url)
with open(fileTest, 'wb') as file:
file.write(response.content)
#time.sleep(1)
if not os.path.exists(fileTest):
j += 1
fileSave(url, fileTest, j)
def fileDownload(album, file, save_dir, string, start, end, i):
if start in string:
if end in string:
url1 = string.split(start, 1)
url2 = url1[1].split(end, 1)
url = url2[0]
string = string.replace(start + url + end, '')
if url != '':
fileTest = save_dir + album + '_' + str(i) + '.jpg'
fileSave(url, fileTest)
i += 1
fileDownload(album, file, save_dir, string, start, end, i)
# создаём папки, даём разрешение на запись
# там где хранятся файлы .html альбомов
dir_file = '/dir/file/';
# куда будем сохранять
dir_image = '/dir/image/';
start = '><img src="'
end = '" alt="'
scan_f = os.listdir(dir_file)
for file in scan_f:
...
[507]

https://github.com/cl0oud/publ ...
python save archive image vk.com
[code]import time
import os
#python -m pip install requests
import requests
import sys
sys.setrecursionlimit(1000000000)
def is_url_image(url, j = 0):
#time.sleep(1)
try:
#image_formats = ("image/png", "image/jpeg", "image/jpg")
#r = requests.head(url)
#if r.headers["content-type"] in image_formats:
# return True
#return False
r = requests.get(url)
if r.status_code == 200:
return True
return False
except:
if j < 3:
j += 1
is_url_image(url, j)
else:
return False
def fileSave(url, fileTest, j = 0):
print(url + ' - ' + fileTest)
if j < 3:
if not os.path.exists(fileTest):
if is_url_image(url):
response = requests.get(url)
with open(fileTest, 'wb') as file:
file.write(response.content)
#time.sleep(1)
if not os.path.exists(fileTest):
j += 1
fileSave(url, fileTest, j)
def fileDownload(album, file, save_dir, string, start, end, i):
if start in string:
if end in string:
url1 = string.split(start, 1)
url2 = url1[1].split(end, 1)
url = url2[0]
string = string.replace(start + url + end, '')
if url != '':
fileTest = save_dir + album + '_' + str(i) + '.jpg'
fileSave(url, fileTest)
i += 1
fileDownload(album, file, save_dir, string, start, end, i)
# создаём папки, даём разрешение на запись
# там где хранятся файлы .html альбомов
dir_file = '/dir/file/';
# куда будем сохранять
dir_image = '/dir/image/';
start = '><img src="'
end = '" alt="'
scan_f = os.listdir(dir_file)
for file in scan_f:
...
загрузить
ещё
ещё