主角出场：Document（libs/core/langchain_core/documents/base.py:288），简单到让人意外：

# libs/core/langchain_core/documents/base.py:288
class Document(BaseMedia):
    """Class for storing a piece of text and associated metadata.

    !!! note
        `Document` is for **retrieval workflows**, not chat I/O. For sending
        text to an LLM in a conversation, use message types from
        `langchain.messages`.

    Example:
        document = Document(
            page_content="Hello, world!", metadata={"source": "https://example.com"}
        )
    """

    page_content: str                        # ★正文：就是一段字符串
    type: Literal["Document"] = "Document"

    def __init__(self, page_content: str, **kwargs: Any) -> None:
        """Pass page_content in as positional or named arg."""
        super().__init__(page_content=page_content, **kwargs)   # 支持位置参数写法

metadata 卡片上通常写什么？没有强制 schema，约定俗成的常客：

再看一个隐藏细节——Document.__str__（libs/core/langchain_core/documents/base.py:331）被特意重写了：

# libs/core/langchain_core/documents/base.py:331
def __str__(self) -> str:
    """Override `__str__` to restrict it to page_content and metadata."""
    # The purpose of this change is to make sure that user code that feeds
    # Document objects directly into prompts remains unchanged due to the
    # addition of the id field (or any other fields in the future).
    if self.metadata:
        return f"page_content='{self.page_content}' metadata={self.metadata}"
    return f"page_content='{self.page_content}'"

搬运工协议 BaseLoader（libs/core/langchain_core/document_loaders/base.py:26）：

# libs/core/langchain_core/document_loaders/base.py:26
class BaseLoader(ABC):
    """Interface for document loader.

    Implementations should implement the lazy-loading method using generators
    to avoid loading all documents into memory at once.

    `load` is provided just for user convenience and should not be overridden."""

    # Sub-classes should not implement this method directly. Instead, they
    # should implement the lazy load method.
    def load(self) -> list[Document]:                 # base.py:37
        """Load data into `Document` objects."""
        return list(self.lazy_load())                 # ★一行：把传送带上的书全收进列表

    async def aload(self) -> list[Document]:
        return [document async for document in self.alazy_load()]

lazy_load 的默认实现（libs/core/langchain_core/document_loaders/base.py:91）藏着一个聪明的互相兜底：

# libs/core/langchain_core/document_loaders/base.py:91
def lazy_load(self) -> Iterator[Document]:
    """A lazy loader for `Document`."""
    if type(self).load != BaseLoader.load:        # ★子类重写了 load（老式写法）？
        return iter(self.load())                  #    → 用它的 load 凑一个迭代器
    msg = f"{self.__class__.__name__} does not implement lazy_load()"
    raise NotImplementedError(msg)                # 两个都没实现 → 明确报错

配角一：load_and_split（libs/core/langchain_core/document_loaders/base.py:53）——加载完顺手切块，预告了明天的主角：

# libs/core/langchain_core/document_loaders/base.py:53
def load_and_split(self, text_splitter: TextSplitter | None = None) -> list[Document]:
    """Load `Document` and split into chunks. Chunks are returned as `Document`.

    !!! danger
        Do not override this method. It should be considered to be deprecated!"""
    if text_splitter is None:
        ...
        text_splitter_: TextSplitter = RecursiveCharacterTextSplitter()   # ★默认切分器
    else:
        text_splitter_ = text_splitter
    docs = self.load()
    return text_splitter_.split_documents(docs)      # 切完还是 list[Document]

配角二：Blob（libs/core/langchain_core/documents/base.py:59）——"还没拆封的生书"：原始二进制 + MIME 类型 + 路径，支持 from_path（base.py:214）延迟读盘、as_string（base.py:158）按编码解码。它把"从哪拿字节"和"怎么解析成 Document"分开，PDF 解析器这类"拆封工"（BlobParser）就能独立复用。