> ## Documentation Index
> Fetch the complete documentation index at: https://cndoc-langchain.site/llms.txt
> Use this file to discover all available pages before exploring further.

# ParallelSearchRetriever 集成

> 使用 LangChain Python 集成 ParallelSearchRetriever 检索器。

[`ParallelSearchRetriever`](https://reference.langchain.com/python/langchain-parallel/retrievers/ParallelSearchRetriever) 是一个由 [Parallel](https://platform.parallel.ai/) 的 [Search API](https://docs.parallel.ai/search/search-quickstart) 支持的 LangChain [`BaseRetriever`](/oss/python/langchain/retrieval)。它返回带有丰富 `metadata`（`url`、`title`、`publish_date`、`search_id`、`excerpts`、`query`）的 `list[Document]`，并可接入任何 RAG 流水线。

<Note>
  寻找一个可由 LLM 调用的工具，该工具返回原始搜索响应而非 `Document`？请参阅 [ParallelSearchTool](/oss/python/integrations/tools/parallel_search)。
</Note>

## 概述

### 集成详情

| 类                                                                                                                         | 包                                                                                  | JS 支持 |                                                                                                                        包最新版本                                                                                                                       |
| :------------------------------------------------------------------------------------------------------------------------ | :--------------------------------------------------------------------------------- | :---: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
| [`ParallelSearchRetriever`](https://reference.langchain.com/python/langchain-parallel/retrievers/ParallelSearchRetriever) | [`langchain-parallel`](https://reference.langchain.com/python/langchain-parallel/) |   ❌   | <a href="https://pypi.org/project/langchain-parallel/" target="_blank"><img src="https://img.shields.io/pypi/v/langchain-parallel?style=flat-square&label=%20&color=orange" alt="PyPI - Latest version" noZoom height="100" class="rounded" /></a> |

## 设置

该集成位于 `langchain-parallel` 包中。

<CodeGroup>
  ```bash pip theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
  pip install -U langchain-parallel
  ```

  ```bash uv theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
  uv add langchain-parallel
  ```
</CodeGroup>

### 凭证

前往 [Parallel](https://platform.parallel.ai) 注册并生成 API 密钥。在您的环境中设置 `PARALLEL_API_KEY`：

```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
import getpass
import os

if not os.environ.get("PARALLEL_API_KEY"):
    os.environ["PARALLEL_API_KEY"] = getpass.getpass("Parallel API key:\n")
```

## 实例化

```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_parallel import ParallelSearchRetriever

retriever = ParallelSearchRetriever(
    max_results=3,
    excerpts={"max_chars_per_result": 800},
)
```

## 用法

每个返回的 `Document` 将其摘录合并到 `page_content` 中，并在 `metadata` 中公开源 URL、标题和发布日期：

```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
docs = retriever.invoke("breakthroughs in fusion energy 2025")
for d in docs:
    print(d.metadata.get("title"), "—", d.metadata.get("url"))
    print(d.page_content[:200], "...\n")
```

```text theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
Net energy gain in fusion: NIF results — https://www.nature.com/articles/...
The National Ignition Facility achieved net energy gain on December 5, 2022 ...

Commonwealth Fusion's SPARC milestone — https://news.mit.edu/...
SPARC is on track for first plasma in 2026 ...
```

### 配置搜索

传递一个 `objective` 以提供比关键字 `search_queries` 更丰富的检索目标。检索器将源和获取策略转发给底层的 Search API。

```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
configured = ParallelSearchRetriever(
    max_results=5,
    excerpts={"max_chars_per_result": 1500},
    mode="basic",  # 'basic' (lower latency) or 'advanced' (higher quality)
    source_policy={
        "include_domains": ["nature.com", "science.org", "iter.org"],
    },
)

docs = configured.invoke(
    "What's the latest peer-reviewed result on net-energy-gain fusion?"
)
```

### 异步

```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
docs = await retriever.ainvoke("Latest GLP-1 trial results 2025")
```

## 在链中使用

`ParallelSearchRetriever` 可接入任何 LangChain 链：

```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain.chat_models import init_chat_model

llm = init_chat_model(model="claude-opus-4-7")

prompt = ChatPromptTemplate.from_messages([
    ("system", "Answer using only the context below. Cite URLs."),
    ("human", "Context:\n{context}\n\nQuestion: {question}"),
])

def format_docs(docs):
    return "\n\n".join(
        f"[{d.metadata.get('url')}] {d.page_content[:500]}" for d in docs
    )

chain = (
    {"context": retriever | format_docs, "question": RunnablePassthrough()}
    | prompt
    | llm
    | StrOutputParser()
)

chain.invoke("What was the most recent fusion energy breakthrough?")
```

## API 参考

有关详细文档，请访问 [`ParallelSearchRetriever`](https://reference.langchain.com/python/langchain-parallel/retrievers/ParallelSearchRetriever) API 参考或 [Parallel Search API 指南](https://docs.parallel.ai/search/search-quickstart)。

***

<div className="source-links">
  <Callout icon="terminal-2">
    [将这些文档连接](/use-these-docs)到 Claude、VSCode 等，通过 MCP 获取实时答案。
  </Callout>

  <Callout icon="edit">
    [在 GitHub 上编辑此页面](https://github.com/langchain-ai/docs/edit/main/src/oss/python/integrations/retrievers/parallel.mdx) 或 [提交问题](https://github.com/langchain-ai/docs/issues/new/choose)。
  </Callout>
</div>