Skip to content

Latest commit

 

History

History
227 lines (182 loc) · 9.7 KB

MIGRATION.md

File metadata and controls

227 lines (182 loc) · 9.7 KB
Raw

Migration Notes

10.0.1 → 11.0.0

  • pwnedPasswordRange now returns an object mapping the matching suffix to a count representing the number of occurrences, rather than an array of objects each containing a matching suffix and its count. Code dependent on parsing the response text will need updated to deal with the new data format: 
    {
  "003D68EB55068C33ACE09247EE4C639306B": 3,
  "012C192B2F16F82EA0EB9EF18D9D539B0DD": 1,
  ...
}

9.0.3 → 10.0.0

  • The production/minified versions of the browser build targets have been renamed:

    • ESM for Browsers (<script type="module">)
      • dist/browser/hibp.esm.min.js → dist/browser/hibp.module.js
    • UMD
      • dist/browser/hibp.umd.min.js → dist/browser/hibp.umd.js

  • The development/non-minified versions of the UMD and ESM for browsers build targets have been removed. If you were using them, please update your imports to use the production/minified versions (see above).

  • The internal directory structure of the source code is now being preserved in the CJS and ESM for bundlers build outputs (dist/cjs and dist/esm). If you were deep importing anything you probably shouldn't have been (:wink:), you may need to update your imports.

  • Support for Node.js version 10.x has been dropped. You must upgrade your Node.js environment to at least v12.16.0.

8.0.1 → 9.0.0

  • Output files for all build targets have been consolidated under the dist directory. This should be transparent if you followed the documentation, but the changes are as follows:

    • CommonJS
      • lib/hibp.js → dist/cjs/hibp.js
    • ECMAScript Modules
      • es/hibp.js → dist/esm/hibp.js
    • ECMAScript Modules for Browsers (development)
      • dist/hibp.mjs → dist/browser/hibp.esm.js
    • ECMAScript Modules for Browsers (production)
      • dist/hibp.min.mjs → dist/browser/hibp.esm.min.js
    • UMD (development)
      • dist/hibp.js → dist/browser/hibp.umd.js
    • UMD (production)
      • dist/hibp.min.js → dist/browser/hibp.umd.min.js
    • TypeScript Declarations
      • types/hibp.d.ts → dist/hibp.d.ts

  • Support for Node.js version 8.x has been dropped. You must upgrade your Node.js environment to at least v10.

7.5.2 → 8.0.0

  • The breachedAccount, pasteAccount, and search modules now have an apiKey option, which is required by v3 of the haveibeenpwned.com API (unless you are proxying your requests through a server that inserts an API key on your behalf via the baseUrl option). You can purchase an API key from Troy at https://haveibeenpwned.com/API/Key. See Troy's blog post for rationale and a full explanation.
  • The default value of the truncate option in the breachedAccount and search modules has been changed from false to true per Troy's recommendation. If you do not specify a value of false explicitly, each Breach result will only contain the breach name (no metadata).
  • The default value of the includeUnverified option in the breachedAccount module has been changed from false to true per Troy's recommendation. Although there are not many unverified breaches in the system, it's possible you will get more breaches back than you did previously. You may explicitly disable this by specifying a value of false for this option.
  • Support for Node.js version 6.x has been dropped. You must upgrade your Node.js environment to at least v8.9.0.

6.0.0 → 7.0.0

  • pwnedPassword now uses the more secure hash range API rather than submitting plain text passwords over the wire. The new remote API no longer makes a distinction between passwords that are hashses vs. plain text, so pwnedPassword no longer takes an options object as the isAHash option has been removed.

  • pwnedPassword now resolves with a number representing the number of times the given password was exposed in a breach. Code using truthy checks should continue to function as before (when it returned a boolean), but explicit checks will need updated.

  • pwnedPasswordRange now returns an array of objects containing the matching suffix and a count representing the number of occurrences, rather than a plain text blob of all the data directly from the remote API response. Code dependent on parsing the response text will need updated to deal with the new data format:

    [
  { suffix: "003D68EB55068C33ACE09247EE4C639306B", count: 3 },
  { suffix: "012C192B2F16F82EA0EB9EF18D9D539B0DD", count: 1 },
  ...
]

5.3.0 → 6.0.0

  • Support for Node.js versions less than 6.x has been dropped. If you are leveraging this library in such an environment, you should restrict the version in your dependencies to ^5.3.0.

4.4.0 → 5.0.0

  • The biggest breaking change in 5.0.0 is the removal of the default export. hibp is designed as a collection of modules to be imported explicitly as needed and exporting a default-named object containing all the modules is arguably an anti-pattern. Instead, an anonymous object of all the named modules is exported, providing better dead code elimination support in order to produce smaller bundles when importing from hibp. The quickest upgrade path (providing invocation syntax equivalence to prior versions) is to change your import statement to import all the modules into a local hibp namespace, but the recommended upgrade path is to import exactly which modules you need and update your calls to remove the preceding hibp references.

    // 4.x
import hibp from 'hibp';
hibp.breachedAccount(/* ... */).then(/* ... */);

// 5.x (upgrade option 1, one-liner quick fix)
import * as hibp from 'hibp';
hibp.breachedAccount(/* ... */).then(/* ... */);

// 5.x (upgrade option 2, more explicit but requires more code changes)
import { breachedAccount } from 'hibp';
breachedAccount(/* ... */).then(/* ... */);

  • The browser entry point field has been removed from package.json as webpack was using it by default when omitting the target option or explicitly using target: 'web' (see issue #8 for details). No <script> tag changes should be necessary, but if you were otherwise relying on the browser field to resolve to the UMD build, you will need to update your configuration accordingly. Also worth noting here is the fact that the non-UMD builds have been updated to target browsers (see issue #9), so bundling them instead of the UMD build when targeting browsers should remain fully compatible while producing smaller bundles.

  • The index.js file has been removed entirely. It's sole purpose was to provide a separate entry point for the CJS/ESM (non-UMD) builds to include the source-map-support module to enable source map support in Node for debugging purposes. Source maps are still generated at build time and included in the package, so debugging is still possible but the responsibility of enabling support for source maps is now on the consumer. If you were importing index.js explicitly rather than relying on the entry point fields in package.json, you will need to replace that with hibp.js.

3.0.0 → 4.0.0

  • Support for Node.js versions less than 4.x has been dropped. It will probably still work (at least for the foreseeable future), but I'm not going out of my way to make sure. If you are leveraging this library in such an environment, you should restrict the version in your dependencies to ^3.0.0.

2.2.0 → 3.0.0

  • The browser (UMD) build output has moved from the lib directory to the dist directory to separate it from the server-side output. A development (non-minified) version is also now included, which was omitted in the past as it would have had the same file name in the same directory as the server-side output.

1.0.8 → 2.0.0

  • All API methods which previously resolved to undefined (upon receiving a 404 Not Found response from the remote endpoint) now resolve to null instead. This may or may not be a breaking change, depending on how strictly you're handling the "no data found" return value. Loose truthy/falsey checks like if (breachData) { ... } will be fine, but strict equality checks like if (breachData === undefined) { ... } will break.

    N.B. This is a philosophical change based on various sources regarding the difference between null and undefined in JavaScript. In the case where a query responds with no data, it is an expected absence of value, as that is how the remote API is documented to respond when there are no relevant objects to return.

    Ryan Morr:

    To distinguish between the two, you may want to think of undefined as representing an unexpected absence of value and null as representing an expected absence of value."

    MDN:

    In APIs, null is often retrieved in place where an object can be expected but no object is relevant.

  • All API methods that previously took optional, positional parameters like domain and truncateResults now take an options object instead. For example:

    1.0.8 (old):

    hibp.breachedAccount(account, 'adobe.com', true).then(/* ... */);

    2.0.0 (new):

    hibp
  .breachedAccount(account, { domain: 'adobe.com', truncate: true })
  .then(/* ... */);

    This change was made to make the API more expressive and reduce ambiguity. See the API documentation (or JSDoc comments) for details.