ben.carrasco/prowler
1
0
Fork 0
mirror of https://github.com/prowler-cloud/prowler.git synced 2026-08-19 09:30:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs(ui): trim the Slack channel MSW comments

Browse Source 
- Keep fixture semantics and handler precedence notes
- Drop narrative around refusal shapes
This commit is contained in:
Pablo F.G
2026-08-18 17:46:46 +02:00
parent 5690d1ad94
commit 5af26a5f53
2 changed files with 35 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ui/__tests__/msw/handlers/slack.fixtures.ts
+24 -44
View File
@@ -57,13 +57,10 @@ export interface SlackChannelFixture {
}
export interface SlackTestMessageFixture {
/** Slack accepted the post. */
accepted: boolean;
/**
* The reason the settled task carries when it did not. The contract asks the
* task to report the same stable reason the synchronous endpoints put in
* `code`, but leaves the result's shape to the cloud lane — so this models
* both what it should carry (a reason token) and what it might (prose).
* Why it did not: the reason `code` would carry, or prose — the contract
* leaves the task result's shape open.
*/
error: string | null;
}
@@ -71,10 +68,6 @@ export interface SlackTestMessageFixture {
/**
* A refusal as the API sends one: the machine-readable reason in `code`, human
* copy in `detail`, and — for a `429` — the wait in `Retry-After`.
*
* All three travel together because that is what makes a test honest: a client
* that switched on `detail` would pass against a fixture carrying only `code`,
* and one that ignored `Retry-After` would pass against a `429` without it.
*/
export interface SlackRefusalFixture {
status: number;
@@ -109,36 +102,30 @@ export interface SlackFixture {
* transport failures. Distinct from `appConfigured: false`, which is a `503`.
*/
oauthUpstreamError: boolean;
/** Every channel the connected workspace exposes to Prowler. */
channels: SlackChannelFixture[];
/**
* Channels per cursor page. Small on purpose: the default workspace spans
* two pages, so a UI that stopped at `data` instead of following `links.next`
* would visibly lose channels.
* Small on purpose: the default workspace spans two pages, so a UI that
* stopped at `data` instead of following `links.next` would lose channels.
*/
channelsPageSize: number;
/** Slack refused the listing outright, with the reason named in `code`. */
channelsRefusal: SlackRefusalFixture | null;
/**
* The cursor `channelsRefusal` starts being answered from. Absent means from
* the first page, so the whole read fails; a page size means the first page
* is served and the refusal lands on the second — the partial read, where the
* picker has something to show and a reason why it is short.
* The cursor the refusal starts at. Absent, the whole read fails; a page
* size serves the first page and refuses the second — the partial read.
*/
channelsRefusalFromCursor?: number;
/**
* Slack refused the channel the user chose, when the `PATCH` validated it.
* Distinct from a listing refusal: the workspace answered the picker fine and
* it is the destination itself that cannot be used.
* Slack refused the chosen channel when the `PATCH` validated it — the
* listing itself answered fine.
*/
channelSaveRefusal: SlackRefusalFixture | null;
/** What the test-message task settles as. */
testMessage: SlackTestMessageFixture;
}
/**
* A UUID, as the API's integration ids are: the id travels in the URL of every
* Slack call the management page makes, and the actions accept no other shape.
* A UUID, as the API's ids are: it travels in the URL of every Slack call and
* the actions accept no other shape.
*/
export const SLACK_INTEGRATION_ID = "7c9e6a1b-2d3f-4e5a-8b6c-9d0e1f2a3b4c";
@@ -197,9 +184,8 @@ export const INTEGRATIONS_SERVER_ERROR_DETAIL = "A server error occurred.";
export const SLACK_MISSING_SCOPE_DETAIL =
"Slack refused the request: missing_scope.";
/**
* Sent for a channel that cannot be used, whichever way it cannot: the same
* sentence for "it is gone" and for "the app was removed from it". Only `code`
* separates them, which is the whole reason a client must read `code`.
* The same sentence for "it is gone" and "the app was removed from it": only
* `code` separates them, which is why a client must read `code`.
*/
export const SLACK_UNKNOWN_CHANNEL_DETAIL =
"That channel is not one Prowler can post to.";
@@ -246,8 +232,8 @@ export const SLACK_MISSING_SCOPE_REFUSAL: SlackRefusalFixture = {
};
/**
* Slack rate limiting Prowler. The endpoint this actually happens on is the
* channel listing: `conversations.list` is tier 2 and paginated.
* Where this really happens is the channel listing: `conversations.list` is
* tier 2 and paginated.
*/
export const SLACK_RATE_LIMITED_REFUSAL: SlackRefusalFixture = {
status: 429,
@@ -273,9 +259,8 @@ export const SLACK_CHANNEL_NOT_FOUND_REFUSAL: SlackRefusalFixture = {
};
/**
* The chosen channel is fine — the Prowler app is simply not in it, which
* someone in Slack fixes with `/invite @Prowler`. Identical `detail` to the
* refusal above, deliberately.
* The channel is fine, the Prowler app is simply not in it — fixed with
* `/invite @Prowler`. Identical `detail` to the refusal above, deliberately.
*/
export const SLACK_NOT_IN_CHANNEL_REFUSAL: SlackRefusalFixture = {
status: 400,
@@ -285,9 +270,8 @@ export const SLACK_NOT_IN_CHANNEL_REFUSAL: SlackRefusalFixture = {
};
/**
* The workspace's channels: two public, and one private the Prowler app has
* been invited to. Ordered so the private one lands on the second cursor page
* at the default page size — following `links.next` is what makes it visible.
* Two public channels and one private the Prowler app was invited to, ordered
* so the private one lands on the second cursor page.
*/
export const SLACK_PUBLIC_CHANNEL: SlackChannelFixture = {
id: "C0123AB",
@@ -317,8 +301,8 @@ export const SLACK_CHANNELS: SlackChannelFixture[] = [
export const SLACK_CHANNELS_PAGE_SIZE = 2;
/**
* The channel a finished install posts to: the first one the picker offers, so
* an install seeded with it always points at a channel the listing really has.
* The first channel the picker offers, so an install seeded with it always
* points at a channel the listing really has.
*/
export const SLACK_DEFAULT_CHANNEL = SLACK_PUBLIC_CHANNEL;
@@ -381,9 +365,8 @@ const configuredInstall = (
});
/**
* The same tenant, with a destination channel already on record — the state a
* second visit starts from, and the one that shows whether a later failure
* disturbs what was already saved.
* The same tenant with a destination channel already on record: the state a
* second visit starts from.
*/
export const slackFixtureWithDefaultChannel = (
channel: SlackChannelFixture = SLACK_PUBLIC_CHANNEL,
@@ -405,11 +388,8 @@ export const unreadableCheckTimeSlackFixture = (): SlackFixture =>
});
/**
* The listing refused partway through: the first cursor page is served and
* Slack rate limits the second. What a workspace larger than one page looks
* like when `conversations.list` runs out of tier-2 budget mid-read — the
* channels already read are still usable, and the refusal is only the reason
* the list stops where it does.
* The first cursor page is served and Slack rate limits the second: what is
* already read stays usable, the refusal only says why the list is short.
*/
export const partiallyReadSlackFixture = (
overrides: Partial<SlackFixture> = {},
ui/__tests__/msw/handlers/slack.ts
+11 -18
View File
@@ -59,9 +59,8 @@ const errorBody = (detail: string, status: number, code?: string) => ({
});
/**
* Answer a fixture's refusal exactly as the API would: its own status, its
* `code` when it names one, and `Retry-After` only where the status carries a
* wait. A consumer that reads any of the three gets the real thing.
* Answer a fixture's refusal as the API would: its own status, its `code`
* when it names one, and `Retry-After` only where the status carries a wait.
*/
const refuse = (refusal: SlackRefusalFixture) =>
HttpResponse.json(
@@ -259,8 +258,7 @@ export const handlersForSlack = (fx: SlackFixture) => {
),
http.get<{ taskId: string }>(`${API}/tasks/:taskId`, ({ params }) => {
// The test message settles as its own task (design D9), reporting only
// whether Slack accepted the post.
// The test message settles as its own task (design D9).
if (params.taskId.startsWith(TEST_MESSAGE_TASK_PREFIX)) {
const { accepted, error } = fx.testMessage;
return HttpResponse.json(
@@ -284,18 +282,14 @@ export const handlersForSlack = (fx: SlackFixture) => {
http.get<{ id: string }>(
`${API}/integrations/:id/slack/channels`,
({ params, request }) => {
// Cursor pagination: the UI follows `links.next` opaquely, so the
// cursor's shape is this fixture's business alone. Read before the
// refusals: which page is being asked for decides whether this one is
// refused at all.
// The UI follows `links.next` opaquely, so the cursor's shape is this
// fixture's business alone. Read first: the page decides the refusal.
const cursor = Number(
new URL(request.url).searchParams.get(CHANNEL_CURSOR_PARAM) ?? "0",
);
// A refusal named for this endpoint wins over the fixture's blanket
// rate limiting, which is the coarser switch of the two. It applies
// from the cursor the fixture names, so a read that fails partway
// through is expressible and not only one that fails outright.
// An endpoint-specific refusal wins over the blanket rate limiting,
// and applies from the named cursor, so a partial read is expressible.
if (
fx.channelsRefusal &&
cursor >= (fx.channelsRefusalFromCursor ?? 0)
@@ -325,9 +319,8 @@ export const handlersForSlack = (fx: SlackFixture) => {
),
/**
* The generic PATCH, recording the default channel. The UI submits only
* `channel_id`; the name here is derived from the channel the id resolves
* to, exactly as the API derives it from Slack (design D6).
* The generic PATCH. The UI submits only `channel_id`; the name is derived
* from it here, as the API derives it from Slack (design D6).
*/
http.patch(`${API}/integrations/:id`, async ({ request }) => {
const body = (await request.json().catch(() => null)) as {
@@ -339,8 +332,8 @@ export const handlersForSlack = (fx: SlackFixture) => {
if (!install) {
return HttpResponse.json(errorBody("Not found.", 404), { status: 404 });
}
// Slack refused the channel while the API validated it: the id resolves
// to a channel the picker offered, and Slack still says no.
// Checked before the id lookup: the picker did offer this channel, and
// Slack refused it anyway when the API validated it.
if (fx.channelSaveRefusal) return refuse(fx.channelSaveRefusal);
if (!channel) {
return HttpResponse.json(errorBody(SLACK_UNKNOWN_CHANNEL_DETAIL, 400), {