Document AI

OCR

With the OCR tool of ComPDFKit Document AI, you can recognize the text from all kinds of images or scanned PDF files as you want. The following examples show how to upload a JPG file and run the OCR processing using Java, PHP, C#, Python, and Swift programming languages. Then, output a JSON file.

Java

// Create a client
CPDFClient client = new CPDFClient(publicKey,secretKey);

// Create a task
// Create an example of a DocumentAI OCR task
CPDFCreateTaskResult result = client.createTask(CPDFDocumentAIEnum.OCR);

// Get a task id
String taskId = result.getTaskId();

// File handling parameter settings
CPDFOcrParameter fileParameter = new CPDFOcrParameter();
fileParameter.setLang("auto");

// Upload files
client.uploadFile(new File("test.jpg"), taskId, fileParameter);

// Execute task
client.executeTask(taskId);

// Query TaskInfo
CPDFTaskInfoResult taskInfo = client.getTaskInfo(taskId);

PHP

// Create a client
$client = new CPDFClient('public_key', 'secret_key');

// Create a task
// Create an example of a DocumentAI OCR task
$taskInfo = $client->createTask(CPDFDocumentAI::OCR);

// File handling parameter settings
$file = $client->addFile('test.jpg')
    ->setLang('auto');

// Upload files
$fileInfo = $file->uploadFile($taskInfo['taskId']);

// Execute task
$client->executeTask($taskInfo['taskId']);

// Query TaskInfo
$taskInfo = $client->getTaskInfo($taskInfo['taskId']);

C#

// Create a client
CPDFClient client = new CPDFClient(publicKey,secretKey);

// Create a task
// Create an example of a DocumentAI OCR task
CPDFCreateTaskResult result = client.CreateTask(CPDFDocumentAIEnum.OCR);

// Get a task id
string taskId = result.TaskId;

// File handling parameter settings
CPDFOcrParameter fileParameter = new CPDFOcrParameter();
fileParameter.Lang = "auto";

// Upload files
client.UploadFile(new FileInfo("test.jpg"), taskId, fileParameter);

// Execute task
client.ExecuteTask(taskId);

// Query TaskInfo
CPDFTaskInfoResult taskInfo = client.GetTaskInfo(taskId);

Python

# Create a client
client = CPDFClient(public_key, secret_key)

# Create a task
# Create an example of a DocumentAI OCR task
create_task_result = client.create_task(CPDFDocumentAIEnum.OCR)

# Get a task id
task_id = create_task_result.task_id

# File handling parameter settings
file_parameter = CPDFOcrParameter()
file_parameter.lang = "auto"

# Upload files
client.upload_file('test.jpg', task_id, file_parameter)

# Execute task
client.execute_task(task_id)

# Query TaskInfo
task_info = client.get_task_info(task_id)

Swift

// Create a client
let client: CPDFClient = CPDFClient(publicKey: public_key, secretKey: secret_key)

Task { @MainActor in
    // Create a task
    // Create an example of a DocumentAI OCR task
    let taskModel = await client.createTask(url: CPDFDocumentAI.OCR)
    
    // Get a task id
    let taskId = taskModel?.taskId ?? ""

    // Upload files
    let path = Bundle.main.path(forResource: "test", ofType: "png")
    let uploadFileModel = await client.uploadFile(filepath: path ?? "", params: [
        CPDFFileUploadParameterKey.lang.string():"auto"
    ], taskId: taskId)
    
    // Execute task
    let _ = await client.processFiles(taskId: taskId)
    
    // Query TaskInfo
    let taskInfoModel = await client.getTaskInfo(taskId: taskId)
}

• lang：Supported types and definitions
  • auto - automatic classification language
  • english - English
  • chinese - Simplified Chinese
  • chinese_tra - Traditional Chinese
  • korean - Korean
  • japanese - Japanese
  • latin - Latin
  • devanagari - Sanskrit alphabet

Result：

File Type	Description
.json	OCR recognition results.

Content:

Parameter	Description
cost	OCR recognition time.
boxes	All detected object box positions of the input image.
text	OCR recognition content.
rec_scores	OCR text recognition score, the higher the score, the more credible the result.