|
| 1 | +# Author: Navonil Das (@NavonilDas) |
| 2 | +import semantic_version |
| 3 | +import re |
| 4 | +import json |
| 5 | +from urllib.request import urlopen |
| 6 | + |
| 7 | +NPM_URL = 'https://registry.npmjs.org{}' |
| 8 | +PAGE = '/-/npm/v1/security/advisories?page=1' |
| 9 | + |
| 10 | +# Removes spaces and v Charecter in front of version |
| 11 | +def remove_spaces(x): |
| 12 | + x = re.sub(r" +"," ",x) # Replace Multiple spaces to one |
| 13 | + # Remove space after the Relational operator |
| 14 | + x = re.sub(r'< +','<',x) |
| 15 | + x = re.sub(r'> +','>',x) |
| 16 | + x = re.sub(r'<= +','<=',x) |
| 17 | + x = re.sub(r'>= +','>=',x) |
| 18 | + # Remove v at starting of version |
| 19 | + x = re.sub(r'>=[vV]','>=',x) |
| 20 | + x = re.sub(r'<=[vV]','<=',x) |
| 21 | + x = re.sub(r'>[vV]','>',x) |
| 22 | + x = re.sub(r'<[vV]','<',x) |
| 23 | + return x |
| 24 | + |
| 25 | + |
| 26 | +# Returns all available for a module |
| 27 | +def get_all_version(package_name): |
| 28 | + package_url = NPM_URL.format('/'+package_name) |
| 29 | + response = urlopen(package_url).read() |
| 30 | + data = json.loads(response) |
| 31 | + versions = data.get('versions',{}) |
| 32 | + all_version = [obj for obj in versions] |
| 33 | + return all_version |
| 34 | + |
| 35 | +# Seperate list of Affected version and fixed version from all version |
| 36 | +# using the range specified |
| 37 | +def extract_version(package_name,aff_version_range,fixed_version_range): |
| 38 | + if aff_version_range == '' or fixed_version_range == '': |
| 39 | + return ([],[]) |
| 40 | + |
| 41 | + aff_spec = semantic_version.NpmSpec(remove_spaces(aff_version_range)) |
| 42 | + fix_spec = semantic_version.NpmSpec(remove_spaces(fixed_version_range)) |
| 43 | + all_ver = get_all_version(package_name) |
| 44 | + aff_ver = [] |
| 45 | + fix_ver = [] |
| 46 | + for ver in all_ver: |
| 47 | + cur_version = semantic_version.Version(ver) |
| 48 | + if cur_version in aff_spec: |
| 49 | + aff_ver.append(ver) |
| 50 | + else: |
| 51 | + if cur_version in fix_spec: |
| 52 | + fix_ver.append(ver) |
| 53 | + |
| 54 | + return (aff_ver,fix_ver) |
| 55 | + |
| 56 | + |
| 57 | +# Extract module name, summary, vulnerability id,severity |
| 58 | +def extract_data(JSON): |
| 59 | + package_vulnerabilities = [] |
| 60 | + for obj in JSON.get('objects',[]): |
| 61 | + if 'module_name' not in obj: |
| 62 | + continue |
| 63 | + package_name = obj['module_name'] |
| 64 | + summary = obj.get('overview','') |
| 65 | + severity = obj.get('severity','') |
| 66 | + |
| 67 | + vulnerability_id = obj.get('cves',[]) |
| 68 | + if len(vulnerability_id) > 0: |
| 69 | + vulnerability_id = vulnerability_id[0] |
| 70 | + else: |
| 71 | + vulnerability_id = '' |
| 72 | + |
| 73 | + affected_version,fixed_version = extract_version( |
| 74 | + package_name, |
| 75 | + obj.get('vulnerable_versions',''), |
| 76 | + obj.get('patched_versions','') |
| 77 | + ) |
| 78 | + |
| 79 | + package_vulnerabilities.append({ |
| 80 | + 'package_name': package_name, |
| 81 | + 'summary': summary, |
| 82 | + 'vulnerability_id':vulnerability_id, |
| 83 | + 'fixed_version':fixed_version, |
| 84 | + 'affected_version':affected_version, |
| 85 | + 'severity':severity |
| 86 | + }) |
| 87 | + return package_vulnerabilities |
| 88 | + |
| 89 | + |
| 90 | +# Extract JSON From NPM registry |
| 91 | +def scrape_vulnerabilities(): |
| 92 | + cururl = NPM_URL.format(PAGE) |
| 93 | + response = urlopen(cururl).read() |
| 94 | + package_vulnerabilities = [] |
| 95 | + while True: |
| 96 | + data = json.loads(response) |
| 97 | + package_vulnerabilities = package_vulnerabilities + extract_data(data) |
| 98 | + next_page = data.get('urls',{}).get('next',False) |
| 99 | + if next_page: |
| 100 | + cururl = NPM_URL.format(next_page) |
| 101 | + response = urlopen(cururl).read() |
| 102 | + else: |
| 103 | + break |
| 104 | + return package_vulnerabilities |
| 105 | + |
0 commit comments