mindee/src/v1/parsing/common/inference.js

Mindee Client Library for Node.js

import { CropperExtra, FullTextOcrExtra, RAGExtra } from "./extras/index.js";
import { MindeeConfigurationError } from "../../../errors/index.js";
/**
 *
 * @typeParam DocT an extension of a `Prediction`. Is generic by default to
 * allow for easier optional `PageT` generic typing.
 * @typeParam PageT an extension of a `DocT` (`Prediction`). Should only be set
 * if a document's pages have specific implementation.
 */
export class Inference {
    constructor(rawPrediction) {
        /** Wrapper for a document's pages prediction. */
        this.pages = [];
        this.isRotationApplied = rawPrediction?.is_rotation_applied ?? undefined;
        this.product = rawPrediction?.product;
        if (rawPrediction["extras"] &&
            Object.keys(rawPrediction["extras"].length > 0)) {
            const extras = {};
            Object.entries(rawPrediction["extras"]).forEach(([extraKey, extraValue]) => {
                switch (extraKey) {
                    case "cropper":
                        extras.cropper = new CropperExtra(extraValue);
                        break;
                    case "full_text_ocr":
                        extras.fullTextOcr = new FullTextOcrExtra(extraValue);
                        break;
                    case "rag":
                        extras.rag = new RAGExtra(extraValue);
                        break;
                }
            });
            this.extras = extras;
        }
    }
    /**
     * Default string representation.
     */
    toString() {
        let pages = "";
        if (this.pages.toString().length > 0) {
            pages = `
Page Predictions
================

${this.pages.map((e) => e.toString() || "").join("\n")}`;
        }
        return `Inference
#########
:Product: ${this.product.name} v${this.product.version}
:Rotation applied: ${this.isRotationApplied ? "Yes" : "No"}

Prediction
==========
${this.prediction.toString().length === 0 ? "" : this.prediction.toString() + "\n"}${pages}`;
    }
    /**
     * Takes in an input string and replaces line breaks with `\n`.
     * @param outStr string to cleanup
     * @returns cleaned out string
     */
    static cleanOutString(outStr) {
        const lines = / \n/gm;
        return outStr.replace(lines, "\n");
    }
}
/**
 * Factory to allow for static-like property access syntax in TypeScript.
 * Used to retrieve endpoint data for standard products.
 */
export class InferenceFactory {
    /**
     * Builds a blank product of the given type & sends back the endpointName & endpointVersion parameters of OTS classes.
     * Note: this is needed to avoid passing anything other than the class of the object to the parse()/enqueue() call.
     * @param inferenceClass Class of the product we are using
     * @returns {Inference} An empty instance of a given product.
     */
    static getEndpoint(inferenceClass) {
        const emptyProduct = new inferenceClass({
            prediction: {},
            pages: [],
        });
        if (!emptyProduct.endpointName ||
            !emptyProduct.endpointVersion ||
            emptyProduct.endpointName.length === 0 ||
            emptyProduct.endpointVersion.length === 0) {
            throw new MindeeConfigurationError(`Error during endpoint verification, no endpoint found for product ${inferenceClass.name}.`);
        }
        return [emptyProduct.endpointName, emptyProduct.endpointVersion];
    }
}