✨ rework confidence scores to be easily comparable

Author: ianardee

lib/mindee/parsing/v2/field/base_field.rb

@@ -20,13 +20,16 @@ class BaseField
           # @param indent_level [Integer] Level of indentation for rst display.
           def initialize(raw_prediction, indent_level = 0)
             @indent_level = indent_level
-            @confidence = raw_prediction.key?('confidence') ? raw_prediction['confidence'] : nil
-            @locations = if raw_prediction.key?('locations')
-                           raw_prediction['locations'].map do |location|
+            confidence = raw_prediction['confidence']
+            @confidence = FieldConfidence.new(confidence) unless confidence.nil? || confidence.empty?
+
+            locations = raw_prediction['locations']
+            @locations = if locations.nil? || locations.empty?
+                           []
+                         else
+                           locations.map do |location|
                              FieldLocation.new(location)
                            end
-                         else
-                           []
                          end
           end
 

lib/mindee/parsing/v2/field/field_confidence.rb

@@ -19,70 +19,106 @@ class FieldConfidence
           LOW = 'Low'
 
           # List of valid values, as frozen strings.
-          VALID_VALUES = [CERTAIN, HIGH, MEDIUM, LOW].freeze
+          VALID_VALUES = ['Certain', 'High', 'Medium', 'Low'].freeze
 
           # @param value [String] The confidence level value.
           # @raise [ArgumentError] If the value is not a valid confidence level.
           def initialize(value)
-            unless VALID_VALUES.include?(value)
+            case value
+            when 'Certain' then @value = CERTAIN
+            when 'High' then @value = HIGH
+            when 'Medium' then @value = MEDIUM
+            when 'Low' then @value = LOW
+            else
               raise ArgumentError,
-                    "Invalid confidence level: #{value}. Must be one of: #{VALID_VALUES.join(', ')}"
+                    "Invalid confidence level: '#{value}'. Must be one of: #{VALID_VALUES.join(', ')}"
             end
 
             @value = value
           end
 
-          # Create a FieldConfidence from a string value.
-          # @param value [String] The confidence level string.
-          # @return [FieldConfidence] The confidence instance.
-          def self.from_string(value)
-            new(value)
+          # String representation of the confidence level.
+          # @return [String] The confidence level value.
+          def to_s
+            @value
           end
 
-          # Check if this is a certain confidence level.
-          # @return [Boolean] `true` if confidence is certain.
-          def certain?
-            @value == CERTAIN
+          # String representation of the confidence level.
+          # @return [Integer] The confidence level value as an integer: 1 is LOW, 4 is HIGH.
+          def to_i
+            val_to_i(@value)
           end
 
-          # Check if this is a high confidence level.
-          # @return [Boolean] `true` if confidence is high.
-          def high?
-            @value == HIGH
+          # Inspect method for debugging.
+          # @return [String] Debug representation.
+          def inspect
+            "#<#{self.class.name}:#{@value}>"
           end
 
-          # Check if this is a medium confidence level.
-          # @return [Boolean] `true` if confidence is medium.
-          def medium?
-            @value == MEDIUM
-          end
+          # rubocop:disable Style/CaseLikeIf
 
-          # Check if this is a low confidence level.
-          # @return [Boolean] `true` if confidence is low.
-          def low?
-            @value == LOW
+          # Equality of two FieldConfidence instances.
+          # @param other [String, Integer, FieldConfidence] The other confidence to compare.
+          # @return [Boolean] `true` if they have the same value.
+          def ==(other)
+            if other.is_a?(FieldConfidence)
+              @value == other.value
+            elsif other.is_a?(String)
+              @value == other
+            elsif other.is_a?(Integer)
+              to_i == other
+            else
+              raise ArgumentError, "Invalid type: #{other.class}"
+            end
           end
 
-          # String representation of the confidence level.
-          # @return [String] The confidence level value.
-          def to_s
-            @value
+          # Greater than or equality of two FieldConfidence instances.
+          # @param other [String, Integer, FieldConfidence] The other confidence to compare.
+          def >=(other)
+            if other.is_a?(FieldConfidence)
+              to_i >= val_to_i(other.value)
+            elsif other.is_a?(String)
+              to_i >= val_to_i(other)
+            elsif other.is_a?(Integer)
+              to_i >= other
+            else
+              raise ArgumentError, "Invalid type: #{other.class}"
+            end
           end
 
-          # Compare two FieldConfidence instances.
-          # @param other [FieldConfidence] The other confidence to compare.
-          # @return [Boolean] `true` if they have the same value.
-          def ==(other)
-            other.is_a?(FieldConfidence) && @value == other.value
+          # less than or equality of two FieldConfidence instances.
+          # # @param other [String, Integer, FieldConfidence] The other confidence to compare.
+          def <=(other)
+            if other.is_a?(FieldConfidence)
+              to_i <= val_to_i(other.value)
+            elsif other.is_a?(String)
+              to_i <= val_to_i(other)
+            elsif other.is_a?(Integer)
+              to_i <= other
+            else
+              raise ArgumentError, "Invalid type: #{other.class}"
+            end
           end
 
-          # Make instances comparable and hashable
+          # rubocop:enable Style/CaseLikeIf
+
+          # Aliases for the comparison operators.
           alias eql? ==
+          alias gteql? >=
+          alias lteql? <=
 
-          # Inspect method for debugging.
-          # @return [String] Debug representation.
-          def inspect
-            "#<#{self.class.name}:#{@value}>"
+          protected
+
+          def val_to_i(value)
+            case value
+            when CERTAIN then 4
+            when HIGH then 3
+            when MEDIUM then 2
+            when LOW then 1
+            else
+              raise ArgumentError,
+                    "Invalid confidence level: '#{value}'. Must be one of: #{VALID_VALUES.join(', ')}"
+            end
           end
         end
       end

sig/mindee/parsing/v2/field/field_confidence.rbs

@@ -12,14 +12,17 @@ module Mindee
           VALID_VALUES: Array[String]
           def initialize: (String) -> void
           def self.from_string: (String) -> FieldConfidence
-          def certain?: -> bool
-          def high?: -> bool
-          def medium?: -> bool
-          def low?: -> bool
+          def to_i: -> Integer
           def to_s: -> String
-          def ==: (FieldConfidence) -> bool
-          def eql?: (FieldConfidence) -> bool
+          def ==: (String | Integer | FieldConfidence) -> bool
+          def eql?: (String | Integer | FieldConfidence) -> bool
+          def <=: (String | Integer | FieldConfidence) -> bool
+          def lteql? : (String | Integer | FieldConfidence) -> bool
+          def >=: (String | Integer | FieldConfidence) -> bool
+          def gteql?: (String | Integer | FieldConfidence) -> bool
           def inspect: -> String
+
+          def val_to_i: (String) -> Integer
         end
       end
     end

spec/parsing/v2/inference_spec.rb

@@ -22,7 +22,7 @@ def load_v2_inference(resource_path)
   simple_field = Mindee::Parsing::V2::Field::SimpleField
   object_field = Mindee::Parsing::V2::Field::ObjectField
   list_field   = Mindee::Parsing::V2::Field::ListField
-  field_conf   = Mindee::Parsing::V2::Field::FieldConfidence
+  field_confidence = Mindee::Parsing::V2::Field::FieldConfidence
 
   describe 'simple' do
     it 'loads a blank inference with valid properties' do
@@ -264,19 +264,20 @@ def load_v2_inference(resource_path)
       expect(polygon[3][0]).to be_within(1e-12).of(0.948849)
       expect(polygon[3][1]).to be_within(1e-12).of(0.244565)
 
-      # Confidence can be a FieldConfidence instance or a String depending on implementation.
-      conf_value =
-        if date_field.confidence.respond_to?(:to_s)
-          date_field.confidence.to_s
-        else
-          date_field.confidence
-        end
-      expect(conf_value).to eq('Medium')
-
-      # Optional strict check if equality supports comparing with FieldConfidence constants:
-      if defined?(field_conf) && field_conf.respond_to?(:from_string)
-        expect(conf_value).to eq(field_conf.from_string('Medium').to_s)
-      end
+      confidence = date_field.confidence
+      expect(confidence).to be_a(field_confidence)
+      # equality
+      expect(confidence).to eq(field_confidence::MEDIUM)
+      expect(confidence).to eq('Medium')
+      expect(confidence).to eq(2)
+      # less than or equal
+      expect(confidence).to be_lteql(field_confidence::HIGH)
+      expect(confidence).to be_lteql('High')
+      expect(confidence).to be_lteql(3)
+      # greater than or equal
+      expect(confidence).to be_gteql(field_confidence::LOW)
+      expect(confidence).to be_gteql('Low')
+      expect(confidence).to be_gteql(1)
     end
   end
 end