从知识库检索分段 / 测试检索

post/datasets/{dataset_id}/retrieve

Request example

curl --request POST \
  --url https://api.dify.ai/v1/datasets/{dataset_id}/retrieve \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
  "attachment_ids": [
    "<string>"
  ],
  "external_retrieval_model": {
    "score_threshold": 1,
    "score_threshold_enabled": true,
    "top_k": 1
  },
  "query": "<string>",
  "retrieval_model": {
    "metadata_filtering_conditions": {
      "conditions": [
        {
          "comparison_operator": "contains",
          "name": "<string>",
          "value": "<string>"
        }
      ],
      "logical_operator": "and"
    },
    "reranking_enable": true,
    "reranking_mode": "reranking_model",
    "reranking_model": {
      "reranking_model_name": "<string>",
      "reranking_provider_name": "<string>"
    },
    "score_threshold": 1,
    "score_threshold_enabled": true,
    "search_method": "keyword_search",
    "top_k": 1,
    "weights": {
      "keyword_setting": {
        "keyword_weight": 1
      },
      "vector_setting": {
        "embedding_model_name": "<string>",
        "embedding_provider_name": "<string>",
        "vector_weight": 1
      },
      "weight_type": "semantic_first"
    }
  }
}'

import requests

url = "https://api.dify.ai/v1/datasets/{dataset_id}/retrieve"
headers = {
    "Authorization": "Bearer <token>",
    "Content-Type": "application/json",
}
payload = {
    "attachment_ids": [
        "<string>"
    ],
    "external_retrieval_model": {
        "score_threshold": 1,
        "score_threshold_enabled": True,
        "top_k": 1
    },
    "query": "<string>",
    "retrieval_model": {
        "metadata_filtering_conditions": {
            "conditions": [
                {
                    "comparison_operator": "contains",
                    "name": "<string>",
                    "value": "<string>"
                }
            ],
            "logical_operator": "and"
        },
        "reranking_enable": True,
        "reranking_mode": "reranking_model",
        "reranking_model": {
            "reranking_model_name": "<string>",
            "reranking_provider_name": "<string>"
        },
        "score_threshold": 1,
        "score_threshold_enabled": True,
        "search_method": "keyword_search",
        "top_k": 1,
        "weights": {
            "keyword_setting": {
                "keyword_weight": 1
            },
            "vector_setting": {
                "embedding_model_name": "<string>",
                "embedding_provider_name": "<string>",
                "vector_weight": 1
            },
            "weight_type": "semantic_first"
        }
    }
}

response = requests.request("POST", url, headers=headers, json=payload)

print(response.text)

const response = await fetch('https://api.dify.ai/v1/datasets/{dataset_id}/retrieve', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer <token>',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    "attachment_ids": [
      "<string>"
    ],
    "external_retrieval_model": {
      "score_threshold": 1,
      "score_threshold_enabled": true,
      "top_k": 1
    },
    "query": "<string>",
    "retrieval_model": {
      "metadata_filtering_conditions": {
        "conditions": [
          {
            "comparison_operator": "contains",
            "name": "<string>",
            "value": "<string>"
          }
        ],
        "logical_operator": "and"
      },
      "reranking_enable": true,
      "reranking_mode": "reranking_model",
      "reranking_model": {
        "reranking_model_name": "<string>",
        "reranking_provider_name": "<string>"
      },
      "score_threshold": 1,
      "score_threshold_enabled": true,
      "search_method": "keyword_search",
      "top_k": 1,
      "weights": {
        "keyword_setting": {
          "keyword_weight": 1
        },
        "vector_setting": {
          "embedding_model_name": "<string>",
          "embedding_provider_name": "<string>",
          "vector_weight": 1
        },
        "weight_type": "semantic_first"
      }
    }
  })
});

const data = await response.json();
console.log(data);

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_URL => 'https://api.dify.ai/v1/datasets/{dataset_id}/retrieve',
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_CUSTOMREQUEST => 'POST',
  CURLOPT_POSTFIELDS => '{
  "attachment_ids": [
    "<string>"
  ],
  "external_retrieval_model": {
    "score_threshold": 1,
    "score_threshold_enabled": true,
    "top_k": 1
  },
  "query": "<string>",
  "retrieval_model": {
    "metadata_filtering_conditions": {
      "conditions": [
        {
          "comparison_operator": "contains",
          "name": "<string>",
          "value": "<string>"
        }
      ],
      "logical_operator": "and"
    },
    "reranking_enable": true,
    "reranking_mode": "reranking_model",
    "reranking_model": {
      "reranking_model_name": "<string>",
      "reranking_provider_name": "<string>"
    },
    "score_threshold": 1,
    "score_threshold_enabled": true,
    "search_method": "keyword_search",
    "top_k": 1,
    "weights": {
      "keyword_setting": {
        "keyword_weight": 1
      },
      "vector_setting": {
        "embedding_model_name": "<string>",
        "embedding_provider_name": "<string>",
        "vector_weight": 1
      },
      "weight_type": "semantic_first"
    }
  }
}',
  CURLOPT_HTTPHEADER => [
    'Authorization: Bearer <token>',
    'Content-Type: application/json',
  ],
]);

$response = curl_exec($curl);
curl_close($curl);

echo $response;

package main

import (
	"fmt"
	"io"
	"net/http"
	"strings"
)

func main() {
	url := "https://api.dify.ai/v1/datasets/{dataset_id}/retrieve"
	payload := strings.NewReader("{\n  \"attachment_ids\": [\n    \"<string>\"\n  ],\n  \"external_retrieval_model\": {\n    \"score_threshold\": 1,\n    \"score_threshold_enabled\": true,\n    \"top_k\": 1\n  },\n  \"query\": \"<string>\",\n  \"retrieval_model\": {\n    \"metadata_filtering_conditions\": {\n      \"conditions\": [\n        {\n          \"comparison_operator\": \"contains\",\n          \"name\": \"<string>\",\n          \"value\": \"<string>\"\n        }\n      ],\n      \"logical_operator\": \"and\"\n    },\n    \"reranking_enable\": true,\n    \"reranking_mode\": \"reranking_model\",\n    \"reranking_model\": {\n      \"reranking_model_name\": \"<string>\",\n      \"reranking_provider_name\": \"<string>\"\n    },\n    \"score_threshold\": 1,\n    \"score_threshold_enabled\": true,\n    \"search_method\": \"keyword_search\",\n    \"top_k\": 1,\n    \"weights\": {\n      \"keyword_setting\": {\n        \"keyword_weight\": 1\n      },\n      \"vector_setting\": {\n        \"embedding_model_name\": \"<string>\",\n        \"embedding_provider_name\": \"<string>\",\n        \"vector_weight\": 1\n      },\n      \"weight_type\": \"semantic_first\"\n    }\n  }\n}")

	req, _ := http.NewRequest("POST", url, payload)
	req.Header.Add("Authorization", "Bearer <token>")
	req.Header.Add("Content-Type", "application/json")

	res, _ := http.DefaultClient.Do(req)
	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))
}

HttpRequest request = HttpRequest.newBuilder()
    .uri(URI.create("https://api.dify.ai/v1/datasets/{dataset_id}/retrieve"))
    .header("Authorization", "Bearer <token>")
    .header("Content-Type", "application/json")
    .method("POST", HttpRequest.BodyPublishers.ofString("{\n  \"attachment_ids\": [\n    \"<string>\"\n  ],\n  \"external_retrieval_model\": {\n    \"score_threshold\": 1,\n    \"score_threshold_enabled\": true,\n    \"top_k\": 1\n  },\n  \"query\": \"<string>\",\n  \"retrieval_model\": {\n    \"metadata_filtering_conditions\": {\n      \"conditions\": [\n        {\n          \"comparison_operator\": \"contains\",\n          \"name\": \"<string>\",\n          \"value\": \"<string>\"\n        }\n      ],\n      \"logical_operator\": \"and\"\n    },\n    \"reranking_enable\": true,\n    \"reranking_mode\": \"reranking_model\",\n    \"reranking_model\": {\n      \"reranking_model_name\": \"<string>\",\n      \"reranking_provider_name\": \"<string>\"\n    },\n    \"score_threshold\": 1,\n    \"score_threshold_enabled\": true,\n    \"search_method\": \"keyword_search\",\n    \"top_k\": 1,\n    \"weights\": {\n      \"keyword_setting\": {\n        \"keyword_weight\": 1\n      },\n      \"vector_setting\": {\n        \"embedding_model_name\": \"<string>\",\n        \"embedding_provider_name\": \"<string>\",\n        \"vector_weight\": 1\n      },\n      \"weight_type\": \"semantic_first\"\n    }\n  }\n}"))
    .build();

HttpResponse<String> response = HttpClient.newHttpClient()
    .send(request, HttpResponse.BodyHandlers.ofString());

System.out.println(response.body());

require 'uri'
require 'net/http'

url = URI('https://api.dify.ai/v1/datasets/{dataset_id}/retrieve')

http = Net::HTTP.new(url.host, url.port)
http.use_ssl = url.scheme == 'https'

request = Net::HTTP::Post.new(url)
request['Authorization'] = 'Bearer <token>'
request['Content-Type'] = 'application/json'
request.body = '{
  "attachment_ids": [
    "<string>"
  ],
  "external_retrieval_model": {
    "score_threshold": 1,
    "score_threshold_enabled": true,
    "top_k": 1
  },
  "query": "<string>",
  "retrieval_model": {
    "metadata_filtering_conditions": {
      "conditions": [
        {
          "comparison_operator": "contains",
          "name": "<string>",
          "value": "<string>"
        }
      ],
      "logical_operator": "and"
    },
    "reranking_enable": true,
    "reranking_mode": "reranking_model",
    "reranking_model": {
      "reranking_model_name": "<string>",
      "reranking_provider_name": "<string>"
    },
    "score_threshold": 1,
    "score_threshold_enabled": true,
    "search_method": "keyword_search",
    "top_k": 1,
    "weights": {
      "keyword_setting": {
        "keyword_weight": 1
      },
      "vector_setting": {
        "embedding_model_name": "<string>",
        "embedding_provider_name": "<string>",
        "vector_weight": 1
      },
      "weight_type": "semantic_first"
    }
  }
}'

response = http.request(request)
puts response.read_body

Response

{
  "query": {
    "content": "What is Dify?"
  },
  "records": [
    {
      "child_chunks": [],
      "files": [],
      "score": 0.92,
      "segment": {
        "answer": "",
        "completed_at": 1741267200,
        "content": "Dify is an open-source LLM app development platform.",
        "created_at": 1741267200,
        "created_by": "ad313dd6-ef04-4dd1-a5b0-c0f0b9e2e7e4",
        "disabled_at": null,
        "disabled_by": null,
        "document": {
          "data_source_type": "upload_file",
          "doc_metadata": null,
          "doc_type": null,
          "id": "a8e0e5b5-78c6-4130-a5ce-25feb0e0b4ac",
          "name": "guide.txt"
        },
        "document_id": "a8e0e5b5-78c6-4130-a5ce-25feb0e0b4ac",
        "enabled": true,
        "error": null,
        "hit_count": 1,
        "id": "f3d1c7be-9f3a-40d8-8eb8-3a1ef9c3f2c1",
        "index_node_hash": "abc123def456",
        "index_node_id": "a1b2c3d4-e5f6-7890-abcd-000000000001",
        "indexing_at": 1741267200,
        "keywords": [
          "dify",
          "platform",
          "llm"
        ],
        "position": 1,
        "sign_content": "",
        "status": "completed",
        "stopped_at": null,
        "tokens": 12,
        "word_count": 9
      },
      "summary": null,
      "tsne_position": null
    }
  ]
}

{
  "code": "completion_request_error",
  "message": "Completion request failed.",
  "status": 400
}

{
  "code": "dataset_not_initialized",
  "message": "The dataset is still being initialized or indexing. Please wait a moment.",
  "status": 400
}

{
  "code": "invalid_param",
  "message": "Invalid parameter value.",
  "status": 400
}

{
  "code": "model_currently_not_support",
  "message": "Dify Hosted OpenAI trial currently not support the GPT-4 model.",
  "status": 400
}

{
  "code": "provider_not_initialize",
  "message": "No valid model provider credentials found. Please go to Settings -> Model Provider to complete your provider credentials.",
  "status": 400
}

{
  "code": "provider_quota_exceeded",
  "message": "Your quota for Dify Hosted Model Provider has been exhausted. Please go to Settings -> Model Provider to complete your own provider credentials.",
  "status": 400
}

{
  "code": "forbidden",
  "message": "Dataset api access is not enabled.",
  "status": 403
}

{
  "code": "forbidden",
  "message": "Sorry, you have reached the knowledge base request rate limit of your subscription.",
  "status": 403
}

{
  "code": "forbidden",
  "message": "The API key is not authorized to access this knowledge base.",
  "status": 403
}

{
  "code": "not_found",
  "message": "Dataset not found.",
  "status": 404
}

{
  "code": "internal_server_error",
  "message": "An internal error occurred.",
  "status": 500
}

post/datasets/{dataset_id}/retrieve

Authorization

Authorizationstringheaderrequired#

每个请求都通过 API Key 认证：Authorization: Bearer {API_KEY}。应用接口使用应用 API Key，知识库接口使用知识库 API Key（快速开始）。

API Key 应保存在服务端，切勿嵌入客户端代码。缺失或无效的 Key 会返回 HTTP 401（unauthorized）。

Path parameters

dataset_idstringrequired#

知识库 ID，通过获取知识库列表获取。使用限定范围的密钥时，从 Dify 页面 URL 复制 ID。

Body

application/json

attachment_idsstring[]#

要包含在检索上下文中的附件 ID 列表。

external_retrieval_modelobject#

外部知识库的检索设置。

Show child attributes

score_thresholdnumber#

用于过滤结果的最小相似度分数阈值。

score_threshold_enabledboolean#

是否启用分数阈值过滤。

top_kinteger#

返回结果的最大数量。

querystringrequired#

搜索查询文本。

Length: max 250

retrieval_modelobject#

Show child attributes

metadata_filtering_conditionsobject#

仅检索文档元数据匹配指定条件的分段。条件由服务端基于文档元数据字段评估。

Show child attributes

conditionsobject[]#

要评估的元数据条件列表。

Show child attributes

comparison_operatorenum<string>required#

按元数据类型选择要应用的比较方式：

字符串或数组类型元数据：contains、not contains、start with、end with、is、is not、empty、not empty、in、not in
数值类型元数据：=、≠、>、<、≥、≤
时间类型元数据：before、after

Options:containsnot containsstart withend withisis notemptynot emptyinnot in=≠><≥≤beforeafter

namestringrequired#

用于比较的元数据字段名称。

valuestring · string[] · number#

用于比较的值。类型取决于 comparison_operator：大多数字符串运算符使用字符串；in 和 not in 使用字符串数组；数值运算符使用数字；empty 和 not empty 时省略。

logical_operatorenum<string>#

多个条件之间的组合方式。

Default: "and"

Options:andor

reranking_enablebooleanrequired#

是否启用重排序。

reranking_modeenum<string>#

重排序模式。当 reranking_enable 为 true 时必填。

Options:reranking_modelweighted_score

reranking_modelobject#

重排序模型配置。

Show child attributes

reranking_model_namestring#

重排序模型名称。

reranking_provider_namestring#

重排序模型供应商标识符，格式为 organization/plugin_name/provider_name（例如 langgenius/cohere/cohere）。cohere 这类简写会展开为 langgenius/<name>/<name>，仅对 langgenius 发布的插件有效。

有效取值从获取可用模型中 model_type=rerank 返回的 provider 字段获取。

score_thresholdnumber#

结果的最低相关性分数。仅在 score_threshold_enabled 为 true 时生效。

score_threshold_enabledbooleanrequired#

是否启用分数阈值过滤。

search_methodenum<string>required#

用于检索的搜索方法。

Options:keyword_searchsemantic_searchfull_text_searchhybrid_search

top_kintegerrequired#

返回的最大结果数。

weightsobject#

混合搜索的权重配置。

Show child attributes

keyword_settingobject#

关键词搜索权重设置。

Show child attributes

keyword_weightnumber#

分配给关键词搜索结果的权重。

vector_settingobject#

语义搜索权重设置。

Show child attributes

embedding_model_namestring#

用于向量搜索的嵌入模型名称。

embedding_provider_namestring#

用于向量搜索的嵌入模型供应商。

vector_weightnumber#

分配给语义（向量）搜索结果的权重。

weight_typeenum<string>#

平衡语义搜索和关键词搜索权重的策略。

Options:semantic_firstkeyword_firstcustomized

Responses

application/json

检索结果。

queryobject#

原始查询对象。

Show child attributes

contentstring#

查询文本。

recordsobject[]#

匹配的检索记录列表。

Show child attributes

child_chunksobject[]#

使用分层索引时，分段内匹配的子分段。

Show child attributes

contentstring#

子分段的文本内容。

idstring#

子分段的唯一标识符。

positioninteger#

子分段在父分段中的位置。

scorenumber#

子分段的相关性得分。

filesobject[]#

附加到该分段的文件。

Show child attributes

extensionstring#

文件扩展名。

idstring#

附件文件标识符。

mime_typestring#

文件的 MIME 类型。

namestring#

原始文件名。

sizeinteger#

文件大小（字节）。

source_urlstring#

访问附件的 URL。

scorenumber#