explore-services/services/cache/mecache/cache.js

'use strict';
const axios = require('axios');
let express = require('express');
let app = express();
const { LRUCache } = require('lru-cache');
const request = require('superagent');
const prom = require('prom-client');
const URL = require('url');
let compression = require("compression");
const PropertiesReader = require('properties-reader');
const properties = PropertiesReader('./properties.file');
const environment = properties.get("environment");
const preloadRequests = properties.get("preloadRequests").split(',');;
var accesslog = require('access-log');

const cacheMaxSize = 1000;
let cors = require('cors');
app.use(cors());
app.use(compression());
const lruCache = new LRUCache({ max: cacheMaxSize });
const register = new prom.Registry();
prom.collectDefaultMetrics({register: register});

const responses = new prom.Counter({
    name: 'cache_http_responses_total',
    help: 'A counter for cache response codes for every API request.',
    labelNames: ['scheme', 'target', 'code'],
    registers: [register]
});

const entries = new prom.Gauge({
    name: 'cache_used_entries',
    help: 'A counter to count cache entries',
    registers: [register]
});

const histogram = new prom.Histogram({
    name: 'cache_http_request_duration_seconds',
    help: 'A Histogram for cache. Providing information about a cache request and load latency in seconds.',
    labelNames: ['scheme', 'target', 'cache'],
    registers: [register],
    buckets: [0.1, 0.2, 0.5, 1, 2]
});

let cache = () => {
    return (req, res, next) => {
        if (req.query.url) {
            let key = req.query.url;
            const url = new URL.parse(req.query.url);
            // console.log(req.headers.origin, req.headers.referrer, req.headers.origin)
            const cacheControlHeader = req.headers['cache-control'];
            // Log the Cache-Control header
            // console.log('Cache-Control header sent by client:', cacheControlHeader);
            let forceReload = req.query.forceReload && req.query.forceReload == 'true'?true:false;
            forceReload = forceReload || (cacheControlHeader && (cacheControlHeader.indexOf("no-cache") || cacheControlHeader.indexOf("no-store") || cacheControlHeader.indexOf("must-revalidate")))?true:false;

            const target = url.host + '/' + url.pathname.split('/')[1];
            const scheme = url.protocol.replace(':', '');
            if (lruCache.has(key) && !forceReload) {
               // console.log( key, "hit")
                const end = histogram.startTimer({scheme: scheme, target: target, cache: 'hit'});
                res.send(JSON.parse(lruCache.get(key)));
                responses.inc({scheme: scheme, target: target, code: res.statusCode});
                end();
            } else {
               // console.log( key, "miss", forceReload)
                    const end = histogram.startTimer({scheme: scheme, target: target, cache: 'miss'});
                    res.sendResponse = res.send;
                    res.send = (body) => {
                        if(isAllowedToBeCached(decodeURI(scheme), decodeURI(target))) {
                            let alreadyCached = lruCache.has(key);
                            entries.set(lruCache.size);
                            if (!alreadyCached) {
                                responses.inc({scheme: scheme, target: target, code: res.statusCode});
                                end();
                            }
                            if (res.statusCode === 200) {
                                lruCache.set(key, body);
                                entries.set(lruCache.size);
                            }
                            res.sendResponse(body);
                        }else{
                            res.statusCode = 405;
                            res.sendResponse(JSON.stringify(  {code: 405, message: "Method not Allowed"}));
                        }

                    };
                accesslogCustomFormat(req, res);
                next();
            }
        } else {
            accesslogCustomFormat(req, res);
            next();
        }
    };
};
function isAllowedToBeCached(scheme, target){
    if(environment != "development"){
        return scheme.indexOf("https")!=-1 && ( target.indexOf(".openaire.eu/") !=-1 || target.indexOf("zenodo.org/api") !=-1 || target.indexOf("lab.idiap.ch/enermaps" != -1))
    } else if(environment == "development"){
        return target.indexOf(".openaire.eu/") !=-1 || target.indexOf(".di.uoa.gr") !=-1 || target.indexOf("zenodo.org/api") !=-1 || target.indexOf("dev-openaire.d4science.org") !=-1 || target.indexOf("lab.idiap.ch/enermaps") != -1
    }
    return true;
}
app.get('/clear', (req, res) => {
    let c = lruCache.size;
    const url = req.query.url;
    let message = "";
    if (url) {
        let key = '__express__' + req.query.url;
        lruCache.delete(key);
        message = "Delete entry with key " + url;
        entries.set(lruCache.size);
    } else {
        clearCache();
        message = "Delete " + c + " entries. Now there are: " + lruCache.size
    }
    res.header("Access-Control-Allow-Headers", "Origin, Content-Type,  Content-Length");
    res.header("Access-Control-Allow-Methods", "GET,  OPTIONS");
    res.header("Access-Control-Allow-Methods", "GET,  OPTIONS");
    res.header("Content-Type", "application/json");
    accesslogCustomFormat(req, res);
    res.status(200).send(getResponse(200, message));

});
app.get('/metrics', (req, res) => {
    res.set('Content-Type', register.contentType);
    accesslogCustomFormat(req, res);
    res.end(register.metrics());
});

app.get('/info', (req, res) => {
    accesslogCustomFormat(req, res);
    res.status(200).send(getResponse(200, {size:lruCache.size, keys: Array.from(lruCache.keys())}));
});

app.get('/get', cache(), cors(), (req, res) => {
    setTimeout(() => {
        const url = (req.query) ? req.query.url : null;
        if (!url) {
            res.status(404).send(getResponse(404, "Not Found ")) //not found
        } else {
            request.get(url, function (err, response) {
                res.header("Access-Control-Allow-Headers", "Origin, Content-Type,  Content-Length");
                res.header("Access-Control-Allow-Methods", "GET,  OPTIONS");
                res.header("Access-Control-Allow-Methods", "GET,  OPTIONS");
                res.header("Content-Type", "application/json");
                if (!response && err) {
                    res.status(500).send(getResponse(500, "An error occurred for " + url))
                } else {
                    res.status(response.status).send(response.body);
                }
            })
        }
    })
});


app.use((req, res) => {
    res.status(404).send(getResponse(404, "Not Found")); //not found
});

const server = app.listen(properties.get('port'), function () {
    console.log(`Example app listening on port`, server.address().port)
    //run the timer
    resetAtMidnight();
    initCache();
});

function getResponse(code, message) {
    var response = {};
    response["code"] = code;
    response["message"] = message;
    return response;
}

function clearCache() {
    console.log("cache is cleared!");
    lruCache.clear();
    entries.set(lruCache.size);
    initCache();
}

async function initCache() {
    try {
        const  requests = await axios.get(properties.get('utilsService') + '/grouped-requests');
        const additionalDataPromises = requests.data.map((url) => axios.get('http://localhost:'+properties.get('port') + '/get?url=' + properties.get('utilsService') + url));
        const additionalDataResponses = await Promise.all(additionalDataPromises);
        console.log("Cache initialized group queries!")
    } catch (error) {
        console.error('Error fetching data: Cache initialize failed', error);
    }
    try{
        const additionalDataPromisesPreloadRequests = preloadRequests.map((url) => axios.get('http://localhost:'+properties.get('port') + '/get?url=' + url));
        const additionalDataResponsesPreloadRequests = await Promise.all(additionalDataPromisesPreloadRequests);
        console.log("Cache initialized preload requests!")
    } catch (error) {
        console.error('Error fetching data: Cache initialize failed', error);
    }
}


function resetAtMidnight() {
    console.log("Run Reset timer");
    var now = new Date();
    var night = new Date(
        now.getFullYear(),
        now.getMonth(),
        now.getDate() + 1, //  the next day, ...
        0, 0, 0 // ...at 00:00:00 hours
    );
    var msToMidnight = night.getTime() - now.getTime();
    setTimeout(function () {
        clearCache();              //      <-- This is the function being called at midnight.
        resetAtMidnight();    //      Then, reset again next midnight.
    }, msToMidnight);
}


function accesslogCustomFormat(req, res){
    accesslog(req, res,
    /*{
        userID: function (req) { return req.user; },
        format : 'url=":url" method=":method" statusCode=":statusCode" delta=":delta" ip=":ip"'
    }*/);
//    url="/clear" method="GET" statusCode="200" delta="2" ip="::ffff:195.134.66.178"

    // ::ffff:195.134.66.178 - - [15/Jul/2024:11:24:35 +0300] "GET /clear HTTP/1.1" 200 59 "-" "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/113.0"

}