Azure AI Formrecognizer Java

1---
2name: azure-ai-formrecognizer-java
3description: Build document analysis applications with Azure Document Intelligence (Form Recognizer) SDK for Java. Use when extracting text, tables, key-value pairs from documents, receipts, invoices, or building custom document models.
4package: com.azure:azure-ai-formrecognizer
5---
6 
7# Azure Document Intelligence (Form Recognizer) SDK for Java
8 
9Build document analysis applications using the Azure AI Document Intelligence SDK for Java.
10 
11## Installation
12 
13```xml
14<dependency>
15    <groupId>com.azure</groupId>
16    <artifactId>azure-ai-formrecognizer</artifactId>
17    <version>4.2.0-beta.1</version>
18</dependency>
19```
20 
21## Client Creation
22 
23### DocumentAnalysisClient
24 
25```java
26import com.azure.ai.formrecognizer.documentanalysis.DocumentAnalysisClient;
27import com.azure.ai.formrecognizer.documentanalysis.DocumentAnalysisClientBuilder;
28import com.azure.core.credential.AzureKeyCredential;
29 
30DocumentAnalysisClient client = new DocumentAnalysisClientBuilder()
31    .credential(new AzureKeyCredential("{key}"))
32    .endpoint("{endpoint}")
33    .buildClient();
34```
35 
36### DocumentModelAdministrationClient
37 
38```java
39import com.azure.ai.formrecognizer.documentanalysis.administration.DocumentModelAdministrationClient;
40import com.azure.ai.formrecognizer.documentanalysis.administration.DocumentModelAdministrationClientBuilder;
41 
42DocumentModelAdministrationClient adminClient = new DocumentModelAdministrationClientBuilder()
43    .credential(new AzureKeyCredential("{key}"))
44    .endpoint("{endpoint}")
45    .buildClient();
46```
47 
48### With DefaultAzureCredential
49 
50```java
51import com.azure.identity.DefaultAzureCredentialBuilder;
52 
53DocumentAnalysisClient client = new DocumentAnalysisClientBuilder()
54    .endpoint("{endpoint}")
55    .credential(new DefaultAzureCredentialBuilder().build())
56    .buildClient();
57```
58 
59## Prebuilt Models
60 
61| Model ID | Purpose |
62|----------|---------|
63| `prebuilt-layout` | Extract text, tables, selection marks |
64| `prebuilt-document` | General document with key-value pairs |
65| `prebuilt-receipt` | Receipt data extraction |
66| `prebuilt-invoice` | Invoice field extraction |
67| `prebuilt-businessCard` | Business card parsing |
68| `prebuilt-idDocument` | ID document (passport, license) |
69| `prebuilt-tax.us.w2` | US W2 tax forms |
70 
71## Core Patterns
72 
73### Extract Layout
74 
75```java
76import com.azure.ai.formrecognizer.documentanalysis.models.*;
77import com.azure.core.util.BinaryData;
78import com.azure.core.util.polling.SyncPoller;
79import java.io.File;
80 
81File document = new File("document.pdf");
82BinaryData documentData = BinaryData.fromFile(document.toPath());
83 
84SyncPoller<OperationResult, AnalyzeResult> poller = 
85    client.beginAnalyzeDocument("prebuilt-layout", documentData);
86 
87AnalyzeResult result = poller.getFinalResult();
88 
89// Process pages
90for (DocumentPage page : result.getPages()) {
91    System.out.printf("Page %d: %.2f x %.2f %s%n",
92        page.getPageNumber(),
93        page.getWidth(),
94        page.getHeight(),
95        page.getUnit());
96    
97    // Lines
98    for (DocumentLine line : page.getLines()) {
99        System.out.println("Line: " + line.getContent());
100    }
101    
102    // Selection marks (checkboxes)
103    for (DocumentSelectionMark mark : page.getSelectionMarks()) {
104        System.out.printf("Checkbox: %s (confidence: %.2f)%n",
105            mark.getSelectionMarkState(),
106            mark.getConfidence());
107    }
108}
109 
110// Tables
111for (DocumentTable table : result.getTables()) {
112    System.out.printf("Table: %d rows x %d columns%n",
113        table.getRowCount(),
114        table.getColumnCount());
115    
116    for (DocumentTableCell cell : table.getCells()) {
117        System.out.printf("Cell[%d,%d]: %s%n",
118            cell.getRowIndex(),
119            cell.getColumnIndex(),
120            cell.getContent());
121    }
122}
123```
124 
125### Analyze from URL
126 
127```java
128String documentUrl = "https://example.com/invoice.pdf";
129 
130SyncPoller<OperationResult, AnalyzeResult> poller = 
131    client.beginAnalyzeDocumentFromUrl("prebuilt-invoice", documentUrl);
132 
133AnalyzeResult result = poller.getFinalResult();
134```
135 
136### Analyze Receipt
137 
138```java
139SyncPoller<OperationResult, AnalyzeResult> poller = 
140    client.beginAnalyzeDocumentFromUrl("prebuilt-receipt", receiptUrl);
141 
142AnalyzeResult result = poller.getFinalResult();
143 
144for (AnalyzedDocument doc : result.getDocuments()) {
145    Map<String, DocumentField> fields = doc.getFields();
146    
147    DocumentField merchantName = fields.get("MerchantName");
148    if (merchantName != null && merchantName.getType() == DocumentFieldType.STRING) {
149        System.out.printf("Merchant: %s (confidence: %.2f)%n",
150            merchantName.getValueAsString(),
151            merchantName.getConfidence());
152    }
153    
154    DocumentField transactionDate = fields.get("TransactionDate");
155    if (transactionDate != null && transactionDate.getType() == DocumentFieldType.DATE) {
156        System.out.printf("Date: %s%n", transactionDate.getValueAsDate());
157    }
158    
159    DocumentField items = fields.get("Items");
160    if (items != null && items.getType() == DocumentFieldType.LIST) {
161        for (DocumentField item : items.getValueAsList()) {
162            Map<String, DocumentField> itemFields = item.getValueAsMap();
163            System.out.printf("Item: %s, Price: %.2f%n",
164                itemFields.get("Name").getValueAsString(),
165                itemFields.get("Price").getValueAsDouble());
166        }
167    }
168}
169```
170 
171### General Document Analysis
172 
173```java
174SyncPoller<OperationResult, AnalyzeResult> poller = 
175    client.beginAnalyzeDocumentFromUrl("prebuilt-document", documentUrl);
176 
177AnalyzeResult result = poller.getFinalResult();
178 
179// Key-value pairs
180for (DocumentKeyValuePair kvp : result.getKeyValuePairs()) {
181    System.out.printf("Key: %s => Value: %s%n",
182        kvp.getKey().getContent(),
183        kvp.getValue() != null ? kvp.getValue().getContent() : "null");
184}
185```
186 
187## Custom Models
188 
189### Build Custom Model
190 
191```java
192import com.azure.ai.formrecognizer.documentanalysis.administration.models.*;
193 
194String blobContainerUrl = "{SAS_URL_of_training_data}";
195String prefix = "training-docs/";
196 
197SyncPoller<OperationResult, DocumentModelDetails> poller = adminClient.beginBuildDocumentModel(
198    blobContainerUrl,
199    DocumentModelBuildMode.TEMPLATE,
200    prefix,
201    new BuildDocumentModelOptions()
202        .setModelId("my-custom-model")
203        .setDescription("Custom invoice model"),
204    Context.NONE);
205 
206DocumentModelDetails model = poller.getFinalResult();
207 
208System.out.println("Model ID: " + model.getModelId());
209System.out.println("Created: " + model.getCreatedOn());
210 
211model.getDocumentTypes().forEach((docType, details) -> {
212    System.out.println("Document type: " + docType);
213    details.getFieldSchema().forEach((field, schema) -> {
214        System.out.printf("  Field: %s (%s)%n", field, schema.getType());
215    });
216});
217```
218 
219### Analyze with Custom Model
220 
221```java
222SyncPoller<OperationResult, AnalyzeResult> poller = 
223    client.beginAnalyzeDocumentFromUrl("my-custom-model", documentUrl);
224 
225AnalyzeResult result = poller.getFinalResult();
226 
227for (AnalyzedDocument doc : result.getDocuments()) {
228    System.out.printf("Document type: %s (confidence: %.2f)%n",
229        doc.getDocType(),
230        doc.getConfidence());
231    
232    doc.getFields().forEach((name, field) -> {
233        System.out.printf("Field '%s': %s (confidence: %.2f)%n",
234            name,
235            field.getContent(),
236            field.getConfidence());
237    });
238}
239```
240 
241### Compose Models
242 
243```java
244List<String> modelIds = Arrays.asList("model-1", "model-2", "model-3");
245 
246SyncPoller<OperationResult, DocumentModelDetails> poller = 
247    adminClient.beginComposeDocumentModel(
248        modelIds,
249        new ComposeDocumentModelOptions()
250            .setModelId("composed-model")
251            .setDescription("Composed from multiple models"));
252 
253DocumentModelDetails composedModel = poller.getFinalResult();
254```
255 
256### Manage Models
257 
258```java
259// List models
260PagedIterable<DocumentModelSummary> models = adminClient.listDocumentModels();
261for (DocumentModelSummary summary : models) {
262    System.out.printf("Model: %s, Created: %s%n",
263        summary.getModelId(),
264        summary.getCreatedOn());
265}
266 
267// Get model details
268DocumentModelDetails model = adminClient.getDocumentModel("model-id");
269 
270// Delete model
271adminClient.deleteDocumentModel("model-id");
272 
273// Check resource limits
274ResourceDetails resources = adminClient.getResourceDetails();
275System.out.printf("Models: %d / %d%n",
276    resources.getCustomDocumentModelCount(),
277    resources.getCustomDocumentModelLimit());
278```
279 
280## Document Classification
281 
282### Build Classifier
283 
284```java
285Map<String, ClassifierDocumentTypeDetails> docTypes = new HashMap<>();
286docTypes.put("invoice", new ClassifierDocumentTypeDetails()
287    .setAzureBlobSource(new AzureBlobContentSource(containerUrl).setPrefix("invoices/")));
288docTypes.put("receipt", new ClassifierDocumentTypeDetails()
289    .setAzureBlobSource(new AzureBlobContentSource(containerUrl).setPrefix("receipts/")));
290 
291SyncPoller<OperationResult, DocumentClassifierDetails> poller = 
292    adminClient.beginBuildDocumentClassifier(docTypes,
293        new BuildDocumentClassifierOptions().setClassifierId("my-classifier"));
294 
295DocumentClassifierDetails classifier = poller.getFinalResult();
296```
297 
298### Classify Document
299 
300```java
301SyncPoller<OperationResult, AnalyzeResult> poller = 
302    client.beginClassifyDocumentFromUrl("my-classifier", documentUrl, Context.NONE);
303 
304AnalyzeResult result = poller.getFinalResult();
305 
306for (AnalyzedDocument doc : result.getDocuments()) {
307    System.out.printf("Classified as: %s (confidence: %.2f)%n",
308        doc.getDocType(),
309        doc.getConfidence());
310}
311```
312 
313## Error Handling
314 
315```java
316import com.azure.core.exception.HttpResponseException;
317 
318try {
319    client.beginAnalyzeDocumentFromUrl("prebuilt-receipt", "invalid-url");
320} catch (HttpResponseException e) {
321    System.out.println("Status: " + e.getResponse().getStatusCode());
322    System.out.println("Error: " + e.getMessage());
323}
324```
325 
326## Environment Variables
327 
328```bash
329FORM_RECOGNIZER_ENDPOINT=https://<resource>.cognitiveservices.azure.com/
330FORM_RECOGNIZER_KEY=<your-api-key>
331```
332 
333## Trigger Phrases
334 
335- "document intelligence Java"
336- "form recognizer SDK"
337- "extract text from PDF"
338- "OCR document Java"
339- "analyze invoice receipt"
340- "custom document model"
341- "document classification"
342