---

name: azure-ai-document-intelligence-ts description: "使用预构建和自定义模型从文档中提取文本、表格和结构化数据。" risk: unknown source: community date_added: "2026-02-27"

Azure 文档智能 REST SDK（TypeScript）

使用预构建和自定义模型从文档中提取文本、表格和结构化数据。

安装

npm install @azure-rest/ai-document-intelligence @azure/identity

环境变量

DOCUMENT_INTELLIGENCE_ENDPOINT=https://<resource>.cognitiveservices.azure.com
DOCUMENT_INTELLIGENCE_API_KEY=<api-key>

认证

重要：这是REST客户端。DocumentIntelligence 是函数，不是类。

DefaultAzureCredential

import DocumentIntelligence from "@azure-rest/ai-document-intelligence";
import { DefaultAzureCredential } from "@azure/identity";

const client = DocumentIntelligence(
  process.env.DOCUMENT_INTELLIGENCE_ENDPOINT!,
  new DefaultAzureCredential()
);

API Key

import DocumentIntelligence from "@azure-rest/ai-document-intelligence";

const client = DocumentIntelligence(
  process.env.DOCUMENT_INTELLIGENCE_ENDPOINT!,
  { key: process.env.DOCUMENT_INTELLIGENCE_API_KEY! }
);

分析文档（URL）

import DocumentIntelligence, {
  isUnexpected,
  getLongRunningPoller,
  AnalyzeOperationOutput
} from "@azure-rest/ai-document-intelligence";

const initialResponse = await client
  .path("/documentModels/{modelId}:analyze", "prebuilt-layout")
  .post({
    contentType: "application/json",
    body: { urlSource: "https://example.com/document.pdf" },
    queryParameters: { locale: "en-US" }
  });

if (isUnexpected(initialResponse)) {
  throw initialResponse.body.error;
}

const poller = getLongRunningPoller(client, initialResponse);
const result = (await poller.pollUntilDone()).body as AnalyzeOperationOutput;
console.log("Pages:", result.analyzeResult?.pages?.length);

分析文档（本地文件）

import { readFile } from "node:fs/promises";

const fileBuffer = await readFile("./document.pdf");
const base64Source = fileBuffer.toString("base64");

const initialResponse = await client
  .path("/documentModels/{modelId}:analyze", "prebuilt-invoice")
  .post({
    contentType: "application/json",
    body: { base64Source }
  });

预构建模型

| 模型ID | 用途 | |--------|------| | prebuilt-layout | 文本、表格、选择标记、图形 | | prebuilt-read | 纯文本提取（OCR） | | prebuilt-invoice | 发票字段提取 | | prebuilt-receipt | 收据字段提取 | | prebuilt-idDocument | 身份证件 | | prebuilt-businessCard | 名片 | | prebuilt-tax.us.w2 | 美国W-2税表 |

处理结果

// 遍历页面
for (const page of result.analyzeResult?.pages ?? []) {
  console.log(`第 ${page.pageNumber} 页: ${page.words?.length} 个词`);
}

// 遍历表格
for (const table of result.analyzeResult?.tables ?? []) {
  console.log(`表格: ${table.rowCount} 行 x ${table.columnCount} 列`);
  for (const cell of table.cells) {
    console.log(`[${cell.rowIndex},${cell.columnIndex}]: ${cell.content}`);
  }
}

常见问题

• 认证失败：确保端点URL不含尾部斜杠
• 模型不支持：检查API版本是否支持所选模型
• 文件过大：最大文件大小为500MB，最多2000页