Skip to content

The client library of the web analytics tool. It is in the top 5 by popularity worldwide.

License

Notifications You must be signed in to change notification settings

yandex/metrica-tag

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Metrica tag

This repository contains source code for Metrica tag. Once built the tag can be placed on a web page, where after initialization it starts gathering analytics information about the page usage.

The codebase contains most of the features of the tag delivered by Metrica with exception for some proprietary stuff. Modular code allows for easily inclusion/exclusion of any feature, thus providing a way to achieve required balance between amount of collected data (affecting analytics quality) and script size.

Documentation

For documentation about Metrica, please visit our website: JS API

Usage

Installation

The project is based on nodejs.

If nvm installed, you can run this command to install and use the required version of the nodejs.

nvm install

To install dependencies use the following command:

npm ci

Build

The project contains a single build target output into _build/public/watch.js file. To build it run:

npm run build

The code is build for web and by default is uglified in order to achieve smaller size. If needed (e.g. for debugging purpose) the script can be build without uglification:

npm run build:raw

Features

The code consists of basic code (always included into the build) and a collection of features. The features are listed in the features.json file. By default all features are included (except for those which are explicitly marked as disabled). In order to exclude a feature from the build add a "disabled": true attribute. By enabling/disabling certain features the build can be tweaked to include only the desired capabilities.

Development

Add a feature

All the features listed in features.json can be divided into two distinct types:

  • A feature adding a large amount of code with isolated functionality (e.g. provider, middleware, transport, etc.). This is the main mechanism for adding new functionality. Those features usually have defined "path" field in their description object.
  • A feature adding a small code block into an existing functions in order to enhance its behavior. Mostly used for small experiments and allows to add/exclude code, e.g. for a local testing build.

In order to add a new provider, do the following:

  1. Add a feature to the features.json file in the following format:
{
    "code": "SOME_NEW_FEATURE", // The name of generated variable to be used in code.
    "path": "someNewProvider", // The path to the folder within src/providers with the module that defined the new provider.
    "desc": "Any meaningful description goes here",
    "disabled": true, // Set only if you want to exclude a feature from build.
    "exp": true, // A semantic attribute indicating an experimental feature. Doesn't affect the build process.
    "weight": 1000, /* A feature weight; defaults to 0 if not specified.
        Used for sorting features within feature initialization code in ascending order
        (the lower the weight, the earlier the provider initializes).
        Used for predictable initialization order.
        For example see the STACK_PROXY_FEATURE that must be initialized in the very end. */
},
  1. Generate new feature set by running:
npm run features
  1. Create a folder within src/providers.
    • The index.ts should contain only the initialization code and extensions to existing types. The file must export an initProvider function:
export const initProvider = () => {
    /* The code is wrapped by a feature flag.
       If the feature is not included into a build, the code is cut off by rollup.
       The resultant empty function is cut off as well */
    if (flags[SOME_NEW_FEATURE]) {
        // Required code
        providersSync.push(someNewProvider);

        // Optional code for any additional setup
        addMiddlewareForProvider(SOME_NEW_PROVIDER, watchSyncFlags(), 1);
        providerMap[SOME_NEW_PROVIDER] = useSenderWatch;
        nameMap[SOME_NEW_PROVIDER] = fullList;
    }
};
- The provider itself and all the utility code is put into separate files within the provider folder.
  1. Write the provider code.
  2. Write the test in the __tests__ folder. Use .spec.ts extension.

Tests

For test coverage the following packages are used:

  • mocha - test runner;
  • sinon - stubs;
  • chai - assertions.

Stubs

sinon is used for stubbing functions within tests. In order to provide the best TypeScript experience make sure all stubs are assigned proper types, but do not include too much detail about stubbed function. Prefer the code style as follows:

import * as sinon from 'sinon';
import defer from 'src/utils/defer/defer';

describe('test', () => {
    const sandbox = sinon.createSandbox();
    let setDeferStub: sinon.SinonStub<
        Parameters<typeof defer.setDefer>,
        ReturnType<typeof defer.setDefer>
    >

    beforeEach(() => {
        setDeferStub = sandbox.stub(defer, 'setDefer');
    });

About

The client library of the web analytics tool. It is in the top 5 by popularity worldwide.

Resources

License

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages