def archivedata(urim, preferences):
httpcache = getURICache(urim)
# TODO: only here because we need to detect NotAMemento, need a better solution
memento = memento_resource_factory(urim, httpcache)
archive = ArchiveResource(urim, httpcache)
if preferences['datauri_favicon'].lower() == 'yes':
archive_favicon = convert_imageuri_to_pngdata_uri(
archive.favicon, httpcache, 16, 16
)
else:
archive_favicon = archive.favicon
output = {}
output['urim'] = urim
output['generation-time'] = datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%SZ")
output['archive-uri'] = archive.home_uri
output['archive-name'] = archive.name
output['archive-favicon'] = archive_favicon
output['archive-collection-id'] = archive.collection_id
output['archive-collection-name'] = archive.collection_name
output['archive-collection-uri'] = archive.collection_uri
response = make_response(json.dumps(output, indent=4))
response.headers['Content-Type'] = 'application/json'
response.headers['Preference-Applied'] = \
"datauri_favicon={}".format(preferences['datauri_favicon'])
return response, 200
def test_favicon_from_html(self):
expected_favicon = "http://myarchive.org/content/favicon.ico"
cachedict = {
"http://myarchive.org":
mock_response(headers={},
content="""<html>
<head>
<title>Is this a good title?</title>
<link rel="icon" href="{}">
</head>
<body>Is this all there is to content?</body>
</html>""".format(expected_favicon),
status=200,
url="testing-url://notused"),
expected_favicon:
mock_response(headers={'content-type': 'image/x-testing'},
content="a",
status=200,
url="testing-url://notused")
}
httpcache = mock_httpcache(cachedict)
urim = "http://myarchive.org/20160518000858/http://example.com/somecontent"
x = ArchiveResource(urim, httpcache)
self.assertEqual(x.favicon, expected_favicon)
def test_simplestuff(self):
httpcache = None
urim = "https://myarchive.org/somememento"
x = ArchiveResource(urim, httpcache)
self.assertEqual(x.scheme, "https")
self.assertEqual(x.domain, "myarchive.org")
self.assertEqual(x.name, "MYARCHIVE.ORG")
self.assertEqual(x.uri, "https://myarchive.org")
def test_collection_data_extraction(self):
urim = "http://wayback.archive-it.org/5728/20160518000858/http://blog.willamette.edu/mba/"
requests_cache.install_cache(cachefile)
httpcache = requests.Session()
httpcache.headers.update({'User-Agent': __useragent__})
x = ArchiveResource(urim, httpcache)
self.assertEqual(x.scheme, "http")
self.assertEqual(x.domain, "wayback.archive-it.org")
self.assertEqual(x.name, "ARCHIVE-IT.ORG")
self.assertEqual(x.uri, "http://wayback.archive-it.org")
self.assertEqual(x.home_uri, "https://archive-it.org")
self.assertEqual(x.collection_id, "5728")
self.assertEqual(x.collection_uri,
"https://archive-it.org/collections/5728")
self.assertEqual(x.collection_name, "Social Media")
def test_link_tag_no_rel(self):
expected_favicon = None
cachedict = {
"http://myarchive.org":
mock_response(headers={},
content="""<html>
<head>
<title>Is this a good title?</title>
<link title="a good title" href="content/favicon.ico">
</head>
<body>Is this all there is to content?</body>
</html>""",
status=200,
url="testing-url://notused"),
expected_favicon:
mock_response(headers={'content-type': 'image/x-testing'},
content="a",
status=200,
url="testing-url://notused"),
"http://myarchive.org/favicon.ico":
mock_response(headers={},
content="not found",
status=404,
url="testing-url://notused"),
"https://www.google.com/s2/favicons?domain=myarchive.org":
mock_response(headers={},
content="not found",
status=404,
url="testing-url://notused")
}
httpcache = mock_httpcache(cachedict)
urim = "http://myarchive.org/20160518000858/http://example.com/somecontent"
x = ArchiveResource(urim, httpcache)
self.assertEqual(x.favicon, expected_favicon)