langchain_community.document_loaders.polars_dataframe

from typing import Any, Iterator

from langchain_core.documents import Document

from langchain_community.document_loaders.dataframe import BaseDataFrameLoader


[docs]class PolarsDataFrameLoader(BaseDataFrameLoader):
    """加载`Polars` DataFrame。"""

[docs]    def __init__(self, data_frame: Any, *, page_content_column: str = "text"):
        """使用dataframe对象进行初始化。

参数：
    data_frame：Polars DataFrame对象。
    page_content_column：包含页面内容的列的名称。默认为"text"。
"""
        import polars as pl

        if not isinstance(data_frame, pl.DataFrame):
            raise ValueError(
                f"Expected data_frame to be a pl.DataFrame, got {type(data_frame)}"
            )
        super().__init__(data_frame, page_content_column=page_content_column)

[docs]    def lazy_load(self) -> Iterator[Document]:
        """从数据框中延迟加载记录。"""

        for row in self.data_frame.iter_rows(named=True):
            text = row[self.page_content_column]
            row.pop(self.page_content_column)
            yield Document(page_content=text, metadata=row)
Source code for langchain_community.document_loaders.polars_dataframe
