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

xfintech/fabric/table/tests/test_class_info_all.py

@@ -0,0 +1,547 @@
+import pytest
+
+from xfintech.fabric.column.info import ColumnInfo
+from xfintech.fabric.column.kind import ColumnKind
+from xfintech.fabric.table.info import TableInfo
+
+
+def test_tableinfo_basic_init():
+    table = TableInfo(
+        name="daily_market",
+        desc=" 日线行情表 ",
+        meta={"source": "tushare"},
+        columns=[
+            ColumnInfo(name="price", kind="Float", desc="收盘价"),
+            ColumnInfo(name="volume", kind="Integer", desc="成交量"),
+        ],
+    )
+    assert table.name == "daily_market"
+    assert table.desc == "日线行情表"
+    assert table.meta == {"source": "tushare"}
+    assert len(table.columns) == 2
+    assert "price" in table.columns
+    assert "volume" in table.columns
+
+
+def test_tableinfo_name_must_be_valid():
+    with pytest.raises(ValueError):
+        TableInfo(name="1invalid")
+
+
+def test_tableinfo_name_is_lowercase():
+    table = TableInfo(name="MyTable")
+    assert table.name == "mytable"
+
+
+def test_tableinfo_name_empty_when_none():
+    table = TableInfo(name=None)
+    assert table.name == ""
+
+
+def test_tableinfo_name_accepts_underscore_hyphen():
+    table1 = TableInfo(name="my_table")
+    assert table1.name == "my_table"
+
+    table2 = TableInfo(name="my-table")
+    assert table2.name == "my-table"
+
+    table3 = TableInfo(name="_private_table")
+    assert table3.name == "_private_table"
+
+
+def test_tableinfo_desc_strip():
+    table = TableInfo(
+        name="test",
+        desc="  Something Here   ",
+    )
+    assert table.desc == "Something Here"
+
+
+def test_tableinfo_desc_empty_when_none():
+    table = TableInfo(name="test", desc=None)
+    assert table.desc == ""
+
+
+def test_tableinfo_meta_bytes_to_str():
+    table = TableInfo(
+        name="test",
+        meta={
+            b"bin": b"yes",
+            "code": b"ok",
+        },
+    )
+    assert table.meta == {"bin": "yes", "code": "ok"}
+
+
+def test_tableinfo_meta_str_normal():
+    table = TableInfo(name="test", meta={"x": 1, "y": "value"})
+    assert table.meta == {"x": 1, "y": "value"}
+
+
+def test_tableinfo_meta_none():
+    table = TableInfo(name="test", meta=None)
+    assert table.meta is None
+
+
+def test_tableinfo_columns_from_columninfo_list():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    assert len(table.columns) == 2
+    assert table.columns["price"].kind == ColumnKind.FLOAT
+    assert table.columns["volume"].kind == ColumnKind.INTEGER
+
+
+def test_tableinfo_columns_from_dict_list():
+    table = TableInfo(
+        name="test",
+        columns=[
+            {"name": "price", "kind": "Float"},
+            {"name": "volume", "kind": "Integer"},
+        ],
+    )
+    assert len(table.columns) == 2
+    assert table.columns["price"].kind == ColumnKind.FLOAT
+    assert table.columns["volume"].kind == ColumnKind.INTEGER
+
+
+def test_tableinfo_columns_mixed_types():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            {"name": "volume", "kind": "Integer"},
+        ],
+    )
+    assert len(table.columns) == 2
+    assert isinstance(table.columns["price"], ColumnInfo)
+    assert isinstance(table.columns["volume"], ColumnInfo)
+
+
+def test_tableinfo_columns_invalid_type():
+    with pytest.raises(TypeError):
+        TableInfo(
+            name="test",
+            columns=["invalid"],
+        )
+
+
+def test_tableinfo_columns_none():
+    table = TableInfo(name="test", columns=None)
+    assert table.columns == {}
+
+
+def test_tableinfo_get_column():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    col = table.get_column("price")
+    assert col is not None
+    assert col.name == "price"
+    assert col.kind == ColumnKind.FLOAT
+
+
+def test_tableinfo_get_column_case_insensitive():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    col = table.get_column("PRICE")
+    assert col is not None
+    assert col.name == "price"
+
+
+def test_tableinfo_get_column_not_exists():
+    table = TableInfo(name="test", columns=[])
+    col = table.get_column("nonexistent")
+    assert col is None
+
+
+def test_tableinfo_add_column():
+    table = TableInfo(name="test")
+    table.add_column(ColumnInfo(name="price", kind="Float"))
+
+    assert len(table.columns) == 1
+    assert "price" in table.columns
+    assert table.columns["price"].kind == ColumnKind.FLOAT
+
+
+def test_tableinfo_add_column_from_dict():
+    table = TableInfo(name="test")
+    table.add_column({"name": "price", "kind": "Float"})
+
+    assert len(table.columns) == 1
+    assert "price" in table.columns
+    assert table.columns["price"].kind == ColumnKind.FLOAT
+
+
+def test_tableinfo_remove_column():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    table.remove_column("price")
+
+    assert len(table.columns) == 1
+    assert "price" not in table.columns
+    assert "volume" in table.columns
+
+
+def test_tableinfo_remove_column_case_insensitive():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table.remove_column("PRICE")
+
+    assert len(table.columns) == 0
+
+
+def test_tableinfo_remove_column_not_exists():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table.remove_column("nonexistent")
+
+    assert len(table.columns) == 1
+
+
+def test_tableinfo_update_column_kind_desc_meta():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(
+                name="price",
+                kind="Float",
+                desc="old",
+                meta={"unit": "CNY"},
+            ),
+        ],
+    )
+    table.update_column(
+        "price",
+        kind="Integer",
+        desc="new-desc",
+        meta={"precision": 2},
+    )
+
+    col = table.get_column("price")
+    assert col.kind == ColumnKind.INTEGER
+    assert col.desc == "new-desc"
+    assert col.meta == {"unit": "CNY", "precision": 2}
+
+
+def test_tableinfo_update_column_rename():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table.update_column("price", new="new_price")
+
+    assert "price" not in table.columns
+    assert "new_price" in table.columns
+    assert table.columns["new_price"].kind == ColumnKind.FLOAT
+
+
+def test_tableinfo_update_column_not_exists():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table.update_column("nonexistent", desc="new")
+
+    assert len(table.columns) == 1
+    assert table.columns["price"].desc == ""
+
+
+def test_tableinfo_rename_column():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="old_name", kind="Float"),
+        ],
+    )
+    table.rename_column("old_name", "new_name")
+
+    assert "old_name" not in table.columns
+    assert "new_name" in table.columns
+    assert table.columns["new_name"].name == "new_name"
+
+
+def test_tableinfo_rename_column_case_insensitive():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="old_name", kind="Float"),
+        ],
+    )
+    table.rename_column("OLD_NAME", "new_name")
+
+    assert "old_name" not in table.columns
+    assert "new_name" in table.columns
+
+
+def test_tableinfo_rename_column_not_exists():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table.rename_column("nonexistent", "new_name")
+
+    assert len(table.columns) == 1
+    assert "new_name" not in table.columns
+
+
+def test_tableinfo_list_columns():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    cols = table.list_columns()
+
+    assert len(cols) == 2
+    assert all(isinstance(c, ColumnInfo) for c in cols)
+    names = [c.name for c in cols]
+    assert "price" in names
+    assert "volume" in names
+
+
+def test_tableinfo_list_columns_empty():
+    table = TableInfo(name="test")
+    cols = table.list_columns()
+
+    assert cols == []
+
+
+def test_tableinfo_full_describe():
+    table = TableInfo(
+        name="test",
+        desc="Test table",
+        meta={"source": "tushare"},
+        columns=[
+            ColumnInfo(name="price", kind="Float", desc="Price"),
+        ],
+    )
+    d = table.describe()
+
+    assert d["name"] == "test"
+    assert d["desc"] == "Test table"
+    assert d["meta"] == {"source": "tushare"}
+    assert len(d["columns"]) == 1
+    assert d["columns"][0]["name"] == "price"
+
+
+def test_tableinfo_partial_describe():
+    table = TableInfo(name="test")
+    d = table.describe()
+    assert "name" in d
+    assert "desc" not in d
+    assert "meta" not in d
+    assert "columns" not in d
+
+
+def test_tableinfo_full_to_dict():
+    table = TableInfo(
+        name="test",
+        desc="Test table",
+        meta={"source": "tushare"},
+        columns=[
+            ColumnInfo(name="price", kind="Float", desc="Price"),
+        ],
+    )
+    d = table.to_dict()
+
+    assert "identifier" in d
+    assert d["name"] == "test"
+    assert d["desc"] == "Test table"
+    assert d["meta"] == {"source": "tushare"}
+    assert len(d["columns"]) == 1
+
+
+def test_tableinfo_from_dict():
+    data = {
+        "name": "test",
+        "desc": "Test table",
+        "meta": {"source": "tushare"},
+        "columns": [
+            {"name": "price", "kind": "Float"},
+            {"name": "volume", "kind": "Integer"},
+        ],
+    }
+    table = TableInfo.from_dict(data)
+
+    assert table.name == "test"
+    assert table.desc == "Test table"
+    assert table.meta == {"source": "tushare"}
+    assert len(table.columns) == 2
+
+
+def test_tableinfo_from_dict_minimal():
+    data = {}
+    table = TableInfo.from_dict(data)
+
+    assert table.name == ""
+    assert table.desc == ""
+    assert table.meta is None
+    assert table.columns == {}
+
+
+def test_tableinfo_str():
+    table = TableInfo(
+        name="test",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    s = str(table)
+    assert "price" in s
+
+
+def test_tableinfo_repr():
+    table = TableInfo(name="test")
+    r = repr(table)
+    assert "TableInfo" in r
+    assert "test" in r
+
+
+def test_tableinfo_identifier_is_deterministic():
+    table1 = TableInfo(
+        name="test",
+        desc="Test table",
+        meta={"source": "tushare"},
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    table2 = TableInfo(
+        name="test",
+        desc="Test table",
+        meta={"source": "tushare"},
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    assert table1.identifier == table2.identifier
+
+
+def test_tableinfo_identifier_not_change_with_meta():
+    table1 = TableInfo(
+        name="test",
+        desc="Test table",
+        meta={"source": "tushare"},
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table2 = TableInfo(
+        name="test",
+        desc="Test table",
+        meta={"source": "yahoo"},
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    assert table1.identifier == table2.identifier
+
+
+def test_tableinfo_identifier_changes_with_name():
+    table1 = TableInfo(
+        name="test1",
+        desc="Test table",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table2 = TableInfo(
+        name="test2",
+        desc="Test table",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    assert table1.identifier != table2.identifier
+
+
+def test_tableinfo_identifier_changes_with_desc():
+    table1 = TableInfo(
+        name="test",
+        desc="Desc1",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table2 = TableInfo(
+        name="test",
+        desc="Desc2",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    assert table1.identifier != table2.identifier
+
+
+def test_tableinfo_identifier_changes_with_columns():
+    table1 = TableInfo(
+        name="test",
+        desc="Test table",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    table2 = TableInfo(
+        name="test",
+        desc="Test table",
+        columns=[
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    assert table1.identifier != table2.identifier
+
+
+def test_tableinfo_identifier_column_order_independent():
+    table1 = TableInfo(
+        name="test",
+        desc="Test table",
+        columns=[
+            ColumnInfo(name="price", kind="Float"),
+            ColumnInfo(name="volume", kind="Integer"),
+        ],
+    )
+    table2 = TableInfo(
+        name="test",
+        desc="Test table",
+        columns=[
+            ColumnInfo(name="volume", kind="Integer"),
+            ColumnInfo(name="price", kind="Float"),
+        ],
+    )
+    assert table1.identifier == table2.identifier

xfintech/serde/__init__.py

@@ -0,0 +1,35 @@
+from .common.dataformat import DataFormat
+from .common.deserialiserlike import DeserialiserLike
+from .common.error import (
+    DeserialiserFailedError,
+    DeserialiserImportError,
+    DeserialiserInputTypeError,
+    DeserialiserNotSupportedError,
+    SerialiserFailedError,
+    SerialiserImportError,
+    SerialiserInputTypeError,
+    SerialiserNotSupportedError,
+)
+from .common.serialiserlike import SerialiserLike
+from .deserialiser.pandas import PandasDeserialiser
+from .deserialiser.python import PythonDeserialiser
+from .serialiser.pandas import PandasSerialiser
+from .serialiser.python import PythonSerialiser
+
+__all__ = [
+    "DataFormat",
+    "DeserialiserLike",
+    "SerialiserLike",
+    "PandasDeserialiser",
+    "PythonDeserialiser",
+    "PandasSerialiser",
+    "PythonSerialiser",
+    "DeserialiserFailedError",
+    "DeserialiserImportError",
+    "DeserialiserInputTypeError",
+    "DeserialiserNotSupportedError",
+    "SerialiserFailedError",
+    "SerialiserImportError",
+    "SerialiserInputTypeError",
+    "SerialiserNotSupportedError",
+]

xfintech/serde/common/__init__.py

@@ -0,0 +1,9 @@
+from .dataformat import DataFormat
+from .deserialiserlike import DeserialiserLike
+from .serialiserlike import SerialiserLike
+
+__all__ = [
+    "DataFormat",
+    "SerialiserLike",
+    "DeserialiserLike",
+]

xfintech/serde/common/dataformat.py

@@ -0,0 +1,78 @@
+from __future__ import annotations
+
+from enum import Enum
+
+
+class DataFormat(str, Enum):
+    """
+    描述:
+    - 序列化和反序列化格式枚举类。
+    - 定义支持的数据格式：Parquet、CSV 和 JSON。
+    - 继承自 str 和 Enum，可以直接与字符串比较。
+
+    属性:
+    - PARQUET: Parquet 格式 ("parquet")
+    - CSV: CSV 格式 ("csv")
+    - JSON: JSON 格式 ("json")
+
+    方法:
+    - from_str(string): 从字符串创建 DataFormat 实例（不区分大小写）。
+    - from_any(value): 从 DataFormat 实例或字符串创建 DataFormat 实例。
+    - list_all(): 列出所有支持的 DataFormat 实例。
+    - __str__(): 返回格式的字符串值。
+    - __repr__(): 返回格式的表示形式。
+
+    例子:
+    ```python
+        from xfintech.serde.common.dataformat import DataFormat
+
+        # 使用枚举值
+        fmt = DataFormat.PARQUET
+        print(fmt)  # parquet
+        print(repr(fmt))  # DataFormat.PARQUET
+
+        # 从字符串创建（不区分大小写）
+        fmt = DataFormat.from_str("csv")
+        print(fmt)  # csv
+
+        fmt = DataFormat.from_str("JSON")
+        print(fmt)  # json
+
+        # 字符串比较
+        if DataFormat.PARQUET == "parquet":
+            print("Matched!")  # 输出: Matched!
+
+        # 无效格式会抛出 ValueError
+        try:
+            DataFormat.from_str("xml")
+        except ValueError as e:
+            print(f"Error: {e}")  # Error: Unknown DataFormat: xml
+    ```
+    """
+
+    PARQUET = "parquet"
+    CSV = "csv"
+    JSON = "json"
+
+    @classmethod
+    def from_str(cls, value: str) -> DataFormat:
+        for fmt in DataFormat:
+            if fmt.value.lower() == value.lower():
+                return fmt
+        raise ValueError(f"Unknown DataFormat: {value}")
+
+    @classmethod
+    def from_any(cls, value: DataFormat | str) -> DataFormat:
+        if isinstance(value, DataFormat):
+            return value
+        return cls.from_str(value)
+
+    @classmethod
+    def list_all(cls) -> list[DataFormat]:
+        return [fmt for fmt in cls]
+
+    def __str__(self) -> str:
+        return self.value
+
+    def __repr__(self) -> str:
+        return f"{self.__class__.__name__}.{self.name}"

xfintech/serde/common/deserialiserlike.py

@@ -0,0 +1,38 @@
+from __future__ import annotations
+
+from typing import Any, Protocol, runtime_checkable
+
+
+@runtime_checkable
+class DeserialiserLike(Protocol):
+    """
+    描述:
+    - 通用反序列化接口协议。
+    - 定义反序列化方法的签名，适用于各种数据格式。
+    - 将字节流转换为数据对象（如 DataFrame）。
+
+    方法:
+    - deserialise(data, format, **kwargs): 通用反序列化方法，返回数据对象。
+
+    例子:
+    ```python
+        from xfintech.serde.common.deserialiserlike import DeserialiserLike
+        import pandas as pd
+
+        class MyDeserialiser:
+            @staticmethod
+            def deserialise(data, format, **kwargs):
+                # 实现反序列化逻辑
+                return pd.DataFrame()
+
+        # 检查是否符合协议
+        assert isinstance(MyDeserialiser(), DeserialiserLike)
+    ```
+    """
+
+    @staticmethod
+    def deserialise(
+        data: Any,
+        format: Any,
+        **kwargs: Any,
+    ) -> Any: ...