mindee 4.12.0 → 5.0.0.beta1

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

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      # 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/v2/parsing/inference_file.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      # 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/v2/parsing/inference_job.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+require_relative 'common_response'
+require_relative 'job'
+
+module Mindee
+  module V2
+    module Parsing
+      # 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/v2/parsing/inference_model.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      # 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/v2/parsing/job.rb

@@ -0,0 +1,93 @@
+# frozen_string_literal: true
+
+require 'date'
+require_relative 'error_response'
+require_relative 'job_webhook'
+
+module Mindee
+  module V2
+    module Parsing
+      # 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.join("\n\n"),
+            ]
+          end
+
+          lines.join("\n")
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require_relative 'common_response'
+require_relative 'job'
+
+module Mindee
+  module V2
+    module Parsing
+      # 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/v2/parsing/job_webhook.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+require 'date'
+require_relative 'error_response'
+
+module Mindee
+  module V2
+    module Parsing
+      # 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/v2/parsing/rag_metadata.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      # 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

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

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      # Raw text extracted from all pages in the document.
+      class RawText
+        # @return [Array[Mindee::V2::Parsing::RawTextPage]] List of pages with their extracted text content.
+        attr_reader :pages
+
+        # @param server_response [Hash] Raw JSON parsed into a Hash.
+        def initialize(server_response)
+          @pages = []
+          server_response.fetch('pages', []).each do |page|
+            @pages.push RawTextPage.new(page)
+          end
+        end
+
+        # String representation.
+        # @return [String]
+        def to_s
+          "#{@pages.join("\n\n")}\n"
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      # Raw text extracted from a single page.
+      class RawTextPage
+        # @return [String] Text content of the page as a single string. '\n' is used to separate lines.
+        attr_reader :content
+
+        # @param server_response [Hash] Raw JSON parsed into a Hash.
+        def initialize(server_response)
+          @content = server_response['content']
+        end
+
+        # String representation.
+        # @return [String]
+        def to_s
+          @content
+        end
+      end
+    end
+  end
+end

data/lib/mindee/v2/parsing/search/pagination_metadata.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      module Search
+        # Pagination Metadata data associated with model search.
+        class PaginationMetadata
+          # @return [Integer] Number of items per page.
+          attr_reader :per_page
+
+          # @return [Integer] 1-indexed page number.
+          attr_reader :page
+
+          # @return [Integer] Total items.
+          attr_reader :total_items
+
+          # @return [Integer] Total number of pages.
+          attr_reader :total_pages
+
+          # @param raw_response [Hash] The parsed JSON payload mapping to pagination metadata.
+          def initialize(raw_response)
+            @per_page = raw_response['per_page']
+            @page = raw_response['page']
+            @total_items = raw_response['total_items']
+            @total_pages = raw_response['total_pages']
+          end
+
+          # String representation of the pagination metadata.
+          # @return [String]
+          def to_s
+            [
+              ":Per Page: #{@per_page}",
+              ":Page: #{@page}",
+              ":Total Items: #{@total_items}",
+              ":Total Pages: #{@total_pages}",
+              '',
+            ].join("\n")
+          end
+        end
+      end
+    end
+  end
+end

data/lib/mindee/v2/parsing/search/search_model.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      module Search
+        # Individual model information.
+        class SearchModel
+          # @return [String] ID of the model.
+          attr_reader :id
+
+          # @return [String] Name of the model.
+          attr_reader :name
+
+          # @return [String] Type of the model.
+          attr_reader :model_type
+
+          # @param payload [Hash] The parsed JSON payload mapping to the search model.
+          def initialize(payload)
+            @id = payload['id']
+            @name = payload['name']
+            @model_type = payload['model_type']
+          end
+
+          # String representation of the model.
+          # @return [String]
+          def to_s
+            [
+              ":Name: #{@name}",
+              ":ID: #{@id}",
+              ":Model Type: #{@model_type}",
+            ].join("\n")
+          end
+        end
+      end
+    end
+  end
+end

data/lib/mindee/v2/parsing/search/search_models.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      module Search
+        # Array of search models.
+        class SearchModels < Array
+          def initialize(prediction)
+            super(prediction.map { |entry| SearchModel.new(entry) })
+          end
+
+          # Default string representation.
+          # @return [String]
+          def to_s
+            return "\n" if empty?
+
+            lines = flat_map do |model|
+              [
+                "* :Name: #{model.name}",
+                "  :ID: #{model.id}",
+                "  :Model Type: #{model.model_type}",
+              ]
+            end
+
+            # Joins all lines with a newline and appends a final newline
+            # to perfectly match the C# StringBuilder output.
+            "#{lines.join("\n")}\n"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/mindee/v2/parsing/search/search_response.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Parsing
+      module Search
+        # Models search response.
+        class SearchResponse < CommonResponse
+          # @return [Search::Search] Parsed search payload.
+          attr_reader :models
+          # @return [Search::Search] Pagination metadata.
+          attr_reader :pagination_metadata
+
+          def initialize(server_response)
+            super
+
+            @models = Search::SearchModels.new(server_response['models'])
+            @pagination_metadata = PaginationMetadata.new(server_response['pagination'])
+          end
+
+          # String representation.
+          # @return [String]
+          def to_s
+            [
+              'Models',
+              '######',
+              models.to_s,
+              'Pagination Metadata',
+              '###################',
+              pagination_metadata.to_s,
+              '',
+            ].join("\n")
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,6 @@
+# frozen_string_literal: true
+
+require_relative 'search/pagination_metadata'
+require_relative 'search/search_model'
+require_relative 'search/search_models'
+require_relative 'search/search_response'

data/lib/mindee/v2/parsing.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+require_relative 'parsing/common_response'
+require_relative 'parsing/error_response'
+require_relative 'parsing/field'
+require_relative 'parsing/inference_file'
+require_relative 'parsing/inference_job'
+require_relative 'parsing/inference_model'
+require_relative 'parsing/inference_active_options'
+require_relative 'parsing/job'
+require_relative 'parsing/job_response'
+require_relative 'parsing/job_webhook'
+require_relative 'parsing/rag_metadata'
+require_relative 'parsing/raw_text'
+require_relative 'parsing/raw_text_page'
+require_relative 'parsing/search'

data/lib/mindee/v2/product/base_product.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+require_relative '../parsing/base_response'
+require_relative '../../input/base_parameters'
+
+module Mindee
+  module V2
+    module Product
+      # Base class for all V2 products.
+      class BaseProduct
+        # @return [String] The slug of the endpoint used for this response
+        @slug = ''
+        # @return [Class<BaseParameters>] The class of the parameters used for this response
+        @params_type = Mindee::Input::BaseParameters
+        # @return [Class<BaseResponse>] The class of the response used for this product
+        @response_type = Mindee::V2::Parsing::BaseResponse
+
+        def initialize
+          raise StandardError, 'Cannot instantiate abstract class.' if instance_of?(BaseProduct)
+        end
+
+        class << self
+          attr_reader :params_type, :slug, :response_type
+        end
+      end
+    end
+  end
+end

data/lib/mindee/v2/product/classification/classification.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+require_relative 'classification_response'
+require_relative 'params/classification_parameters'
+
+module Mindee
+  module V2
+    module Product
+      # Classification module.
+      module Classification
+        # Classification product.
+        class Classification < BaseProduct
+          @slug = 'classification'
+          @params_type = Mindee::V2::Product::Classification::Params::ClassificationParameters
+          @response_type = Mindee::V2::Product::Classification::ClassificationResponse
+        end
+      end
+    end
+  end
+end

data/lib/mindee/v2/product/classification/classification_classifier.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Mindee
+  module V2
+    module Product
+      module Classification
+        # Classification of document type from the source file.
+        class ClassificationClassifier
+          # @return [String] The document type, as identified on given classification values.
+          attr_reader :document_type
+
+          # @param server_response [Hash] Hash representation of the JSON returned by the service.
+          def initialize(server_response)
+            @document_type = server_response['document_type']
+          end
+
+          # @return [String] String representation.
+          def to_s
+            "Document Type: #{@document_type}"
+          end
+        end
+      end
+    end
+  end
+end