soutade
/
iwla
1
0
Fork 0
Code Issues Pull Requests Releases 2 Activity

WIP

This commit is contained in:
Grégory Soutadé 2014-11-26 16:56:33 +01:00
parent b11411f115
commit e6b31fbf8a
4 changed files with 47 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
awstats_data.py
View File

File diff suppressed because one or more lines are too long

4
iwla_convert.pl
View File

@ -24,7 +24,7 @@ sub dumpList {
{
$first = 0;
}
print $FIC "'.*$r.*'";
print $FIC "'$r'";
}
}
@ -44,7 +44,7 @@ sub dumpHash {
{
$first = 0;
}
print $FIC "'.*$k.*' : '$v'";
print $FIC "'$k' : '$v'";
}
}

55
plugins/display/referers.py
View File

@ -13,6 +13,14 @@ class IWLADisplayReferers(IPlugin):
super(IWLADisplayReferers, self).__init__(iwla)
self.API_VERSION = 1
def _getSearchEngine(self, engine):
for (k, e) in self.search_engines.items():
for hashid in e['hashid']:
if hashid.match(engine):
return k
print 'Not found %s' % (engine)
return None
def load(self):
domain_name = self.iwla.getConfValue('domain_name', '')
@ -23,26 +31,40 @@ class IWLADisplayReferers(IPlugin):
self.own_domain_re = re.compile('.*%s.*' % (domain_name))
self.search_engines = {}
for engine in awstats_data.search_engines:
for (engine, known_url) in awstats_data.search_engines_knwown_url.items():
self.search_engines[engine] = {
're' : re.compile(engine, re.IGNORECASE)
'known_url' : re.compile(known_url + '(?P<key_phrase>.+)'),
'hashid' : []
}
for (hashid, engine) in awstats_data.search_engines_hashid.items():
hashid_re = re.compile('.*%s.*' % (hashid))
if not engine in self.search_engines.keys():
self.search_engines[engine] = {
'hashid' : [hashid_re]
}
else:
self.search_engines[engine]['hashid'].append(hashid_re)
print 'Hashid %s => %s' % (engine, hashid)
for (engine, not_engine) in awstats_data.not_search_engines_keys.items():
if not engine in self.search_engines: continue
self.search_engines[engine]['not_search_engine'] = \
re.compile(not_engine, re.IGNORECASE)
not_engine_re = re.compile('.*%s.*' % (not_engine))
key = self._getSearchEngine(engine)
if key:
self.search_engines[key]['not_search_engine'] = not_engine_re
for (engine, name) in awstats_data.search_engines_hashid.items():
if not engine in self.search_engines: continue
self.search_engines[engine]['name'] = name
for engine in awstats_data.search_engines:
engine_re = re.compile('.*%s.*' % (engine), re.IGNORECASE)
key = self._getSearchEngine(engine)
if key:
self.search_engines[key]['re'] = not_engine_re
for (engine, knwown_url) in awstats_data.search_engines_knwown_url.items():
engine = engin[2:-2]
if not engine in self.search_engines: continue
print knwown_url
self.search_engines[engine]['known_url'] = re.compile(known_url + '(?P<key_phrase>.+)')
for (k,e) in self.search_engines.items():
if not 're' in e.keys():
print 'Remove %s' % k
del self.search_engines[k]
print self.search_engines
self.html_parser = HTMLParser.HTMLParser()
@ -51,7 +73,6 @@ class IWLADisplayReferers(IPlugin):
def _extractKeyPhrase(self, key_phrase_re, parameters, key_phrases):
if not parameters or not key_phrase_re: return
for p in parameters.split('&'):
groups = key_phrase_re.match(p)
if groups:
@ -89,8 +110,8 @@ class IWLADisplayReferers(IPlugin):
parameters = r['extract_referer'].get('extract_parameters', None)
key_phrase_re = e.get('known_url', None)
print parameters
print key_phrase_re
# print parameters
# print key_phrase_re
self._extractKeyPhrase(key_phrase_re, parameters, key_phrases)

2
plugins/pre_analysis/robots.py
View File

@ -11,7 +11,7 @@ class IWLAPreAnalysisRobots(IPlugin):
self.API_VERSION = 1
def load(self):
self.awstats_robots = map(lambda (x) : re.compile(x, re.IGNORECASE), awstats_data.robots)
self.awstats_robots = map(lambda (x) : re.compile(('.*%s.*') % (x), re.IGNORECASE), awstats_data.robots)
return True