about summary refs log tree commit diff
path: root/test/python/server_test_case.py
diff options
context:
space:
mode:
authorPjotr Prins <pjotr.public01@thebird.nl>2026-10-05 06:50:30 +0000
committerPjotr <wrk@bosa.lan>2026-10-05 06:50:30 +0000
commit693e7ee89f8618c1abbf8af5c0db23b071700803 (patch)
treee5db71e3f7e8af7342be1004aac9d5dfd0e1db2a /test/python/server_test_case.py
parentadc173c1f3c78da487827122262573e753284260 (diff)
downloadgenenetwork4-693e7ee89f8618c1abbf8af5c0db23b071700803.tar.gz
Add POST /dev/token, a dev-only gn-auth token issuer
Calls gn-auth's own password grant (POST /auth/token) for gn-auth's
well-known development user and returns the result, so local testing
no longer needs a manual curl/gn-auth round trip of its own.

Only ever registered -- not just rejected at request time, genuinely
absent from the router and from /openapi.json -- when two new Config
fields are both set: dev_mode (GENENETWORK4_DEV_MODE) and a loopback
bind (GENENETWORK4_BIND, also new). Both are config, not something
genenetwork4 infers from its actual network exposure, by design: dev
mode alone, or a loopback bind alone, each leave it disabled, and the
check (Config::dev_token_enabled) is a plain function so the condition
itself has direct unit test coverage rather than only being exercised
indirectly through the route.

src/main.rs now binds to config.bind instead of a hardcoded "0.0.0.0",
via a proper SocketAddr rather than string formatting -- the latter
would mishandle an IPv6 bind address like "::1" (needs brackets).

GN_AUTH_DEV_CLIENT_ID/_CLIENT_SECRET/_USERNAME/_PASSWORD default to
gn-auth's own published dev fixtures (flask init-dev-clients,
gn_auth/wsgi.py), so this works unconfigured against an unmodified dev
gn-auth instance -- confirmed live, including using the returned token
against a real gated endpoint (/traits/BXD/10001), not just checking
the token response's shape.

tests/dev_token.rs covers all four combinations of the two settings in
one sequential test function (mutating the same two env vars from more
than one #[tokio::test] in a file would race, since they run
concurrently by default) plus the /openapi.json visibility check.
test/python/test_dev_token.py mirrors this using separate test classes
instead, which is safe here since each spawns its own server
subprocess with its own environment (ServerTestCase gains EXTRA_ENV for
this).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XDDpXvZBQat9uSgDeEMJ8d
Diffstat (limited to 'test/python/server_test_case.py')
-rw-r--r--test/python/server_test_case.py10
1 files changed, 8 insertions, 2 deletions
diff --git a/test/python/server_test_case.py b/test/python/server_test_case.py
index 0025c4b..ec243a2 100644
--- a/test/python/server_test_case.py
+++ b/test/python/server_test_case.py
@@ -26,11 +26,17 @@ def wait_for_port(host, port, timeout=10):
 
 
 class ServerTestCase(unittest.TestCase):
-    """Spawns the genenetwork4 binary for the duration of the test class."""
+    """Spawns the genenetwork4 binary for the duration of the test class.
+
+    Subclasses that need the server started with extra/overridden
+    environment variables (e.g. GENENETWORK4_DEV_MODE) can set EXTRA_ENV.
+    """
+
+    EXTRA_ENV = {}
 
     @classmethod
     def setUpClass(cls):
-        env = dict(os.environ, GENENETWORK4_PORT=PORT)
+        env = dict(os.environ, GENENETWORK4_PORT=PORT, **cls.EXTRA_ENV)
         cls.proc = subprocess.Popen([BINARY], env=env)
         if not wait_for_port(HOST, PORT):
             cls.proc.terminate()