iachat/enterprise/app/jobs/captain/documents/crawl_job.rb
Vishnu Narayanan 990d045cb3
fix: handle empty FIRECRAWL_KEY in captain crawl job (#10936)
- handle empty FIRECRAWL_KEY in captain crawl job
2025-02-20 12:27:37 -08:00

52 lines
1.5 KiB
Ruby

class Captain::Documents::CrawlJob < ApplicationJob
queue_as :low
def perform(document)
if InstallationConfig.find_by(name: 'CAPTAIN_FIRECRAWL_API_KEY')&.value.present?
perform_firecrawl_crawl(document)
else
perform_simple_crawl(document)
end
end
private
include Captain::FirecrawlHelper
def perform_simple_crawl(document)
page_links = Captain::Tools::SimplePageCrawlService.new(document.external_link).page_links
page_links.each do |page_link|
Captain::Tools::SimplePageCrawlParserJob.perform_later(
assistant_id: document.assistant_id,
page_link: page_link
)
end
Captain::Tools::SimplePageCrawlParserJob.perform_later(
assistant_id: document.assistant_id,
page_link: document.external_link
)
end
def perform_firecrawl_crawl(document)
captain_usage_limits = document.account.usage_limits[:captain] || {}
document_limit = captain_usage_limits[:documents] || {}
crawl_limit = [document_limit[:current_available] || 10, 500].min
Captain::Tools::FirecrawlService
.new
.perform(
document.external_link,
firecrawl_webhook_url(document),
crawl_limit
)
end
def firecrawl_webhook_url(document)
webhook_url = Rails.application.routes.url_helpers.enterprise_webhooks_firecrawl_url
"#{webhook_url}?assistant_id=#{document.assistant_id}&token=#{generate_firecrawl_token(document.assistant_id, document.account_id)}"
end
end