3
ܓi! @ s( d dl Z d dlmZ G dd deZdS ) N)SoSMapc
s e Zd ZdZdddgZddgZd#ZdZdZdZ dZ
i Z i Z fddZ
dd Zdd Z fddZdd Zd d Zd d Zd!d" Z ZS )$SoSHostnameMapaL Mapping store for hostnames and domain names
Hostnames are obfuscated using an incrementing counter based on the total
number of hosts matched regardless of domain name.
Domain names are obfuscated based on the host's hostname, plus any user
defined domains passed in by the `--domains` options.
Domains are obfuscated as whole units, meaning the domains 'example.com'
and 'host.foo.example.com' will be separately obfuscated with no relation
for example as 'obfuscatedomdain1.com' and 'obfuscatedomain2.com'.
Top-level domains are left untouched.
Z localhostz.*localdomain.*z^com..*ZwwwZapi.yaml.yml.crt.key.pem.log.repo.rules.conf.cfgTr c s d|kr|j dd}t j|S )zOverride the base get_regex_result() to provide a regex that, if
this is an FQDN or a straight domain, will include an underscore
formatted regex as well.
.z(\.|_))replacesuperget_regex_result)selfitem) __class__ "/usr/lib/python3.6/hostname_map.pyr 5 s z SoSHostnameMap.get_regex_resultc C s y2t | jj ddd jdd }t|d | _W n tk
rF Y nX y s z!SoSHostnameMap.set_initial_countsc sj || j krdS |jd}dj|jddd t|dkrJ|d | jkS t fdd| j D rfdS dS ) zkCheck if a potential domain is in one of the domains we've loaded
and should be obfuscated
Tr r r c 3 s | ]} j |V qd S )N)endswith).0Z_d)no_tldr r ] s z?SoSHostnameMap.domain_name_in_loaded_domains..F)r" r joinlenr any)r r r r )r) r domain_name_in_loaded_domainsR s
z,SoSHostnameMap.domain_name_in_loaded_domainsc s d}d}d }x&|j d r2||d 7 }|dd }qW x&|jd
rZ||d 7 }|dd }q6W || jkrp| j| S | j|j sdj|||gS |j| jrd|jdd
}|j|d}||7 }|| jkrxt | jj
dt dD ]}d}|j|}|jd} | d r | d | j kr d}t |dks|d r6q| rh|d jdsZ|j|rht
j|}P |j|d ry,|j|d d }
|d t
j|
}P W q tk
r Y qX qW |st
j|}|| | S )N r _r r T)r keyF)r r1 )r r1 r+ r+ r+ )
startswithr' datasetr/ lowerr,
strip_extsr r r keysr- r r get Exception) r r prefixsuffixfinalZextZ _existingZ _host_substrZ_testZ_hZitm)r r r r8 b sR
zSoSHostnameMap.getc C s |j d}t|dkr(| j|d j S t|dkr\| j|}tdd |D rX|j }|S t|dkr|d }|dd }t|dkr| j|j }nd}| j|}|| j|< dj||g}tdd |D r|j }|S d S ) Nr r r c s s | ]}|j V qd S )N)isupper)r( r$ r r r r* s z/SoSHostnameMap.sanitize_item..unknownc s s | ]}|j V qd S )N)r> )r( r$ r r r r* s ) r r- sanitize_short_namer5 sanitize_domainallupperr4 r, ) r r r dnamehostnamer Z ob_hostname ob_domainZ_fqdnr r r
sanitize_item s*
z SoSHostnameMap.sanitize_itemc C s` | s|| j kr|S || jkrVd| j }|| j|< | jd7 _|| j|< | j| | j| S )zObfuscate the short name of the host with an incremented counter
based on the total number of obfuscated host names
r r ) skip_keysr4 r r Zadd_regex_item)r rE Zob_hostr r r r@ s
z"SoSHostnameMap.sanitize_short_namec C sz x*| j D ] }tj|dj|rdj|S qW |d j }dj|dd j }| j|}dj||g}|| jdj|< |S )zeObfuscate the domainname, broken out into subdomains. Top-level
domains are ignored.
r r r r+ r+ )ignore_matchesrematchr, r5 _new_obfuscated_domainr4 )r r Z_skipZ
top_domainrD rF r r r rA s
z SoSHostnameMap.sanitize_domainc C s4 || j kr*d| j | j |< | jd7 _| j | S )zDGenerate an obfuscated domain for each subdomain name given
Zobfuscateddomainr )r" r# )r rD r r r rL s
z%SoSHostnameMap._new_obfuscated_domain)
r r r r r r r
r r r
)__name__
__module__ __qualname____doc__rI rH r6 Zignore_short_itemsZmatch_full_words_onlyr r# r" r r r& r/ r8 rG r@ rA rL
__classcell__r r )r r r s, = r )rJ Zsos.cleaner.mappingsr r r r r r s