data_importer/lib/cve_list_importer.rb

96 lines
2.3 KiB
Ruby
Raw Normal View History

require 'git'
require 'json'
require 'date'
require 'bulk_insert'
# This class can be used to import cvelist json data from mitre from their github repo
class CveListImporter
attr_accessor :repo_url, :repo_path
def initialize
@repo_url = 'https://github.com/CVEProject/cvelist.git'
@repo_path = '/data_importer/data/cve_list'
end
def git_clone_repo
Git.clone(repo_url, repo_path)
end
def pull_latest_changes
`cd #{repo_path}; git pull;`
puts "Now pulling latest changes from #{repo_path}"
end
def read_json(filename)
JSON.parse(File.read(filename))
end
def list_jsons_for_year(year)
year_fp = "#{repo_path}/#{year}/*"
Dir["#{year_fp}/*.json"]
end
def read_jsons_for_year(year)
filenames = list_jsons_for_year(year)
filenames.map { |filename| read_json(filename) }
end
def read_all_jsons
(1999..Date.today.year).map do |year|
read_jsons_for_year(year.to_s)
end
end
def cve_attrs_from_item(json)
cve_attrs = {}
cve_attrs[:cve_data_meta] = json['CVE_data_meta']
cve_attrs[:cve_id] = json['CVE_data_meta']['ID']
cve_attrs[:affects] = json['affects']
cve_attrs[:data_format] = json['data_format']
cve_attrs[:data_type] = json['data_type']
cve_attrs[:data_version] = json['data_version']
cve_attrs[:description] = json['description']
cve_attrs[:impact] = json['impact']
cve_attrs[:problemtype] = json['problemtype']
cve_attrs[:references] = json['references']
cve_attrs[:source] = json['source']
cve_attrs
end
# for bulk inserting
def cves_for_year(year)
json_data = read_jsons_for_year(year)
json_data.map do |json_f|
cve_attrs_from_item(json_f)
end
end
def import
if Dir.exist?(repo_path)
pull_latest_changes
else
git_clone_repo
end
(1999..Date.today.year).map do |year|
cves_from_json = cves_for_year(year)
ids = cves_from_json.map { |cve| cve[:cve_id] }
cve_ids_in_db = Cve.where(:cve_id => ids).pluck(:cve_id)
new_cve_ids = ids - cve_ids_in_db
new_cves = cves_from_json.select { |cve| cve if new_cve_ids.include?(cve[:cve_id]) }
puts "Importing any new CVEs from #{year}"
bulk_insert(new_cves)
end
end
def bulk_insert(cves)
Cve.bulk_insert do |worker|
cves.each do |attrs|
worker.add(attrs)
end
end
end
end