def obfuscate_wiki_entry(self, line, user_profile):
fields = line.rstrip('\r\n').decode('utf8').split('\t')
record = ArticleRevisionRecord(*fields)
user_info = {}
if user_profile is not None:
user_id = record.user_id
if user_id != 'NULL':
profile_entry = user_profile.get(user_id)
if profile_entry is None:
log.error("Missing profile entry for user_id %s", user_id)
else:
user_info['name'] = [
profile_entry.name,
]
if record.ip_address != 'NULL' and record.ip_address != 'ip_address':
log.warning("Found non-NULL IP address")
if record.automatic_log != '' and record.automatic_log != 'automatic_log':
log.warning(u"Found non-zero-length automatic_log: %s",
record.automatic_log)
# Can't reset values, so update original fields.
fields[12] = backslash_encode_value(
self.obfuscator.obfuscate_text(
backslash_decode_value(record.content), user_info))
fields[2] = backslash_encode_value(
self.obfuscator.obfuscate_text(
backslash_decode_value(record.user_message), user_info))
return u"\t".join(fields).encode('utf-8')
def filter_row(self, row):
user_id = row[5]
user_info = {}
if user_id != 'NULL':
user_id = int(user_id)
user_info['user_id'] = [user_id, ]
try:
entry = self.user_by_id[user_id]
if 'username' in entry:
user_info['username'] = [entry['username'], ]
if 'name' in entry:
user_info['name'] = [entry['name'], ]
except KeyError:
log.error("Unable to find wiki user_id: %s in the user_by_id map", user_id)
row[2] = '' # user_message
row[3] = '' # automatic_log
row[4] = '' # ip_address
# For user_id, preserve 'NULL' value if present.
if user_id != 'NULL':
row[5] = self.remap_id(user_id)
wiki_content = backslash_decode_value(row[12].decode('utf8'))
cleaned_content = self.obfuscator.obfuscate_text(wiki_content, user_info)
row[12] = backslash_encode_value(cleaned_content).encode('utf8')
return row
def filter_row(self, row):
user_id = row[5]
user_info = {}
if user_id != 'NULL':
user_id = int(user_id)
user_info['user_id'] = [user_id, ]
try:
entry = self.user_by_id[user_id]
if 'username' in entry:
user_info['username'] = [entry['username'], ]
if 'name' in entry:
user_info['name'] = [entry['name'], ]
except KeyError:
log.error("Unable to find wiki user_id: %s in the user_by_id map", user_id)
row[2] = '' # user_message
row[3] = '' # automatic_log
row[4] = '' # ip_address
# For user_id, preserve 'NULL' value if present.
if user_id != 'NULL':
row[5] = self.remap_id(user_id)
wiki_content = backslash_decode_value(row[12].decode('utf8'))
cleaned_content = self.obfuscator.obfuscate_text(wiki_content, user_info)
row[12] = backslash_encode_value(cleaned_content).encode('utf8')
return row
def obfuscate_wiki_entry(self, line, user_profile):
fields = line.rstrip('\r\n').decode('utf8').split('\t')
record = ArticleRevisionRecord(*fields)
user_info = {}
if user_profile is not None:
user_id = record.user_id
if user_id != 'NULL':
profile_entry = user_profile.get(user_id)
if profile_entry is None:
log.error("Missing profile entry for user_id %s", user_id)
else:
user_info['name'] = [profile_entry.name, ]
if record.ip_address != 'NULL' and record.ip_address != 'ip_address':
log.warning("Found non-NULL IP address")
if record.automatic_log != '' and record.automatic_log != 'automatic_log':
log.warning(u"Found non-zero-length automatic_log: %s", record.automatic_log)
# Can't reset values, so update original fields.
fields[12] = backslash_encode_value(self.obfuscator.obfuscate_text(backslash_decode_value(record.content), user_info))
fields[2] = backslash_encode_value(self.obfuscator.obfuscate_text(backslash_decode_value(record.user_message), user_info))
return u"\t".join(fields).encode('utf-8')
def load_user_info(userinfo_path):
"""Reads a custom user-info file from the local fs that contains username, email, user-id, fullname."""
result = {}
with open(userinfo_path, 'r') as infile:
next(infile, None)
for line in infile:
fields = line.rstrip('\r\n').decode('utf8').split('\t')
# Once split, we can clean up the individual entries, and interpret the embedded newlines and tabs.
# We'll also strip here, to remove the additional whitespace on usernames and fullnames.
fields = [backslash_decode_value(field).strip() for field in fields]
record = UserInfoRecord(*fields)
entry = {key: [value, ] for key, value in record.__dict__.iteritems()}
# Store records twice, once with an int key, and once with a string key.
# (They should therefore not collide.)
result[int(record.user_id)] = entry
result[record.username] = entry
return result
def load_user_info(userinfo_path):
"""Reads a custom user-info file from the local fs that contains username, email, user-id, fullname."""
result = {}
with open(userinfo_path, 'r') as infile:
next(infile, None)
for line in infile:
fields = line.rstrip('\r\n').decode('utf8').split('\t')
# Once split, we can clean up the individual entries, and interpret the embedded newlines and tabs.
# We'll also strip here, to remove the additional whitespace on usernames and fullnames.
fields = [
backslash_decode_value(field).strip() for field in fields
]
record = UserInfoRecord(*fields)
entry = {
key: [
value,
]
for key, value in record.__dict__.iteritems()
}
# Store records twice, once with an int key, and once with a string key.
# (They should therefore not collide.)
result[int(record.user_id)] = entry
result[record.username] = entry
return result
def test_decoding(self, text, expected_result):
self.assertEquals(obfuscate_util.backslash_decode_value(text),
expected_result)
def test_encoding_round_trip(self, text):
self.assertEquals(
text,
obfuscate_util.backslash_decode_value(
obfuscate_util.backslash_encode_value(text)))
def test_decoding(self, text, expected_result):
self.assertEquals(obfuscate_util.backslash_decode_value(text), expected_result)
def test_encoding_round_trip(self, text):
self.assertEquals(text, obfuscate_util.backslash_decode_value(obfuscate_util.backslash_encode_value(text)))
