Merge pull request #54 from yusufipk/codex-model-list

Ask Codex itself which models it offers

tests/test_assistant.py

@@ -17,7 +17,8 @@ import unittest
 from unittest import mock
 
 from dikte import assistant
-from tests.support import DikteTest, fake_urlopen, only_these_tools
+from tests.support import (DikteTest, FakeCompleted, fake_urlopen,
+                           only_these_tools)
 
 
 class FakeCli:
@@ -664,5 +665,50 @@ class Ask(DikteTest):
         self.assertEqual(len(calls), 1)
 
 
+class CodexModels(DikteTest):
+    """The model list read off `codex debug models`."""
+
+    CATALOG = {"models": [
+        {"slug": "gpt-6-mini", "visibility": "list", "priority": 9},
+        {"slug": "gpt-6", "visibility": "list", "priority": 1},
+        {"slug": "codex-auto-review", "visibility": "hide", "priority": 3},
+    ]}
+
+    def models(self, reply, code=0):
+        with only_these_tools("codex"), \
+                mock.patch.object(subprocess, "run",
+                                  return_value=FakeCompleted(
+                                      returncode=code, stdout=reply)) as run:
+            found = assistant.codex_models()
+        self.run_call = run
+        return found
+
+    def test_the_catalog_arrives_best_first_without_the_hidden_ones(self):
+        found = self.models(json.dumps(self.CATALOG))
+        self.assertEqual(found, ["gpt-6", "gpt-6-mini"])
+        self.assertEqual(self.run_call.call_args.args[0],
+                         ["codex", "debug", "models"])
+
+    def test_a_codex_that_is_not_installed_is_not_run(self):
+        with only_these_tools(), \
+                mock.patch.object(subprocess, "run") as run:
+            self.assertEqual(assistant.codex_models(), [])
+        run.assert_not_called()
+
+    def test_a_codex_too_old_to_have_the_command(self):
+        self.assertEqual(self.models("error: unknown subcommand", code=2), [])
+
+    def test_a_catalog_that_is_not_what_was_expected(self):
+        self.assertEqual(self.models(json.dumps(["gpt-6"])), [])
+        self.assertEqual(self.models(""), [])
+
+    def test_a_codex_that_hangs_is_given_up_on(self):
+        with only_these_tools("codex"), \
+                mock.patch.object(subprocess, "run",
+                                  side_effect=subprocess.TimeoutExpired(
+                                      ["codex"], 30)):
+            self.assertEqual(assistant.codex_models(), [])
+
+
 if __name__ == "__main__":
     unittest.main()

tests/test_ui.py

@@ -133,6 +133,8 @@ class Settings(DikteTest):
                                             "_load_models"))
         self.enterContext(mock.patch.object(settings_ui.SettingsWindow,
                                             "_load_transcribe_models"))
+        self.enterContext(mock.patch.object(settings_ui.SettingsWindow,
+                                            "_load_codex_models"))
         # The local model boxes fetch their own list the moment they are shown,
         # from a thread, which is nobody's test failing but a real request.
         self.enterContext(mock.patch.object(settings_ui.LocalModelBox,
@@ -269,6 +271,19 @@ class Settings(DikteTest):
                 self.assertEqual(shown, [provider])
                 self.assertFalse(box.isHidden())
 
+    def test_codex_answering_refills_both_of_its_boxes(self):
+        """The list Codex gave replaces the built-in one, in both places, and
+        neither loses what was already picked."""
+        conf = self.config(cleanup_codex_model="my-own-model")
+        window = self.window(conf)
+        window._on_codex_models_loaded(["gpt-6", "gpt-6-mini"])
+        for combo in (window.cleanup_codex_model, window.assistant_codex_model):
+            with self.subTest(combo=combo.objectName() or "combo"):
+                offered = [combo.itemText(i) for i in range(combo.count())]
+                self.assertEqual(offered[1:], ["gpt-6", "gpt-6-mini"])
+        self.assertEqual(window.cleanup_codex_model.currentText(),
+                         "my-own-model")
+
     def test_the_update_line_names_the_version_that_is_running(self):
         window = self.window(cfg.Config())
         self.assertIn(settings_ui.__version__, window.update_status.text())
@@ -984,7 +999,9 @@ class MeetingSources(DikteTest):
                 only_these_tools(), \
                 mock.patch.object(settings_ui.SettingsWindow, "_load_models"), \
                 mock.patch.object(settings_ui.SettingsWindow,
-                                  "_load_transcribe_models"):
+                                  "_load_transcribe_models"), \
+                mock.patch.object(settings_ui.SettingsWindow,
+                                  "_load_codex_models"):
             window = settings_ui.SettingsWindow(cfg.Config())
         self.addCleanup(window.deleteLater)
         self.addCleanup(window.close)
@@ -1013,6 +1030,9 @@ class LocalModels(DikteTest):
         # "nothing can transcribe" question from its real binary and model.
         self.patch_attr(ggml, "BIN_DIR", self.path("bin"))
         self.patch_attr(ggml, "MODELS_DIR", self.path("models"))
+        # And one with Codex on it would ask it for its model list.
+        self.enterContext(mock.patch.object(settings_ui.SettingsWindow,
+                                            "_load_codex_models"))
 
     def window(self, conf):
         window = settings_ui.SettingsWindow(conf)