feat: add support for indicating workplaceTypes (#296)

extractors/hiringcafe/manifest.ts

@@ -68,6 +68,9 @@ export const manifest: ExtractorManifest = {
         single:
           context.settings.searchCities ?? context.settings.jobspyLocation,
       }),
+      workplaceTypes: context.settings.workplaceTypes
+        ? JSON.parse(context.settings.workplaceTypes)
+        : undefined,
       maxJobsPerTerm,
       onProgress: (event) => {
         if (context.shouldCancel?.()) return;

extractors/hiringcafe/src/default-search-state.ts

@@ -35,10 +35,11 @@ export function createDefaultSearchState(args: {
   searchQuery: string;
   location: HiringCafeCountryLocation | null;
   dateFetchedPastNDays: number;
+  workplaceTypes?: Array<"Remote" | "Hybrid" | "Onsite">;
 }): HiringCafeSearchState {
   return {
     locations: args.location ? [args.location] : [],
-    workplaceTypes: ["Remote", "Hybrid", "Onsite"],
+    workplaceTypes: args.workplaceTypes ?? ["Remote", "Hybrid", "Onsite"],
     defaultToUserLocation: false,
     userLocation: null,
     commitmentTypes: [

extractors/hiringcafe/src/main.ts

@@ -68,6 +68,8 @@ interface NominatimResult {
   address?: Record<string, unknown>;
 }
 
+type HiringCafeWorkplaceType = "Remote" | "Hybrid" | "Onsite";
+
 function emitProgress(payload: Record<string, unknown>): void {
   if (process.env.JOBOPS_EMIT_PROGRESS !== "1") return;
   console.log(`${JOBOPS_PROGRESS_PREFIX}${JSON.stringify(payload)}`);
@@ -79,6 +81,36 @@ function parsePositiveInt(input: string | undefined, fallback: number): number {
   return parsed;
 }
 
+function parseWorkplaceTypes(
+  raw: string | undefined,
+): HiringCafeWorkplaceType[] {
+  if (!raw) return ["Remote", "Hybrid", "Onsite"];
+
+  try {
+    const parsed = JSON.parse(raw) as unknown;
+    if (!Array.isArray(parsed)) return ["Remote", "Hybrid", "Onsite"];
+
+    const out: HiringCafeWorkplaceType[] = [];
+    const seen = new Set<HiringCafeWorkplaceType>();
+    for (const value of parsed) {
+      const mapped =
+        value === "remote"
+          ? "Remote"
+          : value === "hybrid"
+            ? "Hybrid"
+            : value === "onsite"
+              ? "Onsite"
+              : null;
+      if (!mapped || seen.has(mapped)) continue;
+      seen.add(mapped);
+      out.push(mapped);
+    }
+    return out.length > 0 ? out : ["Remote", "Hybrid", "Onsite"];
+  } catch {
+    return ["Remote", "Hybrid", "Onsite"];
+  }
+}
+
 function encodeSearchState(searchState: unknown): string {
   const json = JSON.stringify(searchState);
   const urlEncodedJson = encodeURIComponent(json);
@@ -301,6 +333,7 @@ function createCitySearchState(args: {
   searchQuery: string;
   dateFetchedPastNDays: number;
   context: CityLocationContext;
+  workplaceTypes: HiringCafeWorkplaceType[];
 }): Record<string, unknown> {
   return {
     locations: [
@@ -340,7 +373,7 @@ function createCitySearchState(args: {
         },
       },
     ],
-    workplaceTypes: ["Remote", "Hybrid", "Onsite"],
+    workplaceTypes: args.workplaceTypes,
     defaultToUserLocation: true,
     userLocation: null,
     physicalEnvironments: [
@@ -553,6 +586,9 @@ async function run(): Promise<void> {
     process.env.HIRING_CAFE_OUTPUT_JSON ||
     join(__dirname, "../storage/datasets/default/jobs.json");
   const headless = process.env.HIRING_CAFE_HEADLESS !== "false";
+  const workplaceTypes = parseWorkplaceTypes(
+    process.env.HIRING_CAFE_WORKPLACE_TYPES,
+  );
 
   let browser = await firefox.launch(
     await launchOptions({
@@ -620,11 +656,13 @@ async function run(): Promise<void> {
             searchQuery: searchTerm,
             dateFetchedPastNDays,
             context: cityLocationContext,
+            workplaceTypes,
           })
         : createDefaultSearchState({
             searchQuery: searchTerm,
             location: countryLocation,
             dateFetchedPastNDays,
+            workplaceTypes,
           });
       const encodedSearchState = encodeSearchState(searchState);
 

extractors/hiringcafe/src/run.ts

@@ -54,6 +54,7 @@ export interface RunHiringCafeOptions {
   country?: string;
   countryKey?: string;
   locations?: string[];
+  workplaceTypes?: Array<"remote" | "hybrid" | "onsite">;
   locationRadiusMiles?: number;
   maxJobsPerTerm?: number;
   onProgress?: (event: HiringCafeProgressEvent) => void;
@@ -216,6 +217,9 @@ export async function runHiringCafe(
           JOBOPS_EMIT_PROGRESS: "1",
           HIRING_CAFE_SEARCH_TERMS: JSON.stringify(searchTerms),
           HIRING_CAFE_COUNTRY: country,
+          HIRING_CAFE_WORKPLACE_TYPES: JSON.stringify(
+            options.workplaceTypes ?? ["remote", "hybrid", "onsite"],
+          ),
           HIRING_CAFE_MAX_JOBS_PER_TERM: String(maxJobsPerTerm),
           HIRING_CAFE_OUTPUT_JSON: DATASET_PATH,
           HIRING_CAFE_LOCATION_QUERY: strictLocationFilter ? location : "",