237 lines
8.2 KiB
Python
237 lines
8.2 KiB
Python
from django.apps import AppConfig
|
|
from django.http import HttpResponse
|
|
from django.shortcuts import render
|
|
from django.shortcuts import redirect
|
|
#from jinja2 import Environment, FileSystemLoader
|
|
from django.core.files import File
|
|
from kubernetes import client, config
|
|
from celery import shared_task
|
|
|
|
from lxml import html
|
|
import requests
|
|
import os
|
|
import time
|
|
import random
|
|
import mimetypes
|
|
import ruamel.yaml
|
|
import yaml
|
|
from urllib import request as urlrequest
|
|
from urllib import parse as urlparse
|
|
|
|
|
|
|
|
|
|
|
|
proxy_url = 'http://gluetun-svc.gluetun.svc:8888'
|
|
proxy = urlrequest.ProxyHandler({
|
|
'http': proxy_url,
|
|
'https': proxy_url # Add the handler for HTTPS
|
|
})
|
|
|
|
proxies = {
|
|
"http" : proxy_url,
|
|
"https" : proxy_url
|
|
}
|
|
|
|
destination_folder = "/tmp/stash/"
|
|
|
|
def hello(request):
|
|
return HttpResponse('<h1>Hello stash!</h1>')
|
|
def form(request):
|
|
return render(request, 'stash_form.html',{})
|
|
def album(request):
|
|
return render(request, 'album_form.html',{})
|
|
|
|
def check_and_create_destination_folder(name):
|
|
#check if the output directory exist and create if not
|
|
if not os.path.isdir(destination_folder):
|
|
os.makedirs(destination_folder)
|
|
|
|
#check if the named directory exist and create if not
|
|
destination_named_folder = os.path.join(destination_folder,name)
|
|
if not os.path.isdir(destination_named_folder):
|
|
os.makedirs(destination_named_folder)
|
|
|
|
return destination_named_folder
|
|
|
|
def album_form(request):
|
|
name = request.GET.get('name')
|
|
url = request.GET.get('url')
|
|
external_galleries = request.GET.get('external_galleries')
|
|
|
|
album_downloader.delay(name=name,url=url,external_galleries=external_galleries)
|
|
|
|
text_generated = "successfully sent to the album downloader"
|
|
return HttpResponse(text_generated, content_type="text/plain")
|
|
|
|
@shared_task()
|
|
def album_downloader(name,url,external_galleries):
|
|
print(name+" "+url+" "+external_galleries)
|
|
|
|
opener = urlrequest.build_opener(proxy)
|
|
urlrequest.install_opener(opener)
|
|
|
|
req = urlrequest.Request(url)
|
|
urlcurl = urlrequest.urlopen(req).read()
|
|
|
|
page = requests.get(url,proxies=proxies)
|
|
webpage = html.fromstring(page.content)
|
|
|
|
#create the prerequisites folder
|
|
destination_named_folder = check_and_create_destination_folder(name=name)
|
|
|
|
|
|
#get all URLs from the code source (searching for href tag on the code)
|
|
#https://stackoverflow.com/questions/21455349/xpath-query-get-attribute-href-from-a-tag
|
|
|
|
webpage_all_url = webpage.xpath('//a/@href')
|
|
|
|
#search for the href that contains the galleries
|
|
webpage_all_album_url = webpage.xpath("//a[contains(@href,'pornpics.com/galleries/')]/@href")
|
|
#print(webpage_all_album_url[3])
|
|
#print(webpage_all_album_url)
|
|
|
|
|
|
for album in webpage_all_album_url:
|
|
print("Album "+album)
|
|
print("")
|
|
page = requests.get(album,proxies=proxies)
|
|
webpage = html.fromstring(page.content)
|
|
|
|
images_list = webpage.xpath("//a[contains(@href,'cdni.pornpics.com/1280/')]/@href")
|
|
images_downloader.delay(name=name,images_url_list=images_list,destination_named_folder=destination_named_folder)
|
|
print(images_list)
|
|
return 0
|
|
|
|
@shared_task()
|
|
def images_downloader(name,images_url_list,destination_named_folder):
|
|
for image_url in images_url_list:
|
|
split = urlparse.urlsplit(image_url)
|
|
#filename = destination_folder + split.path.split("/")[-1]
|
|
filename = os.path.join(destination_named_folder,split.path.split("/")[-1])
|
|
|
|
|
|
print("downloading: "+filename+" -- "+image_url)
|
|
|
|
opener = urlrequest.build_opener(proxy)
|
|
|
|
urlrequest.install_opener(opener)
|
|
try:
|
|
urlrequest.urlretrieve(image_url, filename )
|
|
except:
|
|
print("Can't download "+filename+" -- "+image_url)
|
|
individual_image_downloader.delay(image_url=image_url, filename=filename)
|
|
|
|
return 0
|
|
|
|
@shared_task()
|
|
def individual_image_downloader(image_url, filename ):
|
|
print("Downloading again: "+filename+" -- "+image_url)
|
|
|
|
opener = urlrequest.build_opener(proxy)
|
|
|
|
urlrequest.install_opener(opener)
|
|
try:
|
|
urlrequest.urlretrieve(image_url, filename )
|
|
except:
|
|
print("Still Can't download "+filename+" -- "+image_url+" == Cancelling")
|
|
return 0
|
|
|
|
|
|
|
|
def incremental(request):
|
|
return render(request, 'incremental_form.html',{})
|
|
|
|
def incremental_form(request):
|
|
name = request.GET.get('name')
|
|
url = request.GET.get('url')
|
|
url_suffix = request.GET.get('url_suffix')
|
|
maximum_increment = int(request.GET.get('maximum_increment'))
|
|
minimum_increment = request.GET.get('minimum_increment')
|
|
|
|
increment_and_download.delay(minimum_increment=minimum_increment,maximum_increment=maximum_increment,url=url,url_suffix=url_suffix,name=name)
|
|
|
|
text_generated = "successfully sent to celery"
|
|
return HttpResponse(text_generated, content_type="text/plain")
|
|
|
|
@shared_task()
|
|
def increment_and_download(minimum_increment,maximum_increment,url,url_suffix,name):
|
|
if minimum_increment == None:
|
|
minimum_increment = 1
|
|
|
|
text_generated = ""
|
|
minimum_increment = int(minimum_increment)
|
|
maximum_increment = int(maximum_increment)
|
|
|
|
#check if the output directory exist and create if not
|
|
if not os.path.isdir(destination_folder):
|
|
os.makedirs(destination_folder)
|
|
|
|
#check if the named directory exist and create if not
|
|
destination_named_folder = os.path.join(destination_folder,name)
|
|
if not os.path.isdir(destination_named_folder):
|
|
os.makedirs(destination_named_folder)
|
|
|
|
print("\n downloading for "+name+" "+url+"XXX"+url_suffix)
|
|
|
|
for increment in range(minimum_increment,maximum_increment):
|
|
print("increment : "+str(increment)+"/"+str(maximum_increment))
|
|
if len(str(increment)) == 3:
|
|
increment_generated ='0' + str(increment)
|
|
elif len(str(increment)) == 2:
|
|
increment_generated ='00' + str(increment)
|
|
elif len(str(increment)) == 1:
|
|
increment_generated ='000' + str(increment)
|
|
|
|
url_generated = url+increment_generated+url_suffix
|
|
destination_filename = increment_generated+url_suffix
|
|
destination_file = os.path.join(destination_named_folder,destination_filename)
|
|
|
|
opener = urlrequest.build_opener(proxy)
|
|
urlrequest.install_opener(opener)
|
|
urlrequest.urlretrieve(url_generated, destination_file )
|
|
|
|
text_generated = "\n downloading for "+name+" "+url+increment_generated+url_suffix
|
|
print("download completed for "+name+" of "+str(maximum_increment)+" images")
|
|
return text_generated
|
|
def handle_survey_form(request):
|
|
config.load_kube_config(config_file='/kubeconfig/config')
|
|
v1 = client.BatchV1Api()
|
|
name = request.GET.get('name')
|
|
url = request.GET.get('url')
|
|
external_galleries = str(request.GET.get('external_galleries'))
|
|
|
|
stash_folder = "/stash"
|
|
|
|
if external_galleries:
|
|
print("external_galleries is not null")
|
|
performer_external_galleries = stash_folder+"/"+name+"_external_galleries.txt"
|
|
with open(performer_external_galleries, "a") as f:
|
|
f.write(external_galleries)
|
|
skip_searching_gallery = True
|
|
else:
|
|
print("external_galleries is null")
|
|
skip_searching_gallery = False
|
|
|
|
with open('django_code/stash_management/templates/template.stash-pornpics-dowloader.job.yml', 'r') as file:
|
|
#get the manifest in a variable template_manifest from the manifest file
|
|
template_manifest = yaml.safe_load(file)
|
|
|
|
#edit the container name
|
|
template_manifest['metadata']['name'] ="stash-pornpics-dowloader-"+name
|
|
template_manifest['spec']['template']['spec']['containers'][0]['name'] ="stash-pornpics-dowloader-"+name
|
|
|
|
#edit the command based on the input
|
|
command=['bash', '/config/download.sh', '--http_proxy', 'http://gluetun-svc.gluetun.svc:8888', '--BASE_URL', url, '--pornstars', name]
|
|
template_manifest['spec']['template']['spec']['containers'][0]['command']=command
|
|
|
|
|
|
#execute based on the value
|
|
api_response = v1.create_namespaced_job(body=template_manifest,namespace='stash')
|
|
return HttpResponse(api_response)
|
|
|
|
|
|
|
|
|
|
|