From 34f45f9899529782198f08638d580599bed2c524 Mon Sep 17 00:00:00 2001
From: Adrian Damian <Adrian.Damian@nrc.ca>
Date: Thu, 28 Mar 2024 12:09:38 -0700
Subject: [PATCH] Streamlined cavern calls

---
 vos/vos/tests/test_vos.py | 11 +++++++++++
 vos/vos/vos.py            | 16 +++++++++++-----
 2 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/vos/vos/tests/test_vos.py b/vos/vos/tests/test_vos.py
index c7c8bf44..56d006ad 100644
--- a/vos/vos/tests/test_vos.py
+++ b/vos/vos/tests/test_vos.py
@@ -149,6 +149,7 @@ def test_get_node_url():
     response.headers = {'Location': transfer_url}
     mock_session = Mock(spec=requests.Session, get=Mock(return_value=response))
     client.get_session = Mock(return_value=mock_session)
+    client._fs_type = False
     assert transfer_url == \
         client.get_node_url('vos://cadc.nrc.ca!vospace/auser',
                             view='header')
@@ -478,11 +479,21 @@ def is_remote_file(uri):
         response.iter_content.return_value = BytesIO(file_content)
         session.get.return_value = response
         test_client.get_session = Mock(return_value=session)
+        # client must be a vault client
+        test_client._fs_type = False
         test_client.copy('{}{}'.format(vospaceLocation,
                                        '[1][10:60]'), osLocation)
         get_node_url_mock.assert_called_once_with(
             vospaceLocation, method='GET', cutout='[1][10:60]', view='cutout')
 
+        # test cavern does not support SODA operations
+        test_client._fs_type = True
+        with pytest.raises(ValueError):
+            test_client.copy('{}{}'.format(vospaceLocation, '[1][10:60]'), osLocation)
+            with pytest.raises(ValueError):
+                test_client.copy(vospaceLocation, osLocation, head=True)
+
+        test_client._fs_type = False
         # copy to vospace when md5 sums are the same -> only update occurs
         get_node_url_mock.reset_mock()
         computed_md5_mock.reset_mock()
diff --git a/vos/vos/vos.py b/vos/vos/vos.py
index 499e84a1..529e0869 100644
--- a/vos/vos/vos.py
+++ b/vos/vos/vos.py
@@ -1526,6 +1526,7 @@ def __init__(self, vospace_certfile=None,
             Client.VOSPACE_CERTFILE or vospace_certfile
         self.vospace_token = vospace_token
         self.insecure = insecure
+        self._fs_type = True  # True - file system type (cavern), False - db type (vault)
 
     def glob(self, pathname):
         """Return a list of paths matching a pathname pattern.
@@ -1690,6 +1691,10 @@ def get_endpoints(self, uri):
 
             else:
                 raise OSError('No scheme in {}'.format(uri))
+        # following is a CADC hack as others can deploy the services under different
+        # resource IDs
+        if 'vault' in resource_id:
+            self._fs_type = False
         if resource_id not in self._endpoints:
             try:
                 self._endpoints[resource_id] = EndPoints(
@@ -1844,6 +1849,8 @@ def copy(self, source, destination, send_md5=False, disposition=False,
 
             get_urls = []
             files_url = None
+            if self._fs_type and (cutout or view == 'header'):
+                raise ValueError('cavern/arc service does not support cutouts or header operations')
             if new_vos:
                 files_url = self.get_node_url(source, method='GET', cutout=cutout,
                                               view=view)
@@ -2419,6 +2426,10 @@ def _get(self, uri):
             if not file_path:
                 return None
             files_url = '{}{}'.format(files_ep, file_path)
+            if self._fs_type:
+                # files_url contains the bytes
+                return files_url
+            # remaining is for vault
             try:
                 response = self.get_session(uri).get(files_url, allow_redirects=False)
                 response.raise_for_status()
@@ -2426,11 +2437,6 @@ def _get(self, uri):
                 return None
             if response.status_code == 303:
                 return response.headers.get('Location', None)
-            elif response.status_code == 200:
-                # TODO this happens for cavern. It is wasteful since the response
-                # already contains the bytes but there's no other way to find out the
-                # actual location of the bytes.
-                return files_url
             return None
 
     def _add_soda_ops(self, url, view=None, cutout=None):