Word filters

master
Bob Mottram 2019-07-14 21:50:27 +01:00
parent 07e751751a
commit e0f3ac4e8c
3 changed files with 93 additions and 1 deletions

View File

@ -32,6 +32,8 @@ from session import createSession
from session import getJson
from blocking import addBlock
from blocking import removeBlock
from filters import addFilter
from filters import removeFilter
import json
import os
import shutil
@ -191,6 +193,10 @@ parser.add_argument('--block', dest='block', type=str,default=None, \
help='Block a particular address')
parser.add_argument('--unblock', dest='unblock', type=str,default=None, \
help='Remove a block on a particular address')
parser.add_argument('--filter', dest='filterStr', type=str,default=None, \
help='Adds a word or phrase which if present will cause a message to be ignored')
parser.add_argument('--unfilter', dest='unfilterStr', type=str,default=None, \
help='Remove a filter on a particular word or phrase')
args = parser.parse_args()
debug=False
@ -614,6 +620,22 @@ if args.unblock:
print('The block on '+args.block+' was removed by '+args.nickname)
sys.exit()
if args.filterStr:
if not args.nickname:
print('Please specify a nickname')
sys.exit()
if addFilter(baseDir,args.nickname,domain,args.filterStr):
print('Filter added: '+args.filterStr)
sys.exit()
if args.unfilterStr:
if not args.nickname:
print('Please specify a nickname')
sys.exit()
if removeFilter(baseDir,args.nickname,domain,args.unfilterStr):
print('Filter removed: '+args.unfilterStr)
sys.exit()
if args.testdata:
useBlurhash=False
nickname='testuser567'

62
filters.py 100644
View File

@ -0,0 +1,62 @@
__filename__ = "filters.py"
__author__ = "Bob Mottram"
__license__ = "AGPL3+"
__version__ = "0.0.1"
__maintainer__ = "Bob Mottram"
__email__ = "bob@freedombone.net"
__status__ = "Production"
import os
def addFilter(baseDir: str,nickname: str,domain: str,words: str) -> bool:
"""Adds a filter for particular words within the content of a incoming posts
"""
filtersFilename=baseDir+'/accounts/'+nickname+'@'+domain+'/filters.txt'
if os.path.isfile(filtersFilename):
if words in open(filtersFilename).read():
return False
filtersFile=open(filtersFilename, "a+")
filtersFile.write(words+'\n')
filtersFile.close()
return True
def removeFilter(baseDir: str,nickname: str,domain: str, \
words: str) -> bool:
"""Removes a word filter
"""
filtersFilename=baseDir+'/accounts/'+nickname+'@'+domain+'/filters.txt'
if os.path.isfile(filtersFilename):
if words in open(filtersFilename).read():
with open(filtersFilename, 'r') as fp:
with open(filtersFilename+'.new', 'w') as fpnew:
for line in fp:
line=line.replace('\n','')
if words not in line:
fpnew.write(line+'\n')
if os.path.isfile(filtersFilename+'.new'):
os.rename(filtersFilename+'.new',filtersFilename)
return True
return False
def isFiltered(baseDir: str,nickname: str,domain: str,content: str) -> bool:
"""Should the given content be filtered out?
This is a simple type of filter which just matches words, not a regex
You can add individual words or use word1+word2 to indicate that two
words must be present although not necessarily adjacent
"""
filtersFilename=baseDir+'/accounts/'+nickname+'@'+domain+'/filters.txt'
if os.path.isfile(filtersFilename):
with open(filtersFilename, 'r') as fp:
for line in fp:
filterStr=line.replace('\n','')
if '+' not in filterStr:
if filterStr in content:
return True
else:
filterWords=filterStr.replace('"','').split('+')
for word in filterWords:
if not word in content:
return False
return True
return False

View File

@ -37,6 +37,7 @@ from capabilities import capabilitiesReceiveUpdate
from like import updateLikesCollection
from like import undoLikesCollectionEntry
from blocking import isBlocked
from filters import isFiltered
def getPersonPubKey(session,personUrl: str,personCache: {},debug: bool) -> str:
if not personUrl:
@ -127,6 +128,13 @@ def savePostToInboxQueue(baseDir: str,httpPrefix: str,nickname: str, domain: str
if isBlocked(baseDir,nickname,domain,postNickname,postDomain):
return None
if postJsonObject.get('object'):
if isinstance(postJsonObject['object'], dict):
if postJsonObject['object'].get('content'):
if isinstance(postJsonObject['object']['content'], str):
if isFiltered(baseDir,nickname,domain,postJsonObject['object']['content']):
return None
if postJsonObject.get('id'):
postId=postJsonObject['id'].replace('/activity','')
else: