Amazon Textract とAWSSDK

入力ドキュメントからテキストを検出します。

    public static void detectDocText(TextractClient textractClient,String sourceDoc) {

        try {

            InputStream sourceStream = new FileInputStream(new File(sourceDoc));
            SdkBytes sourceBytes = SdkBytes.fromInputStream(sourceStream);

            // Get the input Document object as bytes
            Document myDoc = Document.builder()
                    .bytes(sourceBytes)
                    .build();

            DetectDocumentTextRequest detectDocumentTextRequest = DetectDocumentTextRequest.builder()
                    .document(myDoc)
                    .build();

            // Invoke the Detect operation
            DetectDocumentTextResponse textResponse = textractClient.detectDocumentText(detectDocumentTextRequest);

            List<Block> docInfo = textResponse.blocks();

            Iterator<Block> blockIterator = docInfo.iterator();

            while(blockIterator.hasNext()) {
                Block block = blockIterator.next();
                System.out.println("The block type is " +block.blockType().toString());
            }

            DocumentMetadata documentMetadata = textResponse.documentMetadata();
            System.out.println("The number of pages in the document is " +documentMetadata.pages());

        } catch (TextractException | FileNotFoundException e) {

            System.err.println(e.getMessage());
            System.exit(1);
        }
    }

Amazon S3 バケット内のドキュメントからテキストを検出します。

    public static void detectDocTextS3 (TextractClient textractClient, String bucketName, String docName) {

        try {
            S3Object s3Object = S3Object.builder()
                    .bucket(bucketName)
                    .name(docName)
                    .build();

            // Create a Document object and reference the s3Object instance
            Document myDoc = Document.builder()
                    .s3Object(s3Object)
                    .build();

            // Create a DetectDocumentTextRequest object
            DetectDocumentTextRequest detectDocumentTextRequest = DetectDocumentTextRequest.builder()
                    .document(myDoc)
                    .build();

            // Invoke the detectDocumentText method
            DetectDocumentTextResponse textResponse = textractClient.detectDocumentText(detectDocumentTextRequest);

            List<Block> docInfo = textResponse.blocks();

            Iterator<Block> blockIterator = docInfo.iterator();

            while(blockIterator.hasNext()) {
                Block block = blockIterator.next();
                System.out.println("The block type is " +block.blockType().toString());
            }

            DocumentMetadata documentMetadata = textResponse.documentMetadata();
            System.out.println("The number of pages in the document is " +documentMetadata.pages());

        } catch (TextractException e) {

            System.err.println(e.getMessage());
            System.exit(1);
        }
    }

class TextractWrapper:
    """Encapsulates Textract functions."""
    def __init__(self, textract_client, s3_resource, sqs_resource):
        """
        :param textract_client: A Boto3 Textract client.
        :param s3_resource: A Boto3 Amazon S3 resource.
        :param sqs_resource: A Boto3 Amazon SQS resource.
        """
        self.textract_client = textract_client
        self.s3_resource = s3_resource
        self.sqs_resource = sqs_resource

    def detect_file_text(self, *, document_file_name=None, document_bytes=None):
        """
        Detects text elements in a local image file or from in-memory byte data.
        The image must be in PNG or JPG format.

        :param document_file_name: The name of a document image file.
        :param document_bytes: In-memory byte data of a document image.
        :return: The response from Amazon Textract, including a list of blocks
                 that describe elements detected in the image.
        """
        if document_file_name is not None:
            with open(document_file_name, 'rb') as document_file:
                document_bytes = document_file.read()
        try:
            response = self.textract_client.detect_document_text(
                Document={'Bytes': document_bytes})
            logger.info(
                "Detected %s blocks.", len(response['Blocks']))
        except ClientError:
            logger.exception("Couldn't detect text.")
            raise
        else:
            return response