Email Finder
Pricing
from $3.12 / 1,000 api requests
Email Finder tool helps you discover verified professional email addresses using just a person's name and their company domain.
Pricing
from $3.12 / 1,000 api requests
Rating
5.0
(1)
Developer
Tomba io
Maintained by CommunityActor stats
3
Bookmarked
90
Total users
2
Monthly active users
2 days ago
Last modified
Categories
Share
1[settings]2idiomatic_version_file_enable_tools = []3
4[tools."github:apify/apify-cli"]5version = "latest"6bin = "apify"7
8[tools."github:apify/apify-cli".platforms]9linux-x64 = { asset_pattern = "apify-*-linux-x64" }10linux-arm64 = { asset_pattern = "apify-*-linux-arm64" }11macos-x64 = { asset_pattern = "apify-*-darwin-x64" }12macos-arm64 = { asset_pattern = "apify-*-darwin-arm64" }13windows-x64 = { asset_pattern = "apify-*-windows-x64.exe" }14windows-arm64 = { asset_pattern = "apify-*-windows-arm64.exe" }.prettierignore{ "printWidth": 120, "singleQuote": true, "tabWidth": 4}{ "extends": "./tsconfig.json", "compilerOptions": { "noEmit": true }, "include": ["./src/**/*", "./test/**/*"]}# configurations.idea.vscode.zed
# crawlee and apify storage foldersapify_storagecrawlee_storagestorage
# installed filesnode_modules
# git folder.git
# dist folderdisttest# This file tells Git which files shouldn't be added to source control
.idea.vscode.zedstorageapify_storagecrawlee_storagenode_modulesdisttsconfig.tsbuildinfo
# Added by Apify CLI.venv1# Development2
3Notes for maintainers of the Tomba Email Finder Actor. The README is the end-user page shown on Apify Store.4
5## Requirements6
7- Node.js 20+8- [Apify CLI](https://docs.apify.com/cli) for deployment9
10## Scripts11
12```bash13npm install14npm run build # compile TypeScript to dist/15npm run lint # ESLint (src and test)16npm run format # Prettier17npm test # unit + end-to-end tests (node:test)18npm start # run locally with tsx19```20
21## Credentials22
23The Actor uses our Tomba account. Credentials come from environment variables, never from the input:24
25| Variable | Description |26| -------------------- | -------------------------------------------------- |27| `TOMBA_API_KEY` | Tomba API key (`ta_…`) |28| `TOMBA_API_SECRET` | Tomba secret (`ts_…`) |29| `TOMBA_API_ENDPOINT` | Optional API base URL; only used by the test suite |3031`.actor/actor.json` maps the variables to Apify secrets:32
33```bash34apify secrets add tombaApiKey ta_xxxxxxxxxxxxxxxxxxxx35apify secrets add tombaApiSecret ts_xxxxxxxxxxxxxxxxxxxx36apify push37```38
39Run locally:40
41```bash42TOMBA_API_KEY=ta_… TOMBA_API_SECRET=ts_… npm start43```44
45## Pricing (pay per event)46
47In **Apify Console → Publication → Monetization**, choose **Pay per event** and add:48
49| Event | Price | Charged when |50| --------------- | -------- | --------------------------------------------- |51| `tomba-request` | $0.00312 | Tomba returns a billable response (see below) |5253Every Tomba credit is one `tomba-request` event. The count follows Tomba's published rule for Email Finder, **1 search credit, or 6 with phone data**:54
55| Billable response | Events (`count`) |56| ----------------------------------------------------------------- | ----------------------- |57| Without `enrichMobile` | 1 |58| `enrichMobile=true` and `data.phone_data` has at least one number | 6 (1 + `PHONE_CREDITS`) |59| `enrichMobile=true` and `data.phone_data` missing or empty | 1 |6061`main.ts` passes `count = (body) => 1 + (hasPhoneData(body.data) ? PHONE_CREDITS : 0)` to `callTomba()`. Each item reports the events actually charged in `chargedCredits` (`res.chargedCount ?? 0`, so 0 for cache hits and errors) and the number of phones in `phoneNumbers`.62
63`isBillable()` in `src/tomba.ts` mirrors Tomba's billing:64
65| Tomba outcome | Charged |66| ------------------------------------------------------ | ------- |67| JSON with non-empty `data`, including negative answers | Yes |68| Error status (4xx, 5xx, including 422 and 429) | No |69| Success with empty or null `data` | No |70| Success with an `errors` object | No |71| Non-JSON body (reported as 502) | No |72| Cache hit | No |7374A negative answer here is a `data` object with `email: null` (Tomba knows the person but has no email); it is charged. A request needs (`domain` or `company`) and (`fullName` or `firstName` + `lastName`); otherwise it is never sent to Tomba and produces a free error item.75
76## Architecture77
78- `src/tomba.ts`: shared helper, identical in every Tomba Actor. It handles credentials, caching (per-Actor `tomba-cache-<actorId>` key-value store; falls back to an in-run cache if it can't be opened), retries with exponential backoff, pay-per-event charging, budget reservation, the concurrency pool and resume state.79- `src/main.ts`: Actor-specific input handling and output mapping. Requests are normalized (`normalizeDomain`, trimmed and whitespace-collapsed `company`/`fullName`/`firstName`/`lastName`) and deduplicated on `domain|company|fullName|firstName|lastName` (case-insensitive). `maxResults` limits the number of requests processed. Each request calls `Finder.emailFinder({ domain, company, fullName, firstName, lastName, enrichMobile, webhook_url })` (`GET /email-finder?domain=&company=&full_name=&first_name=&last_name=&enrich_mobile=&webhook_url=`); only non-empty fields are sent, `enrich_mobile=true` only when `enrichMobile` is on and `webhook_url` only when `webhookUrl` is set. These parameters are also the cache key, so results with and without phone data are cached separately. The output item is Tomba's `data` object spread (including `phone_data` when returned), plus the input echo (`domain`, `firstName`, `lastName`, and `fullName` / `inputCompany` when given; the input company is `inputCompany` because `company` is Tomba's field), `source`, `phoneNumbers`, `charged`, `chargedCredits`, `cached` and, when there is no result, `error`.80- The `tomba` SDK v1.1.1 resolves every call to `{ data, rateLimit }`, where `data` is the response body. Its `.d.ts` types still declare the old return type, so always go through `callTomba()`.81
82## Tests83
84- `test/tomba.test.ts`: unit tests for the shared helper (identical in every Actor)85- `test/main.test.ts`: end-to-end tests that run `src/main.ts` against a local mock Tomba API86- `test/helpers.ts`: mock server and Actor runner (identical in every Actor)87
88Locally, the Apify SDK prices every event at $1 when `ACTOR_TEST_PAY_PER_EVENT=true`, so the tests use `maxTotalChargeUsd` as an event count.89
90## Standby mode (real-time API)91
92`.actor/actor.json` sets `usesStandbyMode: true` and `webServerSchema: ./web_server_schema.json` (OpenAPI 3).93
94- `src/standby.ts` (shared, identical in every Actor): `runActor()` runs a batch job, or, when `APIFY_META_ORIGIN=STANDBY`, starts an HTTP server on `Actor.config.get('containerPort')`.95 - `GET /` with the `x-apify-container-server-readiness-probe` header, or with no query: readiness / usage.96 - `GET /?…`: one request built by `fromQuery()` from `domain`, `company`, `firstName` (`first_name`), `lastName` (`last_name`), `fullName` (`full_name`), plus `enrichMobile`, `webhookUrl` and `maxResults`.97 - `POST /`: the same JSON input as a batch run (use it for several people).98 - Responses: `200 { items }`, `400` invalid input, `402` max charge limit reached, `404`, `405`.99- `run(input, ctx)` in `src/main.ts` is shared by both modes: `ctx.push()` writes to the dataset in batch runs and to the HTTP response in Standby; `ctx.isDone()`/`ctx.markDone()` persist resume state only in batch runs. Input errors throw `InputError` (failed run in batch, `400` in Standby).100- Caching and pay-per-event charging work the same in both modes.101
102Try it locally:103
104```bash105APIFY_META_ORIGIN=STANDBY ACTOR_WEB_SERVER_PORT=8080 TOMBA_API_KEY=ta_… TOMBA_API_SECRET=ts_… npm start106curl "localhost:8080/?domain=stripe.com&firstName=John&lastName=Doe"107```108
109## Key-value store schema110
111`.actor/key_value_store_schema.json` documents the default key-value store records (`INPUT`, `TOMBA_STATE`). The cross-run cache lives in the separate named store `tomba-cache-<actorId>`, one per Actor: under limited permissions an Actor can only open named storages it created itself, so the Tomba Actors must not share one store. If the store can't be opened, the run logs a warning and caches for this run only.112
113## Memory114
115`defaultMemoryMbytes` is 256: the Actor only makes HTTP calls, so more memory just costs more.1import prettier from 'eslint-config-prettier';2
3import apify from '@apify/eslint-config/ts.js';4import globals from 'globals';5import tsEslint from 'typescript-eslint';6
7// eslint-disable-next-line import/no-default-export8export default [9 { ignores: ['**/dist', 'eslint.config.mjs'] },10 ...apify,11 prettier,12 {13 languageOptions: {14 parser: tsEslint.parser,15 parserOptions: {16 project: 'tsconfig.eslint.json',17 },18 globals: {19 ...globals.node,20 ...globals.jest,21 },22 },23 plugins: {24 '@typescript-eslint': tsEslint.plugin,25 },26 rules: {27 'no-console': 0,28 },29 },30 {31 // node:test's describe/it return promises by design.32 files: ['test/**/*.ts'],33 rules: {34 '@typescript-eslint/no-floating-promises': 0,35 },36 },37];root = true
[*]indent_style = spaceindent_size = 4charset = utf-8trim_trailing_whitespace = trueinsert_final_newline = trueend_of_line = lf{ "extends": "@apify/tsconfig", "compilerOptions": { "module": "NodeNext", "moduleResolution": "NodeNext", "target": "ES2022", "outDir": "dist", "noUnusedLocals": false, "skipLibCheck": true, "lib": ["DOM"] }, "include": ["./src/**/*"]}{ "name": "email-finder", "version": "1.0.0", "type": "module", "description": "Tomba Email Finder Actor for discovering email addresses by combining domain + first name + last name", "engines": { "node": ">=18.0.0" }, "dependencies": { "apify": "^3.7.2", "tomba": "^1.1.1" }, "devDependencies": { "@apify/eslint-config": "^1.0.0", "@apify/tsconfig": "^0.1.1", "@types/node": "^22.15.32", "ajv": "^8.20.0", "eslint": "^9.29.0", "eslint-config-prettier": "^10.1.5", "globals": "^16.2.0", "prettier": "^3.5.3", "standard-version": "^9.5.0", "tsx": "^4.20.3", "typescript": "^5.8.3", "typescript-eslint": "^8.34.1" }, "scripts": { "start": "npm run start:dev", "start:prod": "node dist/main.js", "start:dev": "tsx src/main.ts", "build": "tsc", "lint": "eslint src test", "lint:fix": "eslint src test --fix", "format": "prettier --write .", "format:check": "prettier --check .", "release": "standard-version && git push --follow-tags origin main", "test": "node --import tsx --test test/*.test.ts" }, "author": "It's not you it's me", "license": "ISC"}# Specify the base Docker image. You can read more about# the available images at https://docs.apify.com/sdk/js/docs/guides/docker-images# You can also use any other image from Docker Hub.FROM apify/actor-node:22 AS builder
# Check preinstalled packagesRUN npm ls crawlee apify puppeteer playwright
# Copy just package.json and package-lock.json# to speed up the build using Docker layer cache.COPY package*.json ./
# Install all dependencies. Don't audit to speed up the installation.RUN npm install --include=dev --audit=false
# Next, copy the source files using the user set# in the base image.COPY . ./
# Install all dependencies and build the project.# Don't audit to speed up the installation.RUN npm run build
# Create final imageFROM apify/actor-node:22
# Check preinstalled packagesRUN npm ls crawlee apify puppeteer playwright
# Copy just package.json and package-lock.json# to speed up the build using Docker layer cache.COPY package*.json ./
# Install NPM packages, skip optional and development dependencies to# keep the image small. Avoid logging too much and print the dependency# tree for debuggingRUN npm --quiet set progress=false \ && npm install --omit=dev --omit=optional \ && echo "Installed NPM packages:" \ && (npm list --omit=dev --all || true) \ && echo "Node.js version:" \ && node --version \ && echo "NPM version:" \ && npm --version \ && rm -r ~/.npm
# Copy built JS files from builder imageCOPY /usr/src/app/dist ./dist
# Next, copy the remaining files and directories with the source code.# Since we do this after NPM install, quick build will be really fast# for most source file changes.COPY . ./
# Run the image.CMD npm run start:prod --silent{ "actorSpecification": 1, "fields": { "$schema": "http://json-schema.org/draft-07/schema#", "type": "object", "properties": { "email": { "type": ["string", "null"], "description": "Author's email address" }, "first_name": { "type": ["string", "null"], "description": "Author's first name" }, "last_name": { "type": ["string", "null"], "description": "Author's last name" }, "full_name": { "type": ["string", "null"], "description": "Author's full name" }, "website_url": { "type": ["string", "null"], "description": "Website URL associated with the author" }, "company": { "type": ["string", "null"], "description": "Author's company or organization" }, "position": { "type": ["string", "null"], "description": "Author's job title or position" }, "country": { "type": ["string", "null"], "description": "Author's country" }, "gender": { "type": ["string", "null"], "description": "Author's gender" }, "twitter": { "type": ["string", "null"], "description": "Author's Twitter profile URL" }, "linkedin": { "type": ["string", "null"], "description": "Author's LinkedIn profile URL" }, "score": { "type": ["number", "null"], "description": "Confidence score for the author information (0-100)" }, "accept_all": { "type": ["boolean", "null"], "description": "Whether the domain accepts all emails" }, "phone_number": { "type": ["boolean", "string", "null"], "description": "Whether a phone number is available for this author" }, "info": { "type": ["object", "null"], "description": "Additional detailed information about the author and source", "properties": { "title": { "type": ["string", "null"], "description": "Page title where the author was found" }, "description": { "type": ["string", "null"], "description": "Page description or meta description" }, "first_name": { "type": ["string", "null"], "description": "First name from page metadata" }, "last_name": { "type": ["string", "null"], "description": "Last name from page metadata" }, "full_name": { "type": ["string", "null"], "description": "Full name from page metadata" }, "article_domain": { "type": ["string", "null"], "description": "Domain of the article or page" }, "same_as_domain": { "type": ["string", "null"], "description": "Whether the domain matches expected domain" }, "linkedin": { "type": ["string", "null"], "description": "LinkedIn handle from page metadata" }, "twitter": { "type": ["string", "null"], "description": "Twitter handle from page metadata" }, "phone": { "type": ["string", "null"], "description": "Phone number from page metadata" }, "email": { "type": ["string", "null"], "description": "Email from page metadata" }, "gender": { "type": ["string", "null"], "description": "Gender from page metadata" }, "position": { "type": ["string", "null"], "description": "Position from page metadata" }, "image": { "type": ["string", "null"], "description": "Author image URL from page metadata" }, "gravatar": { "type": ["boolean", "null"], "description": "Whether author has a Gravatar" }, "emails": { "type": ["array", "null"], "description": "Array of additional emails found", "items": { "type": "string" } } } }, "verification": { "type": ["object", "null"], "description": "Email verification information from Tomba", "properties": { "date": { "type": ["string", "null"], "description": "Date when the email was verified (ISO format)" }, "status": { "type": ["string", "null"], "description": "Verification status (valid, invalid, unknown)" } } }, "sources": { "type": ["array", "null"], "description": "Sources where the author information was found", "items": { "type": "object", "properties": { "uri": { "type": ["string", "null"], "description": "URI where the email was found" }, "website_url": { "type": ["string", "null"], "description": "Website URL where the email was found" }, "extracted_on": { "type": ["string", "null"], "description": "Date when the email was extracted (ISO format)" }, "last_seen_on": { "type": ["string", "null"], "description": "Date when the email was last seen (ISO format)" }, "still_on_page": { "type": ["boolean", "null"], "description": "Whether the email is still on the page" } } } }, "phone_data": { "type": ["array", "null"], "description": "Phone numbers found for the person (only when `enrichMobile` is on)", "items": { "type": "object", "properties": { "number": { "type": ["string", "null"], "description": "Phone number" }, "type": { "type": ["string", "null"], "description": "Phone type (for example mobile)" } } } }, "phoneNumbers": { "type": "integer", "description": "Number of phone numbers returned in `phone_data`" }, "charged": { "type": "boolean", "description": "Whether this request was charged (pay-per-event)" }, "chargedCredits": { "type": "integer", "description": "Number of Tomba credits charged for this item (0 when free or cached)" }, "cached": { "type": "boolean", "description": "Whether the result came from the cache (free)" }, "domain": { "type": ["string", "null"], "description": "Input company domain (normalized)" }, "firstName": { "type": ["string", "null"], "description": "Input first name" }, "lastName": { "type": ["string", "null"], "description": "Input last name" }, "fullName": { "type": ["string", "null"], "description": "Input full name (when provided)" }, "inputCompany": { "type": ["string", "null"], "description": "Input company name (when provided)" }, "source": { "type": "string", "description": "Data source identifier (tomba_email_finder)" }, "error": { "type": ["string", "null"], "description": "Error message if the search failed" } }, "required": ["domain", "source"] }, "views": { "overview": { "title": "Overview", "transformation": { "fields": ["email", "full_name", "position", "company", "website_url", "verification.status", "score"] }, "display": { "component": "table", "properties": { "email": { "label": "Email" }, "full_name": { "label": "Full Name" }, "position": { "label": "Position" }, "company": { "label": "Company" }, "website_url": { "label": "Website" }, "verification.status": { "label": "Email Status" }, "score": { "label": "Confidence Score" } } } }, "detailed": { "title": "Detailed View", "transformation": { "fields": [ "domain", "firstName", "lastName", "fullName", "inputCompany", "email", "first_name", "last_name", "full_name", "position", "company", "website_url", "country", "gender", "twitter", "linkedin", "score", "accept_all", "phone_number", "phone_data", "phoneNumbers", "verification", "sources", "source", "error", "charged", "chargedCredits", "cached" ] }, "display": { "component": "table" } }, "sources": { "title": "Source Analysis", "transformation": { "fields": ["email", "full_name", "sources"] }, "display": { "component": "table", "properties": { "email": { "label": "Email" }, "full_name": { "label": "Author" }, "sources": { "label": "Sources Found" } } } } }}{ "openapi": "3.0.0", "info": { "title": "Tomba Email Finder API", "version": "1.0.0", "description": "Real-time API (Apify Standby mode). Find a person's email address (and optionally phone numbers) from a company domain or name and their name. $0.00312 per credit: a lookup with an answer costs 1 credit ($0.00312), or 6 credits ($0.01872) when `enrichMobile` is on and phone numbers are returned. Negative answers (person found, no email) are charged; empty results, errors, incomplete requests and cached results are free." }, "paths": { "/": { "get": { "summary": "Find the email of one person", "description": "Without query parameters, returns a short usage note (also used as the readiness probe).", "parameters": [ { "name": "domain", "in": "query", "description": "Company domain. Required unless `company` is set. URLs are cleaned up.", "required": false, "schema": { "type": "string" }, "example": "stripe.com" }, { "name": "company", "in": "query", "description": "Company name. Alternative to `domain`.", "required": false, "schema": { "type": "string" }, "example": "Stripe" }, { "name": "firstName", "in": "query", "description": "Person's first name. Required with `lastName` unless `fullName` is set.", "required": false, "schema": { "type": "string" }, "example": "John" }, { "name": "first_name", "in": "query", "description": "Alias of `firstName`.", "required": false, "schema": { "type": "string" } }, { "name": "lastName", "in": "query", "description": "Person's last name. Required with `firstName` unless `fullName` is set.", "required": false, "schema": { "type": "string" }, "example": "Doe" }, { "name": "last_name", "in": "query", "description": "Alias of `lastName`.", "required": false, "schema": { "type": "string" } }, { "name": "fullName", "in": "query", "description": "Person's full name. Alternative to `firstName` + `lastName`.", "required": false, "schema": { "type": "string" }, "example": "John Doe" }, { "name": "full_name", "in": "query", "description": "Alias of `fullName`.", "required": false, "schema": { "type": "string" } }, { "name": "enrichMobile", "in": "query", "description": "Also return the person's phone numbers in `phone_data` (6 credits instead of 1 when phone data is returned).", "required": false, "schema": { "type": "boolean", "default": false }, "example": false }, { "name": "webhookUrl", "in": "query", "description": "Optional URL that Tomba notifies when a result is ready (sent to Tomba as `webhook_url`).", "required": false, "schema": { "type": "string", "pattern": "^https?://" } }, { "name": "maxResults", "in": "query", "description": "Maximum number of requests to process (default 50).", "required": false, "schema": { "type": "integer", "minimum": 1, "default": 50 } } ], "responses": { "200": { "$ref": "#/components/responses/Results" }, "400": { "$ref": "#/components/responses/Error" }, "402": { "$ref": "#/components/responses/Error" } } }, "post": { "summary": "Find emails using a JSON input", "description": "Accepts the same JSON input as a normal Actor run.", "requestBody": { "required": true, "content": { "application/json": { "schema": { "type": "object", "properties": { "requests": { "type": "array", "description": "List of people to find. Each request needs a company (`domain` or `company` name) and a name (`fullName`, or `firstName` and `lastName`).", "items": { "type": "object", "properties": { "domain": { "type": "string", "description": "Company domain name (e.g., 'stripe.com'). Required unless `company` is set" }, "company": { "type": "string", "description": "Company name (e.g., 'Stripe'). Alternative to `domain`" }, "firstName": { "type": "string", "description": "Person's first name. Required with `lastName` unless `fullName` is set" }, "lastName": { "type": "string", "description": "Person's last name. Required with `firstName` unless `fullName` is set" }, "fullName": { "type": "string", "description": "Person's full name (e.g., 'John Doe'). Alternative to `firstName` + `lastName`" } }, "additionalProperties": false }, "minItems": 1, "maxItems": 1000, "example": [ { "domain": "stripe.com", "firstName": "John", "lastName": "Doe" }, { "company": "Shopify", "fullName": "Jane Smith" } ] }, "maxResults": { "type": "integer", "description": "Maximum number of requests to process (default: 50, max: 1000)", "default": 50, "minimum": 1, "maximum": 1000, "example": 50 }, "enrichMobile": { "type": "boolean", "description": "Also return the person's phone numbers in `phone_data`. A result with phone data costs 6 credits instead of 1; results without phone data still cost 1 credit.", "default": false }, "webhookUrl": { "type": "string", "description": "Optional URL that Tomba notifies when a result is ready (sent to Tomba as `webhook_url`).", "pattern": "^https?://", "example": "https://example.com/tomba-webhook" } }, "required": ["requests"] } } } }, "responses": { "200": { "$ref": "#/components/responses/Results" }, "400": { "$ref": "#/components/responses/Error" }, "402": { "$ref": "#/components/responses/Error" } } } } }, "components": { "responses": { "Results": { "description": "Email finder results, in the same format as the dataset items.", "content": { "application/json": { "schema": { "type": "object", "properties": { "items": { "type": "array", "items": { "$ref": "#/components/schemas/EmailFinderResult" } } } } } } }, "Error": { "description": "Invalid input (400) or the run's max charge limit was reached (402).", "content": { "application/json": { "schema": { "type": "object", "properties": { "error": { "type": "string" } } } } } } }, "schemas": { "EmailFinderResult": { "type": "object", "properties": { "email": { "type": "string", "nullable": true, "description": "Author's email address" }, "first_name": { "type": "string", "nullable": true, "description": "Author's first name" }, "last_name": { "type": "string", "nullable": true, "description": "Author's last name" }, "full_name": { "type": "string", "nullable": true, "description": "Author's full name" }, "website_url": { "type": "string", "nullable": true, "description": "Website URL associated with the author" }, "company": { "type": "string", "nullable": true, "description": "Author's company or organization" }, "position": { "type": "string", "nullable": true, "description": "Author's job title or position" }, "country": { "type": "string", "nullable": true, "description": "Author's country" }, "gender": { "type": "string", "nullable": true, "description": "Author's gender" }, "twitter": { "type": "string", "nullable": true, "description": "Author's Twitter profile URL" }, "linkedin": { "type": "string", "nullable": true, "description": "Author's LinkedIn profile URL" }, "score": { "type": "number", "nullable": true, "description": "Confidence score for the author information (0-100)" }, "accept_all": { "type": "boolean", "nullable": true, "description": "Whether the domain accepts all emails" }, "phone_number": { "type": "boolean", "nullable": true, "description": "Whether a phone number is available for this author" }, "info": { "type": "object", "nullable": true, "description": "Additional detailed information about the author and source", "properties": { "title": { "type": "string", "nullable": true, "description": "Page title where the author was found" }, "description": { "type": "string", "nullable": true, "description": "Page description or meta description" }, "first_name": { "type": "string", "nullable": true, "description": "First name from page metadata" }, "last_name": { "type": "string", "nullable": true, "description": "Last name from page metadata" }, "full_name": { "type": "string", "nullable": true, "description": "Full name from page metadata" }, "article_domain": { "type": "string", "nullable": true, "description": "Domain of the article or page" }, "same_as_domain": { "type": "string", "nullable": true, "description": "Whether the domain matches expected domain" }, "linkedin": { "type": "string", "nullable": true, "description": "LinkedIn handle from page metadata" }, "twitter": { "type": "string", "nullable": true, "description": "Twitter handle from page metadata" }, "phone": { "type": "string", "nullable": true, "description": "Phone number from page metadata" }, "email": { "type": "string", "nullable": true, "description": "Email from page metadata" }, "gender": { "type": "string", "nullable": true, "description": "Gender from page metadata" }, "position": { "type": "string", "nullable": true, "description": "Position from page metadata" }, "image": { "type": "string", "nullable": true, "description": "Author image URL from page metadata" }, "gravatar": { "type": "boolean", "nullable": true, "description": "Whether author has a Gravatar" }, "emails": { "type": "array", "nullable": true, "description": "Array of additional emails found", "items": { "type": "string" } } } }, "verification": { "type": "object", "nullable": true, "description": "Email verification information from Tomba", "properties": { "date": { "type": "string", "nullable": true, "description": "Date when the email was verified (ISO format)" }, "status": { "type": "string", "nullable": true, "description": "Verification status (valid, invalid, unknown)" } } }, "sources": { "type": "array", "nullable": true, "description": "Sources where the author information was found", "items": { "type": "object", "properties": { "uri": { "type": "string", "nullable": true, "description": "URI where the email was found" }, "website_url": { "type": "string", "nullable": true, "description": "Website URL where the email was found" }, "extracted_on": { "type": "string", "nullable": true, "description": "Date when the email was extracted (ISO format)" }, "last_seen_on": { "type": "string", "nullable": true, "description": "Date when the email was last seen (ISO format)" }, "still_on_page": { "type": "boolean", "nullable": true, "description": "Whether the email is still on the page" } } } }, "phone_data": { "type": "array", "nullable": true, "description": "Phone numbers found for the person (only when `enrichMobile` is on)", "items": { "type": "object", "properties": { "number": { "type": "string", "nullable": true, "description": "Phone number" }, "type": { "type": "string", "nullable": true, "description": "Phone type (for example mobile)" } } } }, "phoneNumbers": { "type": "integer", "description": "Number of phone numbers returned in `phone_data`" }, "charged": { "type": "boolean", "description": "Whether this request was charged (pay-per-event)" }, "chargedCredits": { "type": "integer", "description": "Number of Tomba credits charged for this item (0 when free or cached)" }, "cached": { "type": "boolean", "description": "Whether the result came from the cache (free)" }, "domain": { "type": "string", "nullable": true, "description": "Input company domain (normalized)" }, "firstName": { "type": "string", "nullable": true, "description": "Input first name" }, "lastName": { "type": "string", "nullable": true, "description": "Input last name" }, "fullName": { "type": "string", "nullable": true, "description": "Input full name (when provided)" }, "inputCompany": { "type": "string", "nullable": true, "description": "Input company name (when provided)" }, "source": { "type": "string", "description": "Data source identifier (tomba_email_finder)" }, "error": { "type": "string", "nullable": true, "description": "Error message if the search failed" } } } } }}{ "actorOutputSchemaVersion": 1, "title": "Email Finder Output", "description": "Output from Tomba Email Finder Actor - discovered email addresses with contact information", "type": "object", "properties": { "results": { "type": "string", "title": "Email Discovery Results", "description": "Found email addresses with contact details and verification status", "template": "{{links.apiDefaultDatasetUrl}}/items" } }}{ "actorSpecification": 1, "name": "email-finder", "title": "Tomba Email Finder", "description": "Find the most likely email address (and optionally phone numbers) of a person from a company domain or name and their full name or first and last name using the Tomba API.", "version": "1.0", "buildTag": "latest", "meta": { "templateId": "ts-empty" }, "dockerfile": "../Dockerfile", "input": "./input_schema.json", "output": "./output_schema.json", "storages": { "dataset": "./dataset_schema.json", "keyValueStore": "./key_value_store_schema.json" }, "environmentVariables": { "TOMBA_API_KEY": "@tombaApiKey", "TOMBA_API_SECRET": "@tombaApiSecret" }, "usesStandbyMode": true, "webServerSchema": "./web_server_schema.json", "defaultMemoryMbytes": 256}{ "actorKeyValueStoreSchemaVersion": 1, "title": "Tomba Email Finder storage", "description": "Records the Actor keeps in its default key-value store. Results are in the dataset.", "collections": { "input": { "title": "Input", "description": "The input of this run.", "key": "INPUT", "contentTypes": ["application/json"] }, "progress": { "title": "Progress", "description": "Requests already processed, so a migrated or restarted run continues without charging again.", "key": "TOMBA_STATE", "contentTypes": ["application/json"], "jsonSchema": { "$schema": "http://json-schema.org/draft-07/schema#", "type": "object", "properties": { "done": { "type": "object", "description": "Processed requests (keys: domain|company|fullName|firstName|lastName, lowercase) mapped to true", "additionalProperties": { "type": "boolean" } } }, "required": ["done"] } } }}{ "title": "Tomba Email Finder Input", "type": "object", "schemaVersion": 1, "properties": { "requests": { "title": "Email Finder Requests", "type": "array", "editor": "json", "description": "List of people to find. Each request needs a company (`domain` or `company` name) and a name (`fullName`, or `firstName` and `lastName`).", "items": { "type": "object", "properties": { "domain": { "type": "string", "title": "Domain", "description": "Company domain name (e.g., 'stripe.com'). Required unless `company` is set" }, "company": { "type": "string", "title": "Company", "description": "Company name (e.g., 'Stripe'). Alternative to `domain`" }, "firstName": { "type": "string", "title": "First Name", "description": "Person's first name. Required with `lastName` unless `fullName` is set" }, "lastName": { "type": "string", "title": "Last Name", "description": "Person's last name. Required with `firstName` unless `fullName` is set" }, "fullName": { "type": "string", "title": "Full Name", "description": "Person's full name (e.g., 'John Doe'). Alternative to `firstName` + `lastName`" } }, "additionalProperties": false }, "minItems": 1, "maxItems": 1000, "example": [ { "domain": "stripe.com", "firstName": "John", "lastName": "Doe" }, { "company": "Shopify", "fullName": "Jane Smith" } ], "prefill": [ { "domain": "stripe.com", "firstName": "John", "lastName": "Doe" } ] }, "maxResults": { "title": "Maximum Results", "type": "integer", "description": "Maximum number of requests to process (default: 50, max: 1000)", "editor": "number", "default": 50, "minimum": 1, "maximum": 1000, "example": 50 }, "enrichMobile": { "title": "Find phone numbers", "type": "boolean", "description": "Also return the person's phone numbers in `phone_data`. A result with phone data costs 6 credits instead of 1; results without phone data still cost 1 credit.", "default": false, "editor": "checkbox" }, "webhookUrl": { "title": "Webhook URL", "type": "string", "description": "Optional URL that Tomba notifies when a result is ready (sent to Tomba as `webhook_url`).", "editor": "textfield", "pattern": "^https?://", "example": "https://example.com/tomba-webhook" }, "maxConcurrency": { "title": "Max concurrency", "type": "integer", "description": "How many Tomba requests run in parallel. There is no rate limit, so raise it for large lists.", "minimum": 1, "maximum": 50, "default": 10, "editor": "number", "sectionCaption": "Run options", "sectionDescription": "Concurrency, retry and cache settings. The defaults work well for most runs." }, "maxRetries": { "title": "Max retries", "type": "integer", "description": "How many times to retry a request after a network error, 429 or 5xx response (exponential backoff). Failed requests are never charged.", "minimum": 0, "maximum": 10, "default": 3, "editor": "number" }, "useCache": { "title": "Use cache", "type": "boolean", "description": "Reuse results from previous runs for identical requests. Cached results are free.", "default": true, "editor": "checkbox" }, "cacheTtlHours": { "title": "Cache TTL (hours)", "type": "integer", "description": "How long cached results stay valid. 0 disables the cross-run cache.", "minimum": 0, "maximum": 720, "default": 24, "editor": "number", "unit": "hours" } }, "required": ["requests"]}1// Shared test harness for the Tomba Apify Actors.2// Keep this file identical across all Actors.3import { spawn } from 'node:child_process';4import { readFileSync } from 'node:fs';5import { mkdir, mkdtemp, readdir, readFile, rm, writeFile } from 'node:fs/promises';6import type { IncomingMessage, Server, ServerResponse } from 'node:http';7import { createServer, request as httpRequest } from 'node:http';8import type { AddressInfo } from 'node:net';9import { tmpdir } from 'node:os';10import { join } from 'node:path';11
12import { Ajv } from 'ajv';13
14export interface MockRequest {15 method: string;16 path: string;17 query: Record<string, string>;18 headers: IncomingMessage['headers'];19 body: unknown;20}21
22export interface MockResponse {23 status?: number;24 body?: unknown;25 /** Raw (non-JSON) body. */26 raw?: string;27 headers?: Record<string, string>;28}29
30export type MockHandler = (req: MockRequest) => MockResponse | Promise<MockResponse>;31
32export interface MockServer {33 url: string;34 requests: MockRequest[];35 close: () => Promise<void>;36}37
38/** Start a local HTTP server that impersonates the Tomba API. */39export async function startMockTomba(handler: MockHandler): Promise<MockServer> {40 const requests: MockRequest[] = [];41 const server: Server = createServer((req: IncomingMessage, res: ServerResponse) => {42 const chunks: Buffer[] = [];43 req.on('data', (chunk: Buffer) => chunks.push(chunk));44 req.on('end', async () => {45 const url = new URL(req.url ?? '/', 'http://localhost');46 const text = Buffer.concat(chunks).toString();47 let body: unknown = text;48 try {49 body = text ? JSON.parse(text) : undefined;50 } catch {51 // keep raw text52 }53 const request: MockRequest = {54 method: req.method ?? 'GET',55 path: url.pathname,56 query: Object.fromEntries(url.searchParams),57 headers: req.headers,58 body,59 };60 requests.push(request);61 try {62 const out = await handler(request);63 const status = out.status ?? 200;64 if (out.raw !== undefined) {65 res.writeHead(status, { 'content-type': 'text/html', ...out.headers });66 res.end(out.raw);67 } else {68 res.writeHead(status, { 'content-type': 'application/json', ...out.headers });69 res.end(JSON.stringify(out.body ?? {}));70 }71 } catch (err) {72 res.writeHead(500, { 'content-type': 'application/json' });73 res.end(JSON.stringify({ errors: { message: (err as Error).message } }));74 }75 });76 });77 await new Promise<void>((resolve) => {78 server.listen(0, '127.0.0.1', resolve);79 });80 const { port } = server.address() as AddressInfo;81 return {82 url: `http://127.0.0.1:${port}`,83 requests,84 close: async () =>85 new Promise<void>((resolve) => {86 server.close(() => resolve());87 }),88 };89}90
91export interface RunOptions {92 input: Record<string, unknown>;93 /** Mock server URL; omit to run without an endpoint override. */94 endpoint?: string;95 /** Reuse a storage directory (to test cache and resume). */96 storageDir?: string;97 /** Keep the default storages from the previous run in `storageDir`. */98 keepStorage?: boolean;99 /** Max total charge in USD (locally every event costs $1). */100 maxTotalChargeUsd?: number;101 /** Set to false to run without Tomba credentials. */102 withCredentials?: boolean;103 timeoutMs?: number;104}105
106export interface ChargeLogEntry {107 eventName: string;108 chargedCount?: number;109 [key: string]: unknown;110}111
112export interface RunResult {113 code: number | null;114 output: string;115 items: Record<string, unknown>[];116 charges: ChargeLogEntry[];117 storageDir: string;118 /** Number of charged events per event name. */119 chargeCounts: Record<string, number>;120}121
122/** Run the Actor (`src/main.ts`) in a child process against local storage. */123export async function runActor(options: RunOptions): Promise<RunResult> {124 const storageDir = options.storageDir ?? (await mkdtemp(join(tmpdir(), 'tomba-actor-test-')));125 const inputDir = join(storageDir, 'key_value_stores', 'default');126 await mkdir(inputDir, { recursive: true });127 await writeFile(join(inputDir, 'INPUT.json'), JSON.stringify(options.input));128
129 const env: NodeJS.ProcessEnv = {130 PATH: process.env.PATH,131 HOME: process.env.HOME,132 CRAWLEE_STORAGE_DIR: storageDir,133 CRAWLEE_PURGE_ON_START: options.keepStorage ? '0' : '1',134 APIFY_LOG_LEVEL: 'INFO',135 ACTOR_TEST_PAY_PER_EVENT: 'true',136 ACTOR_USE_CHARGING_LOG_DATASET: 'true',137 };138 if (options.withCredentials !== false) {139 env.TOMBA_API_KEY = 'ta_test_key';140 env.TOMBA_API_SECRET = 'ts_test_secret';141 }142 if (options.endpoint) env.TOMBA_API_ENDPOINT = options.endpoint;143 if (options.maxTotalChargeUsd !== undefined) env.ACTOR_MAX_TOTAL_CHARGE_USD = String(options.maxTotalChargeUsd);144
145 const { code, output } = await new Promise<{ code: number | null; output: string }>((resolve, reject) => {146 const child = spawn(process.execPath, ['--import', 'tsx', 'src/main.ts'], { env, cwd: process.cwd() });147 let out = '';148 child.stdout.on('data', (d: Buffer) => {149 out += d.toString();150 });151 child.stderr.on('data', (d: Buffer) => {152 out += d.toString();153 });154 const timer = setTimeout(() => {155 child.kill('SIGKILL');156 reject(new Error(`Actor run timed out\n${out}`));157 }, options.timeoutMs ?? 60_000);158 child.on('close', (c) => {159 clearTimeout(timer);160 resolve({ code: c, output: out });161 });162 });163
164 const items = await readDataset(storageDir, 'default');165 assertMatchesDatasetSchema(items);166 const charges = (await readDataset(storageDir, 'charging_log')) as ChargeLogEntry[];167 const chargeCounts: Record<string, number> = {};168 for (const c of charges) chargeCounts[c.eventName] = (chargeCounts[c.eventName] ?? 0) + (c.chargedCount ?? 1);169
170 return { code, output, items, charges, storageDir, chargeCounts };171}172
173export async function readDataset(storageDir: string, name: string): Promise<Record<string, unknown>[]> {174 const dir = join(storageDir, 'datasets', name);175 let files: string[];176 try {177 files = (await readdir(dir)).filter((f) => f.endsWith('.json')).sort();178 } catch {179 return [];180 }181 return Promise.all(files.map(async (f) => JSON.parse(await readFile(join(dir, f), 'utf8'))));182}183
184export async function removeStorage(storageDir: string): Promise<void> {185 await rm(storageDir, { recursive: true, force: true });186}187
188/** Total number of charged events across all event names. */189export function totalCharges(result: RunResult): number {190 return Object.values(result.chargeCounts).reduce((a, b) => a + b, 0);191}192
193export interface StandbyActor {194 /** Base URL of the Actor's HTTP server. */195 url: string;196 storageDir: string;197 /** GET or POST a request to the Actor and parse the JSON response. */198 call: (199 path: string,200 init?: { method?: string; body?: unknown; headers?: Record<string, string> },201 ) => Promise<{ status: number; body: Record<string, unknown> }>;202 /** Stop the Actor and return its log output and charges. */203 stop: () => Promise<{ output: string; chargeCounts: Record<string, number> }>;204}205
206async function freePort(): Promise<number> {207 const server = createServer();208 await new Promise<void>((resolve) => {209 server.listen(0, '127.0.0.1', resolve);210 });211 const { port } = server.address() as AddressInfo;212 await new Promise<void>((resolve) => {213 server.close(() => resolve());214 });215 return port;216}217
218async function httpCall(219 url: string,220 init: { method?: string; body?: unknown; headers?: Record<string, string> } = {},221): Promise<{ status: number; body: Record<string, unknown> }> {222 let payload: string | undefined;223 if (typeof init.body === 'string') payload = init.body;224 else if (init.body !== undefined) payload = JSON.stringify(init.body);225 return new Promise((resolve, reject) => {226 const req = httpRequest(227 url,228 {229 method: init.method ?? (payload ? 'POST' : 'GET'),230 headers: { 'content-type': 'application/json', ...init.headers },231 },232 (res) => {233 const chunks: Buffer[] = [];234 res.on('data', (c: Buffer) => chunks.push(c));235 res.on('end', () => {236 const text = Buffer.concat(chunks).toString();237 let body: Record<string, unknown> = {};238 try {239 body = JSON.parse(text);240 } catch {241 body = { raw: text };242 }243 resolve({ status: res.statusCode ?? 0, body });244 });245 },246 );247 req.on('error', reject);248 if (payload) req.write(payload);249 req.end();250 });251}252
253/** Start the Actor in Standby mode (real-time HTTP API) and wait until it is ready. */254export async function startStandbyActor(255 options: { endpoint?: string; maxTotalChargeUsd?: number; withCredentials?: boolean } = {},256): Promise<StandbyActor> {257 const storageDir = await mkdtemp(join(tmpdir(), 'tomba-standby-test-'));258 const port = await freePort();259 const env: NodeJS.ProcessEnv = {260 PATH: process.env.PATH,261 HOME: process.env.HOME,262 CRAWLEE_STORAGE_DIR: storageDir,263 APIFY_LOG_LEVEL: 'INFO',264 APIFY_META_ORIGIN: 'STANDBY',265 ACTOR_WEB_SERVER_PORT: String(port),266 ACTOR_TEST_PAY_PER_EVENT: 'true',267 ACTOR_USE_CHARGING_LOG_DATASET: 'true',268 };269 if (options.withCredentials !== false) {270 env.TOMBA_API_KEY = 'ta_test_key';271 env.TOMBA_API_SECRET = 'ts_test_secret';272 }273 if (options.endpoint) env.TOMBA_API_ENDPOINT = options.endpoint;274 if (options.maxTotalChargeUsd !== undefined) env.ACTOR_MAX_TOTAL_CHARGE_USD = String(options.maxTotalChargeUsd);275
276 const child = spawn(process.execPath, ['--import', 'tsx', 'src/main.ts'], { env, cwd: process.cwd() });277 let output = '';278 child.stdout.on('data', (d: Buffer) => {279 output += d.toString();280 });281 child.stderr.on('data', (d: Buffer) => {282 output += d.toString();283 });284 const exited = new Promise<void>((resolve) => {285 child.on('close', () => resolve());286 });287
288 const url = `http://127.0.0.1:${port}`;289 const deadline = Date.now() + 30_000;290 for (;;) {291 if (child.exitCode !== null) throw new Error(`Standby Actor exited early\n${output}`);292 try {293 const res = await httpCall(`${url}/`, { headers: { 'x-apify-container-server-readiness-probe': '1' } });294 if (res.status === 200) break;295 } catch {296 // not listening yet297 }298 if (Date.now() > deadline) {299 child.kill('SIGKILL');300 throw new Error(`Standby Actor did not become ready\n${output}`);301 }302 await new Promise((r) => {303 setTimeout(r, 100);304 });305 }306
307 return {308 url,309 storageDir,310 call: async (path, init) => httpCall(`${url}${path}`, init),311 stop: async () => {312 child.kill('SIGTERM');313 await exited;314 const charges = (await readDataset(storageDir, 'charging_log')) as ChargeLogEntry[];315 const chargeCounts: Record<string, number> = {};316 for (const c of charges)317 chargeCounts[c.eventName] = (chargeCounts[c.eventName] ?? 0) + (c.chargedCount ?? 1);318 await removeStorage(storageDir);319 return { output, chargeCounts };320 },321 };322}323
324let datasetValidator: ReturnType<Ajv['compile']> | undefined;325
326/**327 * Apify validates every pushed item against `.actor/dataset_schema.json` and fails the run on a mismatch,328 * while local runs don't. Check the items the same way so schema mismatches fail the tests.329 */330export function assertMatchesDatasetSchema(items: Record<string, unknown>[]): void {331 if (!datasetValidator) {332 const { fields } = JSON.parse(readFileSync(join(process.cwd(), '.actor/dataset_schema.json'), 'utf8'));333 delete fields.$schema;334 datasetValidator = new Ajv({ allErrors: true, strict: false }).compile(fields);335 }336 for (const [index, item] of items.entries()) {337 if (!datasetValidator(item)) {338 const errors = (datasetValidator.errors ?? [])339 .map((e) => `${e.instancePath || '/'} ${e.message}`)340 .join('; ');341 throw new Error(342 `Dataset item ${index} doesn't match .actor/dataset_schema.json: ${errors}\n${JSON.stringify(item)}`,343 );344 }345 }346}347
348/**349 * Validate an input against `.actor/input_schema.json` (Apify validates the input before a run or API call350 * starts). Returns the error messages; empty when the input is accepted.351 */352export function inputSchemaErrors(input: Record<string, unknown>): string[] {353 const schema = JSON.parse(readFileSync(join(process.cwd(), '.actor/input_schema.json'), 'utf8'));354 const validate = new Ajv({ allErrors: true, strict: false }).compile({355 type: 'object',356 properties: schema.properties,357 required: schema.required ?? [],358 });359 return validate(input) ? [] : (validate.errors ?? []).map((e) => `${e.instancePath || '/'} ${e.message}`);360}1// Unit tests for src/tomba.ts. Keep this file identical across all Actors.2import assert from 'node:assert/strict';3import { mkdtemp } from 'node:fs/promises';4import { tmpdir } from 'node:os';5import { join } from 'node:path';6import { after, before, describe, it } from 'node:test';7
8process.env.CRAWLEE_STORAGE_DIR = await mkdtemp(join(tmpdir(), 'tomba-unit-'));9process.env.ACTOR_TEST_PAY_PER_EVENT = 'true';10process.env.TOMBA_API_KEY = 'ta_unit';11process.env.TOMBA_API_SECRET = 'ts_unit';12
13const { Actor } = await import('apify');14const { TombaException } = await import('tomba');15const T = await import('../src/tomba.js');16
17const ok = (body: unknown) => async () => ({ data: body, rateLimit: {} });18const httpError = (status: number, body: unknown = {}, headers: Record<string, string> = {}) =>19 new TombaException('error', { status, headers }, body);20const charged = (event = T.EVENT_REQUEST) => Actor.getChargingManager().getChargedEventCount(event);21
22let seq = 0;23const params = () => ({ n: ++seq });24
25before(async () => {26 await Actor.init();27 await T.setupTomba({ maxRetries: 1, useCache: true, cacheTtlHours: 1, maxConcurrency: 4 });28});29
30after(async () => {31 await Actor.exit({ exit: false });32});33
34describe('isBillable', () => {35 const cases: [string, unknown, boolean][] = [36 ['data array', { data: [{ a: 1 }] }, true],37 ['data object', { data: { email: 'a@b.co' } }, true],38 ['negative answer (object with empty list)', { data: { organization: {}, emails: [] } }, true],39 ['negative answer (undeliverable)', { data: { email: { status: 'invalid' } } }, true],40 ['empty array', { data: [] }, false],41 ['empty object', { data: {} }, false],42 ['object with only null values', { data: { person: null, company: null } }, false],43 ['negative answer (email null, other fields set)', { data: { email: null, first_name: 'John' } }, true],44 ['null data', { data: null }, false],45 ['missing data', { meta: {} }, false],46 ['errors object', { data: { a: 1 }, errors: { message: 'x' } }, false],47 ['string body', '<html>', false],48 ['null body', null, false],49 ['array body', [1, 2], false],50 ];51 for (const [name, body, expected] of cases) {52 it(name, () => assert.equal(T.isBillable(body), expected));53 }54});55
56describe('normalizers', () => {57 it('normalizeDomain strips protocol, www, path and case', () => {58 assert.equal(T.normalizeDomain(' https://www.Stripe.com/about?x=1 '), 'stripe.com');59 assert.equal(T.normalizeDomain('http://tomba.io'), 'tomba.io');60 assert.equal(T.normalizeDomain('sub.example.org'), 'sub.example.org');61 });62
63 it('normalizeEmail trims and lowercases', () => {64 assert.equal(T.normalizeEmail(' John@Stripe.COM '), 'john@stripe.com');65 });66
67 it('unique keeps first occurrence and drops empty keys', () => {68 assert.deepEqual(T.unique(['a', 'b', 'a', '', 'c']), ['a', 'b', 'c']);69 assert.deepEqual(70 T.unique([{ id: 1 }, { id: 2 }, { id: 1 }], (x) => String(x.id)),71 [{ id: 1 }, { id: 2 }],72 );73 });74});75
76describe('phone data', () => {77 it('counts phone numbers in phone_data', () => {78 assert.equal(T.phoneDataCount({ phone_data: [{ number: '+1' }, { number: '+2' }] }), 2);79 assert.equal(T.phoneDataCount({ phone_data: [] }), 0);80 assert.equal(T.phoneDataCount({ phone_number: true }), 0);81 assert.equal(T.phoneDataCount(null), 0);82 assert.equal(T.phoneDataCount('x'), 0);83 });84
85 it('detects records with phone data', () => {86 assert.equal(T.hasPhoneData({ phone_data: [{ number: '+1' }] }), true);87 assert.equal(T.hasPhoneData({ phone_data: [] }), false);88 assert.equal(T.hasPhoneData({}), false);89 });90
91 it('prices phone data at 5 credits', () => {92 assert.equal(T.PHONE_CREDITS, 5);93 });94});95
96describe('cacheStoreName', () => {97 it("is unique per Actor so limited-permission runs never open another Actor's store", () => {98 const previous = process.env.ACTOR_ID;99 try {100 process.env.ACTOR_ID = 'AbC123xyz';101 assert.equal(T.cacheStoreName(), 'tomba-cache-AbC123xyz');102 delete process.env.ACTOR_ID;103 assert.equal(T.cacheStoreName(), 'tomba-cache');104 } finally {105 if (previous === undefined) delete process.env.ACTOR_ID;106 else process.env.ACTOR_ID = previous;107 }108 });109});110
111describe('setupTomba without cache permissions', () => {112 it('keeps working with an in-run cache when the cache store cannot be opened', async () => {113 const original = Actor.openKeyValueStore.bind(Actor);114 Actor.openKeyValueStore = async () => {115 throw new Error('Insufficient permissions for the key-value store.');116 };117 try {118 await T.setupTomba({ maxRetries: 1, useCache: true, cacheTtlHours: 1, maxConcurrency: 4 });119 const p = { cachePermissionTest: true };120 let calls = 0;121 const fn = async () => {122 calls++;123 return { data: { data: [1] }, rateLimit: {} };124 };125 const first = await T.callTomba('perm', p, fn);126 const second = await T.callTomba('perm', p, fn);127 assert.equal(first.charged, true);128 assert.equal(second.cached, true);129 assert.equal(calls, 1);130 } finally {131 Actor.openKeyValueStore = original;132 await T.setupTomba({ maxRetries: 1, useCache: true, cacheTtlHours: 1, maxConcurrency: 4 });133 }134 });135});136
137describe('runPool', () => {138 it('processes every item with bounded concurrency', async () => {139 let active = 0;140 let peak = 0;141 const seen: number[] = [];142 await T.runPool(143 Array.from({ length: 20 }, (_, i) => i),144 async (item) => {145 active++;146 peak = Math.max(peak, active);147 await new Promise((r) => {148 setTimeout(r, 5);149 });150 seen.push(item);151 active--;152 },153 3,154 );155 assert.equal(seen.length, 20);156 assert.deepEqual(157 [...seen].sort((a, b) => a - b),158 Array.from({ length: 20 }, (_, i) => i),159 );160 assert.ok(peak <= 3, `peak concurrency ${peak}`);161 });162
163 it('handles an empty list', async () => {164 await T.runPool([], async () => assert.fail('should not run'));165 });166});167
168describe('callTomba', () => {169 it('charges a response with data and unwraps the SDK { data, rateLimit } shape', async () => {170 const start = charged();171 const res = await T.callTomba('test', params(), ok({ data: [{ name: 'React' }] }));172 assert.equal(res.status, 200);173 assert.equal(res.charged, true);174 assert.equal(res.cached, false);175 assert.deepEqual(res.data, [{ name: 'React' }]);176 assert.equal(charged(), start + 1);177 });178
179 it('accepts an SDK that returns the body directly', async () => {180 const res = await T.callTomba('test', params(), async () => ({ data: { x: 1 } }));181 assert.equal(res.charged, true);182 assert.deepEqual(res.data, { x: 1 });183 });184
185 it('charges a negative answer', async () => {186 const res = await T.callTomba('test', params(), ok({ data: { organization: {}, emails: [] } }));187 assert.equal(res.charged, true);188 });189
190 it('serves a repeated request from cache without charging', async () => {191 const p = params();192 let calls = 0;193 const fn = async () => {194 calls++;195 return { data: { data: { v: 1 } }, rateLimit: {} };196 };197 await T.callTomba('test', p, fn);198 const start = charged();199 const res = await T.callTomba('test', { ...p }, fn);200 assert.equal(res.cached, true);201 assert.equal(res.charged, false);202 assert.deepEqual(res.data, { v: 1 });203 assert.equal(calls, 1);204 assert.equal(charged(), start);205 });206
207 it('uses a different cache entry for different params or endpoints', async () => {208 const p = params();209 await T.callTomba('a', p, ok({ data: [1] }));210 const other = await T.callTomba('b', p, ok({ data: [2] }));211 assert.equal(other.cached, false);212 assert.deepEqual(other.data, [2]);213 });214
215 it('does not charge or cache empty data', async () => {216 const p = params();217 const start = charged();218 const res = await T.callTomba('test', p, ok({ data: [] }));219 assert.equal(res.charged, false);220 assert.equal(res.status, 200);221 const again = await T.callTomba('test', p, ok({ data: [] }));222 assert.equal(again.cached, false);223 assert.equal(charged(), start);224 });225
226 it('does not charge a success with an errors object', async () => {227 const res = await T.callTomba('test', params(), ok({ data: { a: 1 }, errors: { message: 'Bad thing' } }));228 assert.equal(res.charged, false);229 assert.equal(res.error, 'Bad thing');230 });231
232 it('reports a non-JSON body as 502 without charging', async () => {233 const res = await T.callTomba('test', params(), ok('<html>oops</html>'));234 assert.equal(res.status, 502);235 assert.equal(res.charged, false);236 });237
238 it('does not retry or charge a 422', async () => {239 let calls = 0;240 const res = await T.callTomba('test', params(), async () => {241 calls++;242 throw httpError(422, { errors: { message: 'Invalid domain' } });243 });244 assert.equal(calls, 1);245 assert.equal(res.status, 422);246 assert.equal(res.charged, false);247 assert.match(res.error ?? '', /Invalid domain/);248 });249
250 it('does not retry SDK input validation errors', async () => {251 let calls = 0;252 const res = await T.callTomba('test', params(), async () => {253 calls++;254 throw new TombaException('Domain is required.');255 });256 assert.equal(calls, 1);257 assert.equal(res.status, 400);258 assert.equal(res.charged, false);259 });260
261 it('retries a 429 honoring Retry-After, then charges the success', async () => {262 let calls = 0;263 const res = await T.callTomba('test', params(), async () => {264 calls++;265 if (calls === 1) throw httpError(429, {}, { 'retry-after': '1' });266 return { data: { data: { ok: true } }, rateLimit: {} };267 });268 assert.equal(calls, 2);269 assert.equal(res.charged, true);270 });271
272 it('retries network errors', async () => {273 let calls = 0;274 const res = await T.callTomba('test', params(), async () => {275 calls++;276 if (calls === 1) throw new TombaException('connect ECONNREFUSED 127.0.0.1:9');277 return { data: { data: [1] }, rateLimit: {} };278 });279 assert.equal(calls, 2);280 assert.equal(res.charged, true);281 });282
283 it('gives up after maxRetries on 5xx without charging', async () => {284 let calls = 0;285 const res = await T.callTomba('test', params(), async () => {286 calls++;287 throw httpError(503);288 });289 assert.equal(calls, 2); // 1 attempt + maxRetries (1)290 assert.equal(res.status, 503);291 assert.equal(res.charged, false);292 });293
294 it('charges the given event name', async () => {295 const start = charged(T.EVENT_PHONE_FINDER);296 await T.callTomba('phone', params(), ok({ data: { phone: '+1' } }), T.EVENT_PHONE_FINDER);297 assert.equal(charged(T.EVENT_PHONE_FINDER), start + 1);298 });299
300 it('charges a fixed count of events', async () => {301 const start = charged();302 const res = await T.callTomba('test', params(), ok({ data: [1] }), T.EVENT_REQUEST, 3);303 assert.equal(res.chargedCount, 3);304 assert.equal(charged(), start + 3);305 });306
307 it('charges a count computed from the response body', async () => {308 const start = charged();309 const res = await T.callTomba(310 'test',311 params(),312 ok({ data: { items: [1, 2, 3, 4] } }),313 T.EVENT_REQUEST,314 (body) => (body.data as { items: unknown[] }).items.length,315 );316 assert.equal(res.chargedCount, 4);317 assert.equal(charged(), start + 4);318 });319
320 it('does not charge a count for a non-billable response', async () => {321 const start = charged();322 const res = await T.callTomba('test', params(), ok({ data: [] }), T.EVENT_REQUEST, 5);323 assert.equal(res.charged, false);324 assert.equal(charged(), start);325 });326
327 it('tracks stats', () => {328 assert.ok(T.stats.requests > 0);329 assert.ok(T.stats.charged > 0);330 assert.ok(T.stats.cached > 0);331 assert.ok(T.stats.retries > 0);332 assert.ok(T.stats.failed > 0);333 });334});1// End-to-end tests: run the Actor against a mock Tomba API.2import assert from 'node:assert/strict';3import { after, afterEach, describe, it } from 'node:test';4
5import type { MockHandler, MockServer } from './helpers.js';6import { removeStorage, runActor, startMockTomba, startStandbyActor, totalCharges } from './helpers.js';7
8const SOURCES = [9 {10 uri: 'https://stripe.com/blog/team',11 website_url: 'stripe.com',12 extracted_on: '2024-09-17T11:26:56+02:00',13 last_seen_on: '2025-09-06T04:51:06+02:00',14 still_on_page: true,15 },16];17
18/** Realistic Tomba email-finder `data` object for a found email. */19function found(domain: string, first: string, last: string): Record<string, unknown> {20 return {21 email: `${first.toLowerCase()}.${last.toLowerCase()}@${domain}`,22 first_name: first,23 last_name: last,24 full_name: `${first} ${last}`,25 country: 'US',26 gender: 'male',27 phone_number: false,28 position: 'Head of Sales',29 twitter: null,30 linkedin: `https://www.linkedin.com/in/${first.toLowerCase()}${last.toLowerCase()}`,31 disposable: false,32 webmail: false,33 accept_all: false,34 company: 'Stripe',35 website_url: domain,36 score: 97,37 verification: { date: '2025-10-17T00:00:00+02:00', status: 'valid' },38 sources: SOURCES,39 };40}41
42/** Phone fixtures returned with `enrich_mobile=true`: Jane has two numbers, Phoneless none, everyone else one. */43const PHONES: Record<string, Record<string, unknown>[]> = {44 Jane: [45 { number: '+14155550123', type: 'mobile' },46 { number: '+14155550199', type: 'work' },47 ],48 Phoneless: [],49};50const phonesFor = (first: string) => PHONES[first] ?? [{ number: '+14155550123', type: 'mobile' }];51
52/**53 * Default Tomba behaviour: known people return an email, `nobody` returns a negative answer,54 * special domains return empty data, 422 or a non-JSON body. Requests by `company` resolve to55 * `<company>.com`, requests by `full_name` are split into first and last name. `phone_data` is56 * only returned when the query has `enrich_mobile=true`.57 */58const tomba: MockHandler = (req) => {59 assert.equal(req.method, 'GET');60 assert.equal(req.path, '/email-finder');61 const { company, full_name: fullName, enrich_mobile: enrichMobile } = req.query;62 const domain = req.query.domain ?? `${company.toLowerCase()}.com`;63 const [first, last] = fullName ? fullName.split(' ') : [req.query.first_name, req.query.last_name];64 const phones = enrichMobile === 'true' ? { phone_data: phonesFor(first) } : {};65 if (domain === 'empty.com') return { body: { data: null } };66 if (domain === 'invalid.com') return { status: 422, body: { errors: { message: 'Invalid domain' } } };67 if (domain === 'html.com') return { raw: '<html>Bad gateway</html>' };68 if (first === 'Nobody') {69 return {70 body: {71 data: {72 email: null,73 first_name: first,74 last_name: last,75 full_name: null,76 website_url: domain,77 score: 0,78 verification: { date: null, status: null },79 sources: [],80 ...phones,81 },82 },83 };84 }85 return { body: { data: { ...found(domain, first, last), ...phones } } };86};87
88const servers: MockServer[] = [];89const dirs: string[] = [];90
91async function mock(handler: MockHandler = tomba): Promise<MockServer> {92 const server = await startMockTomba(handler);93 servers.push(server);94 return server;95}96
97async function run(...args: Parameters<typeof runActor>) {98 const result = await runActor(...args);99 dirs.push(result.storageDir);100 return result;101}102
103const person = (domain: string, firstName = 'John', lastName = 'Doe') => ({ domain, firstName, lastName });104
105afterEach(async () => {106 await Promise.all(servers.splice(0).map(async (s) => s.close()));107});108
109after(async () => {110 await Promise.all(dirs.map(removeStorage));111});112
113describe('email-finder', () => {114 it('returns the email and charges one event per billable request', async () => {115 const server = await mock();116 const result = await run({117 input: { requests: [person('stripe.com'), person('empty.com')] },118 endpoint: server.url,119 });120
121 assert.equal(result.code, 0, result.output);122 const hit = result.items.find((i) => i.domain === 'stripe.com');123 assert.deepEqual(hit, {124 ...found('stripe.com', 'John', 'Doe'),125 domain: 'stripe.com',126 firstName: 'John',127 lastName: 'Doe',128 source: 'tomba_email_finder',129 phoneNumbers: 0,130 charged: true,131 chargedCredits: 1,132 cached: false,133 });134
135 const empty = result.items.find((i) => i.domain === 'empty.com');136 assert.deepEqual(empty, {137 domain: 'empty.com',138 firstName: 'John',139 lastName: 'Doe',140 source: 'tomba_email_finder',141 phoneNumbers: 0,142 charged: false,143 chargedCredits: 0,144 cached: false,145 error: 'No email found for this person',146 });147
148 assert.deepEqual(result.chargeCounts, { 'tomba-request': 1 });149 });150
151 it('sends the built-in credentials and the expected query to Tomba', async () => {152 const server = await mock();153 await run({ input: { requests: [person('stripe.com')] }, endpoint: server.url });154 assert.equal(server.requests.length, 1);155 assert.equal(server.requests[0].headers['x-tomba-key'], 'ta_test_key');156 assert.equal(server.requests[0].headers['x-tomba-secret'], 'ts_test_secret');157 assert.deepEqual(server.requests[0].query, { domain: 'stripe.com', first_name: 'John', last_name: 'Doe' });158 });159
160 it('normalizes domains, trims names and deduplicates requests', async () => {161 const server = await mock();162 const result = await run({163 input: {164 requests: [165 person('https://www.Stripe.com/about', ' John ', ' Doe '),166 person('stripe.com', 'john', 'DOE'),167 person('STRIPE.COM', 'John', 'Doe'),168 ],169 },170 endpoint: server.url,171 });172 assert.deepEqual(173 server.requests.map((r) => r.query),174 [{ domain: 'stripe.com', first_name: 'John', last_name: 'Doe' }],175 );176 assert.equal(result.items.length, 1);177 });178
179 it('charges a negative answer (person found, no email)', async () => {180 const server = await mock();181 const result = await run({182 input: { requests: [person('stripe.com', 'Nobody', 'Here')] },183 endpoint: server.url,184 });185 assert.equal(result.items.length, 1);186 assert.equal(result.items[0].email, null);187 assert.equal(result.items[0].charged, true);188 assert.equal(result.items[0].error, undefined);189 assert.deepEqual(result.chargeCounts, { 'tomba-request': 1 });190 });191
192 it('does not charge Tomba error statuses and does not retry them', async () => {193 const server = await mock();194 const result = await run({ input: { requests: [person('invalid.com')] }, endpoint: server.url });195 assert.equal(result.code, 0, result.output);196 assert.equal(server.requests.length, 1);197 assert.equal(result.items[0].charged, false);198 assert.match(String(result.items[0].error), /422: Invalid domain/);199 assert.equal(totalCharges(result), 0);200 });201
202 it('does not charge a non-JSON body', async () => {203 const server = await mock();204 const result = await run({ input: { requests: [person('html.com')] }, endpoint: server.url });205 assert.equal(result.items[0].charged, false);206 assert.match(String(result.items[0].error), /Invalid response/);207 assert.equal(totalCharges(result), 0);208 });209
210 it('retries 429 and 5xx responses, then charges the success once', async () => {211 let calls = 0;212 const server = await mock(async (req) => {213 calls++;214 if (calls === 1)215 return {216 status: 429,217 body: { errors: { message: 'Too many requests' } },218 headers: { 'retry-after': '1' },219 };220 if (calls === 2) return { status: 503, body: {} };221 return tomba(req);222 });223 const result = await run({ input: { requests: [person('stripe.com')], maxRetries: 3 }, endpoint: server.url });224 assert.equal(server.requests.length, 3);225 assert.equal(result.items[0].charged, true);226 assert.equal(result.items[0].email, 'john.doe@stripe.com');227 assert.deepEqual(result.chargeCounts, { 'tomba-request': 1 });228 });229
230 it('serves repeated runs from the cache for free', async () => {231 const server = await mock();232 const first = await run({ input: { requests: [person('stripe.com')] }, endpoint: server.url });233 const second = await run({234 input: { requests: [person('stripe.com')] },235 endpoint: server.url,236 storageDir: first.storageDir,237 });238
239 assert.equal(server.requests.length, 1);240 assert.equal(second.items.length, 1);241 assert.equal(second.items[0].email, 'john.doe@stripe.com');242 assert.equal(second.items[0].cached, true);243 assert.equal(second.items[0].charged, false);244 assert.equal(totalCharges(second), 0);245 });246
247 it('calls Tomba again when the cache is disabled', async () => {248 const server = await mock();249 const input = { requests: [person('stripe.com')], useCache: false };250 const first = await run({ input, endpoint: server.url });251 await run({ input, endpoint: server.url, storageDir: first.storageDir });252 assert.equal(server.requests.length, 2);253 });254
255 it('stops at the max charge limit and resumes without reprocessing', async () => {256 const server = await mock();257 const domains = ['a.com', 'b.com', 'c.com', 'd.com', 'e.com'];258 const input = { requests: domains.map((d) => person(d)), maxConcurrency: 1, useCache: false };259
260 // Locally every event costs $1, so a $2 budget allows two billable requests.261 const first = await run({ input, endpoint: server.url, maxTotalChargeUsd: 2 });262 assert.equal(first.code, 0, first.output);263 assert.equal(totalCharges(first), 2);264 assert.equal(server.requests.length, 2);265 assert.equal(first.items.length, 2);266
267 const second = await run({ input, endpoint: server.url, storageDir: first.storageDir, keepStorage: true });268 assert.equal(second.code, 0, second.output);269 // The charging log is kept with the storage, so it now holds both runs: 2 + 3 charges.270 assert.equal(totalCharges(second), 5);271 assert.equal(second.items.length, 5);272 assert.deepEqual(273 server.requests.map((r) => r.query.domain),274 domains,275 );276 });277
278 it('respects maxResults', async () => {279 const server = await mock();280 const result = await run({281 input: { requests: [person('a.com'), person('b.com'), person('c.com')], maxResults: 2 },282 endpoint: server.url,283 });284 assert.equal(server.requests.length, 2);285 assert.equal(result.items.length, 2);286 assert.equal(totalCharges(result), 2);287 });288
289 it('runs requests in parallel', async () => {290 let active = 0;291 let peak = 0;292 const server = await mock(async (req) => {293 active++;294 peak = Math.max(peak, active);295 await new Promise((r) => {296 setTimeout(r, 100);297 });298 active--;299 return tomba(req);300 });301 const requests = Array.from({ length: 8 }, (_, i) => person(`site${i}.com`));302 await run({ input: { requests, maxConcurrency: 4 }, endpoint: server.url });303 assert.equal(server.requests.length, 8);304 assert.ok(peak > 1 && peak <= 4, `peak concurrency ${peak}`);305 });306
307 it('fails without Tomba credentials and never calls the API', async () => {308 const server = await mock();309 const result = await run({310 input: { requests: [person('stripe.com')] },311 endpoint: server.url,312 withCredentials: false,313 });314 assert.notEqual(result.code, 0);315 assert.match(result.output, /misconfigured/);316 assert.doesNotMatch(result.output, /ta_test_key|ts_test_secret/);317 assert.equal(server.requests.length, 0);318 });319
320 it('never sends requests with a missing field and does not charge them', async () => {321 const server = await mock();322 const result = await run({323 input: {324 requests: [325 { domain: 'stripe.com', firstName: 'John' },326 { domain: '', firstName: 'Jane', lastName: 'Smith' },327 { firstName: ' ', lastName: 'Smith', domain: 'stripe.com' },328 person('stripe.com', 'Jane', 'Smith'),329 ],330 },331 endpoint: server.url,332 });333 assert.equal(result.code, 0, result.output);334 assert.deepEqual(335 server.requests.map((r) => r.query),336 [{ domain: 'stripe.com', first_name: 'Jane', last_name: 'Smith' }],337 );338 const invalid = result.items.filter((i) => i.error);339 assert.equal(invalid.length, 3);340 for (const item of invalid) {341 assert.equal(item.charged, false);342 assert.match(String(item.error), /Missing required fields/);343 }344 assert.deepEqual(result.chargeCounts, { 'tomba-request': 1 });345 });346
347 it('sends enrich_mobile and webhook_url only when they are set', async () => {348 const server = await mock();349 await run({ input: { requests: [person('stripe.com')], enrichMobile: false }, endpoint: server.url });350 await run({351 input: {352 requests: [person('shopify.com')],353 enrichMobile: true,354 webhookUrl: 'https://hooks.example.com/tomba',355 },356 endpoint: server.url,357 });358 assert.deepEqual(359 server.requests.map((r) => r.query),360 [361 { domain: 'stripe.com', first_name: 'John', last_name: 'Doe' },362 {363 domain: 'shopify.com',364 first_name: 'John',365 last_name: 'Doe',366 enrich_mobile: 'true',367 webhook_url: 'https://hooks.example.com/tomba',368 },369 ],370 );371 });372
373 it('fails on a webhook URL that is not http(s) and never calls the API', async () => {374 const server = await mock();375 const result = await run({376 input: { requests: [person('stripe.com')], webhookUrl: 'ftp://hooks.example.com' },377 endpoint: server.url,378 });379 assert.notEqual(result.code, 0);380 assert.equal(server.requests.length, 0);381 });382
383 it('charges 1 credit without enrichMobile and returns no phone data', async () => {384 const server = await mock();385 const result = await run({386 input: { requests: [person('stripe.com', 'Jane', 'Smith')] },387 endpoint: server.url,388 });389 assert.equal(result.items[0].phone_data, undefined);390 assert.equal(result.items[0].phoneNumbers, 0);391 assert.equal(result.items[0].chargedCredits, 1);392 assert.deepEqual(result.chargeCounts, { 'tomba-request': 1 });393 });394
395 it('charges 6 credits per result with phone data, whatever the number of phones', async () => {396 const server = await mock();397 const result = await run({398 input: {399 requests: [person('stripe.com', 'John', 'Doe'), person('stripe.com', 'Jane', 'Smith')],400 enrichMobile: true,401 },402 endpoint: server.url,403 });404 assert.equal(result.code, 0, result.output);405 const john = result.items.find((i) => i.firstName === 'John');406 const jane = result.items.find((i) => i.firstName === 'Jane');407 assert.deepEqual(john?.phone_data, [{ number: '+14155550123', type: 'mobile' }]);408 assert.equal(john?.phoneNumbers, 1);409 assert.equal(john?.chargedCredits, 6);410 assert.equal(john?.charged, true);411 assert.equal(jane?.phoneNumbers, 2);412 assert.equal(jane?.chargedCredits, 6);413 assert.deepEqual(result.chargeCounts, { 'tomba-request': 12 });414 });415
416 it('charges 1 credit with enrichMobile when no phone number is found', async () => {417 const server = await mock();418 const result = await run({419 input: { requests: [person('stripe.com', 'Phoneless', 'Person')], enrichMobile: true },420 endpoint: server.url,421 });422 assert.deepEqual(result.items[0].phone_data, []);423 assert.equal(result.items[0].phoneNumbers, 0);424 assert.equal(result.items[0].chargedCredits, 1);425 assert.deepEqual(result.chargeCounts, { 'tomba-request': 1 });426 });427
428 it('serves phone data from the cache for free', async () => {429 const server = await mock();430 const input = { requests: [person('stripe.com', 'Jane', 'Smith')], enrichMobile: true };431 const first = await run({ input, endpoint: server.url });432 assert.equal(totalCharges(first), 6);433 const second = await run({ input, endpoint: server.url, storageDir: first.storageDir });434 assert.equal(server.requests.length, 1);435 assert.equal(second.items[0].cached, true);436 assert.equal(second.items[0].charged, false);437 assert.equal(second.items[0].chargedCredits, 0);438 assert.equal(second.items[0].phoneNumbers, 2);439 assert.equal(totalCharges(second), 0);440 });441
442 it('does not reuse a cached result without phone data when enrichMobile is turned on', async () => {443 const server = await mock();444 const first = await run({ input: { requests: [person('stripe.com')] }, endpoint: server.url });445 const second = await run({446 input: { requests: [person('stripe.com')], enrichMobile: true },447 endpoint: server.url,448 storageDir: first.storageDir,449 });450 assert.equal(server.requests.length, 2);451 assert.equal(second.items[0].phoneNumbers, 1);452 assert.equal(totalCharges(second), 6);453 });454
455 it('accepts a company name instead of a domain and a full name instead of first and last name', async () => {456 const server = await mock();457 const result = await run({458 input: {459 requests: [460 { company: ' Stripe ', fullName: ' John Doe ' },461 { domain: 'https://www.shopify.com/', fullName: 'Jane Smith' },462 { company: 'Acme', firstName: 'Bob', lastName: 'Ray' },463 { company: 'stripe', fullName: 'john doe' },464 ],465 maxConcurrency: 1,466 },467 endpoint: server.url,468 });469 assert.equal(result.code, 0, result.output);470 assert.deepEqual(471 server.requests.map((r) => r.query),472 [473 { company: 'Stripe', full_name: 'John Doe' },474 { domain: 'shopify.com', full_name: 'Jane Smith' },475 { company: 'Acme', first_name: 'Bob', last_name: 'Ray' },476 ],477 );478 assert.equal(result.items.length, 3);479 const john = result.items.find((i) => i.fullName === 'John Doe');480 assert.equal(john?.email, 'john.doe@stripe.com');481 assert.equal(john?.inputCompany, 'Stripe');482 assert.equal(john?.company, 'Stripe');483 assert.equal(john?.domain, '');484 const bob = result.items.find((i) => i.inputCompany === 'Acme');485 assert.equal(bob?.email, 'bob.ray@acme.com');486 assert.equal(bob?.fullName, undefined);487 assert.deepEqual(result.chargeCounts, { 'tomba-request': 3 });488 });489
490 it('rejects requests without a company or without a name, for free', async () => {491 const server = await mock();492 const result = await run({493 input: {494 requests: [495 { company: 'Stripe' },496 { fullName: 'John Doe' },497 { company: 'Stripe', firstName: 'John' },498 { company: ' ', domain: '', fullName: 'Jane Doe' },499 { company: 'Stripe', fullName: ' ', lastName: 'Doe' },500 ],501 },502 endpoint: server.url,503 });504 assert.equal(result.code, 0, result.output);505 assert.equal(server.requests.length, 0);506 assert.equal(result.items.length, 5);507 for (const item of result.items) {508 assert.equal(item.charged, false);509 assert.equal(item.chargedCredits, 0);510 assert.equal(item.phoneNumbers, 0);511 assert.match(String(item.error), /Missing required fields/);512 }513 assert.equal(totalCharges(result), 0);514 });515
516 it('fails on empty input', async () => {517 const server = await mock();518 const result = await run({ input: { requests: [] }, endpoint: server.url });519 assert.notEqual(result.code, 0);520 assert.equal(server.requests.length, 0);521 });522});523
524describe('email-finder standby (real-time API)', () => {525 it('answers the readiness probe and a bare GET with usage info', async () => {526 const server = await mock();527 const actor = await startStandbyActor({ endpoint: server.url });528 try {529 const probe = await actor.call('/', { headers: { 'x-apify-container-server-readiness-probe': '1' } });530 assert.equal(probe.status, 200);531 const usage = await actor.call('/');532 assert.equal(usage.status, 200);533 assert.match(String(usage.body.usage), /GET/);534 assert.equal(server.requests.length, 0);535 } finally {536 await actor.stop();537 }538 });539
540 it('looks up one person from GET query parameters and charges per credit', async () => {541 const server = await mock();542 const actor = await startStandbyActor({ endpoint: server.url });543 let stopped;544 try {545 const res = await actor.call('/?domain=https://www.Stripe.com/team&first_name=John&last_name=Doe');546 assert.equal(res.status, 200);547 assert.deepEqual(res.body.items, [548 {549 ...found('stripe.com', 'John', 'Doe'),550 domain: 'stripe.com',551 firstName: 'John',552 lastName: 'Doe',553 source: 'tomba_email_finder',554 phoneNumbers: 0,555 charged: true,556 chargedCredits: 1,557 cached: false,558 },559 ]);560
561 const phones = await actor.call('/?company=Stripe&full_name=Jane%20Smith&enrichMobile=true');562 const [item] = phones.body.items as Record<string, unknown>[];563 assert.equal(item.inputCompany, 'Stripe');564 assert.equal(item.fullName, 'Jane Smith');565 assert.equal(item.phoneNumbers, 2);566 assert.equal(item.chargedCredits, 6);567 assert.equal(server.requests[1].query.enrich_mobile, 'true');568
569 const empty = await actor.call('/?domain=empty.com&firstName=John&lastName=Doe');570 const [miss] = empty.body.items as Record<string, unknown>[];571 assert.equal(miss.error, 'No email found for this person');572 assert.equal(miss.charged, false);573 } finally {574 stopped = await actor.stop();575 }576 assert.deepEqual(stopped.chargeCounts, { 'tomba-request': 7 });577 });578
579 it('accepts a POST with the same JSON input as a normal run', async () => {580 const server = await mock();581 const actor = await startStandbyActor({ endpoint: server.url });582 try {583 const res = await actor.call('/', {584 body: { requests: [person('stripe.com'), person('shopify.com', 'Jane', 'Smith'), { company: 'X' }] },585 });586 assert.equal(res.status, 200);587 const items = res.body.items as Record<string, unknown>[];588 assert.equal(items.length, 3);589 assert.equal(items.filter((i) => i.email).length, 2);590 assert.equal(server.requests.length, 2);591 } finally {592 await actor.stop();593 }594 });595
596 it('serves repeated requests from the cache for free', async () => {597 const server = await mock();598 const actor = await startStandbyActor({ endpoint: server.url });599 let stopped;600 try {601 await actor.call('/?domain=stripe.com&fullName=John%20Doe');602 const second = await actor.call('/?domain=stripe.com&fullName=John%20Doe');603 assert.ok((second.body.items as Record<string, unknown>[]).every((i) => i.cached === true));604 assert.equal(server.requests.length, 1);605 } finally {606 stopped = await actor.stop();607 }608 assert.deepEqual(stopped.chargeCounts, { 'tomba-request': 1 });609 });610
611 it('keeps serving after a request hits maxResults', async () => {612 const server = await mock();613 const actor = await startStandbyActor({ endpoint: server.url });614 try {615 const first = await actor.call('/', {616 body: { requests: [person('a.com'), person('b.com')], maxResults: 1 },617 });618 assert.equal((first.body.items as unknown[]).length, 1);619 const second = await actor.call('/?domain=c.com&firstName=John&lastName=Doe');620 assert.equal((second.body.items as unknown[]).length, 1);621 } finally {622 await actor.stop();623 }624 });625
626 it('rejects invalid input with 400 and unknown paths with 404', async () => {627 const server = await mock();628 const actor = await startStandbyActor({ endpoint: server.url });629 try {630 assert.equal((await actor.call('/', { body: {} })).status, 400);631 assert.equal((await actor.call('/', { body: 'not json' })).status, 400);632 assert.equal((await actor.call('/?maxResults=abc&domain=a.com')).status, 400);633 assert.equal((await actor.call('/?enrichMobile=maybe&domain=a.com&fullName=John%20Doe')).status, 400);634 assert.equal((await actor.call('/?domain=a.com&fullName=John%20Doe&webhookUrl=ftp://x')).status, 400);635 assert.equal((await actor.call('/nope')).status, 404);636 assert.equal((await actor.call('/', { method: 'DELETE' })).status, 405);637 assert.equal(server.requests.length, 0);638 } finally {639 await actor.stop();640 }641 });642
643 it('returns 402 once the max charge limit is reached', async () => {644 const server = await mock();645 const actor = await startStandbyActor({ endpoint: server.url, maxTotalChargeUsd: 1 });646 try {647 const first = await actor.call('/?domain=a.com&firstName=John&lastName=Doe');648 assert.equal(first.status, 200);649 const second = await actor.call('/?domain=b.com&firstName=John&lastName=Doe');650 assert.equal(second.status, 402);651 assert.equal(server.requests.length, 1);652 } finally {653 await actor.stop();654 }655 });656});1// Shared runner for the Tomba Apify Actors: batch runs and Standby (real-time HTTP API) mode.2// Keep this file identical across all Actors.3import type { IncomingMessage, ServerResponse } from 'node:http';4import { createServer } from 'node:http';5
6import { Actor, log } from 'apify';7
8import type { RunOptions } from './tomba.js';9import { isStopped, logSummary, setupTomba, useRunState } from './tomba.js';10
11/** Invalid input: a failed batch run, or `400 Bad Request` in Standby mode. */12export class InputError extends Error {}13
14export type Item = Record<string, unknown>;15
16export interface RunContext {17 /** Save result items (dataset in batch runs, HTTP response in Standby mode). */18 push: (items: Item | Item[]) => Promise<void>;19 /** True when this input was already processed (batch runs resume after a migration or restart). */20 isDone: (key: string) => boolean;21 /** Record a processed input so a resumed run skips it. */22 markDone: (key: string) => void;23 standby: boolean;24}25
26/** Processes one Actor input. Throws InputError for invalid input. */27export type ActorRun<I> = (input: I, ctx: RunContext) => Promise<void>;28
29export interface ActorDefinition<I> {30 /** Name used in logs. */31 title: string;32 run: ActorRun<I>;33 /** Builds the Actor input from `GET /?…` query parameters in Standby mode. */34 fromQuery: (query: URLSearchParams) => I;35 /** Number of inputs, for the batch summary log. */36 count?: (input: I) => number;37}38
39const MAX_BODY_BYTES = 1_000_000;40
41export function isStandby(): boolean {42 return Actor.config.get('metaOrigin') === 'STANDBY';43}44
45/** Entry point: runs the Actor as a batch job, or as an HTTP server in Standby mode. */46export async function runActor<I extends RunOptions>(definition: ActorDefinition<I>): Promise<void> {47 await Actor.init();48
49 if (isStandby()) {50 await setupTomba();51 await startStandbyServer(definition);52 return;53 }54
55 const input = ((await Actor.getInput<I>()) ?? {}) as I;56 await setupTomba(input);57 const state = await useRunState();58 const startedAt = Date.now();59 try {60 await definition.run(input, {61 push: async (items) => Actor.pushData(items),62 isDone: (key) => Boolean(state.done[key]),63 markDone: (key) => {64 state.done[key] = true;65 },66 standby: false,67 });68 } catch (err) {69 if (err instanceof InputError) {70 await Actor.fail(err.message);71 return;72 }73 throw err;74 }75 logSummary(definition.title, definition.count?.(input) ?? 0, startedAt);76 await Actor.exit();77}78
79async function startStandbyServer<I>(definition: ActorDefinition<I>): Promise<void> {80 const server = createServer((req, res) => {81 handleRequest(definition, req, res).catch((err: Error) => {82 log.exception(err, 'Standby request failed');83 send(res, 500, { error: 'Internal error' });84 });85 });86 const port = Actor.config.get('containerPort');87 await new Promise<void>((resolve) => {88 server.listen(port, resolve);89 });90 log.info(`${definition.title} is ready for real-time requests`, { port });91}92
93async function handleRequest<I>(definition: ActorDefinition<I>, req: IncomingMessage, res: ServerResponse) {94 if (req.headers['x-apify-container-server-readiness-probe']) {95 send(res, 200, { status: 'ready' });96 return;97 }98
99 const url = new URL(req.url ?? '/', 'http://localhost');100 if (url.pathname !== '/') {101 send(res, 404, { error: 'Not found. Use GET / with query parameters or POST / with a JSON input.' });102 return;103 }104
105 let input: I;106 try {107 if (req.method === 'GET') {108 if ([...url.searchParams.keys()].length === 0) {109 send(res, 200, {110 status: 'ready',111 usage: 'GET /?<parameters> or POST / with the same JSON input as a normal run.',112 });113 return;114 }115 input = definition.fromQuery(url.searchParams);116 } else if (req.method === 'POST') {117 input = (await readJson(req)) as I;118 } else {119 send(res, 405, { error: 'Method not allowed. Use GET or POST.' });120 return;121 }122 } catch (err) {123 send(res, 400, { error: (err as Error).message });124 return;125 }126
127 if (isStopped()) {128 send(res, 402, { error: 'Max charge limit reached for this Actor run.' });129 return;130 }131
132 const items: Item[] = [];133 try {134 await definition.run(input, {135 push: async (pushed) => {136 items.push(...(Array.isArray(pushed) ? pushed : [pushed]));137 },138 isDone: () => false,139 markDone: () => undefined,140 standby: true,141 });142 } catch (err) {143 if (err instanceof InputError) {144 send(res, 400, { error: err.message });145 return;146 }147 throw err;148 }149 send(res, 200, { items });150}151
152async function readJson(req: IncomingMessage): Promise<unknown> {153 const chunks: Buffer[] = [];154 let size = 0;155 for await (const chunk of req) {156 size += (chunk as Buffer).length;157 if (size > MAX_BODY_BYTES) throw new InputError('Request body is too large (max 1 MB).');158 chunks.push(chunk as Buffer);159 }160 const text = Buffer.concat(chunks).toString('utf8').trim();161 if (!text) return {};162 try {163 const body = JSON.parse(text);164 if (typeof body !== 'object' || body === null || Array.isArray(body)) {165 throw new InputError('Request body must be a JSON object.');166 }167 return body;168 } catch (err) {169 if (err instanceof InputError) throw err;170 throw new InputError('Request body must be valid JSON.');171 }172}173
174function send(res: ServerResponse, status: number, body: unknown): void {175 if (res.headersSent) return;176 res.writeHead(status, { 'content-type': 'application/json; charset=utf-8' });177 res.end(JSON.stringify(body));178}179
180// --- Query string helpers for `fromQuery` ---------------------------------------------------181
182/** Values of a list parameter: repeated (`?domain=a&domain=b`) and/or comma-separated (`?domains=a,b`). */183export function queryList(query: URLSearchParams, ...names: string[]): string[] {184 return names185 .flatMap((name) => query.getAll(name))186 .flatMap((value) => value.split(','))187 .map((value) => value.trim())188 .filter(Boolean);189}190
191export function queryString(query: URLSearchParams, name: string): string | undefined {192 const value = query.get(name)?.trim();193 return value || undefined;194}195
196export function queryBool(query: URLSearchParams, name: string): boolean | undefined {197 const value = query.get(name)?.trim().toLowerCase();198 if (value === undefined || value === '') return undefined;199 if (['1', 'true', 'yes'].includes(value)) return true;200 if (['0', 'false', 'no'].includes(value)) return false;201 throw new InputError(`Query parameter "${name}" must be true or false.`);202}203
204export function queryInt(query: URLSearchParams, name: string): number | undefined {205 const value = query.get(name)?.trim();206 if (value === undefined || value === '') return undefined;207 const number = Number(value);208 if (!Number.isInteger(number)) throw new InputError(`Query parameter "${name}" must be an integer.`);209 return number;210}1// Shared Tomba helper for the Tomba Apify Actors.2// Keep this file identical across all Actors.3import { createHash } from 'node:crypto';4
5import type { KeyValueStore } from 'apify';6import { Actor, log } from 'apify';7import { TombaClient, TombaException } from 'tomba';8
9/** Pay-per-event event names (configured in Apify Console → Monetization). */10export const EVENT_REQUEST = 'tomba-request';11export const EVENT_PHONE_FINDER = 'phone-finder-request';12
13/**14 * Tomba credits added per phone number (or per address with phone data, depending on the endpoint)15 * returned when `enrich_mobile=true`. Every credit is charged as one `tomba-request` event.16 */17export const PHONE_CREDITS = 5;18
19/** Number of phone numbers in a record's `phone_data` array (0 when missing or empty). */20export function phoneDataCount(record: unknown): number {21 if (!isObject(record)) return 0;22 return Array.isArray(record.phone_data) ? record.phone_data.length : 0;23}24
25/** True when a record has at least one phone number in `phone_data`. */26export function hasPhoneData(record: unknown): boolean {27 return phoneDataCount(record) > 0;28}29
30export interface RunOptions {31 maxConcurrency?: number;32 maxRetries?: number;33 useCache?: boolean;34 cacheTtlHours?: number;35}36
37export interface TombaCallResult {38 /** Raw JSON body returned by Tomba (`{ data, meta, ... }`). */39 body?: Record<string, unknown>;40 /** Shortcut to `body.data`. */41 data?: unknown;42 /** HTTP status (0 for network errors, 502 for a non-JSON body). */43 status: number;44 charged: boolean;45 /** Number of events charged for this call (0 when not charged). */46 chargedCount?: number;47 cached: boolean;48 /** True when the call was not made because the user's max charge limit was reached. */49 skipped?: boolean;50 error?: string;51}52
53export const stats = {54 requests: 0,55 charged: 0,56 cached: 0,57 failed: 0,58 retries: 0,59};60
61const CACHE_STORE_PREFIX = 'tomba-cache';62const BASE_DELAY_MS = 1000;63const MAX_DELAY_MS = 30000;64const NETWORK_ERROR = /ECONN|ETIMEDOUT|ENOTFOUND|EAI_AGAIN|EPIPE|socket hang up|timeout|network/i;65
66let client: TombaClient | undefined;67let options: Required<RunOptions> = { maxConcurrency: 10, maxRetries: 3, useCache: true, cacheTtlHours: 24 };68let cacheStore: KeyValueStore | undefined;69const memoryCache = new Map<string, Record<string, unknown>>();70const inFlight = new Map<string, number>();71let stopped = false;72
73/** Initialise the Tomba client from the Actor's environment variables and apply the run options. */74export async function setupTomba(runOptions: RunOptions = {}): Promise<TombaClient> {75 const key = process.env.TOMBA_API_KEY;76 const secret = process.env.TOMBA_API_SECRET;77 if (!key || !secret) {78 await Actor.fail('Actor is misconfigured: Tomba credentials are missing. Please contact the Actor developer.');79 throw new Error('unreachable');80 }81
82 options = {83 maxConcurrency: clamp(runOptions.maxConcurrency ?? 10, 1, 50),84 maxRetries: clamp(runOptions.maxRetries ?? 3, 0, 10),85 useCache: runOptions.useCache ?? true,86 cacheTtlHours: Math.max(0, runOptions.cacheTtlHours ?? 24),87 };88
89 client = new TombaClient();90 // Verification and enrichment can take over a minute, so keep the SDK's 120 s timeout.91 client.setKey(key).setSecret(secret).setTimeout(120_000);92 // Only used by the test suite to point the client at a local mock server.93 if (process.env.TOMBA_API_ENDPOINT) client.setEndpoint(process.env.TOMBA_API_ENDPOINT);94
95 if (options.useCache && options.cacheTtlHours > 0) {96 try {97 cacheStore = await Actor.openKeyValueStore(cacheStoreName());98 } catch (err) {99 // E.g. "Insufficient permissions" under limited permissions: never fail the run because of the cache.100 log.warning('Cross-run cache is unavailable; results are cached for this run only.', {101 error: (err as Error).message,102 });103 cacheStore = undefined;104 }105 }106
107 return client;108}109
110/**111 * Name of the cross-run cache store. It is per Actor: with limited permissions an Actor can only open112 * named storages it created itself, so the Tomba Actors must not share one store.113 */114export function cacheStoreName(): string {115 const actorId = Actor.getEnv().actorId ?? process.env.ACTOR_ID;116 return actorId ? `${CACHE_STORE_PREFIX}-${actorId}` : CACHE_STORE_PREFIX;117}118
119/** The Tomba client created by setupTomba(). */120export function getClient(): TombaClient {121 if (!client) throw new Error('setupTomba() must be called first');122 return client;123}124
125export function getOptions(): Required<RunOptions> {126 return options;127}128
129/** True once the user's max-charge budget is exhausted; workers should stop picking up new items. */130export function isStopped(): boolean {131 return stopped;132}133
134export function stop(): void {135 stopped = true;136}137
138/**139 * Tomba billing rules: charge only when Tomba returns a JSON body without an `errors` object140 * and with a non-empty `data` (a negative answer such as an undeliverable verification141 * or a domain search with no emails is still a non-empty `data` object).142 */143export function isBillable(body: unknown): boolean {144 if (!isObject(body)) return false;145 if (body.errors !== undefined && body.errors !== null) return false;146 const { data } = body;147 if (data === null || data === undefined) return false;148 if (Array.isArray(data)) return data.length > 0;149 // An object whose values are all null (e.g. `{ person: null, company: null }`) is an empty result.150 if (isObject(data)) return Object.values(data).some((v) => v !== null && v !== undefined);151 return data !== '' && data !== false;152}153
154/**155 * Perform one Tomba API call with caching, retry and pay-per-event charging.156 *157 * @param endpoint short endpoint name used for the cache key (e.g. 'technology')158 * @param params request parameters used for the cache key159 * @param fn the SDK call160 * @param eventName pay-per-event name to charge when the response is billable161 * @param count events to charge for a billable response: a number, or a function of the response body162 */163export async function callTomba(164 endpoint: string,165 params: Record<string, unknown>,166 fn: (tomba: TombaClient) => Promise<unknown>,167 eventName: string = EVENT_REQUEST,168 count: number | ((body: Record<string, unknown>) => number) = 1,169): Promise<TombaCallResult> {170 if (!client) throw new Error('setupTomba() must be called first');171
172 const cacheKey = hashKey(endpoint, params);173 const cachedBody = await readCache(cacheKey);174 if (cachedBody) {175 stats.cached++;176 return { body: cachedBody, data: cachedBody.data, status: 200, charged: false, cached: true };177 }178
179 // Reserve budget before calling Tomba so concurrent workers never exceed the user's max charge.180 // A fixed count is known up front, so reserve all of it; a body-dependent count reserves at least 1.181 const reserved = typeof count === 'number' ? Math.max(1, Math.ceil(count)) : 1;182 if (!reserve(eventName, reserved)) {183 stop();184 return { status: 0, charged: false, cached: false, skipped: true, error: 'Max charge limit reached' };185 }186
187 try {188 for (let attempt = 0; ; attempt++) {189 try {190 stats.requests++;191 const body = unwrap(await fn(client));192
193 if (!isObject(body)) {194 stats.failed++;195 return { status: 502, charged: false, cached: false, error: 'Invalid response from Tomba' };196 }197
198 if (!isBillable(body)) {199 const error = errorMessage(body) ?? undefined;200 return { body, data: body.data, status: 200, charged: false, cached: false, error };201 }202
203 const units = Math.max(1, Math.ceil(typeof count === 'function' ? count(body) : count));204 const chargedCount = await charge(eventName, units);205 const charged = chargedCount > 0;206 await writeCache(cacheKey, body);207 return { body, data: body.data, status: 200, charged, chargedCount, cached: false };208 } catch (err) {209 const { status, message, retryAfter } = parseError(err);210 const retryable = status === 0 || status === 429 || status >= 500;211 if (retryable && attempt < options.maxRetries) {212 stats.retries++;213 const delay = retryAfter ?? Math.min(MAX_DELAY_MS, BASE_DELAY_MS * 2 ** attempt);214 const jitter = Math.random() * 250;215 log.warning(`Tomba ${endpoint} returned ${status || 'network error'}, retrying in ${delay}ms`, {216 attempt: attempt + 1,217 });218 await sleep(delay + jitter);219 continue;220 }221 stats.failed++;222 return { status, charged: false, cached: false, error: message };223 }224 }225 } finally {226 release(eventName, reserved);227 }228}229
230/**231 * Run `worker` over `items` with bounded concurrency. Stops picking new items once the charge limit232 * is reached (`isStopped()`) or `shouldStop()` returns true (e.g. maxResults reached for this run).233 */234export async function runPool<T>(235 items: T[],236 worker: (item: T, index: number) => Promise<void>,237 concurrency = options.maxConcurrency,238 shouldStop: () => boolean = () => false,239): Promise<void> {240 let next = 0;241 const runners = Array.from({ length: Math.min(concurrency, items.length) }, async () => {242 while (!stopped && !shouldStop() && next < items.length) {243 const index = next++;244 await worker(items[index], index);245 }246 });247 await Promise.all(runners);248}249
250/** Persisted run state used to resume after a migration, restart or resurrection. */251export async function useRunState(): Promise<{ done: Record<string, true> }> {252 return Actor.useState<{ done: Record<string, true> }>('TOMBA_STATE', { done: {} });253}254
255export function logSummary(title: string, total: number, startedAt: number): void {256 const seconds = Math.round((Date.now() - startedAt) / 1000);257 log.info(`${title} finished`, {258 inputs: total,259 tombaRequests: stats.requests,260 charged: stats.charged,261 cacheHits: stats.cached,262 failed: stats.failed,263 retries: stats.retries,264 seconds,265 stoppedByChargeLimit: stopped,266 });267}268
269export function normalizeDomain(value: string): string {270 return value271 .trim()272 .toLowerCase()273 .replace(/^https?:\/\//, '')274 .replace(/^www\./, '')275 .replace(/\/.*$/, '');276}277
278export function normalizeEmail(value: string): string {279 return value.trim().toLowerCase();280}281
282export function unique<T>(items: T[], key: (item: T) => string = String): T[] {283 const seen = new Set<string>();284 return items.filter((item) => {285 const k = key(item);286 if (!k || seen.has(k)) return false;287 seen.add(k);288 return true;289 });290}291
292// ---------------------------------------------------------------------------293
294async function charge(eventName: string, count: number): Promise<number> {295 const result = await Actor.charge({ eventName, count });296 if (result.chargedCount > 0) stats.charged += result.chargedCount;297 if (result.eventChargeLimitReached) {298 log.info('Max charge limit reached, finishing current items and stopping.');299 stop();300 }301 return result.chargedCount;302}303
304function reserve(eventName: string, units: number): boolean {305 if (stopped) return false;306 const manager = Actor.getChargingManager();307 const remaining = manager.calculateMaxEventChargeCountWithinLimit(eventName);308 const current = inFlight.get(eventName) ?? 0;309 if (remaining - current < units) return false;310 inFlight.set(eventName, current + units);311 return true;312}313
314function release(eventName: string, units: number): void {315 inFlight.set(eventName, Math.max(0, (inFlight.get(eventName) ?? units) - units));316}317
318async function readCache(key: string): Promise<Record<string, unknown> | undefined> {319 if (!options.useCache) return undefined;320 const hit = memoryCache.get(key);321 if (hit) return hit;322 if (!cacheStore) return undefined;323 const entry = await cacheStore.getValue<{ savedAt: number; body: Record<string, unknown> }>(key);324 if (!entry) return undefined;325 if (Date.now() - entry.savedAt > options.cacheTtlHours * 3600_000) return undefined;326 memoryCache.set(key, entry.body);327 return entry.body;328}329
330async function writeCache(key: string, body: Record<string, unknown>): Promise<void> {331 if (!options.useCache) return;332 memoryCache.set(key, body);333 if (cacheStore) {334 try {335 await cacheStore.setValue(key, { savedAt: Date.now(), body });336 } catch (err) {337 log.debug('Failed to write cache entry', { error: (err as Error).message });338 }339 }340}341
342function hashKey(endpoint: string, params: Record<string, unknown>): string {343 const sorted = Object.keys(params)344 .sort()345 .map((k) => [k, params[k]]);346 return `${endpoint}-${createHash('sha256').update(JSON.stringify(sorted)).digest('hex').slice(0, 40)}`;347}348
349/** tomba >= 1.1.0 resolves to `{ data: body, rateLimit }`; older versions resolve to the body itself. */350function unwrap(response: unknown): unknown {351 if (isObject(response) && 'rateLimit' in response && 'data' in response) return response.data;352 return response;353}354
355function parseError(err: unknown): { status: number; message: string; retryAfter?: number } {356 if (err instanceof TombaException) {357 const response = err.code as { status?: number; headers?: Record<string, string> } | undefined;358 // Without an HTTP response it is either a network error (retryable, status 0)359 // or the SDK's own input validation (not retryable, treated as 400).360 const networkError = NETWORK_ERROR.test(err.message ?? '');361 let status = networkError ? 0 : 400;362 if (typeof response?.status === 'number') status = response.status;363 const retryAfterHeader = Number(response?.headers?.['retry-after']);364 const retryAfter =365 Number.isFinite(retryAfterHeader) && retryAfterHeader > 0 ? retryAfterHeader * 1000 : undefined;366 const message = errorMessage(err.response) ?? err.message ?? 'Tomba request failed';367 return { status, message: status ? `${status}: ${message}` : message, retryAfter };368 }369 return { status: 0, message: err instanceof Error ? err.message : String(err) };370}371
372function errorMessage(body: unknown): string | undefined {373 if (!isObject(body)) return typeof body === 'string' && body.length < 300 ? body : undefined;374 const errors = body.errors as Record<string, unknown> | undefined;375 if (isObject(errors) && typeof errors.message === 'string') return errors.message;376 return undefined;377}378
379function isObject(value: unknown): value is Record<string, unknown> {380 return typeof value === 'object' && value !== null && !Array.isArray(value);381}382
383function clamp(value: number, min: number, max: number): number {384 return Math.min(max, Math.max(min, Math.floor(value)));385}386
387async function sleep(ms: number): Promise<void> {388 return new Promise((resolve) => {389 setTimeout(resolve, ms);390 });391}1import { log } from 'apify';2import { Finder } from 'tomba';3
4import { InputError, queryBool, queryInt, queryString, runActor } from './standby.js';5import type { RunOptions } from './tomba.js';6import {7 callTomba,8 getClient,9 hasPhoneData,10 isBillable,11 normalizeDomain,12 PHONE_CREDITS,13 phoneDataCount,14 runPool,15 unique,16} from './tomba.js';17
18interface EmailFinderRequest {19 domain?: string;20 company?: string;21 fullName?: string;22 firstName?: string;23 lastName?: string;24}25
26type NormalizedRequest = Required<EmailFinderRequest>;27
28interface ActorInput extends RunOptions {29 requests?: EmailFinderRequest[];30 maxResults?: number;31 enrichMobile?: boolean;32 webhookUrl?: string;33}34
35/** Parameters accepted by `Finder.emailFinder()` in tomba 1.1.1. */36interface EmailFinderParams extends EmailFinderRequest {37 enrichMobile?: boolean;38 webhook_url?: string;39}40
41const SOURCE = 'tomba_email_finder';42
43const clean = (value: unknown) => (typeof value === 'string' ? value.trim().replace(/\s+/g, ' ') : '');44const requestKey = (r: NormalizedRequest) =>45 [r.domain, r.company, r.fullName, r.firstName, r.lastName].map((v) => v.toLowerCase()).join('|');46
47/** Tomba: 1 search credit, or 6 when phone data is returned (`enrich_mobile=true`). */48const credits = (body: Record<string, unknown>) => 1 + (hasPhoneData(body.data) ? PHONE_CREDITS : 0);49
50await runActor<ActorInput>({51 title: 'Email Finder',52 count: (input) => input.requests?.length ?? 0,53 fromQuery: (query) => {54 const request: EmailFinderRequest = {55 domain: queryString(query, 'domain'),56 company: queryString(query, 'company'),57 firstName: queryString(query, 'firstName') ?? queryString(query, 'first_name'),58 lastName: queryString(query, 'lastName') ?? queryString(query, 'last_name'),59 fullName: queryString(query, 'fullName') ?? queryString(query, 'full_name'),60 };61 const hasRequest = Object.values(request).some(Boolean);62 return {63 requests: hasRequest ? [request] : [],64 enrichMobile: queryBool(query, 'enrichMobile'),65 webhookUrl: queryString(query, 'webhookUrl'),66 maxResults: queryInt(query, 'maxResults'),67 };68 },69 run: async (input, { push, isDone, markDone, standby }) => {70 if (!Array.isArray(input.requests) || !input.requests.length) {71 throw new InputError('Input must contain at least one request in "requests".');72 }73
74 const maxResults = input.maxResults ?? 50;75 const enrichMobile = input.enrichMobile ?? false;76 const webhookUrl = typeof input.webhookUrl === 'string' ? input.webhookUrl.trim() : '';77 if (webhookUrl && !/^https?:\/\//.test(webhookUrl)) {78 throw new InputError('"webhookUrl" must start with http:// or https://.');79 }80
81 const finder = new Finder(getClient());82
83 const requests = unique(84 input.requests.map(85 (r): NormalizedRequest => ({86 domain: normalizeDomain(clean(r?.domain)),87 company: clean(r?.company),88 fullName: clean(r?.fullName),89 firstName: clean(r?.firstName),90 lastName: clean(r?.lastName),91 }),92 ),93 requestKey,94 );95 const doneCount = requests.filter((r) => isDone(requestKey(r))).length;96 const pending = requests.filter((r) => !isDone(requestKey(r))).slice(0, Math.max(0, maxResults - doneCount));97 if (doneCount > 0) {98 log.info(`Resuming: ${doneCount} requests already processed.`);99 }100 if (!standby) {101 log.info(`Finding emails for ${pending.length} requests${enrichMobile ? ' (with phone numbers)' : ''}`);102 }103
104 await runPool(pending, async (request) => {105 const key = requestKey(request);106 const { domain, company, fullName, firstName, lastName } = request;107
108 // Input echo. The input company is `inputCompany` because `company` is Tomba's own output field.109 const echo = {110 domain,111 firstName,112 lastName,113 ...(fullName ? { fullName } : {}),114 ...(company ? { inputCompany: company } : {}),115 };116 const label = `${fullName || `${firstName} ${lastName}`.trim()} @ ${domain || company}`;117
118 if (!(domain || company) || !(fullName || (firstName && lastName))) {119 await push({120 ...echo,121 source: SOURCE,122 phoneNumbers: 0,123 charged: false,124 chargedCredits: 0,125 cached: false,126 error: 'Missing required fields: a domain or company, and a fullName or firstName and lastName are required',127 });128 log.info(`Skipping invalid request: ${label}`);129 markDone(key);130 return;131 }132
133 const params: EmailFinderParams = {};134 if (domain) params.domain = domain;135 if (company) params.company = company;136 if (fullName) params.fullName = fullName;137 if (firstName) params.firstName = firstName;138 if (lastName) params.lastName = lastName;139 if (enrichMobile) params.enrichMobile = true;140 if (webhookUrl) params.webhook_url = webhookUrl;141
142 const res = await callTomba(143 'email-finder',144 { ...params },145 async () => finder.emailFinder(params),146 undefined,147 credits,148 );149 if (res.skipped) return;150
151 const chargedCredits = res.chargedCount ?? 0;152 if (isBillable(res.body)) {153 const data = res.data as Record<string, unknown>;154 const phoneNumbers = phoneDataCount(data);155 await push({156 ...data,157 ...echo,158 source: SOURCE,159 phoneNumbers,160 charged: res.charged,161 chargedCredits,162 cached: res.cached,163 });164 const email = typeof data.email === 'string' && data.email ? data.email : 'no email found';165 const phones = phoneNumbers ? `, ${phoneNumbers} phone numbers` : '';166 log.info(`${label}: ${email}${phones}${res.cached ? ' (cached)' : ''}`);167 } else {168 await push({169 ...echo,170 source: SOURCE,171 phoneNumbers: 0,172 charged: res.charged,173 chargedCredits,174 cached: res.cached,175 error: res.error ?? 'No email found for this person',176 });177 log.info(`${label}: ${res.error ?? 'no email found'}`);178 }179
180 markDone(key);181 });182 },183});name: Test and push latest version
on: push: branches: - main - master pull_request: workflow_dispatch:
concurrency: group: ${{ github.workflow }}-${{ github.ref }} cancel-in-progress: true
jobs: test: name: Format, lint, build and test runs-on: ubuntu-latest steps: - uses: actions/checkout@v4
- name: Set up Node.js uses: actions/setup-node@v4 with: node-version: 22 cache: npm
- name: Install dependencies run: npm ci
- name: Check formatting run: npm run format:check
- name: Lint run: npm run lint
- name: Build run: npm run build
- name: Test run: npm test
push: name: Push Actor to Apify needs: test # Only publish from the default branch, never from pull requests. if: github.event_name != 'pull_request' && (github.ref == 'refs/heads/main' || github.ref == 'refs/heads/master') runs-on: ubuntu-latest steps: - uses: actions/checkout@v4
# .actor/actor.json maps TOMBA_API_KEY / TOMBA_API_SECRET to the Apify CLI secrets # @tombaApiKey / @tombaApiSecret, which `apify push` reads from ~/.apify/secrets.json. - name: Provide Tomba secrets to the Apify CLI env: TOMBA_API_KEY: ${{ secrets.TOMBA_API_KEY }} TOMBA_API_SECRET: ${{ secrets.TOMBA_API_SECRET }} run: | if [ -z "$TOMBA_API_KEY" ] || [ -z "$TOMBA_API_SECRET" ]; then echo "::error::Add the TOMBA_API_KEY and TOMBA_API_SECRET repository secrets." exit 1 fi mkdir -p ~/.apify node -e 'require("fs").writeFileSync(require("os").homedir() + "/.apify/secrets.json", JSON.stringify({ tombaApiKey: process.env.TOMBA_API_KEY, tombaApiSecret: process.env.TOMBA_API_SECRET }))'
- name: Push Actor to Apify id: push uses: apify/push-actor-action@v1 with: token: ${{ secrets.APIFY_TOKEN }}
- name: Print build details run: | echo "Build ${{ steps.push.outputs.build-id }} finished with status ${{ steps.push.outputs.build-status }}"