feat: added webhdfs test without knox

tests/conftest.py

@@ -193,6 +193,11 @@ def knox_gateway(host: Host, user: str) -> Dict[str, str]:
     }
 
 
+@pytest.fixture(scope="session")
+def webhdfs_gateway(host: Host, user: str) -> str:
+    return f"https://{host.backend.get_hosts('hdfs_nn')[0]}:9871"
+
+
 @pytest.fixture(scope="session")
 def hbase_rest(host: Host) -> str:
     return f"https://{host.backend.get_hosts('hbase_rest')[0]}:8080"

tests/test_hdfs.py

@@ -1,8 +1,14 @@
+import json
 import os
-from typing import Dict
+from typing import Callable, Dict, List
+
+import pytest
 
 from testinfra import host
 
+from .conftest import USERS, retry
+
+
 testinfra_hosts = ["edge"]
 
 
@@ -55,3 +61,91 @@ def test_create_temporary_file_in_user_directory(
         )
         assert "No such file or directory" in hdfs_cmd.stderr, hdfs_cmd
         assert hdfs_cmd.stdout == "", hdfs_cmd
+
+
+@pytest.fixture(scope="module")
+def webhdfs_ranger_policy(
+    ranger_policy: Callable[[str, str, dict, List[dict]], dict],
+):
+    resources = {
+        "path": {"values": ["/*"], "isExcludes": False},
+    }
+    policyItems = [
+        {
+            "users": USERS,
+            "accesses": [
+                {"isAllowed": True, "type": "read"},
+                {"isAllowed": True, "type": "write"},
+                {"isAllowed": True, "type": "execute"},
+            ],
+        }
+    ]
+    ranger_policy("webhdfs_test", "hdfs-tdp", resources, policyItems)
+
+
+def test_create_webhdfs_temporary_file_in_user_directory(
+    host: host.Host,
+    user: str,
+    realm: str,
+    user_file: Dict[str, str],
+    webhdfs_gateway: str,
+    webhdfs_ranger_policy: None,
+    curl: Callable,
+):
+    distant_file = user_file["distant_file"]
+    distant_hdfs_path = user_file["distant_hdfs_path"]
+    file_content = user_file["file_content"]
+    webhdfs_url = webhdfs_gateway
+    webhdfs_gateway_url = f"{webhdfs_url}/webhdfs/v1"
+
+    host.run(f"kinit -kt /home/{user}/{user}.keytab {user}@{realm}")
+    curl_result = retry(
+        lambda: curl(
+            f"-L -T {distant_file} --negotiate -u : -X PUT '{webhdfs_gateway_url}/user/{user}/{distant_hdfs_path}?op=CREATE'"
+        )
+    )()
+    assert curl_result["http_status"] == 201, curl_result
+
+    try:
+        curl_result = curl(
+            f"-L -k --negotiate -u : -X GET '{webhdfs_gateway_url}/user/{user}/{distant_hdfs_path}?op=OPEN'"
+        )
+        assert file_content in curl_result["command"].stdout, curl_result
+
+        curl_result = curl(
+            f"-L -k --negotiate -u : -X GET '{webhdfs_gateway_url}/user/{user}/{distant_hdfs_path}?op=LISTSTATUS'"
+        )
+        assert curl_result["http_status"] == 200, curl_result
+
+        liststatus = json.loads(curl_result["command"].stdout)
+        assert "FileStatuses" in liststatus, curl_result
+        assert "FileStatus" in liststatus["FileStatuses"], curl_result
+        assert len(liststatus["FileStatuses"]["FileStatus"]) == 1, curl_result
+
+        filestatus = liststatus["FileStatuses"]["FileStatus"][0]
+        assert filestatus["group"] == user, curl_result
+        assert filestatus["owner"] == user, curl_result
+        assert filestatus["type"] == "FILE", curl_result
+    finally:
+        curl_result = curl(
+            f"-L -k --negotiate -u : -X DELETE '{webhdfs_gateway_url}/user/{user}/{distant_hdfs_path}?op=DELETE'"
+        )
+        assert curl_result["http_status"] == 200, curl_result
+
+    curl_result = curl(
+        f"-L -k --negotiate -u : -X GET '{webhdfs_gateway_url}/user/{user}/{distant_hdfs_path}?op=LISTSTATUS'",
+        check_status_code=False,
+    )
+    liststatus = json.loads(curl_result["command"].stdout)
+    assert "RemoteException" in liststatus, curl_result
+
+    assert "exception" in liststatus["RemoteException"], curl_result
+    assert (
+        "FileNotFoundException" in liststatus["RemoteException"]["exception"]
+    ), curl_result
+
+    assert "message" in liststatus["RemoteException"], curl_result
+    assert (
+        f"File /user/{user}/{distant_hdfs_path} does not exist."
+        in liststatus["RemoteException"]["message"]
+    ), curl_result