|
| 1 | +import os |
| 2 | +import base64 |
| 3 | +from typing import Dict, List, Optional |
| 4 | + |
| 5 | +from veryfi.client_base import Client |
| 6 | + |
| 7 | + |
| 8 | +class Classify: |
| 9 | + def __init__(self, client: Client): |
| 10 | + self.client = client |
| 11 | + |
| 12 | + def classify_document( |
| 13 | + self, |
| 14 | + file_path: str, |
| 15 | + document_types: Optional[List] = None, |
| 16 | + **kwargs, |
| 17 | + ) -> Dict: |
| 18 | + """ |
| 19 | + Classify a document from a file path. |
| 20 | + https://docs.veryfi.com/api/classify/classify-a-document/ |
| 21 | +
|
| 22 | + :param file_path: Path on disk to a file to submit for classification |
| 23 | + :param document_types: Optional list of document types to classify into. If omitted, a preset set of types will be used. |
| 24 | + :param kwargs: Additional body parameters |
| 25 | + :return: Classification result with document type prediction |
| 26 | + """ |
| 27 | + endpoint_name = "/classify/" |
| 28 | + file_name = os.path.basename(file_path) |
| 29 | + with open(file_path, "rb") as image_file: |
| 30 | + base64_encoded_string = base64.b64encode(image_file.read()).decode("utf-8") |
| 31 | + request_arguments = { |
| 32 | + "file_name": file_name, |
| 33 | + "file_data": base64_encoded_string, |
| 34 | + } |
| 35 | + if document_types is not None: |
| 36 | + request_arguments["document_types"] = document_types |
| 37 | + request_arguments.update(kwargs) |
| 38 | + return self.client._request("POST", endpoint_name, request_arguments) |
| 39 | + |
| 40 | + def classify_document_url( |
| 41 | + self, |
| 42 | + file_url: Optional[str] = None, |
| 43 | + file_urls: Optional[List[str]] = None, |
| 44 | + document_types: Optional[List] = None, |
| 45 | + **kwargs, |
| 46 | + ) -> Dict: |
| 47 | + """ |
| 48 | + Classify a document from a URL. |
| 49 | + https://docs.veryfi.com/api/classify/classify-a-document/ |
| 50 | +
|
| 51 | + :param file_url: Required if file_urls isn't specified. Publicly accessible URL to a file, e.g. "https://cdn.example.com/receipt.jpg". |
| 52 | + :param file_urls: Required if file_url isn't specified. List of publicly accessible URLs to multiple files. |
| 53 | + :param document_types: Optional list of document types to classify into. If omitted, a preset set of types will be used. |
| 54 | + :param kwargs: Additional body parameters |
| 55 | + :return: Classification result with document type prediction |
| 56 | + """ |
| 57 | + endpoint_name = "/classify/" |
| 58 | + request_arguments = { |
| 59 | + "file_url": file_url, |
| 60 | + "file_urls": file_urls, |
| 61 | + } |
| 62 | + if document_types is not None: |
| 63 | + request_arguments["document_types"] = document_types |
| 64 | + request_arguments.update(kwargs) |
| 65 | + return self.client._request("POST", endpoint_name, request_arguments) |
0 commit comments