1
0
mirror of https://github.com/webrecorder/pywb.git synced 2025-03-20 18:59:11 +01:00
pywb/webagg/test/test_upstream.py

75 lines
2.7 KiB
Python
Raw Normal View History

2016-03-08 10:29:59 -08:00
import webtest
from io import BytesIO
from webagg.app import ResAggApp
2016-03-08 10:29:59 -08:00
import requests
from webagg.handlers import DefaultResourceHandler
from webagg.aggregator import SimpleAggregator
from webagg.proxyindexsource import ProxyMementoIndexSource, UpstreamAggIndexSource
2016-03-08 10:29:59 -08:00
from pywb.warc.recordloader import ArcWarcRecordLoader
from .testutils import LiveServerTests, BaseTestClass
2016-03-08 10:29:59 -08:00
class TestUpstream(LiveServerTests, BaseTestClass):
2016-03-08 10:29:59 -08:00
def setup(self):
app = ResAggApp()
base_url = 'http://localhost:{0}'.format(self.server.port)
2016-03-08 10:29:59 -08:00
app.add_route('/upstream',
DefaultResourceHandler(SimpleAggregator(
{'upstream': UpstreamAggIndexSource(base_url + '/live')})
)
)
app.add_route('/upstream_opt',
DefaultResourceHandler(SimpleAggregator(
{'upstream_opt': ProxyMementoIndexSource.upstream_resource(base_url + '/live')})
)
)
self.base_url = base_url
self.testapp = webtest.TestApp(app)
2016-03-08 10:29:59 -08:00
def test_live_paths(self):
res = requests.get(self.base_url + '/')
assert set(res.json().keys()) == {'/live/postreq', '/live'}
2016-03-08 10:29:59 -08:00
def test_upstream_paths(self):
res = self.testapp.get('/')
assert set(res.json.keys()) == {'/upstream/postreq', '/upstream', '/upstream_opt', '/upstream_opt/postreq'}
def test_live_1(self):
resp = requests.get(self.base_url + '/live/resource?url=http://httpbin.org/get', stream=True)
assert resp.headers['WebAgg-Source-Coll'] == 'live'
2016-03-08 10:29:59 -08:00
record = ArcWarcRecordLoader().parse_record_stream(resp.raw, no_record_parse=False)
assert record.rec_headers.get_header('WARC-Target-URI') == 'http://httpbin.org/get'
assert record.status_headers.get_header('Date') != ''
def test_upstream_1(self):
resp = self.testapp.get('/upstream/resource?url=http://httpbin.org/get')
assert resp.headers['WebAgg-Source-Coll'] == 'upstream:live'
2016-03-08 10:29:59 -08:00
raw = BytesIO(resp.body)
record = ArcWarcRecordLoader().parse_record_stream(raw, no_record_parse=False)
assert record.rec_headers.get_header('WARC-Target-URI') == 'http://httpbin.org/get'
assert record.status_headers.get_header('Date') != ''
def test_upstream_2(self):
resp = self.testapp.get('/upstream_opt/resource?url=http://httpbin.org/get')
assert resp.headers['WebAgg-Source-Coll'] == 'upstream_opt:live', resp.headers
2016-03-08 10:29:59 -08:00
raw = BytesIO(resp.body)
record = ArcWarcRecordLoader().parse_record_stream(raw, no_record_parse=False)
assert record.rec_headers.get_header('WARC-Target-URI') == 'http://httpbin.org/get'
assert record.status_headers.get_header('Date') != ''