Refactor cassette matching logic to use provider names directly and improve precision in file matching

Author: crmne

lib/tasks/vcr.rake

@@ -29,13 +29,8 @@ def record_for_providers(providers, cassette_dir) # rubocop:disable Metrics/AbcS
     return
   end
 
-  # Get URL patterns from the providers themselves
-  provider_patterns = get_provider_patterns(providers)
-
-  puts "Finding cassettes for providers: #{providers.join(', ')}"
-
   # Find and delete matching cassettes
-  cassettes_to_delete = find_matching_cassettes(cassette_dir, provider_patterns)
+  cassettes_to_delete = find_matching_cassettes(cassette_dir, providers)
 
   if cassettes_to_delete.empty?
     puts 'No cassettes found for the specified providers.'
@@ -51,33 +46,21 @@ def record_for_providers(providers, cassette_dir) # rubocop:disable Metrics/AbcS
   puts 'Please review the updated cassettes for sensitive information.'
 end
 
-def get_provider_patterns(providers) # rubocop:disable Metrics/MethodLength
-  provider_patterns = {}
-
-  providers.each do |provider_name|
-    provider_module = RubyLLM::Provider.providers[provider_name.to_sym]
-    next unless provider_module
-
-    # Extract the base URL from the provider's api_base method
-    api_base = provider_module.api_base.to_s
-
-    # Create a regex pattern from the domain
-    next unless api_base && !api_base.empty?
-
-    domain = URI.parse(api_base).host
-    pattern = Regexp.new(Regexp.escape(domain))
-    provider_patterns[provider_name] = pattern
-  end
-
-  provider_patterns
-end
-
-def find_matching_cassettes(dir, patterns)
+def find_matching_cassettes(dir, providers) # rubocop:disable Metrics/MethodLength
   cassettes = []
 
   Dir.glob("#{dir}/**/*.yml").each do |file|
-    content = File.read(file)
-    cassettes << file if patterns.values.any? { |pattern| content.match?(pattern) }
+    basename = File.basename(file)
+
+    # Precise matching to avoid cross-provider confusion
+    providers.each do |provider|
+      # Match only exact provider prefixes
+      next unless basename =~ /^[^_]*_#{provider}_/ || # For first section like "chat_openai_"
+                  basename =~ /_#{provider}_[^_]+_/    # For middle sections like "_openai_gpt4_"
+
+      cassettes << file
+      break
+    end
   end
 
   cassettes