类class agentscope.rag

ParserBase

class ParserBase(ABC)

源码Source agentscope/rag/_parser/_base.py:22–116

解析器基类，把私有文档转化为可检索的知识。位于 agentscope.rag 包，继承自 ABC。核心成员包括 supported_extensions、parse。

Each subclass handles a single file format (or a related family, e.g. all plain-text MIME types). Subclasses are typically instantiated once and reused across many parse() calls.

Subclasses should be stateless or thread-safe — a single instance may be invoked concurrently from multiple agent runs.

Subclasses must declare :attr:supported_media_types so that the KnowledgeBaseManager can route uploaded files to the right parser based on standard IANA media types (RFC 6838).

属性摘要Attribute Summary

属性Attribute	类型Type	说明Description	值Value
`supported_media_types`	list[str]	`supported_media_types` 字段。	—

方法摘要Method Summary

方法Method	说明Description
`def supported_extensions() -> list[str]`	返回 `supported_extensions` 结果。 Filename extensions (including the leading ``.``) this parser can produce uploads for.
`async def parse(file: bytes \| str, filename: str) -> list[Section]`asyncabstractabstractmethod	执行解析相关操作。 Parse a file into a list of :class:`Section` objects.

属性详情Attribute Detail

supported_media_types

list[str]

supported_media_types 字段。声明为 list[str]。

The supported media types field.

源码Source agentscope/rag/_parser/_base.py:37

方法详情Method Detail

supported_extensions

def supported_extensions() -> list[str]

返回 supported_extensions 结果。

The base implementation derives extensions from :attr:supported_media_types via :func:mimetypes.guess_all_extensions — good enough for clean IANA types like application/pdf.

返回Returns list[str]

返回str列表。`list[str]`:

源码Source agentscope/rag/_parser/_base.py:44–71

parse

async def parse(file: bytes | str, filename: str) -> list[Section]

asyncabstractabstractmethod

执行解析相关操作。

The file argument is a union covering the three call sites a parser sees in practice: - bytes — the raw payload, as handed in by HTTP uploads and blob-store reads.

参数Parameters

名称Name	类型Type	说明Description
`file`	bytes \| str	文件（类型 `bytes \| str`）。The file content or a path to it (see above).
`filename`	str	filename（类型 `str`）。The original filename (e.g. `"report.pdf"`). Used for error messages and copied into each Section's :attr:`Section.source` field for downstream display / citation.

返回Returns list[Section]

返回Section列表。`list[Section]`:

异常Raises

可能抛出 `Exception`。`TypeError`: If the subclass does not accept the supplied
可能抛出 `Exception`。`file` form.
可能抛出 `Exception`。`FileNotFoundError`: If a binary parser is handed a
可能抛出 `Exception`。`str` that does not name an existing file.
可能抛出 `Exception`。`ValueError`: If the file cannot be parsed.

源码Source agentscope/rag/_parser/_base.py:74–116