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

645 lines
27 KiB
TypeScript
Raw Permalink Normal View History

2024-03-14 19:25:27 +00:00
import type { HttpContext } from '@adonisjs/core/http';
// import { RequestContract } from '@ioc:Adonis/Core/Request';
import { Request } from '@adonisjs/core/http';
import { XMLBuilder } from 'xmlbuilder2/lib/interfaces.js';
import { create } from 'xmlbuilder2';
import dayjs, { Dayjs } from 'dayjs';
2024-03-14 19:25:27 +00:00
import utc from 'dayjs/plugin/utc.js';
import timezone from 'dayjs/plugin/timezone.js';
import { readFileSync } from 'fs';
import { StatusCodes } from 'http-status-codes';
2024-03-14 19:25:27 +00:00
import SaxonJS 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 '#models/dataset';
import Collection from '#models/collection';
import { getDomain, preg_match } from '#app/utils/utility-functions';
2024-03-14 19:25:27 +00:00
import XmlModel from '#app/Library/XmlModel';
import logger from '@adonisjs/core/services/logger';
import ResumptionToken from '#app/Library/Oai/ResumptionToken';
// import Config from '@ioc:Adonis/Core/Config';
import config from '@adonisjs/core/services/config'
// import { inject } from '@adonisjs/fold';
import { inject } from '@adonisjs/core'
// import { TokenWorkerContract } from "MyApp/Models/TokenWorker";
2024-03-14 19:25:27 +00:00
import TokenWorkerContract from '#library/Oai/TokenWorkerContract';
import { ModelQueryBuilderContract } from '@adonisjs/lucid/types/model';
interface XslTParameter {
[key: string]: any;
}
interface Dictionary {
[index: string]: string;
}
interface ListParameter {
cursor: number;
totalIds: number;
start: number;
reldocIds: (number | null)[];
metadataPrefix: string;
}
2024-03-14 19:25:27 +00:00
@inject()
export default class OaiController {
private deliveringDocumentStates = ['published', 'deleted'];
private sampleRegEx = /^[A-Za-zäüÄÜß0-9\-_.!~]+$/;
private xsltParameter: XslTParameter;
/**
* Holds xml representation of document information to be processed.
*
* @var xmlbuilder.XMLDocument | null Defaults to null.
*/
private xml: XMLBuilder;
private proc;
2024-03-14 19:25:27 +00:00
constructor(public tokenWorker: TokenWorkerContract) {
// Load the XSLT file
this.proc = readFileSync('public/assets2/datasetxml2oai.sef.json');
dayjs.extend(utc);
dayjs.extend(timezone);
}
2024-03-14 19:25:27 +00:00
public async index({ response, request }: HttpContext): 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");
2024-03-14 19:25:27 +00:00
const xsltParameter: 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.';
}
let earliestDateFromDb;
// const oaiRequest: OaiParameter = request.body;
try {
const firstPublishedDataset: Dataset | null = await Dataset.earliestPublicationDate();
firstPublishedDataset != null &&
(earliestDateFromDb = firstPublishedDataset.server_date_published.toFormat("yyyy-MM-dd'T'HH:mm:ss'Z'"));
this.xsltParameter['earliestDatestamp'] = earliestDateFromDb;
// start the request
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; // = xmlString;
try {
2024-03-14 19:25:27 +00:00
const result = await SaxonJS.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);
}
2024-03-14 19:25:27 +00:00
protected async handleRequest(oaiRequest: Dictionary, request: Request) {
// Setup stylesheet
// $this->loadStyleSheet('datasetxml2oai-pmh.xslt');
// Set response time
const now: Dayjs = dayjs();
this.xsltParameter['responseDate'] = now.format('YYYY-MM-DDTHH:mm:ss[Z]');
this.xsltParameter['unixTimestamp'] = now.unix();
// set OAI base url
const baseDomain = process.env.OAI_BASE_DOMAIN || 'localhost';
this.xsltParameter['baseURL'] = baseDomain + '/oai';
this.xsltParameter['repURL'] = request.protocol() + '://' + request.hostname();
this.xsltParameter['downloadLink'] = request.protocol() + '://' + request.hostname() + '/file/download/';
this.xsltParameter['doiLink'] = 'https://doi.org/';
this.xsltParameter['doiPrefix'] = 'info:eu-repo/semantics/altIdentifier/doi/';
if (oaiRequest['verb']) {
const verb = oaiRequest['verb'];
this.xsltParameter['oai_verb'] = verb;
if (verb === 'Identify') {
this.handleIdentify();
} else if (verb === 'ListMetadataFormats') {
this.handleListMetadataFormats();
} else if (verb == 'GetRecord') {
await this.handleGetRecord(oaiRequest);
} else if (verb == 'ListRecords') {
await this.handleListRecords(oaiRequest);
} else if (verb == 'ListIdentifiers') {
await this.handleListIdentifiers(oaiRequest);
} else if (verb == 'ListSets') {
await this.handleListSets();
} else {
this.handleIllegalVerb();
}
} else {
// // try {
// // console.log("Async code example.")
// const err = new PageNotFoundException("verb not found");
// throw err;
// // } catch (error) { // manually catching
// // next(error); // passing to default middleware error handler
// // }
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'The verb provided in the request is illegal.',
OaiErrorCodes.BADVERB,
);
}
}
protected handleIdentify() {
const email = process.env.OAI_EMAIL || 'repository@geosphere.at';
const repositoryName = 'Tethys RDR';
const repIdentifier = 'tethys.at';
const sampleIdentifier = 'oai:' + repIdentifier + ':1'; //$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',
'openaire_data': "OpenAIRE",
'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', (builder) => {
builder.preload('collectionRole');
})
.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);
}
protected async handleListIdentifiers(oaiRequest: Dictionary) {
!this.tokenWorker.isConnected && (await this.tokenWorker.connect());
2024-03-14 19:25:27 +00:00
const maxIdentifier: number = config.get('oai.max.listidentifiers', 100);
await this.handleLists(oaiRequest, maxIdentifier);
}
2024-03-14 19:25:27 +00:00
protected async handleListRecords(oaiRequest: Dictionary) {
!this.tokenWorker.isConnected && (await this.tokenWorker.connect());
2024-03-14 19:25:27 +00:00
const maxRecords: number = config.get('oai.max.listrecords', 100);
await this.handleLists(oaiRequest, maxRecords);
}
private async handleLists(oaiRequest: Dictionary, maxRecords: number) {
maxRecords = maxRecords || 100;
const repIdentifier = 'tethys.at';
this.xsltParameter['repIdentifier'] = repIdentifier;
const datasetNode = this.xml.root().ele('Datasets');
// list initialisation
const numWrapper: ListParameter = {
cursor: 0,
totalIds: 0,
start: maxRecords + 1,
reldocIds: [],
metadataPrefix: '',
};
// resumptionToken is defined
if ('resumptionToken' in oaiRequest) {
await this.handleResumptionToken(oaiRequest, maxRecords, numWrapper);
} else {
// no resumptionToken is given
await this.handleNoResumptionToken(oaiRequest, numWrapper);
}
// handling of document ids
const restIds = numWrapper.reldocIds as number[];
const workIds = restIds.splice(0, maxRecords) as number[]; // array_splice(restIds, 0, maxRecords);
// no records returned
if (workIds.length == 0) {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'The combination of the given values results in an empty list.',
OaiErrorCodes.NORECORDSMATCH,
);
}
const datasets: Dataset[] = await Dataset.query()
.whereIn('publish_id', workIds)
.preload('xmlCache')
.preload('collections', (builder) => {
builder.preload('collectionRole');
})
.orderBy('publish_id');
for (const dataset of datasets) {
await this.createXmlRecord(dataset, datasetNode);
}
// store the further Ids in a resumption-file
const countRestIds = restIds.length; //84
if (countRestIds > 0) {
const token = new ResumptionToken();
token.startPosition = numWrapper.start; //101
token.totalIds = numWrapper.totalIds; //184
token.documentIds = restIds; //101 -184
token.metadataPrefix = numWrapper.metadataPrefix;
// $tokenWorker->storeResumptionToken($token);
const res: string = await this.tokenWorker.set(token);
// set parameters for the resumptionToken-node
// const res = token.ResumptionId;
this.setParamResumption(res, numWrapper.cursor, numWrapper.totalIds);
}
}
2024-03-14 19:25:27 +00:00
private async handleResumptionToken(oaiRequest: Dictionary, maxRecords: number, numWrapper: ListParameter) {
const resParam = oaiRequest['resumptionToken']; //e.g. "158886496600000"
const token = await this.tokenWorker.get(resParam);
if (!token) {
throw new OaiModelException(StatusCodes.INTERNAL_SERVER_ERROR, 'cache is outdated.', OaiErrorCodes.BADRESUMPTIONTOKEN);
}
numWrapper.cursor = token.startPosition - 1; //startet dann bei Index 10
numWrapper.start = token.startPosition + maxRecords;
numWrapper.totalIds = token.totalIds;
numWrapper.reldocIds = token.documentIds;
numWrapper.metadataPrefix = token.metadataPrefix;
this.xsltParameter['oai_metadataPrefix'] = numWrapper.metadataPrefix;
}
2024-03-14 19:25:27 +00:00
private async handleNoResumptionToken(oaiRequest: Dictionary, numWrapper: ListParameter) {
// no resumptionToken is given
if ('metadataPrefix' in oaiRequest) {
numWrapper.metadataPrefix = oaiRequest['metadataPrefix'];
} else {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'The prefix of the metadata argument is unknown.',
OaiErrorCodes.BADARGUMENT,
);
}
this.xsltParameter['oai_metadataPrefix'] = numWrapper.metadataPrefix;
let finder: ModelQueryBuilderContract<typeof Dataset, Dataset> = Dataset.query();
// add server state restrictions
finder.whereIn('server_state', this.deliveringDocumentStates);
if ('set' in oaiRequest) {
const set = oaiRequest['set'] as string;
const setArray = set.split(':');
if (setArray[0] == 'data-type') {
if (setArray.length == 2 && setArray[1]) {
finder.where('type', setArray[1]);
}
} else if (setArray[0] == 'open_access') {
const openAccessLicences = ['CC-BY-4.0', 'CC-BY-SA-4.0'];
finder.andWhereHas('licenses', (query) => {
query.whereIn('name', openAccessLicences);
});
} else if (setArray[0] == 'ddc') {
if (setArray.length == 2 && setArray[1] != '') {
finder.andWhereHas('collections', (query) => {
query.where('number', setArray[1]);
});
}
}
}
// const timeZone = "Europe/Vienna"; // Canonical time zone name
// &from=2020-09-03&until2020-09-03
// &from=2020-09-11&until=2021-05-11
if ('from' in oaiRequest && 'until' in oaiRequest) {
const from = oaiRequest['from'] as string;
let fromDate = dayjs(from); //.tz(timeZone);
const until = oaiRequest['until'] as string;
let untilDate = dayjs(until); //.tz(timeZone);
if (!fromDate.isValid() || !untilDate.isValid()) {
throw new OaiModelException(StatusCodes.INTERNAL_SERVER_ERROR, 'Date Parameter is not valid.', OaiErrorCodes.BADARGUMENT);
}
fromDate = dayjs.tz(from, 'Europe/Vienna');
untilDate = dayjs.tz(until, 'Europe/Vienna');
if (from.length != until.length) {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'The request has different granularities for the from and until parameters.',
OaiErrorCodes.BADARGUMENT,
);
}
fromDate.hour() == 0 && (fromDate = fromDate.startOf('day'));
untilDate.hour() == 0 && (untilDate = untilDate.endOf('day'));
finder.whereBetween('server_date_published', [fromDate.format('YYYY-MM-DD HH:mm:ss'), untilDate.format('YYYY-MM-DD HH:mm:ss')]);
} else if ('from' in oaiRequest && !('until' in oaiRequest)) {
const from = oaiRequest['from'] as string;
let fromDate = dayjs(from);
if (!fromDate.isValid()) {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'From date parameter is not valid.',
OaiErrorCodes.BADARGUMENT,
);
}
fromDate = dayjs.tz(from, 'Europe/Vienna');
fromDate.hour() == 0 && (fromDate = fromDate.startOf('day'));
const now = dayjs();
if (fromDate.isAfter(now)) {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'Given from date is greater than now. The given values results in an empty list.',
OaiErrorCodes.NORECORDSMATCH,
);
} else {
finder.andWhere('server_date_published', '>=', fromDate.format('YYYY-MM-DD HH:mm:ss'));
}
} else if (!('from' in oaiRequest) && 'until' in oaiRequest) {
const until = oaiRequest['until'] as string;
let untilDate = dayjs(until);
if (!untilDate.isValid()) {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
'Until date parameter is not valid.',
OaiErrorCodes.BADARGUMENT,
);
}
untilDate = dayjs.tz(until, 'Europe/Vienna');
untilDate.hour() == 0 && (untilDate = untilDate.endOf('day'));
const firstPublishedDataset: Dataset = (await Dataset.earliestPublicationDate()) as Dataset;
const earliestPublicationDate = dayjs(firstPublishedDataset.server_date_published.toISO()); //format("YYYY-MM-DDThh:mm:ss[Z]"));
if (earliestPublicationDate.isAfter(untilDate)) {
throw new OaiModelException(
StatusCodes.INTERNAL_SERVER_ERROR,
`earliestDatestamp is greater than given until date.
The given values results in an empty list.`,
OaiErrorCodes.NORECORDSMATCH,
);
} else {
finder.andWhere('server_date_published', '<=', untilDate.format('YYYY-MM-DD HH:mm:ss'));
}
}
let reldocIdsDocs = await finder.select('publish_id').orderBy('publish_id');
numWrapper.reldocIds = reldocIdsDocs.map((dat) => dat.publish_id);
numWrapper.totalIds = numWrapper.reldocIds.length; //212
}
private setParamResumption(res: string, cursor: number, totalIds: number) {
const tomorrow = dayjs().add(1, 'day').format('YYYY-MM-DDThh:mm:ss[Z]');
this.xsltParameter['dateDelete'] = tomorrow;
this.xsltParameter['res'] = res;
this.xsltParameter['cursor'] = cursor;
this.xsltParameter['totalIds'] = totalIds;
}
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 = coll.collectionRole;
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.OAI_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<Dictionary> {
const sets: { [key: string]: string } = {} as Dictionary;
const datasets: Array<Dataset> = await Dataset.query().select('type').where('server_state', '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}].`;
// Log::error("OAI-PMH: $msg");
2024-03-14 19:25:27 +00:00
logger.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.';
}
}