Deprecate and remove `parseUrl` and refactor `getSanitizedUrlString`

[AbhiPrasad]

Description

Now that we support ES2020 (aka not IE11 anymore) and Node.js 18+, we can get rid of parseUrl in favor of a method that just uses the built-in URL object. This will save us some bundle size (given we can remove that regex), and we get performance benefits from using native code.

sentry-javascript/packages/core/src/utils-hoist/url.ts

Line 17 in 3d63621

export function parseUrl(url: string): PartialURL {

Instead of just blanket replacing parseUrl, we'll slowly convert all it's usages to using a new helper, which looks something like so:

/**
 * Parses string to a URL object
 *
 * @param url - The URL to parse
 * @returns The parsed URL object or undefined if the URL is invalid
 */
export function parseStringToURL(url: string): URL | undefined {
  try {
    // Node 20+, Chrome 120+, Firefox 115+, Safari 17+
    if ('canParse' in URL) {
      // Use `canParse` to short-circuit the URL constructor if it's not a valid URL
      // This is faster than trying to construct the URL and catching the error
      return (URL as unknown as URLwithCanParse).canParse(url) ? new URL(url) : undefined;
    }
  } catch {
    // empty body
  }

  return undefined;
}

• Remove all usages in @sentry/core, and implement parseStringToURL
• Replace

  sentry-javascript/packages/browser-utils/src/instrument/xhr.ts

  Line 150 in 3d63621

  function parseUrl(url: string | unknown): string | undefined {

• sentry-javascript/packages/browser/src/integrations/breadcrumbs.ts

  Line 36 in 3d63621

  parseUrl,

• sentry-javascript/packages/opentelemetry/src/utils/parseSpanDescription.ts

  Line 23 in 3d63621

  parseUrl,

• sentry-javascript/packages/node/src/integrations/http/SentryHttpInstrumentation.ts

  Line 17 in 3d63621

  parseUrl,

• sentry-javascript/packages/node/src/integrations/node-fetch/SentryNodeFetchInstrumentation.ts

  Line 6 in 3d63621

  import { addBreadcrumb, getBreadcrumbLogLevelFromHttpStatusCode, getSanitizedUrlString, parseUrl } from '@sentry/core';

• sentry-javascript/packages/browser/src/tracing/request.ts

  Line 22 in 3d63621

  parseUrl,

• sentry-javascript/packages/browser-utils/src/metrics/browserMetrics.ts

  Line 9 in 3d63621

  parseUrl,

With that, we can also refactor getSanitizedUrlString, which should provide some performance benefits.

sentry-javascript/packages/core/src/utils-hoist/url.ts

Line 55 in 3d63621

export function getSanitizedUrlString(url: PartialURL): string {

• sentry-javascript/packages/bun/src/integrations/bunserver.ts

  Line 10 in 3d63621

  getSanitizedUrlString,

• sentry-javascript/packages/opentelemetry/src/utils/getRequestSpanData.ts

  Line 9 in 3d63621

  import { getSanitizedUrlString, parseUrl } from '@sentry/core';

• sentry-javascript/packages/node/src/integrations/node-fetch/SentryNodeFetchInstrumentation.ts

  Line 6 in 3d63621

  import { addBreadcrumb, getBreadcrumbLogLevelFromHttpStatusCode, getSanitizedUrlString, parseUrl } from '@sentry/core';

• sentry-javascript/packages/node/src/integrations/http/SentryHttpInstrumentation.ts

  Line 14 in 3d63621

  getSanitizedUrlString,

• sentry-javascript/packages/opentelemetry/src/utils/parseSpanDescription.ts

  Line 22 in 3d63621

  getSanitizedUrlString,

Reminder list:

• Can we remove extractQueryParamsFromUrl?

[AbhiPrasad]

It's super annoying to remove parseUrl because new URL breaks when you give it relative URLs 😢

I'm gonna try to solve this a different way.

[AbhiPrasad]

landed in a decently elegant solution for relative urls: #15882