custom_ocr.md

NOTE This guide is intended for engineers extending smart-docs-parser for their custom use-case. If you think this could be benificial to the community then please contribute to smart-docs-parser source-code using the generic guide here.

smart-docs-parser supports custom OCR module as API argument. The module is required to implement a function of name extractDocumentText which accepts document image url, API key as input and returns the list of string lines as output.

Usage

Configuration

config/default.json

Add OCR API key (if any) to the configuration

  "smart-docs-parser": {
    "api_keys": {
      "your-library-name": "YourAPIKEY"
    }
  }

Code

// ES6 import statement
import SmartDocuments from 'smart-docs-parser';
import CustomOCR from '/path/to/ocr_file';

// Sample Request
const extractedDocumentDetails = await SmartDocuments.extractDocumentDetailsFromImage({
    document_url: 'https://avatars2.githubusercontent.com/u/20634933?s=40&v=4',
    document_type: 'PAN_CARD',
    ocr_library: 'your-library-name',
    custom_ocr: CustomOCR
});

OCR Implementation

/path/to/ocr_file.ts

Module

Your OCR should expose a function called extractDocumentText

// ******************************************************* //
// Logic for API handlers starts here                      //
// ******************************************************* //
const extractDocumentText = ({ document_url: documentURL, api_key: apiKey }) => {
  ....
  ....
  return {
    	raw_text: rawText
    }
};
// ******************************************************* //
// Logic for API handlers ends here                        //
// ******************************************************* //

export default { extractDocumentText };

Request

Your parser function should accept { document_url: string, api_key: string } as input.

Response

Your parser function should return { raw_text: Array<string> } as output.

Example

Please refer to sample_ocr.ts