tethys.backend/app/Controllers/Http/Oai/OaiController.ts

import type { HttpContextContract } from '@ioc:Adonis/Core/HttpContext';
import { RequestContract } from '@ioc:Adonis/Core/Request';
import { XMLBuilder } from 'xmlbuilder2/lib/interfaces';
import { create } from 'xmlbuilder2';
import dayjs, { Dayjs } from 'dayjs';
import utc from 'dayjs/plugin/utc';
import timezone from 'dayjs/plugin/timezone';
import { readFileSync } from 'fs';
import { StatusCodes } from 'http-status-codes';
import { transform } from 'saxon-js';
// import { Xslt, xmlParse } from 'xslt-processor'
import { OaiErrorCodes, OaiModelError } from 'App/Exceptions/OaiErrorCodes';
import { OaiModelException, BadOaiModelException } from 'App/Exceptions/OaiModelException';
import Dataset from 'App/Models/Dataset';
import Collection from 'App/Models/Collection';
import { getDomain } from 'App/Utils/utility-functions';
import XmlModel from 'App/Library/XmlModel';

interface XslTParameter {
    [key: string]: any;
}

interface Dictionary {
    [index: string]: string;
}

export default class OaiController {
    private deliveringDocumentStates = ['published', 'deleted'];
    // private sampleRegEx = /^[A-Za-zäüÄÜß0-9\-_.!~]+$/;
    private xsltParameter: XslTParameter;
    // private configuration: Configuration;
    // private tokenWorker: TokenWorker;

    /**
     * Holds xml representation of document information to be processed.
     *
     * @var xmlbuilder.XMLDocument | null  Defaults to null.
     */
    private xml: XMLBuilder;
    private proc;

    constructor() {
        // Load the XSLT file
        this.proc = readFileSync('public/assets2/datasetxml2oai.sef.json');
        // tests
        // const xslPath = 'assets/datasetxml2oai-pmh.xslt'; // Replace with the actual path to your XSLT file
        // this.proc = readFileSync(xslPath, 'utf-8');
        // this.configuration = new Configuration();
        dayjs.extend(utc);
        dayjs.extend(timezone);
    }

    public async index({ response, request }: HttpContextContract): Promise<void> {
        this.xml = create({ version: '1.0', encoding: 'UTF-8', standalone: true }, '<root></root>');

        // this.proc = new XSLTProcessor();
        // const stylesheet = readFileSync(__dirname + "/datasetxml2oai.sef.json");
        const xsltParameter = (this.xsltParameter = {});

        let oaiRequest: Dictionary = {};
        if (request.method() === 'POST') {
            oaiRequest = request.body();
        } else if (request.method() === 'GET') {
            oaiRequest = request.qs();
        } else {
            xsltParameter['oai_error_code'] = 'unknown';
            xsltParameter['oai_error_message'] = 'Only POST and GET methods are allowed for OAI-PMH.';
        }
        // const oaiRequest: OaiParameter = request.body;
        try {
            await this.handleRequest(oaiRequest, request);
        } catch (error) {
            if (error instanceof OaiModelException) {
                const code = error.oaiCode;
                let oaiErrorCode: string | undefined = 'Unknown oai error code ' + code;
                if (OaiModelError.has(error.oaiCode) && OaiModelError.get(code) !== undefined) {
                    oaiErrorCode = OaiModelError.get(error.oaiCode);
                }
                this.xsltParameter['oai_error_code'] = oaiErrorCode;
                this.xsltParameter['oai_error_message'] = error.message;
            } else {
                //     // return next(error); // passing to default express middleware error handler
                this.xsltParameter['oai_error_code'] = 'unknown';
                this.xsltParameter['oai_error_message'] = 'An internal error occured.';
            }
        }

        const xmlString = this.xml.end({ prettyPrint: true });

        let xmlOutput;
        try {
            const result = await transform({
                // stylesheetFileName: `${config.TMP_BASE_DIR}/data-quality/rules/iati.sef.json`,
                stylesheetText: this.proc,
                destination: 'serialized',
                // sourceFileName: sourceFile,
                sourceText: xmlString,
                stylesheetParams: xsltParameter,
                // logLevel: 10,
            });
            xmlOutput = result.principalResult;
        } catch (error) {
            return response.status(500).json({
                message: 'An error occurred while creating the user',
                error: error.message,
            });
        }

        response
            .header('Content-Type', 'application/xml; charset=utf-8')
            .header('Access-Control-Allow-Origin', '*')
            .header('Access-Control-Allow-Methods', 'GET,POST');
        response.status(StatusCodes.OK).send(xmlOutput);
    }

    protected async handleRequest(oaiRequest: Dictionary, request: RequestContract) {
        // Setup stylesheet
        // $this->loadStyleSheet('datasetxml2oai-pmh.xslt');

        // Set response time
        const now: Dayjs = dayjs();
        this.xsltParameter['responseDate'] = now.format('YYYY-MM-DDTHH:mm:ss[Z]');
        this.xsltParameter['unixTimestamp'] = now.unix();

        // set OAI base url
        const baseDomain = process.env.BASE_DOMAIN || 'localhost';
        this.xsltParameter['baseURL'] = baseDomain + '/oai';
        this.xsltParameter['repURL'] = request.protocol() + '://' + request.hostname();
        this.xsltParameter['downloadLink'] = request.protocol() + '://' + request.hostname() + '/file/download/';
        this.xsltParameter['doiLink'] = 'https://doi.org/';
        this.xsltParameter['doiPrefix'] = 'info:eu-repo/semantics/altIdentifier/doi/';

        if (oaiRequest['verb']) {
            const verb = oaiRequest['verb'];
            this.xsltParameter['oai_verb'] = verb;
            if (verb === 'Identify') {
                this.handleIdentify();
            } else if (verb === 'ListMetadataFormats') {
                this.handleListMetadataFormats();
            } else if (verb == 'GetRecord') {
                await this.handleGetRecord(oaiRequest);
            }
            // else if (verb == "ListRecords") {
            //     await this.handleListRecords(oaiRequest);
            // } else if (verb == "ListIdentifiers") {
            //     await this.handleListIdentifiers(oaiRequest);
            // }
            else if (verb == 'ListSets') {
                await this.handleListSets();
            } else {
                this.handleIllegalVerb();
            }
        } else {
            // // try {
            // //     console.log("Async code example.")
            // const err = new PageNotFoundException("verb not found");
            // throw err;
            // // } catch (error) { // manually catching
            // //     next(error); // passing to default middleware error handler
            // // }
            throw new OaiModelException(
                StatusCodes.INTERNAL_SERVER_ERROR,
                'The verb provided in the request is illegal.',
                OaiErrorCodes.BADVERB,
            );
        }
    }

    protected handleIdentify() {
        const email = process.env.OAI_EMAIL || 'repository@geosphere.at';
        const repositoryName = 'Tethys RDR';
        const repIdentifier = 'tethys.at';
        const sampleIdentifier = 'oai:' + repIdentifier + ':1'; //$this->_configuration->getSampleIdentifier();

        // Dataset::earliestPublicationDate()->server_date_published->format('Y-m-d\TH:i:s\Z') : null;
        // earliestDateFromDb!= null && (this.xsltParameter['earliestDatestamp'] = earliestDateFromDb?.server_date_published);

        // set parameters for oai-pmh.xslt
        this.xsltParameter['email'] = email;
        this.xsltParameter['repositoryName'] = repositoryName;
        this.xsltParameter['repIdentifier'] = repIdentifier;
        this.xsltParameter['sampleIdentifier'] = sampleIdentifier;
        // $this->proc->setParameter('', 'earliestDatestamp', $earliestDateFromDb);

        this.xml.root().ele('Datasets');
    }

    protected handleListMetadataFormats() {
        this.xml.root().ele('Datasets');
    }

    protected async handleListSets() {
        const repIdentifier = 'tethys.at';
        this.xsltParameter['repIdentifier'] = repIdentifier;
        const datasetElement = this.xml.root().ele('Datasets');

        const sets: { [key: string]: string } = {
            'open_access': 'Set for open access licenses',
            'doc-type:ResearchData': 'Set for document type ResearchData',
            // ...(await this.getSetsForDatasetTypes()),
            ...(await this.getSetsForCollections()),
            // ... await this.getSetsForProjects(),
        } as Dictionary;

        for (const [key, value] of Object.entries(sets)) {
            const setElement = datasetElement.ele('Rdr_Sets');
            setElement.att('Type', key);
            setElement.att('TypeName', value);
        }
    }

    protected async handleGetRecord(oaiRequest: Dictionary) {
        const repIdentifier = 'tethys.at';
        this.xsltParameter['repIdentifier'] = repIdentifier;

        const dataId = this.validateAndGetIdentifier(oaiRequest);
        const dataset = await Dataset.query().where('publish_id', dataId).preload('xmlCache').preload('collections').first();

        if (!dataset || !dataset.publish_id) {
            throw new OaiModelException(
                StatusCodes.INTERNAL_SERVER_ERROR,
                'The value of the identifier argument is unknown or illegal in this repository.',
                OaiErrorCodes.IDDOESNOTEXIST,
            );
        }

        const metadataPrefix = this.validateAndGetMetadataPrefix(oaiRequest);
        this.xsltParameter['oai_metadataPrefix'] = metadataPrefix;
        // do not deliver datasets which are restricted by document state defined in deliveringStates
        this.validateDatasetState(dataset);

        // add xml elements
        const datasetNode = this.xml.root().ele('Datasets');
        await this.createXmlRecord(dataset, datasetNode);
    }

    private validateAndGetIdentifier(oaiRequest: Dictionary): number {
        // Identifier references metadata Urn, not plain Id!
        // Currently implemented as 'oai:foo.bar.de:{docId}' or 'urn:nbn...-123'
        if (!('identifier' in oaiRequest)) {
            throw new BadOaiModelException('The prefix of the identifier argument is unknown.');
        }
        const dataId = Number(this.getDocumentIdByIdentifier(oaiRequest.identifier));
        if (isNaN(dataId)) {
            throw new OaiModelException(
                StatusCodes.INTERNAL_SERVER_ERROR,
                'The value of the identifier argument is illegal in this repository.',
                OaiErrorCodes.BADARGUMENT,
            );
        }
        return dataId;
    }

    private validateAndGetMetadataPrefix(oaiRequest: Dictionary): string {
        let metadataPrefix = '';
        if ('metadataPrefix' in oaiRequest) {
            metadataPrefix = oaiRequest['metadataPrefix'];
        } else {
            throw new OaiModelException(
                StatusCodes.INTERNAL_SERVER_ERROR,
                'The prefix of the metadata argument is unknown.',
                OaiErrorCodes.BADARGUMENT,
            );
        }
        return metadataPrefix;
    }

    private validateDatasetState(dataset: Dataset): void {
        if (dataset.server_state == null || !this.deliveringDocumentStates.includes(dataset.server_state)) {
            throw new OaiModelException(
                StatusCodes.INTERNAL_SERVER_ERROR,
                'Document is not available for OAI export!',
                OaiErrorCodes.NORECORDSMATCH,
            );
        }
    }

    private async createXmlRecord(dataset: Dataset, datasetNode: XMLBuilder) {
        const domNode = await this.getDatasetXmlDomNode(dataset);

        if (domNode) {
            // add frontdoor url and data-type
            dataset.publish_id && this.addLandingPageAttribute(domNode, dataset.publish_id.toString());
            this.addSpecInformation(domNode, 'data-type:' + dataset.type);

            // if (dataset.collections) {
            //     for (const coll of dataset.collections) {
            //         const collRole = await coll.getCollectionRole();
            //         this.addSpecInformation(domNode, collRole.oai_name + ':' + coll.number);
            //     }
            // }

            datasetNode.import(domNode);
        }
    }

    private async getDatasetXmlDomNode(dataset: Dataset) {
        const xmlModel = new XmlModel(dataset);
        // xmlModel.setModel(dataset);
        xmlModel.excludeEmptyFields();
        xmlModel.caching = true;
        // const cache = dataset.xmlCache ? dataset.xmlCache : null;
        // dataset.load('xmlCache');
        if (dataset.xmlCache) {
            xmlModel.xmlCache = dataset.xmlCache;
        }

        // return cache.getDomDocument();
        const domDocument: XMLBuilder | null = await xmlModel.getDomDocument();
        return domDocument;
    }

    private addSpecInformation(domNode: XMLBuilder, information: string) {
        domNode.ele('SetSpec').att('Value', information);
    }

    private addLandingPageAttribute(domNode: XMLBuilder, dataid: string) {
        const baseDomain = process.env.BASE_DOMAIN || 'localhost';
        const url = 'https://' + getDomain(baseDomain) + '/dataset/' + dataid;
        // add attribute du dataset xml element
        domNode.att('landingpage', url);
    }

    private getDocumentIdByIdentifier(oaiIdentifier: string): string {
        const identifierParts: string[] = oaiIdentifier.split(':'); // explode(":", $oaiIdentifier);
        const dataId: string = identifierParts[2];
        // switch (identifierParts[0]) {
        //     case 'oai':
        //         if (isset($identifierParts[2])) {
        //             $dataId = $identifierParts[2];
        //         }
        //         break;
        //     default:
        //         throw new OaiModelException(
        //             'The prefix of the identifier argument is unknown.',
        //             OaiModelError::BADARGUMENT
        //         );
        //         break;
        // }

        // if (empty($dataId) or !preg_match('/^\d+$/', $dataId)) {
        //     throw new OaiModelException(
        //         'The value of the identifier argument is unknown or illegal in this repository.',
        //         OaiModelError::IDDOESNOTEXIST
        //     );

        return dataId;
    }

    private async getSetsForCollections(): Promise<Dictionary> {
        const sets: { [key: string]: string } = {} as Dictionary;

        const collections = await Collection.query()
            .select('name', 'number', 'role_id')
            .whereHas('collectionRole', (query) => {
                query.where('visible_oai', true);
            })
            .preload('collectionRole');

        collections.forEach((collection) => {
            // if collection has a collection role (classification like ddc):
            if (collection.number) {
                // collection.load('collectionRole');
                const setSpec = collection.collectionRole?.oai_name + ':' + collection.number;
                sets[setSpec] = `Set ${collection.number} '${collection.name}'`;
            }
        });
        return sets;
    }

    // private async getSetsForDatasetTypes(): Promise<IDictionary> {
    //     const sets: { [key: string]: string } = {} as IDictionary;

    //     const datasets: Array<Dataset> = await Dataset.findAll({
    //         attributes: ["type"],
    //         where: { server_state: { [Sequelize.Op.eq]: "published" } },
    //     });
    //     datasets.forEach((dataset) => {
    //         if (dataset.type && false == preg_match(this.sampleRegEx, dataset.type)) {
    //             const msg = `Invalid SetSpec (data-type='${dataset.type}').
    //             Allowed characters are [${this.sampleRegEx}].`;
    //             Logger.err(`OAI: ${msg}`);
    //             // Log::error("OAI-PMH: $msg");
    //             return;
    //         }
    //         const setSpec = "data-type:" + dataset.type;
    //         sets[setSpec] = `Set for document type '${dataset.type}'`;
    //     });
    //     return sets;
    // }

    private handleIllegalVerb() {
        this.xsltParameter['oai_error_code'] = 'badVerb';
        this.xsltParameter['oai_error_message'] = 'The verb provided in the request is illegal.';
    }
}