cachito-nexus: Add compatibility toggle for versions that use orientDB

Nexus has changed the embedded database from versions 3.71.0 forward¹.
The older versions used orientDB, which require a small change in the
URL when searching for component data.

This patch adds a configuration option so that Cachito can handle both
versions Nexus.

¹https://help.sonatype.com/en/sonatype-nexus-repository-3-71-0-release-notes.html

Signed-off-by: Bruno Pimentel <[email protected]>

Author: brunoapimentel

cachito/workers/config.py

@@ -50,13 +50,15 @@ class Config(object):
     cachito_nexus_hoster_password: Optional[str] = None
     cachito_nexus_hoster_url: Optional[str] = None
     cachito_nexus_hoster_username: Optional[str] = None
+    cachito_nexus_hoster_is_orient_db = False
     cachito_nexus_js_hosted_repo_name = "cachito-js-hosted"
     cachito_nexus_max_search_attempts = 5
     cachito_nexus_npm_proxy_url = "http://localhost:8081/repository/cachito-js/"
     cachito_nexus_pip_raw_repo_name = "cachito-pip-raw"
     cachito_nexus_rubygems_raw_repo_name = "cachito-rubygems-raw"
     cachito_nexus_proxy_password: Optional[str] = None
     cachito_nexus_proxy_username: Optional[str] = None
+    cachito_nexus_proxy_is_orient_db = False
     cachito_nexus_request_repo_prefix = "cachito-"
     cachito_nexus_timeout = 60
     cachito_nexus_username = "cachito"

cachito/workers/nexus.py

@@ -267,6 +267,30 @@ def get_component_info_from_nexus(
     return None
 
 
+def _get_nexus_orient_db_configuration():
+    config = get_worker_config()
+    return config.cachito_nexus_hoster_is_orient_db, config.cachito_nexus_proxy_is_orient_db
+
+
+def _treat_raw_component_name(name, is_nexus_hoster):
+    """
+    Treats the raw component name based on the use of orientDB by Nexus.
+
+    Starting from v3.71.0, Nexus has dropped the use of orientDB in favor of H2, which forces
+    all raw component names to have a leading slash. This function works as a compatibility layer
+    for orientDB, removing the leading slash.
+    """
+    nexus_hoster_is_orient_db, nexus_proxy_is_orient_db = _get_nexus_orient_db_configuration()
+
+    if is_nexus_hoster and nexus_hoster_is_orient_db:
+        return name[1:]
+
+    if not is_nexus_hoster and nexus_proxy_is_orient_db:
+        return name[1:]
+
+    return name
+
+
 def get_raw_component_asset_url(repository, name, max_attempts=1, from_nexus_hoster=True):
     """
     Get download URL for the asset of a raw component.
@@ -281,8 +305,14 @@ def get_raw_component_asset_url(repository, name, max_attempts=1, from_nexus_hos
         available
     :return: download URL for the asset, or None if component was not found
     """
+    treated_name = _treat_raw_component_name(name, from_nexus_hoster)
+
     component = get_component_info_from_nexus(
-        repository, "raw", name, max_attempts=max_attempts, from_nexus_hoster=from_nexus_hoster
+        repository,
+        "raw",
+        treated_name,
+        max_attempts=max_attempts,
+        from_nexus_hoster=from_nexus_hoster,
     )
     if component is None:
         return None
@@ -291,10 +321,10 @@ def get_raw_component_asset_url(repository, name, max_attempts=1, from_nexus_hos
 
     # Sanity checks, in practice this should not happen
     if not assets:
-        raise RuntimeError(f"Component {name} has no assets")
+        raise RuntimeError(f"Component {treated_name} has no assets")
     if len(assets) > 1:
         log.debug("All assets: %r", assets)
-        raise RuntimeError(f"Component {name} has more than 1 asset")
+        raise RuntimeError(f"Component {treated_name} has more than 1 asset")
 
     return assets[0]["downloadUrl"]
 

tests/test_workers/test_nexus.py

@@ -411,9 +411,29 @@ def test_get_component_info_from_nexus_version_vs_raw(raw, version):
         )
 
 
-@pytest.mark.parametrize("hoster", [True, False])
+@pytest.mark.parametrize(
+    "hoster, hoster_is_orient_db, proxy_is_orient_db, treated_name",
+    [
+        pytest.param(True, True, False, "foo/bar/foobar-1.0.tar.gz", id="from_hoster_is_orient_db"),
+        pytest.param(False, False, True, "foo/bar/foobar-1.0.tar.gz", id="from_proxy_is_orient_db"),
+        pytest.param(
+            False, False, False, "/foo/bar/foobar-1.0.tar.gz", id="from_proxy_is_not_orient_db"
+        ),
+        pytest.param(
+            True, False, True, "/foo/bar/foobar-1.0.tar.gz", id="from_hoster_is_not_orient_db"
+        ),
+    ],
+)
 @mock.patch("cachito.workers.nexus.get_component_info_from_nexus")
-def test_get_raw_component_asset_url(mock_get_component_info, hoster):
+@mock.patch("cachito.workers.nexus._get_nexus_orient_db_configuration")
+def test_get_raw_component_asset_url(
+    mock_get_nexus_orient_db_configuration,
+    mock_get_component_info,
+    hoster,
+    hoster_is_orient_db,
+    proxy_is_orient_db,
+    treated_name,
+):
     mock_get_component_info.return_value = {
         # "id": "1234",
         # "repository": "cachito-pip-raw",
@@ -437,15 +457,17 @@ def test_get_raw_component_asset_url(mock_get_component_info, hoster):
         ]
     }
 
+    mock_get_nexus_orient_db_configuration.return_value = (hoster_is_orient_db, proxy_is_orient_db)
+
     url = nexus.get_raw_component_asset_url(
-        "cachito-pip-raw", "foo/bar/foobar-1.0.tar.gz", from_nexus_hoster=hoster
+        "cachito-pip-raw", "/foo/bar/foobar-1.0.tar.gz", from_nexus_hoster=hoster
     )
     assert url == "http://nexus/repository/cachito-pip-raw/foo/bar/foobar-1.0.tar.gz"
 
     mock_get_component_info.assert_called_once_with(
         "cachito-pip-raw",
         "raw",
-        "foo/bar/foobar-1.0.tar.gz",
+        treated_name,
         max_attempts=1,
         from_nexus_hoster=hoster,
     )