lib-x17-fintech 2.1.3__py3-none-any.whl

xfintech/data/source/tushare/stock/companyoverview/tests/test_class_companyoverview_all.py

@@ -0,0 +1,647 @@
+import pandas as pd
+import pytest
+
+from xfintech.data.source.tushare.stock.companyoverview import CompanyOverview
+from xfintech.data.source.tushare.stock.companyoverview.constant import (
+    KEY,
+    NAME,
+    PAGINATE,
+    SOURCE,
+    TARGET,
+)
+
+# Test Fixtures
+
+
+class FakeConnection:
+    """Fake Tushare connection for testing"""
+
+    def __init__(self, frame: pd.DataFrame):
+        self.frame = frame
+
+    def express_vip(self, **kwargs):
+        """Mock express_vip API call"""
+        return self.frame
+
+
+class FakeSession:
+    """Fake session for testing"""
+
+    def __init__(self, connection: FakeConnection):
+        self.connection = connection
+
+
+@pytest.fixture
+def mock_session():
+    """Create a mock session with empty data"""
+    fake_conn = FakeConnection(frame=pd.DataFrame())
+    return FakeSession(fake_conn)
+
+
+@pytest.fixture
+def sample_source_data():
+    """Sample company overview data in Tushare format"""
+    return pd.DataFrame(
+        {
+            "ts_code": ["000001.SZ", "000001.SZ", "000002.SZ"],
+            "ann_date": ["20200430", "20201031", "20200430"],
+            "end_date": ["20200331", "20200930", "20200331"],
+            "revenue": ["100000000.50", "120000000.75", "50000000.25"],
+            "operate_profit": ["15000000.30", "18000000.40", "6000000.15"],
+            "total_profit": ["16000000.20", "19000000.35", "7000000.10"],
+            "n_income": ["12000000.00", "14000000.00", "5000000.00"],
+            "total_assets": ["500000000.50", "550000000.75", "200000000.25"],
+            "total_hldr_eqy_exc_min_int": ["200000000.50", "220000000.75", "80000000.25"],
+            "diluted_eps": ["0.50", "0.60", "0.25"],
+            "diluted_roe": ["8.5", "9.2", "7.3"],
+            "yoy_net_profit": ["10000000.00", "12000000.00", "4500000.00"],
+            "bps": ["5.5", "6.0", "4.2"],
+            "yoy_sales": ["15.5", "18.2", "12.3"],
+            "yoy_op": ["20.3", "22.5", "15.8"],
+            "yoy_tp": ["18.7", "20.9", "14.2"],
+            "yoy_dedu_np": ["16.5", "18.8", "13.5"],
+            "yoy_eps": ["15.0", "17.5", "12.0"],
+            "yoy_roe": ["1.5", "2.0", "1.0"],
+            "growth_assets": ["10.5", "12.3", "8.7"],
+            "yoy_equity": ["11.2", "13.5", "9.3"],
+            "growth_bps": ["12.0", "14.2", "10.1"],
+            "or_last_year": ["85000000.00", "100000000.00", "44000000.00"],
+            "op_last_year": ["12500000.00", "15000000.00", "5200000.00"],
+            "tp_last_year": ["13500000.00", "16000000.00", "6100000.00"],
+            "np_last_year": ["10300000.00", "12000000.00", "4400000.00"],
+            "eps_last_year": ["0.43", "0.51", "0.22"],
+            "open_net_assets": ["180000000.00", "200000000.00", "73000000.00"],
+            "open_bps": ["4.9", "5.4", "3.8"],
+            "perf_summary": ["业绩稳定增长", "持续盈利", "良好表现"],
+            "is_audit": ["1", "0", "1"],
+            "remark": ["", "预计数据", ""],
+        }
+    )
+
+
+# Initialization Tests
+
+
+def test_companyoverview_init_basic(mock_session):
+    """Test basic initialization with required session"""
+    overview = CompanyOverview(session=mock_session)
+    assert overview.name == NAME
+    assert overview.key == KEY
+    assert overview.source == SOURCE
+    assert overview.target == TARGET
+    assert overview.paginate == PAGINATE
+
+
+def test_companyoverview_init_with_params_dict(mock_session):
+    """Test initialization with params as dict"""
+    params = {"ts_code": "000001.SZ", "start_date": "20200101", "end_date": "20201231"}
+    overview = CompanyOverview(session=mock_session, params=params)
+    assert overview.params.get("ts_code") == "000001.SZ"
+    assert overview.params.get("start_date") == "20200101"
+    assert overview.params.get("end_date") == "20201231"
+
+
+def test_companyoverview_init_with_params_object(mock_session):
+    """Test initialization with params as Params object"""
+    from xfintech.data.common.params import Params
+
+    params = Params(ts_code="000001.SZ", start_date="20200101")
+    overview = CompanyOverview(session=mock_session, params=params)
+    assert overview.params.get("ts_code") == "000001.SZ"
+
+
+def test_companyoverview_init_with_cache_bool_true(mock_session):
+    """Test initialization with cache enabled"""
+    overview = CompanyOverview(session=mock_session, cache=True)
+    assert overview.cache is not None
+
+
+def test_companyoverview_init_with_cache_bool_false(mock_session):
+    """Test initialization with cache disabled"""
+    overview = CompanyOverview(session=mock_session, cache=False)
+    assert overview.cache is None
+
+
+def test_companyoverview_init_with_cache_dict(mock_session):
+    """Test initialization with cache config dict"""
+    cache_config = {"path": "/tmp/cache", "key": "test"}
+    overview = CompanyOverview(session=mock_session, cache=cache_config)
+    assert overview.cache is not None
+
+
+def test_companyoverview_init_default_paginate_limit(mock_session):
+    """Test that default pagination limit is set correctly"""
+    overview = CompanyOverview(session=mock_session)
+    assert overview.paginate.pagesize == 1000
+    assert overview.paginate.pagelimit == 1000
+
+
+def test_companyoverview_init_with_all_params(mock_session):
+    """Test initialization with all parameters"""
+    params = {"ts_code": "000001.SZ", "start_date": "20200101", "end_date": "20201231"}
+    coolant = {"interval": 0.5}
+    retry = {"max_attempts": 3}
+    cache = True
+
+    overview = CompanyOverview(session=mock_session, params=params, coolant=coolant, retry=retry, cache=cache)
+    assert overview.params.get("ts_code") == "000001.SZ"
+    assert overview.cache is not None
+
+
+def test_companyoverview_constants():
+    """Test that constants are correctly defined"""
+    assert NAME == "companyoverview"
+    assert KEY == "/tushare/companyoverview"
+    assert SOURCE.name == "express_vip"
+    assert TARGET.name == "companyoverview"
+    assert PAGINATE.pagesize == 1000
+
+
+# Transform Tests
+
+
+def test_companyoverview_transform_basic(sample_source_data):
+    """Test basic transformation of overview data"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    assert len(result) == 3
+    assert "code" in result.columns
+    assert "date" in result.columns
+    assert "datecode" in result.columns
+
+
+def test_companyoverview_transform_code_mapping(sample_source_data):
+    """Test that ts_code is correctly mapped to code"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    assert result.iloc[0]["code"] == "000001.SZ"
+    assert result.iloc[2]["code"] == "000002.SZ"
+
+
+def test_companyoverview_transform_date_format(sample_source_data):
+    """Test that dates are properly formatted"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    assert pd.api.types.is_datetime64_any_dtype(result["date"]) is False
+    assert result.iloc[0]["date"] == "2020-03-31"
+
+
+def test_companyoverview_transform_datecode_preserved(sample_source_data):
+    """Test that datecode is correctly created from end_date"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    assert result.iloc[0]["datecode"] == "20200331"
+    assert result.iloc[1]["datecode"] == "20200930"
+
+
+def test_companyoverview_transform_numeric_conversions(sample_source_data):
+    """Test that numeric fields are properly converted"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    assert result.iloc[0]["revenue"] == 100000000.50
+    assert result.iloc[0]["operate_profit"] == 15000000.30
+    assert pd.api.types.is_numeric_dtype(result["revenue"])
+
+
+def test_companyoverview_transform_string_fields(sample_source_data):
+    """Test that string fields are properly converted"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    assert result.iloc[0]["perf_summary"] == "业绩稳定增长"
+    assert result.iloc[1]["remark"] == "预计数据"
+
+
+def test_companyoverview_transform_empty_dataframe():
+    """Test transformation with empty dataframe"""
+    fake_conn = FakeConnection(frame=pd.DataFrame())
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(pd.DataFrame())
+
+    assert isinstance(result, pd.DataFrame)
+    assert result.empty
+    assert len(result.columns) == len(TARGET.list_column_names())
+
+
+def test_companyoverview_transform_none_input():
+    """Test transformation with None input"""
+    fake_conn = FakeConnection(frame=pd.DataFrame())
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(None)
+
+    assert isinstance(result, pd.DataFrame)
+    assert result.empty
+
+
+def test_companyoverview_transform_handles_invalid_dates():
+    """Test that invalid dates are handled gracefully"""
+    data = pd.DataFrame(
+        {
+            "ts_code": ["000001.SZ"],
+            "ann_date": ["invalid"],
+            "end_date": ["20200331"],
+            "revenue": ["100000000.00"],
+            "is_audit": ["1"],
+        }
+    )
+
+    fake_conn = FakeConnection(frame=data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(data)
+
+    assert len(result) == 1
+    assert pd.isna(result.iloc[0]["ann_date"])
+
+
+def test_companyoverview_transform_removes_duplicates(sample_source_data):
+    """Test that duplicate rows are removed"""
+    # Create data with duplicates
+    data = pd.concat([sample_source_data, sample_source_data.iloc[[0]]], ignore_index=True)
+
+    fake_conn = FakeConnection(frame=data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(data)
+
+    # Should have 3 unique rows, not 4
+    assert len(result) == 3
+
+
+def test_companyoverview_transform_sorts_by_code_and_date(sample_source_data):
+    """Test that output is sorted by code and date"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    # Check first row is 000001.SZ, 2020-03-31
+    assert result.iloc[0]["code"] == "000001.SZ"
+    assert result.iloc[0]["date"] == "2020-03-31"
+    # Check last row is 000002.SZ
+    assert result.iloc[2]["code"] == "000002.SZ"
+
+
+def test_companyoverview_transform_resets_index():
+    """Test that index is reset after transformation"""
+    data = pd.DataFrame(
+        {
+            "ts_code": ["000001.SZ", "000002.SZ"],
+            "ann_date": ["20200430", "20200430"],
+            "end_date": ["20200331", "20200331"],
+            "revenue": ["100000000.00", "50000000.00"],
+            "is_audit": ["1", "1"],
+        },
+        index=[5, 10],
+    )
+
+    fake_conn = FakeConnection(frame=data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(data)
+
+    assert result.index.tolist() == [0, 1]
+
+
+def test_companyoverview_transform_only_target_columns(sample_source_data):
+    """Test that only target columns are in output"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.transform(sample_source_data)
+
+    expected_cols = set(TARGET.list_column_names())
+    actual_cols = set(result.columns)
+
+    assert actual_cols == expected_cols
+
+
+# Run Method Tests
+
+
+def test_companyoverview_run_with_cache_hit(sample_source_data):
+    """Test that run() returns cached data when available"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"}, cache=True)
+
+    # First run should fetch and cache
+    result1 = overview.run()
+
+    # Second run should return cached data
+    result2 = overview.run()
+
+    pd.testing.assert_frame_equal(result1, result2)
+
+
+def test_companyoverview_run_basic_date_range(sample_source_data):
+    """Test run() with date range parameters"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(
+        session=session,
+        params={"ts_code": "000001.SZ", "start_date": "20200101", "end_date": "20201231"},
+    )
+
+    result = overview.run()
+
+    assert isinstance(result, pd.DataFrame)
+    assert len(result) > 0
+
+
+def test_companyoverview_run_with_period_param(sample_source_data):
+    """Test run() with period parameter"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(
+        session=session,
+        params={"ts_code": "000001.SZ", "period": "2020-3"},
+    )
+    result = overview.run()
+    assert isinstance(result, pd.DataFrame)
+
+
+def test_companyoverview_run_with_datetime_params(sample_source_data):
+    """Test run() with datetime objects as parameters"""
+    from datetime import datetime
+
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(
+        session=session,
+        params={
+            "ts_code": "000001.SZ",
+            "start_date": datetime(2020, 1, 1),
+            "end_date": datetime(2020, 12, 31),
+        },
+    )
+
+    result = overview.run()
+
+    assert isinstance(result, pd.DataFrame)
+
+
+def test_companyoverview_run_adds_fields_param(sample_source_data):
+    """Test that run() adds fields parameter if not provided"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    # Monkey patch _fetchall to capture params
+    overview._fetchall
+    captured_params = {}
+
+    def mock_fetchall(api, **kwargs):
+        captured_params.update(kwargs)
+        return sample_source_data
+
+    overview._fetchall = mock_fetchall
+    overview.run()
+
+    assert "fields" in captured_params
+    assert len(captured_params["fields"]) > 0
+
+
+def test_companyoverview_run_sets_cache(sample_source_data):
+    """Test that run() saves result to cache"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"}, cache=True)
+
+    result = overview.run()
+
+    # Cache should be set after run
+    cached = overview._load_cache()
+    assert cached is not None
+    pd.testing.assert_frame_equal(result, cached)
+
+
+def test_companyoverview_run_calls_transform(sample_source_data):
+    """Test that run() calls transform method"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    # Monkey patch transform to track if it was called
+    original_transform = overview.transform
+    transform_called = {"called": False}
+
+    def mock_transform(data):
+        transform_called["called"] = True
+        return original_transform(data)
+
+    overview.transform = mock_transform
+    overview.run()
+
+    assert transform_called["called"] is True
+
+
+def test_companyoverview_run_uses_express_vip_api(sample_source_data):
+    """Test that run() uses express_vip API"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    # Monkey patch _fetchall to capture API
+    captured_api = {"api": None}
+
+    def mock_fetchall(api, **kwargs):
+        captured_api["api"] = api
+        return sample_source_data
+
+    overview._fetchall = mock_fetchall
+    overview.run()
+
+    assert captured_api["api"] == session.connection.express_vip
+
+
+# Integration Tests
+
+
+def test_companyoverview_full_workflow(sample_source_data):
+    """Test complete workflow from initialization to output"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+
+    overview = CompanyOverview(
+        session=session,
+        params={"ts_code": "000001.SZ", "start_date": "20200101", "end_date": "20201231"},
+        cache=True,
+    )
+
+    result = overview.run()
+
+    # Verify output structure
+    assert isinstance(result, pd.DataFrame)
+    assert len(result) == 3
+    assert "code" in result.columns
+    assert "date" in result.columns
+    assert "revenue" in result.columns
+    assert "n_income" in result.columns
+
+    # Verify data types
+    assert pd.api.types.is_numeric_dtype(result["revenue"])
+    assert result.iloc[0]["code"] == "000001.SZ"
+
+
+def test_companyoverview_cache_persistence(sample_source_data):
+    """Test that cache persists across multiple runs"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"}, cache=True)
+
+    # Run once to populate cache
+    result1 = overview.run()
+
+    # Clear in-memory data to simulate fresh load
+    overview2 = CompanyOverview(session=session, params={"ts_code": "000001.SZ"}, cache=True)
+    result2 = overview2.run()
+
+    pd.testing.assert_frame_equal(result1, result2)
+
+
+def test_companyoverview_params_uniqueness(mock_session):
+    """Test that different params produce different cache keys"""
+    overview1 = CompanyOverview(
+        session=mock_session,
+        params={"ts_code": "000001.SZ", "start_date": "20200101"},
+        cache=True,
+    )
+    overview2 = CompanyOverview(
+        session=mock_session,
+        params={"ts_code": "000001.SZ", "start_date": "20210101"},
+        cache=True,
+    )
+
+    # Test that different params are properly stored
+    assert overview1.params.get("start_date") == "20200101"
+    assert overview2.params.get("start_date") == "20210101"
+    assert overview1.params.get("start_date") != overview2.params.get("start_date")
+
+
+def test_companyoverview_different_stocks(sample_source_data):
+    """Test that different stock codes are handled correctly"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+
+    overview1 = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+    result1 = overview1.run()
+
+    overview2 = CompanyOverview(session=session, params={"ts_code": "000002.SZ"})
+    result2 = overview2.run()
+
+    # Both should have data for their respective stocks
+    assert "000001.SZ" in result1["code"].values
+    assert "000002.SZ" in result2["code"].values
+
+
+def test_companyoverview_empty_result_handling():
+    """Test handling of empty API results"""
+    empty_data = pd.DataFrame()
+    fake_conn = FakeConnection(frame=empty_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.run()
+
+    assert isinstance(result, pd.DataFrame)
+    assert result.empty
+    assert len(result.columns) > 0
+
+
+def test_companyoverview_large_dataset_handling():
+    """Test handling of large datasets"""
+    # Create large dataset
+    large_data = pd.DataFrame(
+        {
+            "ts_code": ["000001.SZ"] * 1000,
+            "ann_date": ["20200430"] * 1000,
+            "end_date": ["20200331"] * 1000,
+            "revenue": ["100000000.00"] * 1000,
+            "is_audit": ["1"] * 1000,
+        }
+    )
+
+    fake_conn = FakeConnection(frame=large_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.run()
+
+    # Should deduplicate to 1 row
+    assert len(result) == 1
+
+
+def test_companyoverview_without_cache(sample_source_data):
+    """Test that class works correctly without caching"""
+    fake_conn = FakeConnection(frame=sample_source_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"}, cache=False)
+
+    result = overview.run()
+
+    assert isinstance(result, pd.DataFrame)
+    assert len(result) > 0
+    assert overview.cache is None
+
+
+def test_companyoverview_handles_missing_numeric_fields():
+    """Test that missing numeric fields don't break transformation"""
+    # Data with only subset of fields
+    minimal_data = pd.DataFrame(
+        {
+            "ts_code": ["000001.SZ"],
+            "ann_date": ["20200430"],
+            "end_date": ["20200331"],
+            "revenue": ["100000000.00"],
+            "is_audit": ["1"],
+        }
+    )
+
+    fake_conn = FakeConnection(frame=minimal_data)
+    session = FakeSession(fake_conn)
+    overview = CompanyOverview(session=session, params={"ts_code": "000001.SZ"})
+
+    result = overview.run()
+
+    # Should complete without error
+    assert len(result) == 1
+    # Missing fields should be NaN
+    assert pd.isna(result.iloc[0]["operate_profit"])
+
+
+def test_companyoverview_api_doc_reference():
+    """Test that class correctly references API documentation"""
+    assert "doc_id=46" in CompanyOverview.__doc__

xfintech/data/source/tushare/stock/companyprofit/__init__.py

@@ -0,0 +1,21 @@
+from __future__ import annotations
+
+from xfintech.data.source.tushare.stock.companyprofit.companyprofit import (
+    CompanyProfit,
+)
+from xfintech.data.source.tushare.stock.companyprofit.constant import (
+    KEY,
+    NAME,
+    PAGINATE,
+    SOURCE,
+    TARGET,
+)
+
+__all__ = [
+    "CompanyProfit",
+    "KEY",
+    "NAME",
+    "PAGINATE",
+    "SOURCE",
+    "TARGET",
+]