mindee 4.12.0 → 5.0.0.beta1

data/lib/mindee/parsing/v2/field/object_field.rb

@@ -1,135 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'base_field'
-require_relative 'inference_fields'
-
-module Mindee
-  module Parsing
-    module V2
-      module Field
-        # A field containing a nested set of inference fields.
-        class ObjectField < BaseField
-          # @return [InferenceFields] Fields contained in the object.
-          attr_reader :fields
-
-          # @param raw_prediction [Hash] Raw server response hash.
-          # @param indent_level [Integer] Level of indentation for rst display.
-          def initialize(raw_prediction, indent_level = 0)
-            super
-
-            inner_fields = raw_prediction.fetch('fields', raw_prediction)
-            @fields = InferenceFields.new(inner_fields, @indent_level + 1)
-          end
-
-          # String representation of the object field.
-          # @return [String] String representation with newline prefix.
-          def to_s
-            return "\n" unless @fields && !@fields.empty?
-
-            "\n#{@fields.to_s(1)}"
-          end
-
-          # String representation suitable for list display.
-          # @return [String] String representation without leading spaces.
-          def to_s_from_list
-            return '' if @fields.nil?
-            return '' unless @fields && !@fields.empty?
-
-            field_str = @fields.to_s(2)
-            field_str.length > 4 ? (field_str[4..] || '') : ''
-          end
-
-          # String representation of a single object field
-          # @return [String] Multi-line string with proper indentation.
-          def single_str
-            return '' unless @fields && !@fields.empty?
-
-            out_str = ''
-            indent = ' ' * @indent_level
-
-            @fields.each do |field_key, field_value|
-              value_str = field_value && !field_value.to_s.empty? ? field_value.to_s : ''
-              out_str += "\n#{indent}  :#{field_key}: #{value_str}"
-            end
-
-            out_str
-          end
-
-          # String representation for multi-object display
-          # @return [String] Formatted string for list context.
-          def multi_str
-            return '' unless @fields && !@fields.empty?
-
-            out_str = ''
-            indent = ' ' * @indent_level
-            first = true
-
-            @fields.each do |field_key, field_value|
-              value_str = field_value ? field_value.to_s : ''
-
-              if first
-                out_str += "#{indent}:#{field_key}: #{value_str}"
-                first = false
-              else
-                out_str += "\n#{indent}    :#{field_key}: #{value_str}"
-              end
-            end
-
-            out_str
-          end
-
-          # Get a field by key and ensure it is a SimpleField.
-          # @param key [String] Field key to retrieve.
-          # @return [SimpleField] The SimpleField.
-          # @raise [TypeError] If the field is not a SimpleField.
-          def get_simple_field(key)
-            @fields.get_simple_field(key)
-          end
-
-          # Get a field by key and ensure it is a ListField.
-          # @param key [String] Field key to retrieve.
-          # @return [ListField] The ListField.
-          # @raise [TypeError] If the field is not a ListField.
-          def get_list_field(key)
-            field = @fields.get_list_field(key)
-            raise TypeError, "Field #{key} is not a ListField" unless field.is_a?(ListField)
-
-            field
-          end
-
-          # Get a field by key and ensure it is an ObjectField.
-          # @param key [String] Field key to retrieve.
-          # @return [ObjectField] The ObjectField.
-          # @raise [TypeError] If the field is not an ObjectField.
-          def get_object_field(key)
-            field = @fields.get_object_field(key)
-            raise TypeError, "Field #{key} is not an ObjectField" unless field.is_a?(ObjectField)
-
-            field
-          end
-
-          # Get all simple fields.
-          # @return [Hash<String, SimpleField>] Simple fields contained in the object.
-          def simple_fields
-            # @type var fields: Hash<String, SimpleField>
-            @fields.select { |_, value| value.is_a?(SimpleField) }
-          end
-
-          # Get all list fields.
-          # @return [Hash<String, ListField>] List fields contained in the object.
-          def list_fields
-            # @type var fields: Hash<String, ListField>
-            @fields.select { |_, value| value.is_a?(ListField) }
-          end
-
-          # Get all object fields.
-          # @return [Hash<String, ObjectField>] Object fields contained in the object.
-          def object_fields
-            # @type var fields: Hash<String, ObjectField>
-            @fields.select { |_, value| value.is_a?(ObjectField) }
-          end
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/field/simple_field.rb

@@ -1,60 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'base_field'
-
-module Mindee
-  module Parsing
-    module V2
-      module Field
-        # A simple field with a scalar value.
-        class SimpleField < BaseField
-          # @return [String, Integer, Float, Boolean, nil] Value contained in the field.
-          attr_reader :value
-
-          # @param server_response [Hash] Raw server response hash.
-          # @param indent_level [Integer] Level of indentation for rst display.
-          def initialize(server_response, indent_level = 0)
-            super
-            value = server_response['value']
-            @value = if value.is_a?(Integer)
-                       value.to_f
-                     else
-                       value
-                     end
-          end
-
-          # String representation of the field value.
-          # @return [String] String representation of the value.
-          def to_s
-            return '' if @value.nil?
-
-            if @value.is_a?(TrueClass) || @value.is_a?(FalseClass)
-              @value ? 'True' : 'False'
-            elsif @value.is_a?(Integer) || @value.is_a?(Float)
-              num = @value # @type var num: Integer | Float
-              format_numeric_value(num)
-            else
-              @value.to_s
-            end
-          end
-
-          private
-
-          # Format numeric values to display '.0' in string reps.
-          # @param value [Numeric] The numeric value to format.
-          # @return [String] Formatted numeric string.
-          def format_numeric_value(value)
-            float_value = value.to_f
-
-            if float_value == float_value.to_i
-              format('%.1f', float_value)
-            else
-              formatted = format('%.5f', float_value)
-              formatted.sub(%r{\.?0+$}, '')
-            end
-          end
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference.rb

@@ -1,57 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'inference_job'
-require_relative 'inference_model'
-require_relative 'inference_file'
-require_relative 'inference_result'
-require_relative 'inference_active_options'
-
-module Mindee
-  module Parsing
-    module V2
-      # Complete data returned by an inference request.
-      class Inference
-        # @return [String] Identifier of the inference (when provided by API).
-        attr_reader :id
-        # @return [InferenceJob] Metadata about the job.
-        attr_reader :job
-        # @return [InferenceModel] Information about the model used.
-        attr_reader :model
-        # @return [InferenceFile] Information about the processed file.
-        attr_reader :file
-        # @return [InferenceActiveOptions] Options which were activated during the inference.
-        attr_reader :active_options
-        # @return [InferenceResult] Result contents.
-        attr_reader :result
-
-        # @param server_response [Hash] Hash representation of the JSON returned by the service.
-        def initialize(server_response)
-          raise ArgumentError, 'server_response must be a Hash' unless server_response.is_a?(Hash)
-
-          @model  = InferenceModel.new(server_response['model'])
-          @job    = InferenceJob.new(server_response['job']) if server_response.key?('job')
-          @file   = InferenceFile.new(server_response['file'])
-          @active_options = InferenceActiveOptions.new(server_response['active_options'])
-          @result = InferenceResult.new(server_response['result'])
-
-          @id = server_response['id']
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          [
-            'Inference',
-            '#########',
-            @job.to_s,
-            @model.to_s,
-            @file.to_s,
-            @active_options.to_s,
-            @result.to_s,
-            '',
-          ].join("\n")
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference_active_options.rb

@@ -1,67 +0,0 @@
-# frozen_string_literal: true
-
-module Mindee
-  module Parsing
-    module V2
-      # Data schema options activated during the inference.
-      class DataSchemaActiveOption
-        # @return [Boolean]
-        attr_reader :replace
-
-        # @param server_response [Hash]
-        def initialize(server_response)
-          @replace = server_response[:replace] || server_response['replace']
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          "Data Schema\n-----------\n:Replace: #{@replace ? 'True' : 'False'}"
-        end
-      end
-
-      # Options which were activated during the inference.
-      class InferenceActiveOptions
-        # @return [Boolean] Whether the Raw Text feature was activated.
-        attr_reader :raw_text
-        # @return [Boolean] Whether the polygon feature was activated.
-        attr_reader :polygon
-        # @return [Boolean] Whether the confidence feature was activated.
-        attr_reader :confidence
-        # @return [Boolean] Whether the Retrieval-Augmented Generation feature was activated.
-        attr_reader :rag
-        # @return [Boolean] Whether the text context feature was activated.
-        attr_reader :text_context
-        # @return [DataSchemaActiveOption]
-        attr_reader :data_schema
-
-        # @param server_response [Hash] Raw JSON parsed into a Hash.
-        def initialize(server_response)
-          @raw_text = server_response['raw_text']
-          @polygon = server_response['polygon']
-          @confidence = server_response['confidence']
-          @rag = server_response['rag']
-          @text_context = server_response['text_context']
-          @data_schema = DataSchemaActiveOption.new(server_response['data_schema'])
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          parts = [
-            'Active Options',
-            '==============',
-            ":Raw Text: #{@raw_text ? 'True' : 'False'}",
-            ":Polygon: #{@polygon ? 'True' : 'False'}",
-            ":Confidence: #{@confidence ? 'True' : 'False'}",
-            ":RAG: #{@rag ? 'True' : 'False'}",
-            ":Text Context: #{@text_context ? 'True' : 'False'}\n",
-            @data_schema.to_s,
-            '',
-          ]
-          parts.join("\n")
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference_file.rb

@@ -1,38 +0,0 @@
-# frozen_string_literal: true
-
-module Mindee
-  module Parsing
-    module V2
-      # Information about a file returned in an inference.
-      class InferenceFile
-        # @return [String] File name.
-        attr_reader :name
-        # @return [String, nil] Optional alias for the file.
-        attr_reader :file_alias
-        # @return [Integer] Page count.
-        attr_reader :page_count
-        # @return [String] MIME type.
-        attr_reader :mime_type
-
-        # @param server_response [Hash] Raw JSON parsed into a Hash.
-        def initialize(server_response)
-          @name       = server_response['name']
-          @file_alias = server_response['alias']
-          @page_count = server_response['page_count']
-          @mime_type  = server_response['mime_type']
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          "File\n" \
-            "====\n" \
-            ":Name: #{@name}\n" \
-            ":Alias:#{" #{@file_alias}" if @file_alias}\n" \
-            ":Page Count: #{@page_count}\n" \
-            ":MIME Type: #{@mime_type}\n"
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference_job.rb

@@ -1,25 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'common_response'
-require_relative 'job'
-
-module Mindee
-  module Parsing
-    module V2
-      # HTTP response wrapper that embeds a V2 job.
-      class InferenceJob
-        # @return [String] UUID of the Job.
-        attr_reader :id
-
-        def initialize(server_response)
-          @id = server_response['id']
-        end
-
-        # @return [String] String representation of the job.
-        def to_s
-          "Job\n===\n:ID: #{@id}\n"
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference_model.rb

@@ -1,30 +0,0 @@
-# frozen_string_literal: true
-
-module Mindee
-  module Parsing
-    module V2
-      # ID of the model that produced the inference.
-      class InferenceModel
-        # @return [String] Identifier of the model.
-        attr_reader :id
-
-        # @param server_response [Hash] Raw JSON parsed into a Hash.
-        def initialize(server_response)
-          @id = server_response['id']
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          parts = [
-            'Model',
-            '=====',
-            ":ID: #{@id}",
-            '',
-          ]
-          parts.join("\n")
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference_response.rb

@@ -1,30 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'common_response'
-require_relative 'inference'
-
-module Mindee
-  module Parsing
-    module V2
-      # HTTP response wrapper that embeds a V2 Inference.
-      class InferenceResponse < CommonResponse
-        # @return [Inference] Parsed inference payload.
-        attr_reader :inference
-
-        # @param server_response [Hash] Hash parsed from the API JSON response.
-        def initialize(server_response)
-          # CommonResponse takes care of the generic metadata (status, etc.)
-          super
-
-          @inference = Inference.new(server_response['inference'])
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          @inference.to_s
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/inference_result.rb

@@ -1,42 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'field/inference_fields'
-require_relative 'raw_text'
-require_relative 'rag_metadata'
-
-module Mindee
-  module Parsing
-    module V2
-      # Wrapper for the result of a V2 inference request.
-      class InferenceResult
-        # @return [Mindee::Parsing::V2::Field::InferenceFields] Fields produced by the model.
-        attr_reader :fields
-        # @return [Mindee::Parsing::V2::RawText, nil] Optional extra data.
-        attr_reader :raw_text
-        # @return [Mindee::Parsing::V2::RAGMetadata, nil] Optional RAG metadata.
-        attr_reader :rag
-
-        # @param server_response [Hash] Hash version of the JSON returned by the API.
-        def initialize(server_response)
-          raise ArgumentError, 'server_response must be a Hash' unless server_response.is_a?(Hash)
-
-          @fields = Field::InferenceFields.new(server_response['fields'])
-
-          @raw_text = server_response['raw_text'] ? RawText.new(server_response['raw_text']) : nil
-          @rag = (V2::RAGMetadata.new(server_response['rag']) if server_response.key?('rag') && server_response['rag'])
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          parts = [
-            'Fields',
-            '======',
-            @fields.to_s,
-          ]
-          parts.join("\n")
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/job.rb

@@ -1,93 +0,0 @@
-# frozen_string_literal: true
-
-require 'date'
-require_relative 'error_response'
-require_relative 'job_webhook'
-
-module Mindee
-  module Parsing
-    module V2
-      # Metadata returned when polling a job (asynchronous request).
-      class Job
-        # @return [String] Unique job identifier.
-        attr_reader :id
-        # @return [Time] Date and time of the Job creation.
-        attr_reader :created_at
-        # @return [Time, nil] Date and time of the Job completion. Filled once processing is finished.
-        attr_reader :completed_at
-        # @return [String] Identifier of the model used.
-        attr_reader :model_id
-        # @return [String] Name of the processed file.
-        attr_reader :filename
-        # @return [String] Optional alias for the file.
-        attr_reader :alias
-        # @return [String, nil] Current status (submitted, processing, done, …).
-        attr_reader :status
-        # @return [String] URL to query for updated status.
-        attr_reader :polling_url
-        # @return [String, nil] URL that redirects to the final result once ready.
-        attr_reader :result_url
-        # @return [Array<JobWebhook>] Webhooks triggered by the job.
-        attr_reader :webhooks
-        # @return [ErrorResponse, nil] Error details when the job failed.
-        attr_reader :error
-
-        # rubocop:disable Metrics/CyclomaticComplexity
-        # @param server_response [Hash] Parsed JSON payload from the API.
-        def initialize(server_response)
-          raise ArgumentError, 'server_response must be a Hash' unless server_response.is_a?(Hash)
-
-          @id          = server_response['id']
-          @status      = server_response['status'] if server_response.key?('status')
-          unless server_response['error'].nil? || server_response['error'].empty?
-            @error = ErrorResponse.new(server_response['error'])
-          end
-          @created_at = Time.iso8601(server_response['created_at'])
-          if server_response.key?('completed_at') && !server_response['completed_at'].nil?
-            @completed_at = Time.iso8601(server_response['completed_at'])
-          end
-          @model_id    = server_response['model_id']
-          @polling_url = server_response['polling_url']
-          @filename    = server_response['filename']
-          @result_url  = server_response['result_url']
-          @alias       = server_response['alias']
-          @webhooks = []
-          server_response['webhooks'].each do |webhook|
-            @webhooks.push JobWebhook.new(webhook)
-          end
-        end
-        # rubocop:enable Metrics/CyclomaticComplexity
-
-        # String representation.
-        # @return [String]
-        def to_s
-          lines = [
-            'Job',
-            '###',
-            ":ID: #{@id}",
-            ":CreatedAt: #{@created_at}",
-            ":ModelID: #{@model_id}",
-            ":Filename: #{@filename}",
-            ":Alias: #{@alias}",
-            ":Status: #{@status}",
-            ":PollingURL: #{@polling_url}",
-            ":ResultURL: #{@result_url}",
-          ]
-
-          lines << @error.to_s if @error
-
-          unless @webhooks.empty?
-            lines += [
-              '',
-              'Webhooks',
-              '=========',
-              @webhooks.map(&:to_s).join("\n\n"),
-            ]
-          end
-
-          lines.join("\n")
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/job_response.rb

@@ -1,30 +0,0 @@
-# frozen_string_literal: true
-
-require_relative 'common_response'
-require_relative 'job'
-
-module Mindee
-  module Parsing
-    module V2
-      # HTTP response wrapper that embeds a V2 job.
-      class JobResponse < CommonResponse
-        # @return [Job] Parsed job payload.
-        attr_reader :job
-
-        # @param server_response [Hash] Hash parsed from the API JSON response.
-        def initialize(server_response)
-          # CommonResponse takes care of the generic metadata (status, etc.)
-          super
-
-          @job = Job.new(server_response['job'])
-        end
-
-        # String representation.
-        # @return [String]
-        def to_s
-          @job.to_s
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/job_webhook.rb

@@ -1,60 +0,0 @@
-# lib/mindee/parsing/v2/job_response_webhook.rb
-# frozen_string_literal: true
-
-require 'date'
-require_relative 'error_response'
-
-module Mindee
-  module Parsing
-    module V2
-      # Information about a webhook created for a job response.
-      class JobWebhook
-        # @return [String] Identifier of the webhook.
-        attr_reader :id
-        # @return [DateTime, nil] Creation timestamp (or +nil+ when absent).
-        attr_reader :created_at
-        # @return [String] Webhook status.
-        attr_reader :status
-        # @return [ErrorResponse, nil] Error information when something failed.
-        attr_reader :error
-
-        # @param server_response [Hash] Raw JSON parsed into a Hash.
-        def initialize(server_response)
-          @id = server_response['id']
-          @created_at = parse_date(server_response['created_at'])
-          @status     = server_response['status']
-          @error = ErrorResponse.new(server_response['error']) unless server_response['error'].nil?
-        end
-
-        # RST display.
-        #
-        # @return [String]
-        def to_s
-          parts = [
-            'JobResponseWebhook',
-            '##################',
-            ":ID: #{@id}",
-            ":CreatedAt: #{@created_at}",
-            ":Status: #{@status}",
-          ]
-
-          parts << @error.to_s if @error
-          parts.join("\n")
-        end
-
-        private
-
-        # Safely parse an ISO-8601 date/time string to DateTime.
-        # @param str [String, nil] The date string.
-        # @return [DateTime, nil]
-        def parse_date(str)
-          return nil if str.to_s.empty?
-
-          DateTime.iso8601(str || '')
-        rescue ArgumentError
-          nil
-        end
-      end
-    end
-  end
-end

data/lib/mindee/parsing/v2/rag_metadata.rb

@@ -1,17 +0,0 @@
-# frozen_string_literal: true
-
-module Mindee
-  module Parsing
-    module V2
-      # Metadata about the RAG operation.
-      class RAGMetadata
-        # The UUID of the matched document used during the RAG operation.
-        attr_accessor :retrieved_document_id
-
-        def initialize(server_response)
-          @retrieved_document_id = server_response.fetch('retrieved_document_id', nil)
-        end
-      end
-    end
-  end
-end