Google Document AI

[Alucor Rpa]

Hi, I was trying to implement Invoice parser using python on GCP Console. I am able to run the program but I see that it works just like form parser. I cannot find the fields like invoice_no, date etc being extracted like the Invoice schema in demo. How do I access the invoice schema? Below is the code that I used: 

# TODO(developer): Uncomment these variables before running the sample.

project_id= ''

location = 'us' # Format is 'us' or 'eu'

processor_id = '' # Create processor in Cloud Console

file_path = 'invoice.pdf'

def process_document_sample(

    project_id: str, location: str, processor_id: str, file_path: str

):

    from google.cloud import documentai_v1beta3 as documentai

    # You must set the api_endpoint if you use a location other than 'us', e.g.:

    opts = {}

    if location == "eu":

        opts = {"api_endpoint": "eu-documentai.googleapis.com"}

    client = documentai.DocumentProcessorServiceClient(client_options=opts)

    # The full resource name of the processor, e.g.:

    # projects/project-id/locations/location/processor/processor-id

    # You must create new processors in the Cloud Console first

    name = f"projects/{project_id}/locations/{location}/processors/{processor_id}"

    with open(file_path, "rb") as image:

        image_content = image.read()

    # Read the file into memory

    document = {"content": image_content, "mime_type": "application/pdf"}

    # Configure the process request

    request = {"name": name, "raw_document": document}

    # Recognizes text entities in the PDF document

    result = client.process_document(request=request)

    document = result.document

    print("Document processing complete.")

    

    # For a full list of Document object attributes, please reference this page: https://googleapis.dev/python/documentai/latest/_modules/google/cloud/documentai_v1beta3/types/document.html#Document

    document_pages = document.pages

    file = open("sample.txt", "w")

    str_dictionary = repr(document_pages)

    file.write(str_dictionary)

    # Read the text recognition output from the processor

    print("The document contains the following paragraphs:")

   

    for page in document_pages:

        print("Page Number:{}".format(page.page_number))

        for form_field in page.form_fields:

            fieldName=get_text(form_field.field_name,document)

            

            fieldValue = get_text(form_field.field_value,document)

            

            print(fieldName+" : "+fieldValue )

# Extract shards from the text field

def get_text(doc_element: dict, document: dict):

    """

    Document AI identifies form fields by their offsets

    in document text. This function converts offsets

    to text snippets.

    """

    response = ""

    # If a text segment spans several lines, it will

    # be stored in different text segments.

    for segment in doc_element.text_anchor.text_segments:

        start_index = (

            int(segment.start_index)

            if segment in doc_element.text_anchor.text_segments

            else 0

        )

        end_index = int(segment.end_index)

        response += document.text[start_index:end_index]

    return response

process_document_sample(project_id, location, processor_id, file_path)

[michaill]

Hello,

From the documentation here [1] are the fields that can get extracted with the invoice parser processor. 

Let me know if this answers your question or if you have any further questions.

Best regards.

---

[1] https://cloud.google.com/document-ai/docs/invoice-parser#annotations

[Alucor Rpa]

Thank you. That was very helpful.