基础URL

https://gateway.maton.ai/reducto/{native-api-path}

将{native-api-path}替换为实际的Reducto API端点路径。网关将请求代理到platform.reducto.ai并自动注入您的API密钥。

认证

所有请求都需要在Authorization头中包含Maton API密钥：

Authorization: Bearer $MATONAPIKEY

环境变量： 将您的API密钥设置为MATONAPIKEY：

bash
export MATONAPIKEY=YOURAPIKEY

获取您的API密钥

1. 1. 在maton.ai登录或创建账户
2. 前往maton.ai/settings
3. 复制您的API密钥

连接管理

在https://ctrl.maton.ai管理您的Reducto API密钥连接。

列出连接

bash
python < import urllib.request, os, json
req = urllib.request.Request(https://ctrl.maton.ai/connections?app=reducto&status=ACTIVE)
req.addheader(Authorization, fBearer {os.environ[MATONAPI_KEY]})
print(json.dumps(json.load(urllib.request.urlopen(req)), indent=2))
EOF

创建连接

bash
python < import urllib.request, os, json
data = json.dumps({app: reducto}).encode()
req = urllib.request.Request(https://ctrl.maton.ai/connections, data=data, method=POST)
req.addheader(Authorization, fBearer {os.environ[MATONAPI_KEY]})
req.add_header(Content-Type, application/json)
print(json.dumps(json.load(urllib.request.urlopen(req)), indent=2))
EOF

获取连接

bash
python < import urllib.request, os, json
req = urllib.request.Request(https://ctrl.maton.ai/connections/{connection_id})
req.addheader(Authorization, fBearer {os.environ[MATONAPI_KEY]})
print(json.dumps(json.load(urllib.request.urlopen(req)), indent=2))
EOF

响应：
json
{
connection: {
connection_id: f7579208-276c-455f-9962-4635fca739b9,
status: ACTIVE,
creation_time: 2026-02-28T00:12:24.797884Z,
lastupdatedtime: 2026-02-28T00:16:13.509841Z,
url: https://connect.maton.ai/?session_token=...,
app: reducto,
metadata: {},
method: API_KEY
}
}

在浏览器中打开返回的url以输入您的Reducto API密钥。

删除连接

bash
python < import urllib.request, os, json
req = urllib.request.Request(https://ctrl.maton.ai/connections/{connection_id}, method=DELETE)
req.addheader(Authorization, fBearer {os.environ[MATONAPI_KEY]})
print(json.dumps(json.load(urllib.request.urlopen(req)), indent=2))
EOF

指定连接

如果您有多个Reducto连接，请使用Maton-Connection头指定要使用的连接：

bash
python < import urllib.request, os, json
req = urllib.request.Request(https://gateway.maton.ai/reducto/parse)
req.addheader(Authorization, fBearer {os.environ[MATONAPI_KEY]})
req.add_header(Maton-Connection, f7579208-276c-455f-9962-4635fca739b9)
print(json.dumps(json.load(urllib.request.urlopen(req)), indent=2))
EOF

如果省略，网关将使用默认（最早创建的）活动连接。

API参考

解析文档

解析文档并提取结构化内容（文本、表格、图表）。

同步解析

bash
POST /reducto/parse
Content-Type: application/json

{
document_url: https://example.com/document.pdf
}

响应：
json
{
job_id: 04b8aa38-7eb3-4151-98b0-dbaea71358d9,
duration: 17.85,
pdf_url: https://...,
studio_link: https://studio.reducto.ai/job/...,
usage: {
num_pages: 15,
credits: 15.0
},
result: {
chunks: [
{
content: 提取的文本内容...,
blocks: [...]
}
]
}
}

异步解析

对于长文档，使用异步以避免超时：

bash
POST /reducto/parse_async
Content-Type: application/json

{
document_url: https://example.com/document.pdf
}

响应：
json
{
job_id: e234ba95-410a-4dd0-8a14-743dbfc49470
}

使用GET /reducto/job/{job_id}轮询任务状态。

提取数据

使用JSON模式从文档中提取特定字段。

同步提取

bash
POST /reducto/extract
Content-Type: application/json

{
document_url: https://example.com/document.pdf,
schema: {
type: object,
properties: {
title: {type: string, description: 文档标题},
authors: {type: array, items: {type: string}, description: 作者姓名列表}
}
}
}

响应：
json
{
job_id: 36f01a34-7ef6-40da-9e74-7c14902b6182,
usage: {
num_pages: 15,
num_fields: 9,
credits: 45.0
},
studio_link: https://studio.reducto.ai/job/...,
result: [
{
title: 文档标题,
authors: [作者一, 作者二]
}
]
}

异步提取

bash
POST /reducto/extract_async
Content-Type: application/json

{
document_url: https://example.com/document.pdf,
schema: {
type: object,
properties: {
title: {type: string}
}
}
}

响应：
json
{
job_id: 0cdb6a50-df92-438b-875b-8b5c72d5b089
}

分割文档

根据内容类别将文档划分为逻辑章节。

同步分割

bash
POST /reducto/split
Content-Type: application/json

{
document_url: https://example.com/document.pdf,
split_description: [
{name: abstract, description: 摘要部分},
{name: introduction, description: 引言部分},
{name: conclusion, description: 结论部分}
]
}

响应：
json
{
usage: {
num_pages: 15,
credits: 15.0
},
result: {
section_mapping: {
abstract: [1],
introduction: [1, 2],
conclusion: [14, 15]
},
splits: [
{
name: abstract,
pages: [1],
conf: high
}
]
}
}

异步分割

bash
POST /reducto/split_async
Content-Type: application/json

{
document_url: https://example.com/document.pdf,
split_description: [
{name: abstract, description: 摘要部分}
]
}

响应：
json
{
job_id: 381