/ecoindex-audit

Primary LanguageHTMLMIT LicenseMIT

Eco Index Audit

Actions Status [npm version](https://badge.fury.io/jcd ..s/eco-index-audit)

This tool is the CLI version of this website

Installation

If you just need to run an audit, you can use NPX

npx @cnumr/eco-index-audit --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=table

npx @cnumr/eco-index-audit --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=csv

npx @cnumr/eco-index-audit --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=json

npx @cnumr/eco-index-audit --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=sonar --sonarFilePath=index.html --outputPathDir=./reports

But you can also clone, install and run the project locally.

git clone https://github.com/cnumr/eco-index-audit
cd eco-index-audit
npm i

Usage/Examples

node ./src/cli.js --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=table
node ./src/cli.js --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=csv
node ./src/cli.js --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=json

CI Integration

You can execute this module directly from your CI. Here is an example for Github Actions :

name: Running Eco Index via a Github Action

on:
  workflow_dispatch:

jobs:
  build:
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v3
      - name: Use Node.js
        uses: actions/setup-node@v3
        with:
          node-version: 16
      - run: npx @cnumr/eco-index-audit --url=https://www.google.com/ --ecoIndex=90 --output=table

Integration with Sonar

Since the version 3.3.0 the CLI can generate a external sonar report you can add to the Sonar configuration (via the sonar.externalIssuesReportPaths option).

You need to define the path to one of your file managed by Sonar, in order to make the rule visible in Sonar Cloud.

npx @cnumr/eco-index-audit --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=sonar --sonarFilePath=index.html --outputPathDir=./reports

Integration with Cypress

For the moment, you can only use eco-index-audit inside a Cypress test suites running on a Chromium-based browser.

npx cypress run -b chrome

You are able to run this module during your Cypress test. The first step is to define a new task in the cypress.config.js file.

const { defineConfig } = require("cypress");
const {
  prepareAudit,
  checkEcoIndex,
} = require("eco-@cnumr/eco-index-audit/src/cypress");

module.exports = defineConfig({
  e2e: {
    setupNodeEvents(on) {
      on("before:browser:launch", (_browser, launchOptions) => {
        prepareAudit(launchOptions);
      });

      on("task", {
        checkEcoIndex: ({ url, options }) =>
          checkEcoIndex({
            url,
            options: {
              headless: false,
              globals: { data: "data" },
              beforeScript: (globals) => console.log(globals),
              afterScript: (globals) => console.log(globals),
              cookies: [{name:"mycookie", value:"mycookieValue", domain:"mydomain.com", httpOnly:false}]
            },
          }),
      });
    },
  },
});

If the headless parameter is set to false, the UI will opened with the Devtools enabled and will automatically stopped running everything after loading the page (using a debugger statement),

The globals object can be used if you need to share some data during the execution of the beforeScript and afterScript hooks. This object will be available as a parameter to these two hooks.

Inside a Cypress test, eco-index-audit will reuse the same Chromium-based browser used by Cypress.

And then use this task inside your test. Inside your test, you can check if the ecoIndex is below a threshold.

describe("Cypress test", () => {
  const url = "https://google.com";
  beforeEach(() => {
    cy.visit(url);
  });

  it("should have a good ecoindex", () => {
    const threshold = 50;
    cy.task("checkEcoIndex", { url })
      .its("ecoIndex", { timeout: 0 })
      .should("be.greaterThan", threshold);
  });
});

If you have interceptors inside your Cypress test, your eco-index will be better than expected. In order to be as closed as the reality, you can update the initial values of the metrics we used to. For example, if you intercept 4 HTTP requests of 1024 bytes

describe("Cypress test", () => {
  const url = "https://google.com";
  beforeEach(() => {
    cy.visit(url);
  });

  it("should have a good ecoindex", () => {
    const threshold = 50;
    cy.task("checkEcoIndex", {
      url,
      initialValues: {
        numberOfRequests: 4,
        sizeOfRequests: 4 * 1024,
      },
    })
      .its("ecoIndex", { timeout: 0 })
      .should("be.greaterThan", threshold);
  });
});

You can also define the outputPathDir option in order to save the result in a file. These properties are only use for the json and csv formats.

const path = require("path");

describe("Cypress test", () => {
  const url = "https://google.com";
  beforeEach(() => {
    cy.visit(url);
  });

  const outputPathDir = path.join(__dirname, "reports");

  it("should have a good ecoindex", () => {
    const threshold = 50;
    cy.task("checkEcoIndex", {
      url,
      options: {
        output: "json",
        outputPathDir,
      },
    })
      .its("ecoIndex", { timeout: 0 })
      .should("be.greaterThan", threshold);
  });
});

You can also add a timeout before closing the Puppeteer page in order to debug the page thank to the beforeClosingPageTimeout options. The value is in ms

describe("Cypress test", () => {
  const url = "https://google.com";
  beforeEach(() => {
    cy.visit(url);
  });

  it("should have a good ecoindex", () => {
    const threshold = 50;
    cy.task("checkEcoIndex", {
      url,
      options: {
        beforeClosingPageTimeout: 10000,
      },
    })
      .its("ecoIndex", { timeout: 0 })
      .should("be.greaterThan", threshold);
  });
});

Integration with Playwright

We can also run the eco-index-audit module inside a Playwright test. In this repository, we have a working demo. you just need to run the following command after cloning the source code.

npx playwright test

In order to be able to run this module inside a Playwright test suites, you need to add this configuration in your playwright.config.ts

{
  name: 'chromium',
  use: {
    ...devices['Desktop Chrome'],
    launchOptions: {
      args: ['--remote-debugging-port=9222'],
    }
  },

}

An then, in the configuration of eco-index-audit, you need to reuse the same 9222 HTTP port.

import { test, expect } from "@playwright/test";
const check = require("../src/main");

test("get started link", async ({ page, launchOptions }) => {
  const url = "https://playwright.dev/";

  await page.goto(url);

  // Click the get started link.
  await page.getByRole("link", { name: "Get started" }).click();

  // Expects the URL to contain intro.
  await expect(page).toHaveURL(/.*intro/);

  const result = await check(
    {
      url,
      remote_debugging_port: 9222,
    },
    true
  );

  expect(result.ecoIndex).toBe(100);
});

Configure via the eco-index-audit.js file

You can also configure this module via a eco-index-audit.js configuration file. This file should be located in the root folder of your project.

module.exports = {
  url: ["https://www.google.com/"],
  output: ["table"],
};

Thanks to this file, you can enable dynamic output in order to archive results into any external service. In the following snippet, we will sotre the result into Elasticsearch.

module.exports = {
  url: ["https://www.google.com/"],
  output: [
    "table",
    (results) => {
      const { Client } = require("@elastic/elasticsearch");
      const client = new Client({
        cloud: {
          id: "...",
        },
        auth: {
          username: "...",
          password: "...",
        },
      });
      client.index({
        index: "eco-index",
        document: results,
      });
    },
  ],
};

Environment Variables

You can add environment variables in order to configure eco-index-audit:

  • ECOINDEX_VERBOSE: if set to true, will log more information in the terminal.
  • ECOINDEX_DISPLAY_HTML: if set to true, will log the HTML of the audited page.
ECOINDEX_VERBOSE=true npx eco-index-audit --url=https://www.google.com/ --ecoIndex=50 --visits=2000 --output=table

Tests

Update snapshots with npm test -- -u

Tech Stack

Typescript, Puppeteer, Cypress

Resources

Authors

Disclaimer

The LCA values used by ecoindex_api to evaluate environmental impacts are not under free license - ©Frédéric Bordage Please also refer to the mentions provided in the code files for specifics on the IP regime.