Azure @azure-tools/test-recorder library for JavaScript

The Azure SDK for JavaScript is composed of a multitude of libraries that attempt to deliver a common, homogenous SDK to make use of all of the services that Azure can provide. Among the challenges of such a goal, we have some that are specific to tests, some of which we can summarize in the following questions:

How to write live tests that can work as unit tests?
How to ensure that tests are as fast as they can be?
How to avoid writing mocked versions of our HTTP API?
How to protect sensitive data from our live tests?
How to write tests that support parallelism?
How to write isomorphic tests for NodeJS and the Browsers?

Our recorder tool package @azure-tools/test-recorder attempts to provide an answer for those questions.

Note 1: In case you're depending on @azure-tools/test-recorder@1.x.y and want to migrate your tests to version 3, follow the migration guide to recorder v3 from v1

Note 2: If you're looking to onboard to the asset-sync workflow to push out the test recordings to Azure/azure-sdk-assets repository, refer to asset-sync-migration.

Note 3: Refer to testing-commands if you need help on commands to run during testing.

This library provides interfaces and helper methods to equip the SDKs in the azure-sdk-for-js repo with the recording and playback capabilities for the tests, it targets HTTP requests in both Node.js and the Browsers.

@azure-tools/test-recorder, as part of the Test Utils available in this repository, it is supposed to be added only as a devDependency and should be used only for the tests of an sdk.

This tool helps to record and playback the tests in the JS repo by leveraging the unified out-of-process test proxy server.

Index

Key concepts
Getting started
Onboard to asset-sync workflow
Using the Recorder
Examples
Troubleshooting
Next steps
Contributing

Key concepts

To record means to intercept any HTTP request, store it in a file, then store the response received from the live resource that was originally targeted. We leverage the unified out-of-process test proxy server that is built for this use case. The output files are stored in recordings/node/* and in recordings/browsers/*, which are relative to the root of the project you're working on.
To playback means to intercept any HTTP request and to respond it with the stored response of a previously recorded matching request.
Sensitive information means content that should not be shared publicly. Content like passwords, unique identifiers or personal information should be cleaned up from the recordings. Some functionality is provided to fix this problem. You can read more at securing sensitive data.

Getting started

We're about to go through how to set up your project to use the @azure-tools/test-recorder package.

Note: If you're new to this repository, follow the ContributingGuide.md to learn how to setup/build the repo and to create/test an SDK in the Azure/azure-sdk-for-js repository.

Installing the package

From this point forward, we'll assume that you're developing (perhaps contributing!) to one of the azure-sdk-for-js's libraries. So, your next step is to change directory to the path relevant to your project. Let's say you want to add the @azure-tools/test-recorder package to @azure/data-tables (it already uses test-recorder, but bear with us), you'll be doing the following to install the package:

cd sdk/tables/data-tables
rush add -p @azure-tools/test-recorder@^3.0.0 --dev

If you are using @azure/identity in your tests, also install "@azure-tools/test-credential" package.

rush add -p @azure-tools/test-credential@^1.0.0 --dev

With a following rush update, you may see something like below.

{
  // ... your package.json properties
  "devDependencies": {
    // ... your devDependencies
    "@azure-tools/test-credential": "^1.0.0", // If you are using `@azure/identity` in your tests
    "@azure-tools/test-recorder": "^3.0.0"
    // ... more of your devDependencies
  }
  // ... more of your package.json properties
}

And you're ready! Now you can use the test recorder in your code, as shown below:

import { Recorder } from "@azure-tools/test-recorder";

Or, if you know what you want to import, you can also do the following:

import { Recorder, RecorderStartOptions, env, SanitizerOptions } from "@azure-tools/test-recorder";

Configuring your project

Having the recorder as a devDependency means that you'll be able to start recording tests right away by using the Recorder class.

The test-recorder provides the Recorder class that deals with recording and playing back the network requests, depending on the value assigned to the TEST_MODE environment variable.

If TEST_MODE equals to record, it will automatically store network requests in a plain text file in the folder recordings at the root of your library (it is sdk/tables/data-tables in our example).
This package assumes that the tests in the sdk are leveraging mocha and rollup (and karma test runner for browser tests) as suggested by the template package in the repo.

package.json scripts

For the unified recorder client library to work, the test proxy server must be active while you are running your tests. Helpers have been added to the dev-tool package which manage starting and stopping the test proxy server before and after your tests are run.

Your test scripts (in package.json) should be based on the following examples:

script name	command
`unit-test:browser`	`dev-tool run test:browser`
`unit-test:node`	`dev-tool run test:node-ts-input -- --timeout 1200000 --exclude 'test/*/browser/.spec.ts' 'test/*/.spec.ts'`
`integration-test:browser`	`dev-tool run test:browser`
`integration-test:node`	`dev-tool run test:node-js-input -- --timeout 5000000 'dist-esm/test/*/.spec.js'`

Note the difference between the dev-tool node-ts-input and node-js-input commands:

node-ts-input runs the tests using ts-node, without code coverage.
node-js-input runs the tests using the built JavaScript output, and generates coverage reporting using nyc.

TEST_MODE

By using recorder with your clients, the requests are redirected to the test-proxy tool to either save them or replay them. Interactions with the test-proxy tool vary based on what the TEST_MODE environment variable is.

TEST_MODE	What?
`record`	Stores network requests with the help of test-proxy tool in a plain text file in the folder `recordings` at the root of your repository (example: root of the `sdk/tables/data-tables` project)
`playback`	Stored requests/responses are utilized by the test-proxy tool when the requests are redirected to it instead of reaching the service
`live`	Recorder and its methods are no-ops here, requests directly reach the service instead of being redirected at the test-proxy tool layer

Onboard to asset-sync workflow

This section assumes that your package is new to the JS repo and that you're trying to onboard your tests with recorder, and the asset-sync workflow.

From the root of the repo, navigate to your package

cd sdk/<service-folder>/<package-name>

Generate an sdk/<service-folder>/<package-name>/assets.json file by running the following command.

npx dev-tool test-proxy init

This command would generate an assets.json file with an empty tag. Once you generate the recordings for your tests and push them to the assets repo, the tag gets populated here.

For further understanding, please read the asset sync migration and workflow.

Using the `Recorder`

Inside a mocha test (either in the beforeEach or in the test body itself), you will need to instantiate the Recorder as below to leverage its functionalities.

let recorder: Recorder;

beforeEach(async function (this: Mocha.Context) {
  recorder = new Recorder(this.currentTest);
});

The tested client needs to install the recording policy that redirects requests to the test-proxy tool first before they go to the service. This is done by simply passing the client options bag through the recorder.configureClientOptions helper:

const client = new AnyCoreV2Client(/** args **/, recorder.configureClientOptions(/** client options **/));

Recording starts with the recorder.start() method.

await recorder.start(/** recorderOptions go here **/);

Recorder options will typically contain the environment setup needed for the playback mode, and the sanitizers that help with masking the sensitive information in the recordings, more on the recorder options below.

Any requests that are made using the above client (AnyCoreV2Client) will be redirected to the test-proxy tool before they reach the service, the requests and responses will be recorded and saved when recorder.stop() is called in record mode.

Likewise, in playback mode, the saved responses are utilized by the test-proxy tool when the requests are redirected to it instead of reaching the service.

await recorder.stop();

Call this method to ping the test-proxy tool with a stop request, this helps to stop recording, saves the recording file in record mode.

Note: Instantiating, starting, and stopping the recorder all have no effect in the live mode (TEST_MODE=live). In live mode, the redirection to the test-proxy tool doesn't happen and the requests are sent to the services as usual.

Recorder#variable()

To handle the dynamic/generated values for testing that are created as part of the tests, to make sure the requests in the playback mode match the ones in the record mode, you can leverage the Recorder#variable function.

For example, when resources are created dynamically, the name of the resource that is generated would vary in record and playback modes. This is not ideal for playing back the requests/responses because the requests wouldn't match with what was saved in the recording in record mode.

For such cases, you can leverage the recorder.variable() method. It acts differently based on what the TEST_MODE is.

// variable method
recorder.variable("table-name", `table${Math.ceil(Math.random() * 1000 + 1000)}`)

// Shows up in the recording as
  "Variables": {
    "table-name": "table1662"
  }

Lets you register a variable to be stored with the recording. The behavior of this function depends on whether the recorder is in record/live mode or in playback mode.
In record mode, the function will store the value provided with the recording as a variable and return that value.
In playback mode, the function will fetch the value from the variables stored as part of the recording and return the retrieved variable, throwing an error if it is not found.
In live mode, no recordings are saved, just returns the provided value.

Environment variables

@azure-tools/test-recorder exports env which loads the environment variables from the correct location (using process.env and dotenv in Node, and using window.__env__ via karma in the browser), and also means that the environment variables set in envSetupForPlayback are used in playback mode.

recorder.start() internally sets up the environment variables for playback. So, make sure to have the recorder.start() call before you use any environment variables in your tests.
To use an environment variable in a test, just do env["NAME_OF_THE_VARIABLE"].
Recorder also exports a assertEnvironmentVariable global method, which can be used to retrieve the environment variables. The function assertEnvironmentVariable("NAME_OF_THE_VARIABLE") either returns the value or throws an error saying the variable is not defined in your environment. (This function comes handy when your function args expect a non-undefined value but the environment variable may not be defined in the runtime.)

`@azure-tools/test-credential` package and the NoOpCredential

We do not record the AAD traffic since it is typically noise that is not needed for testing the SDK(unless we are testing the @azure/identity package directly which uses the @azure-tools/test-recorder differently to record the tests).

Tests with clients using AAD should make use of the new @azure-tools/test-credential package.
This package provides a NoOpCredential implementation of TokenCredential which makes no network requests, and should be used in playback mode.
The provided createTestCredential helper will handle switching between NoOpCredential in playback and ClientSecretCredential when recording for you:

import { createTestCredential } from "@azure-tools/test-credential";

const credential = createTestCredential();

// Create your client using the test credential.
new MyServiceClient(<endpoint>, credential);

Since AAD traffic is not recorded by the new recorder, there are no AAD credentials to remove from the recording using a sanitizer.

karma.conf - for the browser tests

When running browser tests, the recorder relies on an environment variable to determine where to save the recordings. Add this snippet to your karma.conf.js:

const { relativeRecordingsPath } = require("@azure-tools/test-recorder");

process.env.RECORDINGS_RELATIVE_PATH = relativeRecordingsPath();

And then, again in karma.conf.js, add the variable to the list of environment variables:

module.exports = function (config) {
  config.set({
    /* ... */

    envPreprocessor: [
      // variables
      "RECORDINGS_RELATIVE_PATH", // Add this!
      "TEST_PROXY_HTTP_PORT", // Optional (Incase you need a port other than 5000)
      // more variables
    ],

    /* ... */
  });
};

Onboard to asset-sync workflow

Examples

How to record

To record your tests,

make sure to set the environment variable TEST_MODE to record

# Windows
set TEST_MODE=record

# Linux / Mac
export TEST_MODE=record

then in your code, instantiate the Recorder
call #start() function with the recorder options
modify the client options using the #configureClientOptions() method
then make calls with your client as needed
call #stop() function to save the recording in a file

In the following example, we'll use the recorder with the client from @azure/data-tables:

Example from the template project if you want to check out.

import { RecorderStartOptions, Recorder, env } from "@azure-tools/test-recorder";
import { createTestCredential } from "@azure-tools/test-credential";
import { TableServiceClient } from "@azure/data-tables";

const recorderOptions: RecorderStartOptions = {
  envSetupForPlayback: {
    TABLES_URL: "https://fakeaccount.table.core.windows.net",
  },
  sanitizerOptions: {
    bodySanitizers: [
      {
        target: encodeURIComponent(env.TABLES_URL ?? ""),
        value: encodeURIComponent(`https://fakeaccount.table.core.windows.net`),
      },
    ],
  },
};

describe(`TableServiceClient tests`, () => {
  let recorder: Recorder;
  let credential;

  beforeEach(async function () {
    recorder = new Recorder(this.currentTest);
    await recorder.start(recorderOptions);
    credential = createTestCredential();
  });

  afterEach(async function () {
    await recorder.stop();
  });

  it("should create new table, then delete", async () => {
    const tableName = recorder.variable(
      "table-name",
      `table${Math.ceil(Math.random() * 1000 + 1000)}`
    );
    const client = new TableServiceClient(
      env["TABLES_URL"],
      credential,
      recorder.configureClientOptions({})
    );
    await client.createTable(tableName);
    await client.deleteTable(tableName);
  });
});

After running this test with the TEST_MODE environment variable set to record, the recorder assisted by the test-proxy tool will create a recording file with the contents of the HTTP requests as well as the responses.

If the package has been onboarded to asset-sync workflow, the recording will be loacted under the .assets/ at the root of the repository.
- To view the recording, refer to .assets/.breadcrumb to find the entry that matches your SDK. This will give you the name of the directory within .assets that your recordings are located in.
- Refer to asset sync workflow for more understanding and further steps.
Otherwise, the recording will be located at recordings/node/tableserviceclient_tests/recording_should_create_new_table_then_delete.json.
You'll see in the code above that we're invoking recorder.stop. This is so that, after each test, we can stop recording and the test file can be generated.
We recommend instantiating/starting the new recorder in the beforeEach block and stopping the recorder on afterEach to make sure that the generated files are smaller and easier to understand than by having them all in one chunk.

Securing Sensitive Data

Live tests need to do sensitive operations, like authenticating with your Azure endpoints, keys, secrets, etc. These are generally contained in the environment variables which are used as part of the tests.

We must secure them and not let them leak into our recordings. To avoid storing the sensitive info in the recordings, we use the sanitizers to mask the values with the fake ones or remove them, RecorderStartOptions helps us here.

`RecorderStartOptions`

RecorderStartOptions has two components, envSetupForPlayback and the sanitizers which you'd have seen in the previous snippet.

For a live test to be run, we typically need the test secrets, which are usally stored as Environment variables.

And since in playback mode, the requests don't reach the service, we don't actually need to have/share the test secrets to run the tests in playback mode.

Another angle to this is that the recordings store the requests, which would contain the sensitive information related to the endpoints, tokens, keys, secrets, credentials, etc that are parts of the supposedly secretive environment variables or derivatives of them.

We try our best to make sure the sensitive information is not leaked anywhere with the help of envSetupForPlayback and Sanitizers.

`envSetupForPlayback`

envSetupForPlayback expects key-value pairs, with keys signifying the names of the environment variables, and the values would be the fake ones that you'd like to map/swap the originals with.

  envSetupForPlayback: {
    TABLES_URL: "https://fakeaccount.table.core.windows.net",
  }

Used in record and playback modes. No effect in live mode.

The key-value pairs will be used as the environment variables in playback mode.
If the environment variables are present in the recordings as plain strings, they will be replaced with the provided values in record mode.

`Sanitizers`

Sanitizers	How does it look? Example??	What does it do?
`generalSanitizer`	`{ regex: true, target: "abc+def", value: "fakeValue" }`	Offers a general regex replace across request/response Body, Headers, and URI. For the body, this means regex applying to the raw JSON.
`bodySanitizer`	`{ regex: true, target: "(.)&SECRET=(?<secret_content>[^&])&(.*)", value: fakeSecretValue, groupForReplace: "secret_content" }`	Offers regex replace within a returned body. Specifically, this means regex applying to the raw JSON. If you are attempting to simply replace a specific key, the `bodyKeySanitizer` is probably the way to go.
`headerSanitizer`	`{ key: "your_uuid", value: sanitizedValue }`	Can be used for multiple purposes: 1) To replace a key with a specific value, do not set "regex" value. 2) To do a simple regex replace operation, define arguments "key", "value", and "regex" 3) To do a targeted substitution of a specific group, define all arguments "key", "value", and "regex"
`uriSanitizer`	`{ target: secretEndpoint, value: fakeEndpoint}`	General use sanitizer for cleaning URIs via regex. Runs a regex replace on the member of your choice.
`connectionStringSanitizer`	`{` `actualConnString: env.TABLES_CONN_STRING,` `fakeConnString: "Endpoint=https://fakeaccountname.net/;SharedAccessSignature=st=2021-08-03&sig=fakesigval"` `}`	Internally, - connection strings are parsed and - each part of the connection string is mapped with its corresponding fake value - `generalRegexSanitizer` is applied for each of the parts with the real and fake values that are parsed
`bodyKeySanitizer`	`{ jsonPath: "$.bodyProvided.secret_info", regex: secretValue, value: fakeSecretValue }`	This sanitizer offers regex update of a specific JTokenPath. EG: "TableName" within a json response body having its value replaced by whatever substitution is offered. This simply means that if you are attempting to replace a specific key wholesale, this sanitizer will be simpler than configuring a BodyRegexSanitizer that has to match against the full "KeyName": "Value" that is part of the json structure. Further reading is available here. If the body is NOT a JSON object, this sanitizer will NOT be applied.
`removeHeaderSanitizer`	`{ headersForRemoval: ["X-Content-Type-Options"] }`	A simple sanitizer that should be used to clean out one or multiple headers by their key. Removes headers from before saving a recording.
`resetSanitizer`	`true`	This clears the sanitizers that are added.

Sanitizers can be added in two different ways.

Pass them as part of the recorder.start({ envSetupForPlayback, sanitizerOptions }) call. Sanitizers are applied on the recordings in record mode before they are saved.
Call recorder.addSanitizers(sanitizerOptions, ["record", "playback"]). This way, the same sanitizer would be applied in both record and playback modes.

How to playback

Once you have recorded something, you can run your tests again with TEST_MODE set to playback.

You'll notice how the tests succeed much faster. That's because the requests don't reach the service endpoints, the recorder assisted by the test-proxy tool will respond every request with their matching copy stored in the recordings.

Update existing recordings

Once you have your recorded files, to update them after changing one of the tests, simply re-run the tests with TEST_MODE set to record. This will override previously existing files.

Note: If you rename the file of the test, or the name of the test, the path of the recording will change. Make sure to delete the recordings corresponding to the deleted tests. If at any point in time you lose your recordings, don't worry. Running your tests with TEST_MODE=record will re-generate them.

Skipping tests

Writing live tests can take considerable time, specially since each time you want to check that everything works fine, you potentially need to run again every test. You can specify what test to run by following Mocha's approach of setting certain tests to it.only, and also to skip specific tests with it.skip.

If you launch the recorder in record mode with some of these changes (and given that you activate the recorder on beforeEach), only the files that relate to the changed tests will be updated. Skipped tests won't update their recordings.

This way, you can focus on fixing a specific set of tests with .only, then remove all the .only calls and trust that the playback will keep confirming that the unaffected tests are fine and green.

You can also skip specific tests with the following.

import { isLiveMode } from "@azure-tools/test-recorder";

it("test-title", function (this: Mocha.Context) {
  // isPlaybackMode() and isRecordMode() methods are also available from recorder.
  if (!isLiveMode()) this.skip(); // This skips the test in record and playback modes
  // Test goes here...
  // ...
});

Supporting parallelism

A common issue while running integration tests is that, sometimes two individuals or machines might try to run the same set of tests against the same resource.

This is not directly related to the @azure-tools/test-recorder package, but if you're getting into issues because of concurrent conflicting requests, we strongly suggest using randomly generated strings as prefixes or suffixes for the resources you create.

Refer to Recorder#variable() section to handle the dynamic/generated values for testing that are created as part of the tests, to make sure the requests in the playback mode match the ones in the record mode.

Since new resources are likely to get accumulated because some tests would crash or fail for any reason, make sure you delete the resources that are not cleared.

Isomorphic tests

@azure/test-utils-recorder does support running tests in the browser. If you use Karma, as long as your karma configuration is correct, your tests should work both on NodeJS and in the browsers!

Troubleshooting

Besides the usual debugging of your code and tests, if you ever encounter a problem while recording your tests, make sure to read the output in the recordings. If the output is not what you expected, please follow up the contributing guidelines on how to write an issue for us. We'll make sure to handle it as soon as we find the time.

If you run into issues while running the tests in record/playback modes, some of the following troubleshooting steps may help:

Viewing test proxy log output

dev-tool by default outputs logs from the test proxy to test-proxy-output.log in your package's root directory. These logs can be inspected to see what requests were made to the proxy tool.

Switching ports

If port 5000 is already being used in your machine, you can specify the environment variable TEST_PROXY_HTTP_PORT and point to the port number that you wish. (Example, export TEST_PROXY_HTTP_PORT=2345)

Inspecting recordings

Refer to asset sync workflow - inspect recordings.

Next steps

Almost all the libraries in the azure-sdk-for-js repository leverage test-recorder(v3.0). If you want to refer to the tests that leverage this package, go through the following search link: https://github.com/Azure/azure-sdk-for-js/search?q=test-recorder

Contributing

If you'd like to contribute to this library, please read the contributing guide to learn more about how to build and test the code.

Troubleshooting

Logging

Enabling logging may help uncover useful information about failures. In order to see logs from the recorder client, set the AZURE_LOG_LEVEL environment variable to info. Alternatively, logging can be enabled at runtime by calling the setLogLevel function in the @azure/logger package.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

README.md

README.md

Azure @azure-tools/test-recorder library for JavaScript

Index

Key concepts

Getting started

Installing the package

Configuring your project

package.json scripts

TEST_MODE

Onboard to asset-sync workflow

Using the `Recorder`

Recorder#variable()

Environment variables

`@azure-tools/test-credential` package and the NoOpCredential

karma.conf - for the browser tests

Onboard to asset-sync workflow

Examples

How to record

Securing Sensitive Data

`RecorderStartOptions`

`envSetupForPlayback`

`Sanitizers`

How to playback

Update existing recordings

Skipping tests

Supporting parallelism

Isomorphic tests

Troubleshooting

Viewing test proxy log output

Switching ports

Inspecting recordings

Next steps

Contributing

Troubleshooting

Logging

Files

README.md

Latest commit

History

README.md

File metadata and controls

Azure @azure-tools/test-recorder library for JavaScript

Index

Key concepts

Getting started

Installing the package

Configuring your project

package.json scripts

TEST_MODE

Onboard to asset-sync workflow

Using the Recorder

Recorder#variable()

Environment variables

@azure-tools/test-credential package and the NoOpCredential

karma.conf - for the browser tests

Onboard to asset-sync workflow

Examples

How to record

Securing Sensitive Data

RecorderStartOptions

envSetupForPlayback

Sanitizers

How to playback

Update existing recordings

Skipping tests

Supporting parallelism

Isomorphic tests

Troubleshooting

Viewing test proxy log output

Switching ports

Inspecting recordings

Next steps

Contributing

Troubleshooting

Logging

Using the `Recorder`

`@azure-tools/test-credential` package and the NoOpCredential

`RecorderStartOptions`

`envSetupForPlayback`

`Sanitizers`