config.example.yaml

# config.example.yaml — LeadHunter Pro Enricher Configuration
output_format: "xlsx"

http_timeout:           [4, 6]
playwright_timeout:     8000
browser_restart_every:  150
stop_at:                ""        # "" = disabled; set to "HH:MM" (24h) to auto-stop e.g. "23:00"
autosave_interval:      60

# [IMP-1] Concurrent worker count for HTTP enrichment
enricher_workers: 5

rate_limit:
  min_seconds: 0.1
  max_seconds: 0.5

# [IMP-2] TLDs that should be flagged as geo-suspect (e.g. ['in', 'pk', 'ru'])
GEO_SUSPECT_TLDS: []

# [IMP-3] Keywords that boost the lead score if found in the URL
score_boost_keywords: []

skip_email_keywords:
  - noreply
  - no-reply
  - donotreply
  - privacy
  - dataprotection
  - data-protection
  - gdpr
  - unsubscribe
  - postmaster
  - webmaster
  - bounce
  - complaints
  - legal
  - abuse
  - spam
  - newsletter

generic_email_keywords:
  - info
  - admin
  - hello
  - contact
  - enquiries
  - enquiry
  - office
  - mail
  - email
  - team
  - support
  - help
  - sales
  - accounts
  - finance
  - general
  - service
  - post

junk_email_domains:
  - sentry.io
  - wixpress.com
  - example.com
  - schema.org
  - w3.org
  - googleapis.com
  - cloudflare.com
  - jquery.com
  # [IMP-4] Placeholder domains
  - placeholder.com
  - domain.com
  - email.com
  - doe.com
  - test.com
  - mailinator.com
  - yopmail.com
  - tempmail.com
  - trashmail.com

contact_paths:
  - /contact
  - /contact-us
  - /about
  - /about-us

locale: "en-US"

cookie_selectors:
  - 'button:has-text("Accept all")'
  - 'button:has-text("Accept cookies")'
  - 'button:has-text("Accept")'
  - 'button:has-text("I Accept")'
  - 'button:has-text("Allow all")'
  - 'button:has-text("OK")'
  - 'button:has-text("Got it")'
  - '[id*="accept"]'
  - '[aria-label*="Accept"]'