|
1 | 1 | #NOTE: these ranges are passed to re.compile
|
2 | 2 |
|
3 |
| -gov_misc = ['USAISC', 'DoD', r'\bNASA', 'National Aeronautics', 'Air Force Systems', 'Navy Network Information Center', 'DARPA', r'\bUSDA', '^State of ', '(?i)department', '(?i)agency', 'Shared Services Canada', 'CSRA LLC'] |
| 3 | +gov_misc = ['USAISC', 'DoD', r'\bNASA', 'National Aeronautics', 'Air Force Systems', 'Navy Network Information Center', 'DARPA', r'\bUSDA', '^State of ', '(?i)department', '(?i)agency', 'Shared Services Canada', 'CSRA LLC', 'General Services Commission'] |
4 | 4 |
|
5 |
| -uni_edu = ['(?i)university', '(?i)research', '(?i)academ(y|ic)', '(?i)institute', '(?i)education', '(?i)universidad', '(?i)science', '(?i)scientific', r'(?i)college\b'] |
| 5 | +uni_edu = ['(?i)university', '(?i)research', '(?i)academ(y|ic)', '(?i)institute', '(?i)education', '(?i)universidad', '(?i)science', '(?i)scientific', r'(?i)college\b', '(?i)universitaet'] |
6 | 6 |
|
7 | 7 | nerns = ['WiscNet', 'BELNET', 'CESNET', 'UNINETT', 'SURFnet', '^ARNES$', 'Consortium GARR|Cineca Consorzio',
|
8 | 8 | 'Renater', 'ACONET', 'Tieteen tietotekniikan', '^SUNET ', 'Jisc Services Limited', 'Fundacao para a Ciencia',
|
9 | 9 | 'Entidad Publica Empresarial Red.es', '^BCnet', 'AARNet', 'National Center for High-performance Computing',
|
10 | 10 | 'Deutschen Forschungsnetzes', '^ESnet', 'HEAnet', 'Merit Network', 'Rede Nacional de Ensino e Pesquisa',
|
11 |
| - '[(]RISQ', 'RESTENA', 'Zdruzenie pouzivatelov Slovenskej akademickej', '^SWITCH$'] |
| 11 | + '[(]RISQ', 'RESTENA', 'Zdruzenie pouzivatelov Slovenskej akademickej', '^SWITCH$', |
| 12 | + 'Societe Internationale de Telecommunications Aeronautiques', '^MCNC$', 'OARnet'] |
12 | 13 |
|
13 | 14 | #CDNs, nothing interesting there
|
14 | 15 | cdns = ['Cloudflare', 'Incapsula', 'Akamai', 'Fastly', 'Content Delivery Network Ltd']
|
15 | 16 |
|
16 | 17 | #some hosting providers (kinda), websites + some classic dbs here but not devices. note "EGIHosting" doesn't like being scanned
|
17 |
| -hosting = ['(?i)host', 'Enzu Inc', 'GoDaddy.com', 'PEG TECH INC', 'Strato AG', '(?i)unified layer', '(?i)1&1 internet'] |
| 18 | +hosting = ['(?i)host', 'Enzu Inc', 'GoDaddy.com', 'PEG TECH INC', 'Strato AG', '(?i)unified layer', '(?i)1&1 Ionos'] |
18 | 19 |
|
19 | 20 | #largest ones,
|
20 | 21 | cloud_providers = ['Digital ?Ocean', 'Google (LLC|Ireland|Switz)', 'Amazon([.]com| Data Services)', '(?i)rackspace',
|
21 |
| - 'Microsoft', 'OVH SAS', 'SoftLayer Technologies', 'Linode, LLC', '(?i)hetzner', 'Hangzhou Alibaba Advertising'] |
| 22 | + 'Microsoft', 'OVH SAS', 'SoftLayer Technologies', 'Linode, LLC', '(?i)hetzner', 'Shenzhen Tencent', |
| 23 | + 'Oracle Corporation', r'Leaseweb\b', r'Alibaba\b'] |
| 24 | + |
| 25 | +corporate = ['Daimler AG', 'Apple Inc', 'Eli Lilly and Company', 'Hewlett-Packard Company', 'Alcatel-Lucent', 'NCR Corporation', |
| 26 | + 'The Procter and Gamble Company', 'FUJITSU LIMITED', 'ORANGE BUSINESS SERVICES', 'Wal-Mart Stores', |
| 27 | + 'Cisco Systems', 'SamsungSDS'] |
22 | 28 |
|
23 | 29 | misc = ['^HT$']
|
24 | 30 |
|
|
28 | 34 |
|
29 | 35 | #by default we ignore all of the ranges above as originally this was for devices only
|
30 | 36 |
|
31 |
| -ignore_patterns = gov_misc + uni_edu + nerns + cdns + hosting + cloud_providers + misc |
| 37 | +ignore_patterns = gov_misc + uni_edu + nerns + cdns + hosting + cloud_providers + corporate + misc |
32 | 38 |
|
33 | 39 | #groups are optional, for statistics only
|
34 |
| -ignore_pattern_groups = [['gov + defense', gov_misc], ['universities', uni_edu], ['NERN', nerns], ['CDN', cdns], ['hosting', hosting], ['cloud', cloud_providers]] |
| 40 | +ignore_pattern_groups = [['gov + defense', gov_misc], ['universities', uni_edu], ['NERN', nerns], ['CDN', cdns], ['hosting', hosting], ['cloud', cloud_providers], ['corporate', corporate]] |
35 | 41 |
|
36 | 42 | ####
|
37 | 43 | #### include_patterns is only for include-networks.py
|
|
0 commit comments