3
ܓi! @ s6 d dl Z d dlZd dlZd dlmZ G dd dZdS ) N)Pathc @ s e Zd ZdZg Zg ZdZdZdZdd Z dd Z
dd Z d
d Z d d
Z
dd Zdd Zdd Zdd Zdd Zdd Zdd Zd S ) SoSMapzStandardized way to store items with their obfuscated counterparts.
Each type of sanitization that SoSCleaner supports should have a
corresponding SoSMap() object, to allow for easy retrieval of obfuscated
items.
TFc C sJ i | _ t | _g | _| jjj | _|| _t j
j | jd| j| _ | j
d S )NZ
cleaner_cache)datasetset
_regexes_madecompiled_regexes __class____name__lowerZcnameworkdirospathjoin cache_dir load_entries)selfr r /usr/lib/python3.6/__init__.py__init__ s
zSoSMap.__init__c C s" t | jjddd | jd dS )ai Load cached entries from the disk. This method must be called when
we initialize a Map instance and whenever we want to retrieve
self.dataset (e.g. to store default_mapping file). The later is
essential since a concurrent Map can add more objects to the cache,
so we need to update self.dataset up to date.
Keep in mind that size of self.dataset is usually bigger than number
of files in the corresponding cleaner's directory: directory contains
just whole items (e.g. IP addresses) while dataset contains more
derived objects (e.g. subnets).
T)parentsexist_ok N)r r mkdirload_new_entries_from_dir)r r r r r , s
zSoSMap.load_entriesc C s\ | s0|| j ks0|| jj ks0| jr4t|dkr4dS x"| jD ]}tj||tjr s z'SoSMap.add_regex_item..T)keyreverse)r% r r
r= r appendget_regex_resultsort)r r# r r r r+ s
zSoSMap.add_regex_itemc C s2 | j rdtj| d}n
tj|}tj|tjS )a9 Generate the object/value that is used by the parser when iterating
over pre-generated regexes during parse_line(). For most parsers this
will simply be a ``re.Pattern()`` object, but for more complex parsers
this can be overridden to provide a different object, e.g. a tuple,
for that parer's specific iteration needs.
:param item: The unobfuscated string to generate the regex for
:type item: ``str``
:returns: A compiled regex pattern for the item
:rtype: ``re.Pattern``
z
(? s