初始提交：飞球 ETL 系统全量代码

tests/unit/test_cli_args.py

@@ -0,0 +1,137 @@
+# -*- coding: utf-8 -*-
+"""CLI 参数解析单元测试
+
+验证 --data-source 新参数、--pipeline-flow 弃用映射、
+--pipeline + --tasks 同时使用、以及 build_cli_overrides 集成行为。
+
+需求: 3.1, 3.3, 3.5
+"""
+import warnings
+from argparse import Namespace
+from unittest.mock import patch
+
+import pytest
+
+from cli.main import parse_args, resolve_data_source, build_cli_overrides
+
+
+# ---------------------------------------------------------------------------
+# 1. --data-source 新参数解析
+# ---------------------------------------------------------------------------
+class TestDataSourceArg:
+    """--data-source 新参数测试"""
+
+    @pytest.mark.parametrize("value", ["online", "offline", "hybrid"])
+    def test_data_source_valid_values(self, value):
+        with patch("sys.argv", ["cli", "--data-source", value]):
+            args = parse_args()
+        assert args.data_source == value
+
+    def test_data_source_default_is_none(self):
+        with patch("sys.argv", ["cli"]):
+            args = parse_args()
+        assert args.data_source is None
+
+
+# ---------------------------------------------------------------------------
+# 2. resolve_data_source() 弃用映射
+# ---------------------------------------------------------------------------
+class TestResolveDataSource:
+    """resolve_data_source() 弃用映射测试"""
+
+    def test_explicit_data_source_returns_directly(self):
+        args = Namespace(data_source="online", pipeline_flow=None)
+        assert resolve_data_source(args) == "online"
+
+    def test_data_source_takes_priority_over_pipeline_flow(self):
+        """--data-source 优先于 --pipeline-flow"""
+        args = Namespace(data_source="online", pipeline_flow="FULL")
+        assert resolve_data_source(args) == "online"
+
+
+    @pytest.mark.parametrize(
+        "flow, expected",
+        [
+            ("FULL", "hybrid"),
+            ("FETCH_ONLY", "online"),
+            ("INGEST_ONLY", "offline"),
+        ],
+    )
+    def test_pipeline_flow_maps_with_deprecation_warning(self, flow, expected):
+        """旧参数 --pipeline-flow 映射到正确的 data_source 并发出弃用警告"""
+        args = Namespace(data_source=None, pipeline_flow=flow)
+        with pytest.warns(DeprecationWarning, match="--pipeline-flow 已弃用"):
+            result = resolve_data_source(args)
+        assert result == expected
+
+    def test_neither_arg_defaults_to_hybrid(self):
+        """两个参数都未指定时，默认返回 hybrid"""
+        args = Namespace(data_source=None, pipeline_flow=None)
+        assert resolve_data_source(args) == "hybrid"
+
+
+# ---------------------------------------------------------------------------
+# 3. build_cli_overrides() 集成
+# ---------------------------------------------------------------------------
+class TestBuildCliOverrides:
+    """build_cli_overrides() 集成测试"""
+
+    def _make_args(self, **kwargs):
+        """构造最小 Namespace，未指定的参数设为 None/False"""
+        defaults = dict(
+            store_id=None, tasks=None, dry_run=False,
+            pipeline=None, processing_mode="increment_only",
+            fetch_before_verify=False, verify_tables=None,
+            window_split="none", lookback_hours=24, overlap_seconds=3600,
+            pg_dsn=None, pg_host=None, pg_port=None, pg_name=None,
+            pg_user=None, pg_password=None,
+            api_base=None, api_token=None, api_timeout=None,
+            api_page_size=None, api_retry_max=None,
+            window_start=None, window_end=None,
+            force_window_override=False,
+            window_split_unit=None, window_split_days=None,
+            window_compensation_hours=None,
+            export_root=None, log_root=None,
+            data_source=None, pipeline_flow=None,
+            fetch_root=None, ingest_source=None, write_pretty_json=False,
+            idle_start=None, idle_end=None, allow_empty_advance=False,
+        )
+        defaults.update(kwargs)
+        return Namespace(**defaults)
+
+    def test_data_source_online_sets_run_key(self):
+        args = self._make_args(data_source="online")
+        overrides = build_cli_overrides(args)
+        assert overrides["run"]["data_source"] == "online"
+
+    def test_pipeline_flow_sets_both_keys(self):
+        """旧参数同时写入 pipeline.flow 和 run.data_source"""
+        args = self._make_args(pipeline_flow="FULL")
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore", DeprecationWarning)
+            overrides = build_cli_overrides(args)
+        assert overrides["pipeline"]["flow"] == "FULL"
+        assert overrides["run"]["data_source"] == "hybrid"
+
+    def test_default_data_source_is_hybrid(self):
+        """无 --data-source 也无 --pipeline-flow 时，run.data_source 默认 hybrid"""
+        args = self._make_args()
+        overrides = build_cli_overrides(args)
+        assert overrides["run"]["data_source"] == "hybrid"
+
+
+# ---------------------------------------------------------------------------
+# 4. --pipeline + --tasks 同时使用
+# ---------------------------------------------------------------------------
+class TestPipelineAndTasks:
+    """--pipeline + --tasks 同时使用时的行为"""
+
+    def test_pipeline_and_tasks_both_parsed(self):
+        with patch("sys.argv", [
+            "cli",
+            "--pipeline", "api_full",
+            "--tasks", "ODS_MEMBER,ODS_ORDER",
+        ]):
+            args = parse_args()
+        assert args.pipeline == "api_full"
+        assert args.tasks == "ODS_MEMBER,ODS_ORDER"