tethys.backend/start/rules/referenceValidation.ts

import { FieldContext } from '@vinejs/vine/types';
import vine from '@vinejs/vine';
import { VineString } from '@vinejs/vine';
import { default as axios } from 'axios';
import { ReferenceIdentifierTypes } from '#contracts/enums';

type Options = {
    typeField: string;
};

// Function to check if DOI exists using the DOI API
async function checkDoiExists(doi: string): Promise<boolean> {
    try {
        const response = await axios.get(`${doi}`);
        return response.status === 200; // If status is 200, DOI is valid
    } catch (error) {
        return false; // If request fails, DOI does not exist
    }
}

// Function to check if ISBN exists using the Open Library API
// async function checkIsbnExists(isbn: string): Promise<boolean> {
//     try {
//         const response = await axios.get(`https://isbnsearch.org/isbn/${isbn}`);
//         return response.status === 200 && response.data.includes('ISBN'); // Check if response contains ISBN information
//     } catch (error) {
//         return false; // If request fails, ISBN does not exist
//     }
// }

async function checkIsbnExists(isbn: string): Promise<boolean> {
    // Try Open Library first
    try {
        const response = await axios.get(`https://openlibrary.org/api/books?bibkeys=ISBN:${isbn}&format=json&jscmd=data`);
        const data = response.data;
        if (Object.keys(data).length > 0) {
            return true;
        }
    } catch (error) {
        // If an error occurs, continue to the next API
    }

    // Fallback to Google Books API
    try {
        const response = await axios.get(`https://www.googleapis.com/books/v1/volumes?q=isbn:${isbn}`);
        const data = response.data;
        if (data.totalItems > 0) {
            return true;
        }
    } catch (error) {
        // If an error occurs, continue to the next API
    }

    // Lastly use the Koha library by scraping HTML
    try {
        const response = await axios.get(`https://bibliothek.geosphere.at/cgi-bin/koha/opac-search.pl?idx=nb&q=${isbn}`);
        const html = response.data;
        // Check if zero results are explicitly indicated (German or English)
        if (html.includes('Keine Treffer gefunden!') || html.includes('Your search returned 0 results')) {
            return false;
        }
        // Try to extract the count from German message
        let match = html.match(/Ihre Suche erzielte\s*(\d+)\s*Treffer/);

        // If not found, try the English equivalent
        if (!match) {
            match = html.match(/Your search returned\s*(\d+)\s*results/);
        }

        if (match && match[1]) {
            const count = parseInt(match[1], 10);
            return count > 0;
        }

        // Fallback: if no match is found, return false
        return false;
    } catch (error) {
        return false;
    }
}

async function validateReference(value: unknown, options: Options, field: FieldContext) {
    if (typeof value !== 'string') {
        return;
    }

    const type = field.parent[options.typeField];

    if (type === ReferenceIdentifierTypes.URL) {
        if (!/^https?:\/\/[^\s$.?#].[^\s]*$/.test(value)) {
            field.report('The {{ field }} must be a valid URL', 'validateReference', field);
        } else {
            try {
                const exists = await checkDoiExists(value);
                if (!exists) {
                    field.report('The {{ field }} must be an existing URL', 'validateReference', field);
                }
            } catch (error) {
                field.report('Error checking URL existence: ' + error.message, 'validateReference', field);
            }
        }
    }
    // Check if the value does not match the DOI pattern
    // The regex pattern ^10.\d{4,9}\/[-._;()/:a-zA-Z0-9]+$ is designed to match valid DOI formats.
    // - ^10. ensures that the string starts with "10."
    // - \d{4,9} matches a sequence of 4 to 9 digits.
    // - \/ matches the literal forward slash character.
    // - [-._;()/:a-zA-Z0-9]+ matches one or more characters that can include uppercase and lowercase letters, digits, and a set of special characters.
    // The i flag at the end of the regex makes the matching case-insensitive, meaning it will match both uppercase and lowercase letters.
    // If the value does not match this pattern, the code inside the if block will execute.
    else if (type === ReferenceIdentifierTypes.DOI) {
        // Extract the DOI from the URL if it starts with 'https://doi.org/'
        const doiPattern = /^https:\/\/doi\.org\/(10.\d{4,9}\/[-._;()/:a-zA-Z0-9]+)$/i;
        const match = value.match(doiPattern);
        const doi = match ? match[1] : value;
        // Check if the extracted DOI or the value itself matches the DOI patter
        if (!/^10.\d{4,9}\/[-._;()/:a-zA-Z0-9]+$/i.test(doi)) {
            field.report('The {{ field }} must be a valid DOI', 'validateReference', field);
        } else {
            try {
                const exists = await checkDoiExists(value);
                if (!exists) {
                    field.report('The {{ field }} must be an existing DOI', 'validateReference', field);
                }
            } catch (error) {
                field.report('Error checking DOI existence: ' + error.message, 'validateReference', field);
            }
        }
    } else if (type === ReferenceIdentifierTypes.ISBN) {
        const isbnRegex = /^(?:\d{1,5}-\d{1,7}-\d{1,7}-[\dX]$|97[89]-\d{1,5}-\d{1,7}-\d{1,7}-\d)$/;
        if (!isbnRegex.test(value)) {
            field.report('Invalid {{ field }}. Expected format: 978-3-16-148410-0 or similar.', 'validateReference', field);
        } else {
            try {
                const exists = await checkIsbnExists(value);
                if (!exists) {
                    field.report('The {{ field }} must be an existing ISBN', 'validateReference', field);
                }
            } catch (error) {
                field.report('Error checking ISBN existence: ' + error.message, 'validateReference', field);
            }
        }
    } else if (type === ReferenceIdentifierTypes.Handle && !/^\d{2,}.\d{4,9}\/[-._;()/:a-zA-Z0-9]+$/.test(value)) {
        /// Extract the Handle from the URL if it contains '/handle/'
        field.report('The {{ field }} must be a valid Handle', 'validateReference', field);
    } else if (
        type === ReferenceIdentifierTypes.URN &&
        !/^urn:[a-zA-Z0-9][a-zA-Z0-9-]{0,31}:[a-zA-Z0-9()+,\-.:=@;$_!*'%/?#]+$/.test(value)
    ) {
        field.report('The {{ field }} must be a valid URN', 'validateReference', field);
    } else if (type === ReferenceIdentifierTypes.ISSN && !/^\d{4}-\d{3}[\dxX]$/.test(value)) {
        field.report('The {{ field }} must be a valid ISSN', 'validateReference', field);
    }
}

export const validateReferenceRule = vine.createRule(validateReference);

declare module '@vinejs/vine' {
    interface VineString {
        validateReference(options: Options): this;
    }
}

VineString.macro('validateReference', function (this: VineString, options: Options) {
    return this.use(validateReferenceRule(options));
});