feat: minor improvements to clearbit, diffbot, perigon, and proxycurl

src/services/clearbit-client.ts

@@ -2,7 +2,13 @@ import defaultKy from 'ky'
 import pThrottle from 'p-throttle'
 
 import type { DeepNullable, KyInstance } from '../types.js'
-import { assert, delay, getEnv, throttleKy } from '../utils.js'
+import {
+  assert,
+  delay,
+  getEnv,
+  sanitizeSearchParams,
+  throttleKy
+} from '../utils.js'
 
 export namespace clearbit {
   // Allow up to 600 requests per minute by default.
@@ -11,6 +17,8 @@ export namespace clearbit {
     interval: 60 * 1000
   })
 
+  export const MAX_PAGE_SIZE = 100
+
   export interface CompanyEnrichmentOptions {
     domain: string
     webhook_url?: string
@@ -358,19 +366,8 @@ export namespace clearbit {
     role: string
     seniority: string
   }
-}
 
-/**
+  export const PersonRoles = [
- * The Clearbit API helps with resolving and enriching people and company data.
- *
- * @see https://dashboard.clearbit.com/docs
- */
-export class ClearbitClient {
-  protected readonly ky: KyInstance
-  protected readonly apiKey: string
-  protected readonly _maxPageSize = 100
-
-  static readonly PersonRoles = [
     'communications',
     'customer_service',
     'education',
@@ -391,7 +388,7 @@ export class ClearbitClient {
     'sales'
   ]
 
-  static readonly SenioritiesV2 = [
+  export const SenioritiesV2 = [
     'Executive',
     'VP',
     'Owner',
@@ -402,9 +399,9 @@ export class ClearbitClient {
     'Entry'
   ]
 
-  static readonly Seniorities = ['executive', 'director', 'manager']
+  export const Seniorities = ['executive', 'director', 'manager']
 
-  static readonly SubIndustries = [
+  export const SubIndustries = [
     'Automotive',
     'Consumer Discretionary',
     'Consumer Goods',
@@ -516,6 +513,16 @@ export class ClearbitClient {
     'Energy',
     'Utilities'
   ]
+}
+
+/**
+ * The Clearbit API helps with resolving and enriching people and company data.
+ *
+ * @see https://dashboard.clearbit.com/docs
+ */
+export class ClearbitClient {
+  protected readonly ky: KyInstance
+  protected readonly apiKey: string
 
   constructor({
     apiKey = getEnv('CLEARBIT_API_KEY'),
@@ -573,14 +580,13 @@ export class ClearbitClient {
   async prospectorPeopleV2(options: clearbit.PeopleSearchOptionsV2) {
     return this.ky
       .get('https://prospector.clearbit.com/v2/people/search', {
-        // @ts-expect-error location is a string[] and searchparams shows a TS error heres
+        searchParams: sanitizeSearchParams({
-        searchParams: {
           ...options,
           page_size: Math.min(
-            this._maxPageSize,
+            clearbit.MAX_PAGE_SIZE,
-            options.page_size || this._maxPageSize
+            options.page_size || clearbit.MAX_PAGE_SIZE
           )
-        }
+        })
       })
       .json<clearbit.ProspectorResponseV2>()
   }
@@ -588,15 +594,14 @@ export class ClearbitClient {
   async prospectorPeopleV1(options: clearbit.PeopleSearchOptionsV1) {
     return this.ky
       .get('https://prospector.clearbit.com/v1/people/search', {
-        // @ts-expect-error location is a string[] and searchparams shows a TS error heres
+        searchParams: sanitizeSearchParams({
-        searchParams: {
           email: false,
           ...options,
           page_size: Math.min(
-            this._maxPageSize,
+            clearbit.MAX_PAGE_SIZE,
-            options.page_size || this._maxPageSize
+            options.page_size || clearbit.MAX_PAGE_SIZE
           )
-        }
+        })
       })
       .json<clearbit.ProspectorResponseV1>()
   }

src/services/diffbot-client.ts

@@ -286,7 +286,7 @@ export namespace diffbot {
       .optional()
       .describe('Name of the entity'),
     url: z
-      .array(z.string())
+      .union([z.string(), z.array(z.string())])
       .optional()
       .describe('Origin or homepage URL of the entity'),
     phone: z.string().optional().describe('Phone number of the entity'),

src/services/perigon-client.ts

@@ -168,7 +168,7 @@ export namespace perigon {
       'Labels to filter by, could be "Opinion", "Paid-news", "Non-news", etc. If multiple parameters are passed, they will be applied as OR operations.'
     ),
     excludeLabel: z
-      .union([ArticleLabelSchema, z.literal('Low Content')])
+      .array(z.union([ArticleLabelSchema, z.literal('Low Content')]))
       .optional()
       .describe(
         'Exclude results that include specific labels ("Opinion", "Non-news", "Paid News", etc.). You can filter multiple by repeating the parameter.'
@@ -681,19 +681,20 @@ export class PerigonClient extends AIFunctionsProvider {
     })
   })
   async searchArticles(opts: perigon.ArticlesSearchOptions) {
-    return this.ky
+    const searchParams = sanitizeSearchParams({
-      .get('all', {
+      sortBy: 'relevance',
-        searchParams: sanitizeSearchParams({
       ...opts,
       apiKey: this.apiKey,
       size: Math.max(
         1,
-            Math.min(
+        Math.min(perigon.MAX_PAGE_SIZE, opts.size || perigon.DEFAULT_PAGE_SIZE)
-              perigon.MAX_PAGE_SIZE,
-              opts.size || perigon.DEFAULT_PAGE_SIZE
-            )
       )
     })
+    console.log('perigon.searchArticles', searchParams)
+
+    return this.ky
+      .get('all', {
+        searchParams
       })
       .json<perigon.ArticlesSearchResponse>()
   }
@@ -720,19 +721,20 @@ export class PerigonClient extends AIFunctionsProvider {
     })
   })
   async searchStories(opts: perigon.StoriesSearchOptions) {
-    return this.ky
+    const searchParams = sanitizeSearchParams({
-      .get('stories/all', {
+      sortBy: 'relevance',
-        searchParams: sanitizeSearchParams({
       ...opts,
       apiKey: this.apiKey,
       size: Math.max(
         1,
-            Math.min(
+        Math.min(perigon.MAX_PAGE_SIZE, opts.size || perigon.DEFAULT_PAGE_SIZE)
-              perigon.MAX_PAGE_SIZE,
-              opts.size || perigon.DEFAULT_PAGE_SIZE
-            )
       )
     })
+    console.log('perigon.searchStories', searchParams)
+
+    return this.ky
+      .get('stories/all', {
+        searchParams
       })
       .json<perigon.StoriesSearchResponse>()
   }

src/services/proxycurl-client.ts

@@ -1,3 +1,4 @@
+import type { Simplify } from 'type-fest'
 import defaultKy, { type KyInstance } from 'ky'
 import pThrottle from 'p-throttle'
 import { z } from 'zod'
@@ -52,8 +53,13 @@ export namespace proxycurl {
     typeof CompanyProfileEndpointParamsQueryClassSchema
   >
 
+  /**
+   * Requires one of:
+   * - `facebook_profile_url`
+   * - `linkedin_profile_url`
+   * - `twitter_profile_url`
+   */
   export const PersonProfileEndpointParamsQueryClassSchema = z.object({
-    // requires one of `facebook_profile_url`, `linkedin_profile_url`, or `twitter_profile_url`
     facebook_profile_url: z.string().optional(),
     linkedin_profile_url: z.string().optional(),
     twitter_profile_url: z.string().optional(),
@@ -68,8 +74,8 @@ export namespace proxycurl {
     fallback_to_cache: FallbackToCacheSchema,
     use_cache: UseCacheSchema
   })
-  export type PersonProfileEndpointParamsQueryClass = z.infer<
+  export type PersonProfileEndpointParamsQueryClass = Simplify<
-    typeof PersonProfileEndpointParamsQueryClassSchema
+    z.infer<typeof PersonProfileEndpointParamsQueryClassSchema>
   >
 
   export const PersonLookupEndpointParamsQueryClassSchema = z.object({