def test_server_xml_catalina_fallback(docker_cli, image):
environment = {
'CATALINA_CONNECTOR_PROXYNAME': 'crowd.atlassian.com',
'CATALINA_CONNECTOR_PROXYPORT': '443',
'CATALINA_CONNECTOR_SECURE': 'true',
'CATALINA_CONNECTOR_SCHEME': 'https',
'CATALINA_CONTEXT_PATH': '/mycrowd',
}
container = run_image(docker_cli, image, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(
container,
f'{get_app_install_dir(container)}/apache-tomcat/conf/server.xml')
connector = xml.find('.//Connector')
context = xml.find('.//Context')
assert connector.get('proxyName') == environment.get(
'CATALINA_CONNECTOR_PROXYNAME')
assert connector.get('proxyPort') == environment.get(
'CATALINA_CONNECTOR_PROXYPORT')
assert connector.get('secure') == environment.get(
'CATALINA_CONNECTOR_SECURE')
assert connector.get('scheme') == environment.get(
'CATALINA_CONNECTOR_SCHEME')
def test_confluence_xml_cluster_multicast(docker_cli, image, run_user):
environment = {
'ATL_CLUSTER_TYPE': 'multicast',
'ATL_CLUSTER_NAME': 'atl_cluster_name',
'ATL_CLUSTER_TTL': 'atl_cluster_ttl',
'ATL_CLUSTER_ADDRESS': '99.99.99.99'
}
container = run_image(docker_cli,
image,
user=run_user,
environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall(
'.//property[@name="confluence.cluster"]')[0].text == "true"
assert xml.findall('.//property[@name="confluence.cluster.join.type"]'
)[0].text == "multicast"
assert xml.findall('.//property[@name="confluence.cluster.name"]'
)[0].text == "atl_cluster_name"
assert xml.findall('.//property[@name="confluence.cluster.ttl"]'
)[0].text == "atl_cluster_ttl"
assert xml.findall('.//property[@name="confluence.cluster.address"]'
)[0].text == "99.99.99.99"
def test_dbconfig_xml_defaults(docker_cli, image):
environment = {
'ATL_DB_TYPE': 'postgres72',
'ATL_DB_DRIVER': 'org.postgresql.Driver',
'ATL_JDBC_URL':
'jdbc:postgresql://mypostgres.mycompany.org:5432/jiradb',
'ATL_JDBC_USER': '******',
'ATL_JDBC_PASSWORD': '******',
}
container = run_image(docker_cli, image, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/dbconfig.xml')
assert xml.findtext('.//pool-min-size') == '20'
assert xml.findtext('.//pool-max-size') == '100'
assert xml.findtext('.//pool-min-idle') == '10'
assert xml.findtext('.//pool-max-idle') == '20'
assert xml.findtext('.//pool-max-wait') == '30000'
assert xml.findtext('.//validation-query') == 'select 1'
assert xml.findtext('.//time-between-eviction-runs-millis') == '30000'
assert xml.findtext('.//min-evictable-idle-time-millis') == '5000'
assert xml.findtext('.//pool-remove-abandoned') == 'true'
assert xml.findtext('.//pool-remove-abandoned-timeout') == '300'
assert xml.findtext('.//pool-test-while-idle') == 'true'
assert xml.findtext('.//pool-test-on-borrow') == 'false'
def _http_format_output(self, response_content, response_headers):
"""Formats the request response to the desired type"""
try:
if self.output_format == 'text':
output = response_content.decode('utf-8', 'ignore')
elif self.output_format == 'dict':
output = helpers.convert_xml_to_dict(response_content)
elif self.output_format == 'json':
output = helpers.convert_xml_to_json(response_content)
elif self.output_format == 'xml':
output = helpers.parse_xml(response_content)
else:
output = response_content
if self.callback:
return self.callback(output)
if self.return_type:
return output, response_headers['Content-Type']
return output
except Exception as e:
if not self._silent:
logger.warn(
u"Failed format response from uri %s to %s error %s" %
(self.uri, self.output_format, e))
return None
def test_confluence_xml_default(docker_cli, image):
container = run_image(docker_cli, image)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall('.//buildNumber')[0].text == "0"
assert xml.findall('.//property[@name="hibernate.connection.url"]') == []
assert xml.findall('.//property[@name="confluence.cluster.home"]') == []
def test_seraph_defaults(docker_cli, image):
container = run_image(docker_cli, image)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(
container,
f'{get_app_install_dir(container)}/confluence/WEB-INF/classes/seraph-config.xml'
)
param = xml.findall('.//param-name[.="autologin.cookie.age"]') == []
def test_confluence_lucene_index(docker_cli, image):
container = run_image(
docker_cli,
image,
environment={'ATL_LUCENE_INDEX_DIR': '/some/other/dir'})
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall(
'.//property[@name="lucene.index.dir"]')[0].text == '/some/other/dir'
def test_seraph_login_set(docker_cli, image):
container = run_image(docker_cli,
image,
environment={"ATL_AUTOLOGIN_COOKIE_AGE": "TEST_VAL"})
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(
container,
f'{get_app_install_dir(container)}/confluence/WEB-INF/classes/seraph-config.xml'
)
assert xml.findall('.//param-value[.="TEST_VAL"]')[0].text == "TEST_VAL"
def test_seraph_xml_defaults(docker_cli, image):
container = run_image(docker_cli, image)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(
container,
f'{get_app_install_dir(container)}/atlassian-jira/WEB-INF/classes/seraph-config.xml'
)
assert [
el.findtext('.//param-value') for el in xml.findall('.//init-param')
if el.findtext('.//param-name') == 'autologin.cookie.age'
][0] == '1209600'
def test_seraph_xml_params(docker_cli, image):
environment = {'ATL_AUTOLOGIN_COOKIE_AGE': '9001'}
container = run_image(docker_cli, image, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(
container,
f'{get_app_install_dir(container)}/atlassian-jira/WEB-INF/classes/seraph-config.xml'
)
assert [
el.findtext('.//param-value') for el in xml.findall('.//init-param')
if el.findtext('.//param-name') == 'autologin.cookie.age'
][0] == environment.get('ATL_AUTOLOGIN_COOKIE_AGE')
def test_server_xml_access_log(docker_cli, image):
environment = {
'ATL_TOMCAT_ACCESS_LOG': 'true',
'ATL_TOMCAT_PROXY_INTERNAL_IPS': '192.168.1.1',
}
container = run_image(docker_cli, image, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container,
f'{get_app_install_dir(container)}/conf/server.xml')
valve = xml.find(
'.//Context/Valve[@className="org.apache.catalina.valves.RemoteIpValve"]'
)
assert valve.get('internalProxies') == environment.get(
'ATL_TOMCAT_PROXY_INTERNAL_IPS')
def test_server_xml_params(docker_cli, image):
environment = {
'ATL_TOMCAT_MGMT_PORT': '8006',
'ATL_TOMCAT_PORT': '9090',
'ATL_TOMCAT_MAXTHREADS': '201',
'ATL_TOMCAT_MINSPARETHREADS': '11',
'ATL_TOMCAT_CONNECTIONTIMEOUT': '20001',
'ATL_TOMCAT_ENABLELOOKUPS': 'true',
'ATL_TOMCAT_PROTOCOL': 'HTTP/2',
'ATL_TOMCAT_ACCEPTCOUNT': '11',
'ATL_TOMCAT_SECURE': 'true',
'ATL_TOMCAT_SCHEME': 'https',
'ATL_PROXY_NAME': 'conf.atlassian.com',
'ATL_PROXY_PORT': '443',
'ATL_TOMCAT_MAXHTTPHEADERSIZE': '8193',
'ATL_TOMCAT_CONTEXTPATH': '/myconf',
}
container = run_image(docker_cli, image, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container,
f'{get_app_install_dir(container)}/conf/server.xml')
connector = xml.find('.//Connector')
context = xml.find('.//Context')
assert xml.get('port') == environment.get('ATL_TOMCAT_MGMT_PORT')
assert connector.get('port') == environment.get('ATL_TOMCAT_PORT')
assert connector.get('maxThreads') == environment.get(
'ATL_TOMCAT_MAXTHREADS')
assert connector.get('minSpareThreads') == environment.get(
'ATL_TOMCAT_MINSPARETHREADS')
assert connector.get('connectionTimeout') == environment.get(
'ATL_TOMCAT_CONNECTIONTIMEOUT')
assert connector.get('enableLookups') == environment.get(
'ATL_TOMCAT_ENABLELOOKUPS')
assert connector.get('protocol') == environment.get('ATL_TOMCAT_PROTOCOL')
assert connector.get('acceptCount') == environment.get(
'ATL_TOMCAT_ACCEPTCOUNT')
assert connector.get('secure') == environment.get('ATL_TOMCAT_SECURE')
assert connector.get('scheme') == environment.get('ATL_TOMCAT_SCHEME')
assert connector.get('proxyName') == environment.get('ATL_PROXY_NAME')
assert connector.get('proxyPort') == environment.get('ATL_PROXY_PORT')
assert connector.get('maxHttpHeaderSize') == environment.get(
'ATL_TOMCAT_MAXHTTPHEADERSIZE')
assert context.get('path') == environment.get('ATL_TOMCAT_CONTEXTPATH')
def test_dbconfig_xml_params(docker_cli, image, run_user):
environment = {
'ATL_DB_TYPE': 'postgres72',
'ATL_DB_DRIVER': 'org.postgresql.Driver',
'ATL_JDBC_URL': 'jdbc:postgresql://mypostgres.mycompany.org:5432/jiradb',
'ATL_JDBC_USER': '******',
'ATL_JDBC_PASSWORD': '******',
'ATL_DB_SCHEMA_NAME': 'private',
'ATL_DB_MAXIDLE': '21',
'ATL_DB_MAXWAITMILLIS': '30001',
'ATL_DB_MINEVICTABLEIDLETIMEMILLIS': '5001',
'ATL_DB_MINIDLE': '11',
'ATL_DB_POOLMAXSIZE': '101',
'ATL_DB_POOLMINSIZE': '21',
'ATL_DB_REMOVEABANDONED': 'false',
'ATL_DB_REMOVEABANDONEDTIMEOUT': '301',
'ATL_DB_TESTONBORROW': 'true',
'ATL_DB_TESTWHILEIDLE': 'false',
'ATL_DB_TIMEBETWEENEVICTIONRUNSMILLIS': '30001',
'ATL_DB_VALIDATIONQUERY': 'select 2',
}
container = run_image(docker_cli, image, user=run_user, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/dbconfig.xml')
assert xml.findtext('.//database-type') == environment.get('ATL_DB_TYPE')
assert xml.findtext('.//driver-class') == environment.get('ATL_DB_DRIVER')
assert xml.findtext('.//url') == environment.get('ATL_JDBC_URL')
assert xml.findtext('.//username') == environment.get('ATL_JDBC_USER')
assert xml.findtext('.//password') == environment.get('ATL_JDBC_PASSWORD')
assert xml.findtext('.//schema-name') == environment.get('ATL_DB_SCHEMA_NAME')
assert xml.findtext('.//pool-min-size') == environment.get('ATL_DB_POOLMINSIZE')
assert xml.findtext('.//pool-max-size') == environment.get('ATL_DB_POOLMAXSIZE')
assert xml.findtext('.//pool-min-idle') == environment.get('ATL_DB_MINIDLE')
assert xml.findtext('.//pool-max-idle') == environment.get('ATL_DB_MAXIDLE')
assert xml.findtext('.//pool-max-wait') == environment.get('ATL_DB_MAXWAITMILLIS')
assert xml.findtext('.//validation-query') == environment.get('ATL_DB_VALIDATIONQUERY')
assert xml.findtext('.//time-between-eviction-runs-millis') == environment.get('ATL_DB_TIMEBETWEENEVICTIONRUNSMILLIS')
assert xml.findtext('.//min-evictable-idle-time-millis') == environment.get('ATL_DB_MINEVICTABLEIDLETIMEMILLIS')
assert xml.findtext('.//pool-remove-abandoned') == environment.get('ATL_DB_REMOVEABANDONED')
assert xml.findtext('.//pool-remove-abandoned-timeout') == environment.get('ATL_DB_REMOVEABANDONEDTIMEOUT')
assert xml.findtext('.//pool-test-while-idle') == environment.get('ATL_DB_TESTWHILEIDLE')
assert xml.findtext('.//pool-test-on-borrow') == environment.get('ATL_DB_TESTONBORROW')
def test_dbconfig_xml_default_schema_names(docker_cli, image, run_user, atl_db_type):
default_schema_names = {
'mssql': 'dbo',
'mysql': 'public',
'oracle10g': '',
'postgres72': 'public',
}
schema_name = default_schema_names.get(atl_db_type, '')
environment = {
'ATL_DB_TYPE': atl_db_type,
}
container = run_image(docker_cli, image, user=run_user, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/dbconfig.xml')
assert xml.findtext('.//schema-name') == schema_name
def test_confluence_xml_cluster_aws(docker_cli, image, run_user):
environment = {
'ATL_CLUSTER_TYPE': 'aws',
'ATL_HAZELCAST_NETWORK_AWS_IAM_ROLE':
'atl_hazelcast_network_aws_iam_role',
'ATL_HAZELCAST_NETWORK_AWS_IAM_REGION':
'atl_hazelcast_network_aws_iam_region',
'ATL_HAZELCAST_NETWORK_AWS_HOST_HEADER':
'atl_hazelcast_network_aws_host_header',
'ATL_HAZELCAST_NETWORK_AWS_TAG_KEY':
'atl_hazelcast_network_aws_tag_key',
'ATL_HAZELCAST_NETWORK_AWS_TAG_VALUE':
'atl_hazelcast_network_aws_tag_value',
'ATL_CLUSTER_NAME': 'atl_cluster_name',
'ATL_CLUSTER_TTL': 'atl_cluster_ttl'
}
container = run_image(docker_cli,
image,
user=run_user,
environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall(
'.//property[@name="confluence.cluster"]')[0].text == "true"
assert xml.findall(
'.//property[@name="confluence.cluster.join.type"]')[0].text == "aws"
assert xml.findall('.//property[@name="confluence.cluster.aws.iam.role"]'
)[0].text == "atl_hazelcast_network_aws_iam_role"
assert xml.findall('.//property[@name="confluence.cluster.aws.region"]'
)[0].text == "atl_hazelcast_network_aws_iam_region"
assert xml.findall(
'.//property[@name="confluence.cluster.aws.host.header"]'
)[0].text == "atl_hazelcast_network_aws_host_header"
assert xml.findall('.//property[@name="confluence.cluster.aws.tag.key"]'
)[0].text == "atl_hazelcast_network_aws_tag_key"
assert xml.findall('.//property[@name="confluence.cluster.aws.tag.value"]'
)[0].text == "atl_hazelcast_network_aws_tag_value"
assert xml.findall('.//property[@name="confluence.cluster.name"]'
)[0].text == "atl_cluster_name"
assert xml.findall('.//property[@name="confluence.cluster.ttl"]'
)[0].text == "atl_cluster_ttl"
def test_confluence_xml_postgres(docker_cli, image, run_user):
environment = {
'ATL_DB_TYPE': 'postgresql',
'ATL_JDBC_URL': 'atl_jdbc_url',
'ATL_JDBC_USER': '******',
'ATL_JDBC_PASSWORD': '******'
}
container = run_image(docker_cli,
image,
user=run_user,
environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall('.//property[@name="hibernate.connection.url"]'
)[0].text == "atl_jdbc_url"
assert xml.findall('.//property[@name="hibernate.connection.username"]'
)[0].text == "atl_jdbc_user"
assert xml.findall('.//property[@name="hibernate.connection.password"]'
)[0].text == "atl_jdbc_password"
assert xml.findall('.//property[@name="confluence.database.choice"]'
)[0].text == "postgresql"
assert xml.findall(
'.//property[@name="hibernate.dialect"]'
)[0].text == "com.atlassian.confluence.impl.hibernate.dialect.PostgreSQLDialect"
assert xml.findall('.//property[@name="hibernate.connection.driver_class"]'
)[0].text == "org.postgresql.Driver"
assert xml.findall(
'.//property[@name="hibernate.c3p0.min_size"]')[0].text == "20"
assert xml.findall(
'.//property[@name="hibernate.c3p0.max_size"]')[0].text == "100"
assert xml.findall(
'.//property[@name="hibernate.c3p0.timeout"]')[0].text == "30"
assert xml.findall('.//property[@name="hibernate.c3p0.idle_test_period"]'
)[0].text == "100"
assert xml.findall(
'.//property[@name="hibernate.c3p0.max_statements"]')[0].text == "0"
assert xml.findall(
'.//property[@name="hibernate.c3p0.validate"]')[0].text == "false"
assert xml.findall(
'.//property[@name="hibernate.c3p0.acquire_increment"]')[0].text == "1"
assert xml.findall('.//property[@name="hibernate.c3p0.preferredTestQuery"]'
)[0].text == "select 1"
def test_confluence_xml_postgres_all_set(docker_cli, image, run_user):
environment = {
'ATL_DB_TYPE': 'postgresql',
'ATL_JDBC_URL': 'atl_jdbc_url',
'ATL_JDBC_USER': '******',
'ATL_JDBC_PASSWORD': '******',
'ATL_DB_POOLMAXSIZE': 'x100',
'ATL_DB_POOLMINSIZE': 'x20',
'ATL_DB_TIMEOUT': 'x30',
'ATL_DB_IDLETESTPERIOD': 'x100',
'ATL_DB_MAXSTATEMENTS': 'x0',
'ATL_DB_VALIDATE': 'xfalse',
'ATL_DB_ACQUIREINCREMENT': 'x1',
'ATL_DB_VALIDATIONQUERY': 'xselect 1'
}
container = run_image(docker_cli,
image,
user=run_user,
environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall('.//property[@name="hibernate.connection.driver_class"]'
)[0].text == "org.postgresql.Driver"
assert xml.findall(
'.//property[@name="hibernate.dialect"]'
)[0].text == "com.atlassian.confluence.impl.hibernate.dialect.PostgreSQLDialect"
assert xml.findall(
'.//property[@name="hibernate.c3p0.min_size"]')[0].text == "x20"
assert xml.findall(
'.//property[@name="hibernate.c3p0.max_size"]')[0].text == "x100"
assert xml.findall(
'.//property[@name="hibernate.c3p0.timeout"]')[0].text == "x30"
assert xml.findall('.//property[@name="hibernate.c3p0.idle_test_period"]'
)[0].text == "x100"
assert xml.findall(
'.//property[@name="hibernate.c3p0.max_statements"]')[0].text == "x0"
assert xml.findall(
'.//property[@name="hibernate.c3p0.validate"]')[0].text == "xfalse"
assert xml.findall('.//property[@name="hibernate.c3p0.acquire_increment"]'
)[0].text == "x1"
assert xml.findall('.//property[@name="hibernate.c3p0.preferredTestQuery"]'
)[0].text == "xselect 1"
def test_server_xml_defaults(docker_cli, image):
container = run_image(docker_cli, image)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_install_dir(container)}/conf/server.xml')
connector = xml.find('.//Connector')
context = xml.find('.//Context')
assert connector.get('port') == '8080'
assert connector.get('maxThreads') == '100'
assert connector.get('minSpareThreads') == '10'
assert connector.get('connectionTimeout') == '20000'
assert connector.get('enableLookups') == 'false'
assert connector.get('protocol') == 'HTTP/1.1'
assert connector.get('acceptCount') == '10'
assert connector.get('secure') == 'false'
assert connector.get('scheme') == 'http'
assert connector.get('proxyName') == ''
assert connector.get('proxyPort') == ''
def test_server_xml_catalina_fallback(docker_cli, image):
environment = {
'CATALINA_CONNECTOR_PROXYNAME': 'PROXYNAME',
'CATALINA_CONNECTOR_PROXYPORT': 'PROXYPORT',
'CATALINA_CONNECTOR_SECURE': 'SECURE',
'CATALINA_CONNECTOR_SCHEME': 'SCHEME',
'CATALINA_CONTEXT_PATH': 'CONTEXT'
}
container = run_image(docker_cli, image, environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container,
f'{get_app_install_dir(container)}/conf/server.xml')
connector = xml.find('.//Connector')
context = xml.find('.//Context')
assert connector.get('proxyName') == 'PROXYNAME'
assert connector.get('proxyPort') == 'PROXYPORT'
assert connector.get('scheme') == 'SCHEME'
assert connector.get('secure') == 'SECURE'
assert context.get('path') == 'CONTEXT'
def test_confluence_xml_cluster_tcp(docker_cli, image, run_user):
environment = {
'ATL_CLUSTER_TYPE': 'tcp_ip',
'ATL_CLUSTER_PEERS': '1.1.1.1,99.99.99.99',
'ATL_CLUSTER_NAME': 'atl_cluster_name',
}
container = run_image(docker_cli,
image,
user=run_user,
environment=environment)
_jvm = wait_for_proc(container, get_bootstrap_proc(container))
xml = parse_xml(container, f'{get_app_home(container)}/confluence.cfg.xml')
assert xml.findall(
'.//property[@name="confluence.cluster"]')[0].text == "true"
assert xml.findall('.//property[@name="confluence.cluster.join.type"]'
)[0].text == "tcp_ip"
assert xml.findall('.//property[@name="confluence.cluster.name"]'
)[0].text == "atl_cluster_name"
assert xml.findall('.//property[@name="confluence.cluster.peers"]'
)[0].text == "1.1.1.1,99.99.99.99"
def wikify(xml_data_set):
if xml_data_set:
test_dict = helpers.parse_xml(xml_test_data)
whole_set = []
corpus = test_dict['corpus']['@lang']
whole_set.append(corpus)
# loop the every word from xml file
for word in test_dict['corpus']['lexelt']:
lexelt = word['@item']
lex_set = []
lex_set.append(lexelt)
# loop every instance in the current lexeme
for instance in word['instance']:
instance_set = []
word_to_be_disambiguated = instance['context']['head']
#print("word_to_be_disambiguated: ", word_to_be_disambiguated)
#print("\n")
correct_sense = instance['answer']['@senseid']
#print("correct_sense: ", correct_sense)
#print("\n")
context = instance['context']['#text']
#print("Context: ", context)
#print("\n")
# Remove html entities
context = html.unescape(context)
#print("Unescape context: ", context)
#print("\n")
# Remove tags
context = re.sub(r"\[(.*?)\]+", "", context)
#print("Regex context: ", context)
#print("\n")
# Tokenize
context = tokenizer.tokenize(context)
#print("Tokenized context: ", context)
#print("\n")
# Remove stopwords
context = [x for x in context if x not in stop]
print('Stopwordless context: ', context)
#print("\n")
links = linkify(context, word_to_be_disambiguated)
#print("links: ", links)
#print("\n")
instance_set.append(correct_sense)
instance_set.append(links)
lex_set.append(instance_set)
# TODO: find strings in original context and add wiki url?
whole_set.append(lex_set)
return whole_set
else:
print('No XML file provided')
def make_request(
self,
uri=None,
proto="HTTP",
request_type="GET",
headers=None,
output_format="raw",
return_type=False,
no_token=False,
timeout=None,
):
if timeout is None:
timeout = plexpy.CONFIG.PMS_TIMEOUT
valid_request_types = ["GET", "POST", "PUT", "DELETE"]
if request_type.upper() not in valid_request_types:
logger.debug(u"HTTP request made but unsupported request type given.")
return None
if uri:
if proto.upper() == "HTTPS":
if not self.ssl_verify and hasattr(ssl, "_create_unverified_context"):
context = ssl._create_unverified_context()
handler = HTTPSConnection(host=self.host, port=self.port, timeout=timeout, context=context)
logger.warn(u"PlexPy HTTP Handler :: Unverified HTTPS request made. This connection is not secure.")
else:
handler = HTTPSConnection(host=self.host, port=self.port, timeout=timeout)
else:
handler = HTTPConnection(host=self.host, port=self.port, timeout=timeout)
if not no_token:
if headers:
headers.update({"X-Plex-Token": self.token})
else:
headers = {"X-Plex-Token": self.token}
try:
if headers:
handler.request(request_type, uri, headers=headers)
else:
handler.request(request_type, uri)
response = handler.getresponse()
request_status = response.status
request_content = response.read()
content_type = response.getheader("content-type")
except IOError as e:
logger.warn(u"Failed to access uri endpoint %s with error %s" % (uri, e))
return None
except Exception as e:
logger.warn(
u"Failed to access uri endpoint %s. Is your server maybe accepting SSL connections only? %s"
% (uri, e)
)
return None
except:
logger.warn(u"Failed to access uri endpoint %s with Uncaught exception." % uri)
return None
if request_status in (200, 201):
try:
if output_format == "dict":
output = helpers.convert_xml_to_dict(request_content)
elif output_format == "json":
output = helpers.convert_xml_to_json(request_content)
elif output_format == "xml":
output = helpers.parse_xml(request_content)
else:
output = request_content
if return_type:
return output, content_type
return output
except Exception as e:
logger.warn(u"Failed format response from uri %s to %s error %s" % (uri, output_format, e))
return None
else:
logger.warn(u"Failed to access uri endpoint %s. Status code %r" % (uri, request_status))
return None
else:
logger.debug(u"HTTP request made but no enpoint given.")
return None
return
finally:
for line in f.readlines():
tokens = line.rstrip().split('\t')
if tokens[0] == word1 and tokens[1] == word2:
cooccurrences = int(tokens[2])
# print(word1, word2, cooccurrences)
# only one instance of a word pair in a file
f.close()
return cooccurrences
return cooccurrences
if __name__ == "__main__":
test_dict = helpers.parse_xml(xml_test_data)
# loop the every word from xml file
for word in test_dict['corpus']['lexelt']:
# loop every instance in the current lexeme
for instance in word['instance']:
word_to_be_disambiguated = instance['context']['head']
context = instance['context']['#text']
correct_sense = instance['answer']['@senseid']
# TODO remove tags and quotes from context
# can be produce errors by giving weigth to senses that happen to have the keyword
# format context to a list
context = tokenizer.tokenize(context)
# remove the word to be disambiguated: it doesn't provide any information for the context
def make_request(self,
uri=None,
proto='HTTP',
request_type='GET',
headers=None,
output_format='raw',
return_type=False,
no_token=False,
timeout=None):
if timeout is None:
timeout = plexpy.CONFIG.PMS_TIMEOUT
valid_request_types = ['GET', 'POST', 'PUT', 'DELETE']
if request_type.upper() not in valid_request_types:
logger.debug(
u"HTTP request made but unsupported request type given.")
return None
if uri:
if proto.upper() == 'HTTPS':
if not self.ssl_verify and hasattr(
ssl, '_create_unverified_context'):
context = ssl._create_unverified_context()
handler = HTTPSConnection(host=self.host,
port=self.port,
timeout=timeout,
context=context)
logger.warn(
u"PlexPy HTTP Handler :: Unverified HTTPS request made. This connection is not secure."
)
else:
handler = HTTPSConnection(host=self.host,
port=self.port,
timeout=timeout)
else:
handler = HTTPConnection(host=self.host,
port=self.port,
timeout=timeout)
if not no_token:
if headers:
headers.update({'X-Plex-Token': self.token})
else:
headers = {'X-Plex-Token': self.token}
try:
if headers:
handler.request(request_type, uri, headers=headers)
else:
handler.request(request_type, uri)
response = handler.getresponse()
request_status = response.status
request_content = response.read()
content_type = response.getheader('content-type')
except IOError as e:
logger.warn(u"Failed to access uri endpoint %s with error %s" %
(uri, e))
return None
except Exception as e:
logger.warn(
u"Failed to access uri endpoint %s. Is your server maybe accepting SSL connections only? %s"
% (uri, e))
return None
except:
logger.warn(
u"Failed to access uri endpoint %s with Uncaught exception."
% uri)
return None
if request_status in (200, 201):
try:
if output_format == 'dict':
output = helpers.convert_xml_to_dict(request_content)
elif output_format == 'json':
output = helpers.convert_xml_to_json(request_content)
elif output_format == 'xml':
output = helpers.parse_xml(request_content)
else:
output = request_content
if return_type:
return output, content_type
return output
except Exception as e:
logger.warn(
u"Failed format response from uri %s to %s error %s" %
(uri, output_format, e))
return None
else:
logger.warn(
u"Failed to access uri endpoint %s. Status code %r" %
(uri, request_status))
return None
else:
logger.debug(u"HTTP request made but no enpoint given.")
return None
