From 3bae05c4a4cd7d6d9b892b952b7ca35454319479 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Fri, 11 Jun 2021 15:03:32 -0700 Subject: update access redirect URL endpoints --- tests/test_web.py | 39 ++++++++++++++++++++------------------- 1 file changed, 20 insertions(+), 19 deletions(-) (limited to 'tests') diff --git a/tests/test_web.py b/tests/test_web.py index ee11ee6..7da5880 100644 --- a/tests/test_web.py +++ b/tests/test_web.py @@ -125,11 +125,8 @@ def test_basic_work_landing_page(client: Any, mocker: Any) -> None: def test_basic_access_redirect(client: Any, mocker: Any) -> None: - """ - NOTE: DEPRECATED - """ - with open("tests/files/elastic_fulltext_search.json") as f: + with open("tests/files/elastic_fulltext_get.json") as f: elastic_resp = json.loads(f.read()) es_raw = mocker.patch( @@ -141,7 +138,7 @@ def test_basic_access_redirect(client: Any, mocker: Any) -> None: ] rv = client.get( - "/access-redirect/f81f84e23c9ba5d364c70f01fa26e645d29c0427.pdf", + "/work/2x5qvct2dnhrbctqa2q2uyut6a/access/wayback/https://www.federalreserve.gov/econresdata/feds/2015/files/2015118pap.pdf", allow_redirects=False, ) assert rv.status_code == 302 @@ -150,39 +147,43 @@ def test_basic_access_redirect(client: Any, mocker: Any) -> None: == "https://web.archive.org/web/20200206164725id_/https://www.federalreserve.gov/econresdata/feds/2015/files/2015118pap.pdf" ) + # check that URL is validated rv = client.get( - "/access-redirect/aaaaaaaaaaaaaaaaaaaaaa01fa26e645d29c0427.pdf", + "/work/2x5qvct2dnhrbctqa2q2uyut6a/access/wayback/https://www.federalreserve.gov/econresdata/feds/2015/files/2015118pap.pdf.DUMMY", allow_redirects=False, ) assert rv.status_code == 404 -def test_access_redirects(client: Any, mocker: Any) -> None: +def test_access_redirect_encoding(client: Any, mocker: Any) -> None: - # tricky "URL encoding in archive.org path" case - rv = client.get( - "/access/ia_file/crossref-pre-1909-scholarly-works/10.1016%252Fs0140-6736%252802%252912493-7.zip/10.1016%252Fs0140-6736%252802%252912928-x.pdf", - allow_redirects=False, - ) - assert rv.status_code == 302 - assert ( - rv.headers["Location"] - == "https://archive.org/download/crossref-pre-1909-scholarly-works/10.1016%252Fs0140-6736%252802%252912493-7.zip/10.1016%252Fs0140-6736%252802%252912928-x.pdf" + with open("tests/files/elastic_get_work_a6gvpil4brdgzhqyaog3ftngqe.json") as f: + elastic_ia_resp = json.loads(f.read()) + with open("tests/files/elastic_get_work_ao5l3ykgbvg2vfpqe2y5qold5y.json") as f: + elastic_wayback_resp = json.loads(f.read()) + + es_raw = mocker.patch( + "elasticsearch.connection.Urllib3HttpConnection.perform_request" ) + es_raw.side_effect = [ + (200, {}, json.dumps(elastic_ia_resp)), + (200, {}, json.dumps(elastic_wayback_resp)), + ] + # tricky "URL encoding in archive.org path" case rv = client.get( - "/access/wayback/20170814015956/https://epub.uni-regensburg.de/21901/1/lorenz73.pdf", + "/work/a6gvpil4brdgzhqyaog3ftngqe/access/ia_file/crossref-pre-1909-scholarly-works/10.1016%252Fs0140-6736%252802%252912493-7.zip/10.1016%252Fs0140-6736%252802%252912928-x.pdf", allow_redirects=False, ) assert rv.status_code == 302 assert ( rv.headers["Location"] - == "https://web.archive.org/web/20170814015956id_/https://epub.uni-regensburg.de/21901/1/lorenz73.pdf" + == "https://archive.org/download/crossref-pre-1909-scholarly-works/10.1016%252Fs0140-6736%252802%252912493-7.zip/10.1016%252Fs0140-6736%252802%252912928-x.pdf" ) # spaces ("%20" vs "+") rv = client.get( - "/access/wayback/20170811115414/http://sudjms.net/issues/5-4/pdf/8)A%20comparison%20study%20of%20histochemical%20staining%20of%20various%20tissues%20after.pdf", + "/work/ao5l3ykgbvg2vfpqe2y5qold5y/access/wayback/http://sudjms.net/issues/5-4/pdf/8)A%20comparison%20study%20of%20histochemical%20staining%20of%20various%20tissues%20after.pdf", allow_redirects=False, ) assert rv.status_code == 302 -- cgit v1.2.3