htm 0.0.30 → 0.0.32

data/examples/examples_helper.rb

@@ -72,6 +72,31 @@ end
 
 # Module with helper methods for examples
 module ExamplesHelper
+  # Reset examples database if --reset flag is present in ARGV
+  #
+  # Drops and recreates htm_examples, then runs all migrations.
+  # Removes --reset from ARGV so downstream option parsers are unaffected.
+  #
+  # @return [void]
+  def self.reset_if_requested!
+    return unless ARGV.delete('--reset')
+
+    db_name = "#{ENV['HTM_SERVICE__NAME'] || 'htm'}_examples"
+    puts "Resetting examples database '#{db_name}'..."
+
+    HTM::SequelConfig.disconnect! if HTM::SequelConfig.db
+
+    abort "ERROR: Failed to drop '#{db_name}'" unless system("dropdb --if-exists #{db_name}")
+    abort "ERROR: Failed to create '#{db_name}'" unless system("createdb #{db_name}")
+
+    HTM::SequelConfig.establish_connection!(load_models: false)
+    HTM::Database.setup
+    HTM::SequelConfig.ensure_models_loaded!
+
+    puts "Database reset complete."
+    puts
+  end
+
   # Check if database is available and configured
   #
   # @return [Boolean] true if database is ready

data/lib/htm/circuit_breaker.rb

@@ -158,7 +158,7 @@ class HTM
           end
         when :closed
           # Reset failure count on success in closed state
-          @failure_count = 0 if @failure_count > 0
+          @failure_count = 0 if @failure_count.positive?
         end
       end
     end
@@ -190,11 +190,10 @@ class HTM
       return unless @state == :open && @last_failure_time
 
       elapsed = Time.now - @last_failure_time
-      if elapsed >= @reset_timeout
-        @state = :half_open
-        @success_count = 0
-        HTM.logger.info "CircuitBreaker[#{@name}]: Reset timeout elapsed (#{@reset_timeout}s), circuit HALF-OPEN"
-      end
+      return unless elapsed >= @reset_timeout
+      @state = :half_open
+      @success_count = 0
+      HTM.logger.info "CircuitBreaker[#{@name}]: Reset timeout elapsed (#{@reset_timeout}s), circuit HALF-OPEN"
     end
   end
 end

data/lib/htm/config/builder.rb

@@ -31,7 +31,7 @@ class HTM
           response = RubyLLM.embed(text, model: model)
           embedding = extract_embedding_from_response(response)
 
-          unless embedding.is_a?(Array) && embedding.all? { |v| v.is_a?(Numeric) }
+          unless embedding.is_a?(Array) && embedding.all?(Numeric)
             raise HTM::EmbeddingError, "Invalid embedding response format from #{embedding_provider}"
           end
 
@@ -119,17 +119,17 @@ class HTM
 
       def parse_tag_response(text)
         tags = text.to_s.split("\n").map(&:strip).reject(&:empty?)
-        valid_tags = tags.select { |tag| tag =~ /^[a-z0-9\-]+(:[a-z0-9\-]+)*$/ }
+        valid_tags = tags.grep(/^[a-z0-9-]+(:[a-z0-9-]+)*$/)
         valid_tags.select { |tag| tag.count(':') < max_tag_depth }
       end
 
       def parse_proposition_response(text)
         text.to_s
-          .split("\n")
-          .map(&:strip)
-          .map { |line| line.sub(/^[-*]\s*/, '') }
-          .map(&:strip)
-          .reject(&:empty?)
+            .split("\n")
+            .map(&:strip)
+            .map { |line| line.sub(/^[-*]\s*/, '') }
+            .map(&:strip)
+            .reject(&:empty?)
       end
 
       # ==========================================================================
@@ -138,11 +138,11 @@ class HTM
 
       def build_tag_extraction_prompt(text, existing_ontology)
         taxonomy_context = if existing_ontology.any?
-          sample_tags = existing_ontology.sample([existing_ontology.size, 20].min)
-          tag.taxonomy_context_existing % { sample_tags: sample_tags.join(', ') }
-        else
-          tag.taxonomy_context_empty
-        end
+                             sample_tags = existing_ontology.sample([existing_ontology.size, 20].min)
+                             tag.taxonomy_context_existing % { sample_tags: sample_tags.join(', ') }
+                           else
+                             tag.taxonomy_context_empty
+                           end
 
         tag.user_prompt_template % {
           text: text,

data/lib/htm/config/database.rb

@@ -87,9 +87,9 @@ class HTM
 
         unless database_configured?
           raise HTM::ConfigurationError,
-            "No database configured for environment '#{environment}'. " \
-            "Set HTM_DATABASE__URL or HTM_DATABASE__NAME, " \
-            "or add database.name to the '#{environment}:' section in your config."
+                "No database configured for environment '#{environment}'. " \
+                "Set HTM_DATABASE__URL or HTM_DATABASE__NAME, " \
+                "or add database.name to the '#{environment}:' section in your config."
         end
 
         true
@@ -165,16 +165,16 @@ class HTM
         return true if actual == expected
 
         raise HTM::ConfigurationError,
-          "Database name '#{actual}' does not match expected '#{expected}'.\n" \
-          "Database names must follow the convention: {service_name}_{environment}\n" \
-          "  Service name: #{service_name}\n" \
-          "  Environment:  #{environment}\n" \
-          "  Expected:     #{expected}\n" \
-          "  Actual:       #{actual}\n\n" \
-          "Either:\n" \
-          "  - Set HTM_DATABASE__URL to point to '#{expected}'\n" \
-          "  - Set HTM_DATABASE__NAME=#{expected}\n" \
-          "  - Change HTM_ENV to match the database suffix"
+              "Database name '#{actual}' does not match expected '#{expected}'.\n" \
+              "Database names must follow the convention: {service_name}_{environment}\n  " \
+              "Service name: #{service_name}\n  " \
+              "Environment:  #{environment}\n  " \
+              "Expected:     #{expected}\n  " \
+              "Actual:       #{actual}\n\n" \
+              "Either:\n  " \
+              "- Set HTM_DATABASE__URL to point to '#{expected}'\n  " \
+              "- Set HTM_DATABASE__NAME=#{expected}\n  " \
+              "- Change HTM_ENV to match the database suffix"
       end
 
       # Check if the database name matches the expected convention
@@ -216,23 +216,17 @@ class HTM
 
       def build_database_url
         return nil unless database.name && !database.name.empty?
-
-        auth = if database.user && !database.user.empty?
-          database.password && !database.password.empty? ? "#{database.user}:#{database.password}@" : "#{database.user}@"
-        else
-          ''
-        end
-
-        url = "postgresql://#{auth}#{database.host}:#{database.port}/#{database.name}"
-
-        # Add sslmode as query parameter if set
-        if database.sslmode && !database.sslmode.empty?
-          url += "?sslmode=#{database.sslmode}"
-        end
-
+        url = "postgresql://#{database_auth_segment}#{database.host}:#{database.port}/#{database.name}"
+        url += "?sslmode=#{database.sslmode}" if database.sslmode && !database.sslmode.empty?
         url
       end
 
+      def database_auth_segment
+        return '' unless database.user && !database.user.empty?
+        return "#{database.user}@" unless database.password && !database.password.empty?
+        "#{database.user}:#{database.password}@"
+      end
+
       # ==========================================================================
       # Database Configuration Reconciliation
       # ==========================================================================

data/lib/htm/config/defaults.yml

@@ -49,7 +49,7 @@ defaults:
   # Access: HTM.config.embedding.provider, HTM.config.embedding.model, etc.
   # ---------------------------------------------------------------------------
   embedding:
-    provider: ollama
+    provider: :ollama
     model: nomic-embed-text:latest
     dimensions: 768
     timeout: 120
@@ -65,7 +65,7 @@ defaults:
   #   %{taxonomy_context} - existing taxonomy info or new taxonomy message
   # ---------------------------------------------------------------------------
   tag:
-    provider: ollama
+    provider: :ollama
     model: gemma3:latest
     timeout: 180
     max_depth: 4
@@ -110,7 +110,7 @@ defaults:
   #   %{text} - the content to extract propositions from
   # ---------------------------------------------------------------------------
   proposition:
-    provider: ollama
+    provider: :ollama
     model: gemma3:latest
     timeout: 180
     enabled: true
@@ -164,11 +164,12 @@ defaults:
 
   # ---------------------------------------------------------------------------
   # Chunking Configuration (for file loading)
-  # Access: HTM.config.chunking.size, HTM.config.chunking.overlap
+  # Access: HTM.config.chunking.chunk_size, HTM.config.chunking.chunk_overlap
+  # Note: Using chunk_size/chunk_overlap to avoid collision with Enumerable#size
   # ---------------------------------------------------------------------------
   chunking:
-    size: 1024
-    overlap: 64
+    chunk_size: 1024
+    chunk_overlap: 64
 
   # ---------------------------------------------------------------------------
   # Circuit Breaker Configuration
@@ -195,16 +196,16 @@ defaults:
   # Access: HTM.config.job.backend
   # ---------------------------------------------------------------------------
   job:
-    backend: fiber
+    backend: :fiber
 
   # ---------------------------------------------------------------------------
   # General Settings
   # Access: HTM.config.week_start, HTM.config.connection_timeout, etc.
   # ---------------------------------------------------------------------------
-  week_start: sunday
+  week_start: :sunday
   connection_timeout: 60
   telemetry_enabled: false
-  log_level: info
+  log_level: :info
 
   # ---------------------------------------------------------------------------
   # Provider Credentials
@@ -250,7 +251,7 @@ defaults:
 development:
   database:
     name: htm_development
-  log_level: debug
+  log_level: :debug
 
 # =============================================================================
 # Test Environment Overrides
@@ -259,8 +260,8 @@ test:
   database:
     name: htm_test
   job:
-    backend: inline
-  log_level: warn
+    backend: :inline
+  log_level: :warn
   telemetry_enabled: false
 
 # =============================================================================
@@ -270,5 +271,16 @@ production:
   database:
     pool_size: 25
     sslmode: require
-  log_level: warn
+  log_level: :warn
   telemetry_enabled: true
+
+# =============================================================================
+# Examples Environment Overrides (for running example scripts)
+# =============================================================================
+examples:
+  database:
+    name: htm_examples
+  job:
+    backend: :inline
+  log_level: :info
+  telemetry_enabled: false

data/lib/htm/config/validator.rb

@@ -27,32 +27,28 @@ class HTM
       def validate_provider(name, value)
         return if value.nil?
 
-        unless SUPPORTED_PROVIDERS.include?(value)
-          raise_validation_error("#{name} must be one of: #{SUPPORTED_PROVIDERS.join(', ')} (got #{value.inspect})")
-        end
+        return if SUPPORTED_PROVIDERS.include?(value)
+        raise_validation_error("#{name} must be one of: #{SUPPORTED_PROVIDERS.join(', ')} (got #{value.inspect})")
       end
 
       def validate_job_backend
         return unless job_backend
 
-        unless SUPPORTED_JOB_BACKENDS.include?(job_backend)
-          raise_validation_error("job.backend must be one of: #{SUPPORTED_JOB_BACKENDS.join(', ')} (got #{job_backend.inspect})")
-        end
+        return if SUPPORTED_JOB_BACKENDS.include?(job_backend)
+        raise_validation_error("job.backend must be one of: #{SUPPORTED_JOB_BACKENDS.join(', ')} (got #{job_backend.inspect})")
       end
 
       def validate_week_start
-        unless SUPPORTED_WEEK_STARTS.include?(week_start)
-          raise_validation_error("week_start must be one of: #{SUPPORTED_WEEK_STARTS.join(', ')} (got #{week_start.inspect})")
-        end
+        return if SUPPORTED_WEEK_STARTS.include?(week_start)
+        raise_validation_error("week_start must be one of: #{SUPPORTED_WEEK_STARTS.join(', ')} (got #{week_start.inspect})")
       end
 
       def validate_relevance_weights
         total = relevance_semantic_weight + relevance_tag_weight +
                 relevance_recency_weight + relevance_access_weight
 
-        unless (0.99..1.01).cover?(total)
-          raise_validation_error("relevance weights must sum to 1.0 (got #{total})")
-        end
+        return if (0.99..1.01).cover?(total)
+        raise_validation_error("relevance weights must sum to 1.0 (got #{total})")
       end
 
       def validate_callables
@@ -68,15 +64,13 @@ class HTM
           raise HTM::ValidationError, "proposition_extractor must be callable"
         end
 
-        unless @token_counter.respond_to?(:call)
-          raise HTM::ValidationError, "token_counter must be callable"
-        end
+        return if @token_counter.respond_to?(:call)
+        raise HTM::ValidationError, "token_counter must be callable"
       end
 
       def validate_logger
-        unless @logger.respond_to?(:info) && @logger.respond_to?(:warn) && @logger.respond_to?(:error)
-          raise HTM::ValidationError, "logger must respond to :info, :warn, and :error"
-        end
+        return if @logger.respond_to?(:info) && @logger.respond_to?(:warn) && @logger.respond_to?(:error)
+        raise HTM::ValidationError, "logger must respond to :info, :warn, and :error"
       end
     end
   end