retab 0.1.8 → 0.1.10

data/lib/retab/mime.rb

@@ -69,7 +69,22 @@ module Retab
     end
 
     # Single entry point used by every generator-emitted resource method.
-    def self.coerce(input)
+    #
+    # A file-id document input (a `Retab::FileRef` object, or a Hash
+    # carrying an `:id` key) is resolved CLIENT-SIDE into URL-backed
+    # MimeData by calling the Files download-link endpoint — the document
+    # routes now accept only URL-backed MimeData `{filename, url}` and 422
+    # a file-id wire body. Mirrors the Node SDK and Go CLI
+    # (`resolveFileIDToMIMEData`). `client:` is the `Retab::Client`
+    # threaded through from the resource method body; it must be present to
+    # resolve a file-id input.
+    def self.coerce(input = nil, client: nil, **rest)
+      # Backward-compatible bare-keyword form: `coerce(filename: ..., url: ...)`
+      # slurps the keywords (minus `client:`) into the Hash input shape.
+      input = rest if input.nil? && !rest.empty?
+
+      return resolve_file_ref(input, client: client) if file_ref_like?(input)
+
       case input
       when MimeData
         input
@@ -89,14 +104,12 @@ module Retab
       end
     end
 
-    def self.coerce_document_map(input)
+    def self.coerce_document_map(input, client: nil)
       unless input.respond_to?(:transform_values)
         raise ArgumentError, "cannot coerce #{input.class} to a Retab::MimeData document map"
       end
 
-      input.transform_values do |document|
-        file_ref_like?(document) ? document : coerce(document)
-      end
+      input.transform_values { |document| coerce(document, client: client) }
     end
 
     def to_h
@@ -140,13 +153,71 @@ module Retab
       )
     end
 
+    # A file-id document input: either a spec-derived FileRef-like model
+    # exposing `#id`, or a Hash carrying an `:id` / `'id'` key. These no
+    # longer serialize over the wire — `coerce` resolves them into
+    # URL-backed MimeData via the Files download-link endpoint.
     private_class_method def self.file_ref_like?(input)
-      return true if defined?(Retab::FileRef) && input.is_a?(Retab::FileRef)
+      return false if input.is_a?(MimeData)
+      return true if input.respond_to?(:id) && !input.is_a?(Hash) && !input.is_a?(String)
       return false unless input.is_a?(Hash)
 
       input.key?(:id) || input.key?("id")
     end
 
+    private_class_method def self.file_ref_id(input)
+      if input.is_a?(Hash)
+        sym = input.transform_keys { |k| k.respond_to?(:to_sym) ? k.to_sym : k }
+        sym[:id]
+      elsif input.respond_to?(:id)
+        input.id
+      end
+    end
+
+    private_class_method def self.file_ref_filename(input)
+      if input.is_a?(Hash)
+        sym = input.transform_keys { |k| k.respond_to?(:to_sym) ? k.to_sym : k }
+        sym[:filename]
+      elsif input.respond_to?(:filename)
+        input.filename
+      end
+    end
+
+    # Resolve a file-id document input into URL-backed MimeData by calling
+    # the Files download-link endpoint. Mirrors the Node SDK and the Go CLI
+    # `resolveFileIDToMIMEData`: prefer the link's durable `mime_data`, then
+    # fall back to its `download_url`; filename precedence is
+    # input.filename → mime_data.filename → link.filename → 'document'.
+    private_class_method def self.resolve_file_ref(input, client:)
+      file_id = file_ref_id(input)
+      if file_id.nil? || file_id.to_s.empty?
+        raise ArgumentError, "file-id document input is missing an id"
+      end
+
+      if client.nil?
+        raise(
+          ArgumentError,
+          "cannot resolve file-id #{file_id} document input without a client; " \
+            "pass URL-backed MimeData instead"
+        )
+      end
+
+      link = client.files.get_download_link(file_id: file_id)
+      mime_data = link.mime_data
+      input_filename = file_ref_filename(input)
+
+      filename = input_filename ||
+        (mime_data && mime_data.filename) ||
+        link.filename ||
+        "document"
+      url = (mime_data && mime_data.url) || link.download_url
+      if url.nil? || url.to_s.empty?
+        raise ArgumentError, "file-id #{file_id}: server returned no download URL"
+      end
+
+      new(filename: filename, url: url)
+    end
+
     private_class_method def self.from_string(str)
       if str.match?(%r{\A(https?|gs|data):}i)
         # It's a URL.

data/lib/retab/parses/parse.rb

@@ -10,7 +10,6 @@ module Retab
       file: :file,
       model: :model,
       table_parsing_format: :table_parsing_format,
-      image_resolution_dpi: :image_resolution_dpi,
       instructions: :instructions,
       output: :output,
       status: :status,
@@ -24,7 +23,6 @@ module Retab
       :file,
       :model,
       :table_parsing_format,
-      :image_resolution_dpi,
       :instructions,
       :output,
       :status,
@@ -40,7 +38,6 @@ module Retab
       @file = hash[:file] ? Retab::FileRef.new(hash[:file]) : nil
       @model = hash[:model]
       @table_parsing_format = hash[:table_parsing_format]
-      @image_resolution_dpi = hash[:image_resolution_dpi]
       @instructions = hash[:instructions]
       @output = hash[:output] ? Retab::ParseOutput.new(hash[:output]) : nil
       @status = hash[:status].nil? ? "pending" : hash[:status]

data/lib/retab/parses/parse_request.rb

@@ -9,7 +9,6 @@ module Retab
       document: :document,
       model: :model,
       table_parsing_format: :table_parsing_format,
-      image_resolution_dpi: :image_resolution_dpi,
       instructions: :instructions,
       bust_cache: :bust_cache,
       background: :background
@@ -19,7 +18,6 @@ module Retab
       :document,
       :model,
       :table_parsing_format,
-      :image_resolution_dpi,
       :instructions,
       :bust_cache,
       :background
@@ -31,7 +29,6 @@ module Retab
       @document = hash[:document] ? Retab::MimeData.new(hash[:document]) : nil
       @model = hash[:model].nil? ? "retab-small" : hash[:model]
       @table_parsing_format = hash[:table_parsing_format].nil? ? "html" : hash[:table_parsing_format]
-      @image_resolution_dpi = hash[:image_resolution_dpi]
       @instructions = hash[:instructions]
       @bust_cache = hash[:bust_cache].nil? ? false : hash[:bust_cache]
       @background = hash[:background].nil? ? false : hash[:background]

data/lib/retab/parses.rb

@@ -77,7 +77,6 @@ module Retab
     # @param document [Retab::MimeData, Pathname, IO, String, Hash] The document to parse
     # @param model [String, nil] The model to use for parsing
     # @param table_parsing_format [Retab::Types::ParseRequestTableParsingFormat, nil] Format used to render tables extracted from the document
-    # @param image_resolution_dpi [Integer, nil] DPI used when rasterizing pages for the parser
     # @param instructions [String, nil] Free-form instructions appended to the system prompt to steer the parse.
     # @param bust_cache [Boolean, nil] If true, skip the LLM cache and force a fresh completion
     # @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.
@@ -87,18 +86,16 @@ module Retab
       document:,
       model: nil,
       table_parsing_format: nil,
-      image_resolution_dpi: nil,
       instructions: nil,
       bust_cache: nil,
       background: nil,
       request_options: {}
     )
-      document = Retab::MimeData.coerce(document) unless document.nil?
+      document = Retab::MimeData.coerce(document, client: @client) unless document.nil?
       body = {
         "document" => document,
         "model" => model,
         "table_parsing_format" => table_parsing_format,
-        "image_resolution_dpi" => image_resolution_dpi,
         "instructions" => instructions,
         "bust_cache" => bust_cache,
         "background" => background

data/lib/retab/partitions.rb

@@ -100,7 +100,7 @@ module Retab
       background: nil,
       request_options: {}
     )
-      document = Retab::MimeData.coerce(document) unless document.nil?
+      document = Retab::MimeData.coerce(document, client: @client) unless document.nil?
       body = {
         "document" => document,
         "key" => key,

data/lib/retab/schemas/generate_schema_request.rb

@@ -9,7 +9,6 @@ module Retab
       documents: :documents,
       model: :model,
       instructions: :instructions,
-      image_resolution_dpi: :image_resolution_dpi,
       background: :background
     }.freeze
 
@@ -17,7 +16,6 @@ module Retab
       :documents,
       :model,
       :instructions,
-      :image_resolution_dpi,
       :background
     )
 
@@ -27,7 +25,6 @@ module Retab
       @documents = (hash[:documents] || []).map { |item| item ? Retab::MimeData.new(item) : nil }
       @model = hash[:model].nil? ? "retab-small" : hash[:model]
       @instructions = hash[:instructions]
-      @image_resolution_dpi = hash[:image_resolution_dpi]
       @background = hash[:background].nil? ? false : hash[:background]
     end
   end

data/lib/retab/schemas.rb

@@ -14,7 +14,6 @@ module Retab
     # @param documents [Array<Retab::MimeData, Pathname, IO, String, Hash>]
     # @param model [String, nil]
     # @param instructions [String, nil]
-    # @param image_resolution_dpi [Integer, nil] Resolution of the image sent to the LLM
     # @param background [Boolean, nil] If true, run asynchronously: returns immediately with status 'queued'. Poll GET /v1/schemas/generate/{schema_generation_id} until status is terminal.
     # @param request_options [Hash] (see Retab::Types::RequestOptions)
     # @return [Retab::SchemaGeneration]
@@ -22,16 +21,14 @@ module Retab
       documents:,
       model: nil,
       instructions: nil,
-      image_resolution_dpi: nil,
       background: nil,
       request_options: {}
     )
-      documents = documents.map { |d| Retab::MimeData.coerce(d) } unless documents.nil?
+      documents = documents.map { |d| Retab::MimeData.coerce(d, client: @client) } unless documents.nil?
       body = {
         "documents" => documents,
         "model" => model,
         "instructions" => instructions,
-        "image_resolution_dpi" => image_resolution_dpi,
         "background" => background
       }.compact
       response = @client.request(

data/lib/retab/splits.rb

@@ -98,7 +98,7 @@ module Retab
       background: nil,
       request_options: {}
     )
-      document = Retab::MimeData.coerce(document) unless document.nil?
+      document = Retab::MimeData.coerce(document, client: @client) unless document.nil?
       body = {
         "document" => document,
         "subdocuments" => subdocuments,

data/lib/retab/types/{workflow_test_run_results_order.rb → eval_run_freshness_reasons.rb}

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    WorkflowTestRunResultsOrder = ClassificationsOrder
+    EvalRunFreshnessReasons = ArtifactFreshnessReasons
   end
 end

data/lib/retab/types/eval_run_freshness_status.rb

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

data/lib/retab/types/eval_run_trigger_type.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  module Types
+    class EvalRunTriggerType
+      MANUAL = "manual"
+      API = "api"
+      SCHEDULE = "schedule"
+      WEBHOOK = "webhook"
+      EMAIL = "email"
+      CUSTOM = "custom"
+      RESTART = "restart"
+      ALL = [MANUAL, API, SCHEDULE, WEBHOOK, EMAIL, CUSTOM, RESTART].freeze
+    end
+  end
+end

data/lib/retab/types/{latest_block_test_run_summary_outcome.rb → latest_block_eval_run_summary_outcome.rb}

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    LatestBlockTestRunSummaryOutcome = AssertionOutcome
+    LatestBlockEvalRunSummaryOutcome = AssertionOutcome
   end
 end

data/lib/retab/types/{latest_block_test_run_summary_status.rb → latest_block_eval_run_summary_status.rb}

@@ -4,7 +4,7 @@
 
 module Retab
   module Types
-    class LatestBlockTestRunSummaryStatus
+    class LatestBlockEvalRunSummaryStatus
       PENDING = "pending"
       QUEUED = "queued"
       RUNNING = "running"

data/lib/retab/types/trigger_info_type.rb

@@ -4,15 +4,6 @@
 
 module Retab
   module Types
-    class TriggerInfoType
-      MANUAL = "manual"
-      API = "api"
-      SCHEDULE = "schedule"
-      WEBHOOK = "webhook"
-      EMAIL = "email"
-      CUSTOM = "custom"
-      RESTART = "restart"
-      ALL = [MANUAL, API, SCHEDULE, WEBHOOK, EMAIL, CUSTOM, RESTART].freeze
-    end
+    TriggerInfoType = EvalRunTriggerType
   end
 end

data/lib/retab/types/{workflow_test_runs_order.rb → workflow_eval_result_verdict.rb}

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    WorkflowTestRunsOrder = ClassificationsOrder
+    WorkflowEvalResultVerdict = AssertionOutcome
   end
 end

data/lib/retab/types/workflow_eval_run_results_order.rb

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

data/lib/retab/types/{workflow_test_result_verdict.rb → workflow_eval_runs_order.rb}

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    WorkflowTestResultVerdict = AssertionOutcome
+    WorkflowEvalRunsOrder = ClassificationsOrder
   end
 end

data/lib/retab/types/workflow_eval_schema_drift.rb

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

data/lib/retab/types/{workflow_tests_order.rb → workflow_evals_order.rb}

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    WorkflowTestsOrder = ClassificationsOrder
+    WorkflowEvalsOrder = ClassificationsOrder
   end
 end

data/lib/retab/types/workflow_experiment_freshness_state.rb

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

data/lib/retab/types/workflow_experiment_run_plan_mode.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+# This file is auto-generated by oagen. Do not edit.
+
+module Retab
+  module Types
+    class WorkflowExperimentRunPlanMode
+      RUN = "run"
+      NOOP = "noop"
+      CONFLICT = "conflict"
+      UNKNOWN = "unknown"
+      ALL = [RUN, NOOP, CONFLICT, UNKNOWN].freeze
+    end
+  end
+end

data/lib/retab/types/workflow_experiments_exclude_status.rb

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    WorkflowExperimentsExcludeStatus = LatestBlockTestRunSummaryStatus
+    WorkflowExperimentsExcludeStatus = LatestBlockEvalRunSummaryStatus
   end
 end

data/lib/retab/types/workflow_experiments_status.rb

@@ -4,6 +4,6 @@
 
 module Retab
   module Types
-    WorkflowExperimentsStatus = LatestBlockTestRunSummaryStatus
+    WorkflowExperimentsStatus = LatestBlockEvalRunSummaryStatus
   end
 end

data/lib/retab/version.rb

@@ -5,5 +5,5 @@
 # @oagen-ignore-file
 
 module Retab
-  VERSION = "0.1.0"
+  VERSION = "0.1.10"
 end

data/lib/retab/workflow_artifacts/extraction_workflow_artifact.rb

@@ -11,7 +11,6 @@ module Retab
       model: :model,
       json_schema: :json_schema,
       n_consensus: :n_consensus,
-      image_resolution_dpi: :image_resolution_dpi,
       instructions: :instructions,
       output: :output,
       status: :status,
@@ -29,7 +28,6 @@ module Retab
       :model,
       :json_schema,
       :n_consensus,
-      :image_resolution_dpi,
       :instructions,
       :output,
       :status,
@@ -49,7 +47,6 @@ module Retab
       @model = hash[:model]
       @json_schema = hash[:json_schema] || {}
       @n_consensus = hash[:n_consensus]
-      @image_resolution_dpi = hash[:image_resolution_dpi]
       @instructions = hash[:instructions]
       @output = hash[:output] || {}
       @status = hash[:status].nil? ? "pending" : hash[:status]

data/lib/retab/workflow_artifacts/parse_workflow_artifact.rb

@@ -10,7 +10,6 @@ module Retab
       file: :file,
       model: :model,
       table_parsing_format: :table_parsing_format,
-      image_resolution_dpi: :image_resolution_dpi,
       instructions: :instructions,
       output: :output,
       status: :status,
@@ -25,7 +24,6 @@ module Retab
       :file,
       :model,
       :table_parsing_format,
-      :image_resolution_dpi,
       :instructions,
       :output,
       :status,
@@ -42,7 +40,6 @@ module Retab
       @file = hash[:file] ? Retab::FileRef.new(hash[:file]) : nil
       @model = hash[:model]
       @table_parsing_format = hash[:table_parsing_format]
-      @image_resolution_dpi = hash[:image_resolution_dpi]
       @instructions = hash[:instructions]
       @output = hash[:output] ? Retab::ParseOutput.new(hash[:output]) : nil
       @status = hash[:status].nil? ? "pending" : hash[:status]

data/lib/retab/workflow_artifacts/workflow_artifact.rb

@@ -7,12 +7,14 @@ module Retab
 
     HASH_ATTRS = {
       operation: :operation,
-      id: :id
+      id: :id,
+      additionalProperties: :additional_properties
     }.freeze
 
     attr_accessor(
       :operation,
-      :id
+      :id,
+      :additional_properties
     )
 
     def initialize(json)
@@ -20,6 +22,7 @@ module Retab
       hash = self.class.normalize(json)
       @operation = hash[:operation]
       @id = hash[:id]
+      @additional_properties = hash[:additionalProperties] || {}
     end
   end
 end

data/lib/retab/{workflow_experiments/file_handle_input.rb → workflow_block_executions/block_exec_file_handle_input.rb}

@@ -3,23 +3,23 @@
 # This file is auto-generated by oagen. Do not edit.
 
 module Retab
-  class FileHandleInput < Retab::Types::BaseModel
+  class BlockExecFileHandleInput < Retab::Types::BaseModel
 
     HASH_ATTRS = {
-      type: :type,
-      document: :document
+      document: :document,
+      type: :type
     }.freeze
 
     attr_accessor(
-      :type,
-      :document
+      :document,
+      :type
     )
 
     def initialize(json)
       super()
       hash = self.class.normalize(json)
+      @document = hash[:document] ? Retab::BlockExecFileRef.new(hash[:document]) : nil
       @type = hash[:type].nil? ? "file" : hash[:type]
-      @document = hash[:document] ? Retab::FileRef.new(hash[:document]) : nil
     end
   end
 end

data/lib/retab/workflow_block_executions/block_exec_file_ref.rb

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

data/lib/retab/{workflow_experiments/json_handle_input.rb → workflow_block_executions/block_exec_json_handle_input.rb}

@@ -3,23 +3,23 @@
 # This file is auto-generated by oagen. Do not edit.
 
 module Retab
-  class JsonHandleInput < Retab::Types::BaseModel
+  class BlockExecJsonHandleInput < Retab::Types::BaseModel
 
     HASH_ATTRS = {
-      type: :type,
-      data: :data
+      data: :data,
+      type: :type
     }.freeze
 
     attr_accessor(
-      :type,
-      :data
+      :data,
+      :type
     )
 
     def initialize(json)
       super()
       hash = self.class.normalize(json)
-      @type = hash[:type].nil? ? "json" : hash[:type]
       @data = hash[:data]
+      @type = hash[:type].nil? ? "json" : hash[:type]
     end
   end
 end

data/lib/retab/workflow_block_executions/cancelled_block_execution_lifecycle.rb

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

data/lib/retab/workflow_block_executions/pending_block_execution_lifecycle.rb

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

data/lib/retab/workflow_block_executions/queued_block_execution_lifecycle.rb

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