download_latest_java.py

#!/usr/bin/env python3
# encoding: utf-8


import os
import urllib
import urllib.request
import re

def regex_websearch(url, pattern):
	print(urllib)
	print(urllib.request)
	resp = urllib.request.urlopen(url)
	content = resp.read().decode('UTF-8')
	resp.close()
	match = re.search(pattern, content)
	return match
		

"""
this function downloads the lates Java 


thx to n0ts for the idea
https://gist.github.com/n0ts/40dd9bd45578556f93e7


"""


def download_latest_java(destination, java_version=8, package='server-jre', extension='tar.gz', architecture='linux-x64'): 

	valid_packages = ['jre', 'server-jre', 'jdk']
	if package not in valid_packages:
		print('Invalid Java package selection, valid packages are:')
		for valid_package in valid_packages:
			print('\t' + valid_package)
		return None
	
	
	url = "http://www.oracle.com"

	url_1 = url + "/technetwork/java/javase/downloads/index.html"
	pattern_1 = '\/technetwork\/java/\javase\/downloads\/'+ package + str(java_version) + '-downloads-.+?\.html'
	match = regex_websearch(url_1, pattern_1)
	if match == None:
		print('Unable to download Java from ' + url_1)
		print('Website is down or script is outdated')
		return None

	url_2 = url + match.group(0)
	pattern_2 = "http\:\/\/download.oracle\.com\/otn-pub\/java\/jdk\/[7-9]u[0-9]+?-.+?\/" + package + "-[7-9]u[0-9]+?-" + architecture + "." + extension
	match = regex_websearch(url_2, pattern_2)
	if match == None:
		print('Selected architecture.extension \"' + architecture + '.' + extension + '\" is not available')
		print('Visit \"' + url_2 + '\" to see available architectures and extensions')
		return None

	
	req = urllib.request.Request(match.group(0))
	req.add_header('Cookie', 'oraclelicense=accept-securebackup-cookie')
	if(os.path.exists(destination) and os.path.isdir(destination)):
		destination = destination + '/' + os.path.basename(req.full_url)

	print('Downloading ' + match.group(0) + ' to ' + destination)

	with urllib.request.urlopen(req) as f:
		open(destination, "wb").write(f.read())

	
	return destination

gistfile1.sh

#!/bin/bash

# usage: get_jdk.sh <jdk_version> <rpm|tar>
# jdk_version: default 8
# rpm

JDK_VERSION="8"
EXT="rpm"

if [ -n "$1" ]; then
  if [ "$1" == "7" ]; then
    JDK_VERSION="7"
  fi
fi

if [ -n "$2" ]; then
  if [ "$2" == "tar" ]; then
    EXT="tar.gz"
  fi
fi

URL="http://www.oracle.com"
JDK_DOWNLOAD_URL1="${URL}/technetwork/java/javase/downloads/index.html"
JDK_DOWNLOAD_URL2=`curl -s $JDK_DOWNLOAD_URL1 | grep -Po "\/technetwork\/java/\javase\/downloads\/jdk${JDK_VERSION}-downloads-.+?\.html" | head -1`

if [ -z "$JDK_DOWNLOAD_URL2" ]; then
  echo "Could not get jdk download url - $JDK_DOWNLOAD_URL1"
  exit 1
fi

JDK_DOWNLOAD_URL3="${URL}${JDK_DOWNLOAD_URL2}"

JDK_DOWNLOAD_URL4=`curl -s $JDK_DOWNLOAD_URL3 | egrep -o "http\:\/\/download.oracle\.com\/otn-pub\/java\/jdk\/[7-8]u[0-9]+\-(.*)+\/jdk-[7-8]u[0-9]+(.*)linux-x64.${EXT}"`

wget --no-cookies \
  --no-check-certificate \
  --header "Cookie: oraclelicense=accept-securebackup-cookie" \
  $JDK_DOWNLOAD_URL4