@mindconnect/mindconnect-nodejs
Version:
NodeJS Library for MindSphere Connectivity - TypeScript SDK for MindSphere - MindSphere Command Line Interface - MindSphere Development Proxy
198 lines • 11.7 kB
JavaScript
;
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
return new (P || (P = Promise))(function (resolve, reject) {
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
step((generator = generator.apply(thisArg, _arguments || [])).next());
});
};
Object.defineProperty(exports, "__esModule", { value: true });
const console_1 = require("console");
const fs = require("fs");
const path = require("path");
const utils_1 = require("../../api/utils");
const command_utils_1 = require("./command-utils");
let color = (0, command_utils_1.getColor)("blue");
exports.default = (program) => {
program
.command("anomaly-detection")
.alias("ad")
.option("-m, --mode [template|train|detect]", "mode [template | train | detect]", "train")
.option("-o, --on [data|asset]", // NOTE: 29/04/2021 - batch are excluded for now
"on [data | asset]", // NOTE: 29/04/2021 - batch are excluded for now
"data")
.option("-d, --data <data>", "time series data file", "timeseries.mdsp.json")
.option("-e, --epsilon <epsilon>", "threshold distance")
.option("-s, --clustersize <clustersize>", "minimum cluster size")
.option("-a, --algorithm [EUCLIDEAN|MANHATTAN|CHEBYSHEV]", "distance measure algorithm [EUCLIDEAN | MANHATTAN | CHEBYSHEV]")
.option("-n, --modelname <modelname>", "human-friendly name of the model")
.option("-i, --modelid <modelid>", "mindsphere model id ")
.option("-i, --assetid <assetid>", "mindsphere asset id ")
.option("-n, --aspectname <aspectname>", "mindsphere aspect name")
.option("-f, --from <from>", "begining of the time range")
.option("-u, --to <to>", "end of the time range")
.option("-k, --passkey <passkey>", "passkey")
.option("-y, --retry <number>", "retry attempts before giving up", "3")
.option("-v, --verbose", "verbose output")
.description(color(`train anomaly detection models and detect timeseries anomalies *`))
.action((options) => {
(() => __awaiter(void 0, void 0, void 0, function* () {
try {
checkRequiredParameters(options);
const sdk = (0, command_utils_1.getSdk)(options);
color = (0, command_utils_1.adjustColor)(color, options, true);
(0, command_utils_1.homeDirLog)(options.verbose, color);
(0, command_utils_1.proxyLog)(options.verbose, color);
switch (options.mode) {
case "template":
yield createTemplate(options, sdk);
console.log("Edit the files before submitting them to mindsphere.");
break;
case "train":
yield trainNewModel(options, sdk);
break;
case "detect":
yield detectAnomalies(options, sdk);
break;
default:
throw Error(`no such option: ${options.mode}`);
}
}
catch (err) {
(0, command_utils_1.errorLog)(err, options.verbose);
}
}))();
})
.on("--help", () => printHelp());
};
function printHelp() {
(0, console_1.log)("\n Examples:\n");
(0, console_1.log)(` mc ad --mode template --data timeseries.data.mdsp.json \n \
creates a template for a time series data file`);
(0, console_1.log)(` mc ad --mode train --on data --data timeseries.data.mdsp.json --epsilon 0.5 \n
trains a model on the timeserie specified in the data file`);
(0, console_1.log)(` mc ad --mode detect --on data --data timeseries.data.mdsp.json --modelid <modelid>\n \
detects anomalities of the timeseries in the data file using the model with specified id`);
(0, console_1.log)(` mc ad --mode train --on asset --assetid <assetid> --aspectname Environment --epsilon 0.5\n\
trains a model on the time series of the aspect "Environment" of the asset with the id <assetid>`);
(0, console_1.log)(` mc ad --mode detect --on asset --modelid <modelid> --assetid <assetid> --aspectname Environment --epsilon 0.5\n\
detect anomalities of the timeseries on the specified asset and aspect with selected model`);
(0, command_utils_1.serviceCredentialLog)();
}
function checkRequiredParameters(options) {
options.mode === "template" &&
!options.data &&
(0, command_utils_1.errorLog)("you have to specify the output file of timeserie data", true);
options.mode === "detect" && !options.modelid && (0, command_utils_1.errorLog)("you have to specify the modelid", true);
options.mode === "train" && !options.epsilon && (0, command_utils_1.errorLog)("you have to specify the threshold distance", true);
options.on === "data" && !options.data && (0, command_utils_1.errorLog)("you have to specify the timeserie data file", true);
options.on === "asset" && !options.assetid && (0, command_utils_1.errorLog)("you have to specify the targeted assetId", true);
options.on === "asset" && !options.aspectname && (0, command_utils_1.errorLog)("you have to specify the targeted aspect name", true);
}
function createTemplate(options, sdk) {
return __awaiter(this, void 0, void 0, function* () {
const generatedData = (0, command_utils_1.generateTestData)(10, (x) => {
return 80 + Math.random() * 20 * Math.sin(x);
}, "Acceleration", "number");
const fileName = options.data || "timeseries.data.mdsp.json";
fs.writeFileSync(fileName, JSON.stringify(generatedData, null, 2));
console.log(`The time series data was written into ${color(fileName)}.\nRun \n\n\tmc ad --mode train --on data --data ${fileName} --epsilon 50.0 \n\nto create the model.\n`);
});
}
function trainNewModel(options, sdk) {
return __awaiter(this, void 0, void 0, function* () {
const anomalyDetectionClient = sdk.GetAnomalyDetectionClient();
const tenant = sdk.GetTenant();
const timeOffset = new Date().getTime();
const on = options.on || "data";
const epsilon = options.epsilon;
const clustersize = options.clustersize || 2;
const algorithm = options.algorithm || "EUCLIDEAN";
const modelname = options.modelname || `Generated_by_CLI_${tenant}_${timeOffset}`;
switch (on) {
case "data":
// read the data file content
const filePath = path.resolve(options.data);
!fs.existsSync(filePath) && (0, command_utils_1.errorLog)(`the metadata file ${filePath} doesn't exist!`, true);
const filecontent = fs.readFileSync(filePath);
const filedata = JSON.parse(filecontent.toString());
const result = (yield (0, utils_1.retry)(options.retry, () => __awaiter(this, void 0, void 0, function* () { return anomalyDetectionClient.PostModel(filedata, epsilon, clustersize, algorithm, modelname); })));
console.log(`Model with modelid ${color(result.id)} and name ${color(result.name)} was created.`);
break;
case "asset":
const assetid = options.assetid;
const aspectname = options.aspectname;
const now = new Date();
const lastMonth = new Date();
lastMonth.setDate(lastMonth.getDate() - 7);
const fromLastMonth = new Date(lastMonth.getUTCFullYear(), lastMonth.getUTCMonth(), lastMonth.getUTCDate());
const toNow = new Date(now.getUTCFullYear(), now.getUTCMonth(), now.getUTCDate());
let from = fromLastMonth;
try {
from = options.from ? new Date(options.from) : fromLastMonth;
}
catch (error) { }
let to = toNow;
try {
to = options.to ? new Date(options.to) : toNow;
}
catch (error) { }
const result_asset = (yield (0, utils_1.retry)(options.retry, () => __awaiter(this, void 0, void 0, function* () {
return anomalyDetectionClient.PostModelDirect(epsilon, clustersize, assetid, aspectname, from, to, algorithm, modelname);
})));
console.log(`Model with modelid ${color(result_asset.id)} and name ${color(result_asset.name)} was created.`);
default:
break;
}
});
}
function detectAnomalies(options, sdk) {
var _a, _b;
return __awaiter(this, void 0, void 0, function* () {
const anomalyDetectionClient = sdk.GetAnomalyDetectionClient();
const modelid = options.modelid;
const on = options.on || "data";
let result = [];
switch (on) {
case "data":
// read the data file content
const filePath = path.resolve(options.data);
!fs.existsSync(filePath) && (0, command_utils_1.errorLog)(`the metadata file ${filePath} doesn't exist!`, true);
const filecontent = fs.readFileSync(filePath);
const filedata = JSON.parse(filecontent.toString());
result = (yield (0, utils_1.retry)(options.retry, () => __awaiter(this, void 0, void 0, function* () { return anomalyDetectionClient.DetectAnomalies(filedata, modelid); })));
break;
case "asset":
const assetid = options.assetid;
const aspectname = options.aspectname;
const now = new Date();
const lastMonth = new Date();
lastMonth.setDate(lastMonth.getDate() - 7);
const fromLastMonth = new Date(lastMonth.getUTCFullYear(), lastMonth.getUTCMonth(), lastMonth.getUTCDate());
const toNow = new Date(now.getUTCFullYear(), now.getUTCMonth(), now.getUTCDate());
let from = fromLastMonth;
try {
from = options.from ? new Date(options.from) : fromLastMonth;
}
catch (error) { }
let to = toNow;
try {
to = options.to ? new Date(options.to) : toNow;
}
catch (error) { }
result = (yield (0, utils_1.retry)(options.retry, () => __awaiter(this, void 0, void 0, function* () { return anomalyDetectionClient.DetectAnomaliesDirect(modelid, assetid, aspectname, from, to); })));
default:
break;
}
console.log(`${color((_a = (result || [])) === null || _a === void 0 ? void 0 : _a.length)} anomalies found.\n`);
((_b = (result || [])) === null || _b === void 0 ? void 0 : _b.length) > 0 && printDetectedAnomalies(result, options);
});
}
function printDetectedAnomalies(anomalies, options) {
console.log("\nDetected anomalies:");
console.table(anomalies || [], ["_time", "anomalyExtent"]);
(0, command_utils_1.verboseLog)(JSON.stringify(anomalies, null, 2), options.verbose);
}
//# sourceMappingURL=anomaly-detection.js.map