fix: Captain response builder not getting triggered (#12729)

## Summary
- Fix captain response builder not getting triggered for cases where
responses are created as completed.

## Testing Instructions 
- Test articles with firecrawl
- Test articles without firecrawl
- Test PDF documents

---------

Co-authored-by: Pranav <pranav@chatwoot.com>

enterprise/app/jobs/captain/tools/firecrawl_parser_job.rb

@@ -5,11 +5,13 @@ class Captain::Tools::FirecrawlParserJob < ApplicationJob
     assistant = Captain::Assistant.find(assistant_id)
     metadata = payload[:metadata]
 
+    canonical_url = normalize_link(metadata['url'])
     document = assistant.documents.find_or_initialize_by(
-      external_link: metadata['url']
+      external_link: canonical_url
     )
 
     document.update!(
+      external_link: canonical_url,
       content: payload[:markdown],
       name: metadata['title'],
       status: :available
@@ -17,4 +19,10 @@ class Captain::Tools::FirecrawlParserJob < ApplicationJob
   rescue StandardError => e
     raise "Failed to parse FireCrawl data: #{e.message}"
   end
+
+  private
+
+  def normalize_link(raw_url)
+    raw_url.to_s.delete_suffix('/')
+  end
 end

enterprise/app/jobs/captain/tools/simple_page_crawl_parser_job.rb

@@ -15,11 +15,11 @@ class Captain::Tools::SimplePageCrawlParserJob < ApplicationJob
     page_title = crawler.page_title || ''
     content = crawler.body_text_content || ''
 
-    document = assistant.documents.find_or_initialize_by(
-      external_link: page_link
-    )
+    normalized_link = normalize_link(page_link)
+    document = assistant.documents.find_or_initialize_by(external_link: normalized_link)
 
     document.update!(
+      external_link: normalized_link,
       name: page_title[0..254], content: content[0..14_999], status: :available
     )
   rescue StandardError => e
@@ -28,6 +28,10 @@ class Captain::Tools::SimplePageCrawlParserJob < ApplicationJob
 
   private
 
+  def normalize_link(raw_link)
+    raw_link.to_s.delete_suffix('/')
+  end
+
   def limit_exceeded?(account)
     limits = account.usage_limits[:captain][:documents]
     limits[:current_available].negative? || limits[:current_available].zero?

enterprise/app/models/captain/document.rb

@@ -37,6 +37,7 @@ class Captain::Document < ApplicationRecord
   validate :validate_file_attachment, if: -> { pdf_file.attached? }
   before_validation :ensure_account_id
   before_validation :set_external_link_for_pdf
+  before_validation :normalize_external_link
 
   enum status: {
     in_progress: 0,
@@ -47,7 +48,7 @@ class Captain::Document < ApplicationRecord
   after_create_commit :enqueue_crawl_job
   after_create_commit :update_document_usage
   after_destroy :update_document_usage
-  after_commit :enqueue_response_builder_job, on: :update, if: :should_enqueue_response_builder?
+  after_commit :enqueue_response_builder_job
   scope :ordered, -> { order(created_at: :desc) }
 
   scope :for_account, ->(account_id) { where(account_id: account_id) }
@@ -94,15 +95,18 @@ class Captain::Document < ApplicationRecord
   end
 
   def enqueue_response_builder_job
-    return if status != 'available'
+    return unless should_enqueue_response_builder?
 
     Captain::Documents::ResponseBuilderJob.perform_later(self)
   end
 
   def should_enqueue_response_builder?
-    # Only enqueue when status changes to available
-    # Avoid re-enqueueing when metadata is updated by the job itself
-    saved_change_to_status? && status == 'available'
+    return false if destroyed?
+    return false unless available?
+
+    return saved_change_to_status? if pdf_document?
+
+    (saved_change_to_status? || saved_change_to_content?) && content.present?
   end
 
   def update_document_usage
@@ -140,4 +144,11 @@ class Captain::Document < ApplicationRecord
     timestamp = Time.current.strftime('%Y%m%d%H%M%S')
     self.external_link = "PDF: #{pdf_file.filename.base}_#{timestamp}"
   end
+
+  def normalize_external_link
+    return if external_link.blank?
+    return if pdf_document?
+
+    self.external_link = external_link.delete_suffix('/')
+  end
 end