retab 0.1.10 → 0.1.12

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6d8f00bc26ed73aa5d559ff12a8903bc8e0902d16aa47ee5bacdc174face30ad
-  data.tar.gz: c152d02486ec700e26c0a59c1ff51e07f15567516bfd06e19899ce1619771e30
+  metadata.gz: 42b59c6b5a798c010207cd26417fffd501f4fcfab09c0c26c72e4b8a913774d4
+  data.tar.gz: c72d4edba45cb7a74dc0d730d9d5eb9bea9070119e7dfee6fa79825c55ca87ea
 SHA512:
-  metadata.gz: b89d753a806f26903c1f1bd3c005a80f92d8b7d940eba048d6f78de16e5241a27a628505eb2c320144dc1eed41579eefd750488007a345ced2368d4e24dec5fa
-  data.tar.gz: 4472ad621a070d8e80e703f3f4f4740d212ff7481966ce3094cb082361f74c648a8f0712bd43a1c34997d91ed8205c5146149b614252b4b607124ab33dba6054
+  metadata.gz: 11f68093fdd2b39eccde81b522e0d876af633a41d81ffe4b987b7b43575c3ee0c311f64b89955610068bafe261268a10907c80a0c219b31cc2ee45507fc7e00f
+  data.tar.gz: 6d6a30db0bbd8a0a23c232535299850755cd46baf4fd2fa953441435d1e36d989972d41faa8228573705179645f94fec6e40c4b8733ecf38e92eaaa49a104d86

data/lib/retab/client.rb

@@ -9,6 +9,10 @@ module Retab
       @classifications ||= Retab::Classifications.new(self)
     end
 
+    def consensus
+      @consensus ||= Retab::Consensus.new(self)
+    end
+
     def edits
       @edits ||= Retab::Edits.new(self)
     end
@@ -45,6 +49,10 @@ module Retab
       @tables ||= Retab::Tables.new(self)
     end
 
+    def usage
+      @usage ||= Retab::Usage.new(self)
+    end
+
     def workflows
       @workflows ||= Retab::Workflows.new(self)
     end

data/lib/retab/consensus/create_consensus_request.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class CreateConsensusRequest < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      include_alignment: :include_alignment,
+      inputs: :inputs,
+      json_schema: :json_schema
+    }.freeze
+
+    attr_accessor(
+      :include_alignment,
+      :inputs,
+      :json_schema
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @include_alignment = hash[:include_alignment]
+      @inputs = (hash[:inputs] || []).map { |item| item || {} }
+      @json_schema = hash[:json_schema] || {}
+    end
+  end
+end

data/lib/retab/consensus/reconciliation_alignment.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconciliationAlignment < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      aligned_inputs: :aligned_inputs,
+      path_alignments: :path_alignments,
+      reference_index: :reference_index
+    }.freeze
+
+    attr_accessor(
+      :aligned_inputs,
+      :path_alignments,
+      :reference_index
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @aligned_inputs = (hash[:aligned_inputs] || []).map { |item| item || {} }
+      @path_alignments = (hash[:path_alignments] || []).map { |item|
+        item ? Retab::ReconciliationPathAlignment.new(item) : nil
+      }
+      @reference_index = hash[:reference_index]
+    end
+  end
+end

data/lib/retab/consensus/reconciliation_field.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconciliationField < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      likelihood: :likelihood,
+      path: :path,
+      supporting_input_count: :supporting_input_count,
+      total_input_count: :total_input_count,
+      value: :value
+    }.freeze
+
+    attr_accessor(
+      :likelihood,
+      :path,
+      :supporting_input_count,
+      :total_input_count,
+      :value
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @likelihood = hash[:likelihood]
+      @path = hash[:path]
+      @supporting_input_count = hash[:supporting_input_count]
+      @total_input_count = hash[:total_input_count]
+      @value = hash[:value]
+    end
+  end
+end

data/lib/retab/consensus/reconciliation_path_alignment.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconciliationPathAlignment < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      canonical_path: :canonical_path,
+      source_paths: :source_paths
+    }.freeze
+
+    attr_accessor(
+      :canonical_path,
+      :source_paths
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @canonical_path = hash[:canonical_path]
+      @source_paths = (hash[:source_paths] || [])
+    end
+  end
+end

data/lib/retab/consensus/reconciliation_result.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconciliationResult < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      alignment: :alignment,
+      consensus: :consensus,
+      fields: :fields,
+      likelihoods: :likelihoods
+    }.freeze
+
+    attr_accessor(
+      :alignment,
+      :consensus,
+      :fields,
+      :likelihoods
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @alignment = hash[:alignment] ? Retab::ReconciliationAlignment.new(hash[:alignment]) : nil
+      @consensus = hash[:consensus] || {}
+      @fields = (hash[:fields] || []).map { |item| item ? Retab::ReconciliationField.new(item) : nil }
+      @likelihoods = hash[:likelihoods] || {}
+    end
+  end
+end

data/lib/retab/consensus.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+require "json"
+
+module Retab
+  class Consensus
+    def initialize(client)
+      @client = client
+    end
+
+    # Create Consensus
+    # @param include_alignment [Boolean, nil]
+    # @param inputs [Array<Hash{String => Object}>, nil]
+    # @param json_schema [Hash{String => Object}, nil]
+    # @param request_options [Hash] (see Retab::Types::RequestOptions)
+    # @return [Retab::ReconciliationResult]
+    def create(
+      inputs:,
+      include_alignment: nil,
+      json_schema: nil,
+      request_options: {}
+    )
+      body = {
+        "include_alignment" => include_alignment,
+        "inputs" => inputs,
+        "json_schema" => json_schema
+      }.compact
+      response = @client.request(
+        method: :post,
+        path: "/v1/consensus",
+        auth: true,
+        body: body,
+        request_options: request_options
+      )
+      result = Retab::ReconciliationResult.new(response.body)
+      result.last_response = Retab::Types::ApiResponse.new(
+        http_status: response.code.to_i,
+        http_headers: response.each_header.to_h,
+        request_id: response["x-request-id"]
+      )
+      result
+    end
+  end
+end

data/lib/retab/experiment_run_metrics/experiment_summary_metrics_response.rb

@@ -15,6 +15,8 @@ module Retab
       score: :score,
       prior_score: :prior_score,
       documents: :documents,
+      scored_document_count: :scored_document_count,
+      total_document_count: :total_document_count,
       aggregate: :aggregate,
       prior_run_id: :prior_run_id
     }.freeze
@@ -29,6 +31,8 @@ module Retab
       :score,
       :prior_score,
       :documents,
+      :scored_document_count,
+      :total_document_count,
       :aggregate,
       :prior_run_id
     )
@@ -47,6 +51,8 @@ module Retab
       @documents = (hash[:documents] || []).map { |item|
         item ? Retab::ExperimentSummaryMetricDocument.new(item) : nil
       }
+      @scored_document_count = hash[:scored_document_count]
+      @total_document_count = hash[:total_document_count]
       @aggregate = hash[:aggregate] ? Retab::ExperimentExtractSummaryAggregate.new(hash[:aggregate]) : nil
       @prior_run_id = hash[:prior_run_id]
     end

data/lib/retab/extractions/extraction_request.rb

@@ -16,7 +16,7 @@ module Retab
       bust_cache: :bust_cache,
       stream: :stream,
       background: :background,
-      chunking_keys: :chunking_keys
+      deep_extraction: :deep_extraction
     }.freeze
 
     attr_accessor(
@@ -30,7 +30,7 @@ module Retab
       :bust_cache,
       :stream,
       :background,
-      :chunking_keys
+      :deep_extraction
     )
 
     def initialize(json)
@@ -46,7 +46,7 @@ module Retab
       @bust_cache = hash[:bust_cache].nil? ? false : hash[:bust_cache]
       @stream = hash[:stream].nil? ? false : hash[:stream]
       @background = hash[:background].nil? ? false : hash[:background]
-      @chunking_keys = hash[:chunking_keys] || {}
+      @deep_extraction = hash[:deep_extraction].nil? ? false : hash[:deep_extraction]
     end
   end
 end

data/lib/retab/extractions.rb

@@ -109,7 +109,7 @@ module Retab
     # @param bust_cache [Boolean, nil] If true, skip the LLM cache and force a fresh completion
     # @param stream [Boolean, nil]
     # @param background [Boolean, nil] If true, run asynchronously: returns immediately with status 'queued' and an empty output. Poll GET /v1/<primitive>/{id} until status is terminal. Mutually exclusive with stream.
-    # @param chunking_keys [Hash{String => String}, nil]
+    # @param deep_extraction [Boolean, nil] Optimizes for accuracy over latency in documents with very large arrays.
     # @param request_options [Hash] (see Retab::Types::RequestOptions)
     # @return [Retab::Extraction]
     def create(
@@ -123,7 +123,7 @@ module Retab
       bust_cache: nil,
       stream: nil,
       background: nil,
-      chunking_keys: nil,
+      deep_extraction: nil,
       request_options: {}
     )
       document = Retab::MimeData.coerce(document, client: @client) unless document.nil?
@@ -138,7 +138,7 @@ module Retab
         "bust_cache" => bust_cache,
         "stream" => stream,
         "background" => background,
-        "chunking_keys" => chunking_keys
+        "deep_extraction" => deep_extraction
       }.compact
       response = @client.request(
         method: :post,
@@ -167,7 +167,7 @@ module Retab
     # @param bust_cache [Boolean, nil] If true, skip the LLM cache and force a fresh completion
     # @param stream [Boolean, nil]
     # @param background [Boolean, nil] If true, run asynchronously: returns immediately with status 'queued' and an empty output. Poll GET /v1/<primitive>/{id} until status is terminal. Mutually exclusive with stream.
-    # @param chunking_keys [Hash{String => String}, nil]
+    # @param deep_extraction [Boolean, nil] Optimizes for accuracy over latency in documents with very large arrays.
     # @param request_options [Hash] (see Retab::Types::RequestOptions)
     # @return [void]
     def create_stream(
@@ -181,7 +181,7 @@ module Retab
       bust_cache: nil,
       stream: nil,
       background: nil,
-      chunking_keys: nil,
+      deep_extraction: nil,
       request_options: {}
     )
       document = Retab::MimeData.coerce(document, client: @client) unless document.nil?
@@ -196,7 +196,7 @@ module Retab
         "bust_cache" => bust_cache,
         "stream" => stream,
         "background" => background,
-        "chunking_keys" => chunking_keys
+        "deep_extraction" => deep_extraction
       }.compact
       @client.request(
         method: :post,

data/lib/retab/paginated_list.rb

@@ -5,7 +5,7 @@
 # @oagen-ignore-file
 #
 # Cross-language paginated-list wrapper. See
-# .notes/blueprints/sdk-pagination-contract.md for the full contract.
+# docs/blueprints/sdk-pagination-contract.md for the full contract.
 #
 # Every `def list` method on a Retab service returns one of these. The
 # class includes `Enumerable`, so callers can iterate every page

data/lib/retab/splits/reconstruct_document_ref.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructDocumentRef < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      filename: :filename,
+      id: :id,
+      mime_type: :mime_type
+    }.freeze
+
+    attr_accessor(
+      :filename,
+      :id,
+      :mime_type
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @filename = hash[:filename]
+      @id = hash[:id]
+      @mime_type = hash[:mime_type]
+    end
+  end
+end

data/lib/retab/splits/reconstruct_enriched_row.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructEnrichedRow < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      cells: :cells
+    }.freeze
+
+    attr_accessor :cells
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @cells = (hash[:cells] || [])
+    end
+  end
+end

data/lib/retab/splits/reconstruct_enriched_table.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructEnrichedTable < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      csv: :csv,
+      header: :header,
+      label: :label,
+      rows: :rows
+    }.freeze
+
+    attr_accessor(
+      :csv,
+      :header,
+      :label,
+      :rows
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @csv = hash[:csv]
+      @header = (hash[:header] || [])
+      @label = hash[:label]
+      @rows = (hash[:rows] || []).map { |item| item ? Retab::ReconstructEnrichedRow.new(item) : nil }
+    end
+  end
+end

data/lib/retab/splits/reconstruct_enrichment_options.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructEnrichmentOptions < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      fill_key_spans: :fill_key_spans,
+      flatten_header: :flatten_header,
+      promote_sections: :promote_sections
+    }.freeze
+
+    attr_accessor(
+      :fill_key_spans,
+      :flatten_header,
+      :promote_sections
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @fill_key_spans = hash[:fill_key_spans]
+      @flatten_header = hash[:flatten_header]
+      @promote_sections = hash[:promote_sections]
+    end
+  end
+end

data/lib/retab/splits/reconstruct_request.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructRequest < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      document: :document,
+      subdocuments: :subdocuments
+    }.freeze
+
+    attr_accessor(
+      :document,
+      :subdocuments
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @document = hash[:document] ? Retab::ReconstructDocumentRef.new(hash[:document]) : nil
+      @subdocuments = (hash[:subdocuments] || []).map { |item| item ? Retab::ReconstructSubdocument.new(item) : nil }
+    end
+  end
+end

data/lib/retab/splits/reconstruct_response.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructResponse < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      tables: :tables
+    }.freeze
+
+    attr_accessor :tables
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @tables = (hash[:tables] || []).map { |item| item ? Retab::ReconstructEnrichedTable.new(item) : nil }
+    end
+  end
+end

data/lib/retab/splits/reconstruct_subdocument.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class ReconstructSubdocument < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      enrichment: :enrichment,
+      name: :name,
+      partition_key: :partition_key,
+      regions: :regions
+    }.freeze
+
+    attr_accessor(
+      :enrichment,
+      :name,
+      :partition_key,
+      :regions
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @enrichment = hash[:enrichment] ? Retab::ReconstructEnrichmentOptions.new(hash[:enrichment]) : nil
+      @name = hash[:name]
+      @partition_key = hash[:partition_key]
+      @regions = (hash[:regions] || []).map { |item| item ? Retab::SheetRegion.new(item) : nil }
+    end
+  end
+end

data/lib/retab/splits/sheet_region.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  class SheetRegion < Retab::Types::BaseModel
+
+    HASH_ATTRS = {
+      col_end: :col_end,
+      col_start: :col_start,
+      header_rows: :header_rows,
+      row_end: :row_end,
+      row_start: :row_start,
+      sheet_index: :sheet_index,
+      sheet_name: :sheet_name
+    }.freeze
+
+    attr_accessor(
+      :col_end,
+      :col_start,
+      :header_rows,
+      :row_end,
+      :row_start,
+      :sheet_index,
+      :sheet_name
+    )
+
+    def initialize(json)
+      super()
+      hash = self.class.normalize(json)
+      @col_end = hash[:col_end]
+      @col_start = hash[:col_start]
+      @header_rows = (hash[:header_rows] || [])
+      @row_end = hash[:row_end]
+      @row_start = hash[:row_start]
+      @sheet_index = hash[:sheet_index]
+      @sheet_name = hash[:sheet_name]
+    end
+  end
+end

data/lib/retab/splits/split_result.rb

@@ -7,12 +7,14 @@ module Retab
 
     HASH_ATTRS = {
       name: :name,
-      pages: :pages
+      pages: :pages,
+      regions: :regions
     }.freeze
 
     attr_accessor(
       :name,
-      :pages
+      :pages,
+      :regions
     )
 
     def initialize(json)
@@ -20,6 +22,7 @@ module Retab
       hash = self.class.normalize(json)
       @name = hash[:name]
       @pages = (hash[:pages] || [])
+      @regions = (hash[:regions] || []).map { |item| item ? Retab::SheetRegion.new(item) : nil }
     end
   end
 end

data/lib/retab/splits.rb

@@ -124,6 +124,36 @@ module Retab
       result
     end
 
+    # Reconstruct Split
+    # @param document [Retab::ReconstructDocumentRef]
+    # @param subdocuments [Array<Retab::ReconstructSubdocument>]
+    # @param request_options [Hash] (see Retab::Types::RequestOptions)
+    # @return [Retab::ReconstructResponse]
+    def create_reconstruct(
+      document:,
+      subdocuments:,
+      request_options: {}
+    )
+      body = {
+        "document" => document,
+        "subdocuments" => subdocuments
+      }
+      response = @client.request(
+        method: :post,
+        path: "/v1/splits/reconstruct",
+        auth: true,
+        body: body,
+        request_options: request_options
+      )
+      result = Retab::ReconstructResponse.new(response.body)
+      result.last_response = Retab::Types::ApiResponse.new(
+        http_status: response.code.to_i,
+        http_headers: response.each_header.to_h,
+        request_id: response["x-request-id"]
+      )
+      result
+    end
+
     # Get Split
     # @param split_id [String]
     # @param include_output [Boolean, nil] When false, returns a cheap status-only projection (no output), served from cache for in-flight background runs.

data/lib/retab/types/usage_order.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  module Types
+    UsageOrder = ClassificationsOrder
+  end
+end