Author SHA1 Message Date
ilia 22ecdbe819 fix(hiringcafe): use Next.js SSR endpoint after search API auth broke
Hiring Cafe now returns 401/404 on the legacy /api/search-jobs endpoints.
Switch the extractor to the site's _next/data SSR transport so discovery
runs succeed again without manual cookie injection.
2026-06-10 14:19:14 -04:00
+73 -54
View File
@@ -111,12 +111,6 @@ function parseWorkplaceTypes(
} }
} }
function encodeSearchState(searchState: unknown): string {
const json = JSON.stringify(searchState);
const urlEncodedJson = encodeURIComponent(json);
return Buffer.from(urlEncodedJson, "utf-8").toString("base64");
}
function asRecord(value: unknown): Record<string, unknown> | null { function asRecord(value: unknown): Record<string, unknown> | null {
if (!value || typeof value !== "object" || Array.isArray(value)) return null; if (!value || typeof value !== "object" || Array.isArray(value)) return null;
return value as Record<string, unknown>; return value as Record<string, unknown>;
@@ -230,7 +224,18 @@ function extractResultsBatch(payload: unknown): RawHiringCafeJob[] {
} }
const payloadRecord = asRecord(payload); const payloadRecord = asRecord(payload);
const results = payloadRecord?.results; if (!payloadRecord) return [];
const pageProps = asRecord(payloadRecord.pageProps);
const ssrHits = pageProps?.ssrHits ?? payloadRecord.ssrHits;
if (Array.isArray(ssrHits)) {
return ssrHits.filter(
(item): item is RawHiringCafeJob =>
Boolean(item) && typeof item === "object" && !Array.isArray(item),
);
}
const results = payloadRecord.results;
if (!Array.isArray(results)) return []; if (!Array.isArray(results)) return [];
return results.filter( return results.filter(
@@ -242,7 +247,22 @@ function extractResultsBatch(payload: unknown): RawHiringCafeJob[] {
function parseTotalCount(payload: unknown): number | null { function parseTotalCount(payload: unknown): number | null {
const payloadRecord = asRecord(payload); const payloadRecord = asRecord(payload);
if (!payloadRecord) return null; if (!payloadRecord) return null;
return toNumberOrNull(payloadRecord.total);
const pageProps = asRecord(payloadRecord.pageProps);
return (
toNumberOrNull(pageProps?.ssrTotalCount) ??
toNumberOrNull(payloadRecord.ssrTotalCount) ??
toNumberOrNull(payloadRecord.total)
);
}
function isLastSsrPage(payload: unknown): boolean {
const payloadRecord = asRecord(payload);
if (!payloadRecord) return true;
const pageProps = asRecord(payloadRecord.pageProps);
const isLastPage = pageProps?.ssrIsLastPage ?? payloadRecord.ssrIsLastPage;
return isLastPage === true;
} }
function buildCityLocationId(input: string): string { function buildCityLocationId(input: string): string {
@@ -501,23 +521,37 @@ function createCitySearchState(args: {
}; };
} }
async function callHiringCafeApi( async function readBuildIdFromPage(page: Page): Promise<string> {
page: Page, const buildId = await page.evaluate(() => {
endpoint: string, const element = document.querySelector("#__NEXT_DATA__");
params: Record<string, string>, if (!element?.textContent) return null;
): Promise<unknown> { const data = JSON.parse(element.textContent) as { buildId?: string };
const response = await page.evaluate( return data.buildId ?? null;
async ({ endpointArg, paramsArg }) => { });
const url = new URL(endpointArg, window.location.origin); if (!buildId) {
for (const [key, value] of Object.entries(paramsArg)) { throw new Error("Hiring Cafe page did not expose Next.js buildId");
url.searchParams.set(key, value); }
return buildId;
} }
const res = await fetch(url.toString(), { async function fetchSsrSearchPage(
method: "GET", page: Page,
buildId: string,
searchState: Record<string, unknown>,
pageNo: number,
): Promise<unknown> {
const response = await page.evaluate(
async ({ buildIdArg, searchStateArg, pageNoArg }) => {
const params = new URLSearchParams();
params.set("searchState", JSON.stringify(searchStateArg));
params.set("page", String(pageNoArg));
const url = `/_next/data/${buildIdArg}/index.json?${params.toString()}`;
const res = await fetch(url, {
credentials: "include", credentials: "include",
headers: { headers: {
Accept: "application/json, text/plain, */*", Accept: "application/json",
"x-nextjs-data": "1",
}, },
}); });
@@ -539,7 +573,7 @@ async function callHiringCafeApi(
return output; return output;
}, },
{ endpointArg: endpoint, paramsArg: params }, { buildIdArg: buildId, searchStateArg: searchState, pageNoArg: pageNo },
); );
const result = response as BrowserApiResponse; const result = response as BrowserApiResponse;
@@ -547,14 +581,14 @@ async function callHiringCafeApi(
if (!result.ok) { if (!result.ok) {
const snippet = result.responseText.slice(0, 250); const snippet = result.responseText.slice(0, 250);
throw new Error( throw new Error(
`Hiring Cafe API ${endpoint} failed (${result.status} ${result.statusText}): ${snippet}`, `Hiring Cafe SSR search failed (${result.status} ${result.statusText}): ${snippet}`,
); );
} }
if (result.data === null) { if (result.data === null) {
const snippet = result.responseText.slice(0, 250); const snippet = result.responseText.slice(0, 250);
throw new Error( throw new Error(
`Hiring Cafe API ${endpoint} returned non-JSON response: ${snippet}`, `Hiring Cafe SSR search returned non-JSON response: ${snippet}`,
); );
} }
@@ -626,6 +660,8 @@ async function run(): Promise<void> {
await initializePage(); await initializePage();
} }
const buildId = await readBuildIdFromPage(page);
const countryLocation = resolveHiringCafeCountryLocation(country); const countryLocation = resolveHiringCafeCountryLocation(country);
const countryLong = const countryLong =
countryLocation?.address_components[0]?.long_name ?? "United Kingdom"; countryLocation?.address_components[0]?.long_name ?? "United Kingdom";
@@ -664,40 +700,23 @@ async function run(): Promise<void> {
dateFetchedPastNDays, dateFetchedPastNDays,
workplaceTypes, workplaceTypes,
}); });
const encodedSearchState = encodeSearchState(searchState); let termTarget = maxJobsPerTerm;
let totalAvailable: number | null = null;
try {
const countPayload = await callHiringCafeApi(
page,
"/api/search-jobs/get-total-count",
{
s: encodedSearchState,
},
);
totalAvailable = parseTotalCount(countPayload);
} catch (error) {
const message = error instanceof Error ? error.message : String(error);
console.warn(
`Hiring Cafe count request failed for term '${searchTerm}': ${message}`,
);
}
const termTarget =
totalAvailable !== null
? Math.min(maxJobsPerTerm, totalAvailable)
: maxJobsPerTerm;
let pageNo = 0; let pageNo = 0;
let termCollected = 0; let termCollected = 0;
while (termCollected < termTarget && pageNo < PAGE_LIMIT) { while (termCollected < termTarget && pageNo < PAGE_LIMIT) {
const size = Math.min(1000, termTarget - termCollected); const jobsPayload = await fetchSsrSearchPage(
const jobsPayload = await callHiringCafeApi(page, "/api/search-jobs", { page,
size: String(size), buildId,
page: String(pageNo), searchState,
s: encodedSearchState, pageNo,
}); );
const totalAvailable = parseTotalCount(jobsPayload);
if (totalAvailable !== null) {
termTarget = Math.min(maxJobsPerTerm, totalAvailable);
}
const batch = extractResultsBatch(jobsPayload); const batch = extractResultsBatch(jobsPayload);
if (batch.length === 0) break; if (batch.length === 0) break;
@@ -727,7 +746,7 @@ async function run(): Promise<void> {
totalCollected: termCollected, totalCollected: termCollected,
}); });
if (batch.length < size) break; if (isLastSsrPage(jobsPayload)) break;
pageNo += 1; pageNo += 1;
} }