xberg 1.2.5 → 1.2.6

data/sig/types.rbs

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:099b158c0f39786295393233e710e06294b87b228d769cfd93e37625226b9e92
+# alef:hash:f06d266849d9d28303cb88810861ceeedbdf08525bd64d878a8bab9ec1b2e606
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
@@ -10,10 +10,10 @@ module Xberg
 type json_value = Hash[String, untyped] | Array[untyped] | String | Integer | Float | bool | nil
 
     class AccelerationConfig
-        attr_reader provider: ExecutionProviderType
+        attr_reader provider: enum_ExecutionProviderType
         attr_reader device_id: Integer
 
-    def initialize: (?provider: ExecutionProviderType, ?device_id: Integer) -> void
+    def initialize: (?provider: enum_ExecutionProviderType, ?device_id: Integer) -> void
     end
 
     class ArchiveEntry
@@ -92,11 +92,11 @@ def initialize: (?entry_count: Integer, ?citation_keys: Array[String], ?authors:
     end
 
     class BrowserConfig
-        attr_reader mode: BrowserMode
-        attr_reader backend: BrowserBackend
+        attr_reader mode: enum_BrowserMode
+        attr_reader backend: enum_BrowserBackend
         attr_reader endpoint: String?
         attr_reader timeout: Integer
-        attr_reader wait: BrowserWait
+        attr_reader wait: enum_BrowserWait
         attr_reader wait_selector: String?
         attr_reader extra_wait: Integer?
         attr_reader proxy: ProxyConfig?
@@ -106,7 +106,7 @@ def initialize: (?entry_count: Integer, ?citation_keys: Array[String], ?authors:
         attr_reader capture_network_events: bool
         attr_reader session_affinity: bool
 
-def initialize: (?mode: BrowserMode, ?backend: BrowserBackend, ?endpoint: String, ?timeout: Integer, ?wait: BrowserWait, ?wait_selector: String, ?extra_wait: Integer, ?proxy: ProxyConfig, ?block_url_patterns: Array[String], ?eval_script: String, ?robots_user_agent: String, ?capture_network_events: bool, ?session_affinity: bool) -> void
+def initialize: (?mode: enum_BrowserMode, ?backend: enum_BrowserBackend, ?endpoint: String, ?timeout: Integer, ?wait: enum_BrowserWait, ?wait_selector: String, ?extra_wait: Integer, ?proxy: ProxyConfig, ?block_url_patterns: Array[String], ?eval_script: String, ?robots_user_agent: String, ?capture_network_events: bool, ?session_affinity: bool) -> void
     end
 
     class CacheStats
@@ -138,13 +138,13 @@ def initialize: (?mode: BrowserMode, ?backend: BrowserBackend, ?endpoint: String
 
     class Chunk
         attr_reader content: String
-        attr_reader chunk_type: ChunkType
+        attr_reader chunk_type: enum_ChunkType
     attr_reader embedding: Array[Float]?
         attr_reader sparse_embedding: SparseEmbedding?
         attr_reader late_interaction: MultiVectorEmbedding?
         attr_reader metadata: ChunkMetadata
 
-def initialize: (content: String, chunk_type: ChunkType, ?embedding: Array[Float], ?sparse_embedding: SparseEmbedding, ?late_interaction: MultiVectorEmbedding, metadata: ChunkMetadata) -> void
+def initialize: (content: String, chunk_type: enum_ChunkType, ?embedding: Array[Float], ?sparse_embedding: SparseEmbedding, ?late_interaction: MultiVectorEmbedding, metadata: ChunkMetadata) -> void
     end
 
     class ChunkClassificationConfig
@@ -200,16 +200,16 @@ def initialize: (byte_start: Integer, byte_end: Integer, ?token_count: Integer,
         attr_reader max_characters: Integer
         attr_reader overlap: Integer
         attr_reader trim: bool
-        attr_reader chunker_type: ChunkerType
+        attr_reader chunker_type: enum_ChunkerType
         attr_reader embedding: EmbeddingConfig?
         attr_reader sparse_embedding: SparseEmbeddingConfig?
         attr_reader late_interaction: LateInteractionConfig?
         attr_reader preset: String?
         attr_reader sizing: ChunkSizing
         attr_reader topic_threshold: Float?
-        attr_reader table_chunking: TableChunkingMode
+        attr_reader table_chunking: enum_TableChunkingMode
 
-    def initialize: (?max_characters: Integer, ?overlap: Integer, ?trim: bool, ?chunker_type: ChunkerType, ?embedding: EmbeddingConfig, ?sparse_embedding: SparseEmbeddingConfig, ?late_interaction: LateInteractionConfig, ?preset: String, ?sizing: ChunkSizing, ?topic_threshold: Float, ?table_chunking: TableChunkingMode) -> void
+    def initialize: (?max_characters: Integer, ?overlap: Integer, ?trim: bool, ?chunker_type: enum_ChunkerType, ?embedding: EmbeddingConfig, ?sparse_embedding: SparseEmbeddingConfig, ?late_interaction: LateInteractionConfig, ?preset: String, ?sizing: ChunkSizing, ?topic_threshold: Float, ?table_chunking: enum_TableChunkingMode) -> void
     end
 
     class Citation
@@ -259,7 +259,7 @@ def initialize: (text: String, context_path: Array[String], node_types: Array[St
     end
 
     class CodeDataNode
-        attr_reader kind: CodeDataNodeKind
+        attr_reader kind: enum_CodeDataNodeKind
         attr_reader key: String?
         attr_reader value: String?
     attr_reader attributes: Array[CodeDataAttribute]
@@ -267,7 +267,7 @@ def initialize: (text: String, context_path: Array[String], node_types: Array[St
         attr_reader byte_start: Integer
         attr_reader byte_end: Integer
 
-def initialize: (kind: CodeDataNodeKind, ?key: String, ?value: String, attributes: Array[CodeDataAttribute], children: Array[CodeDataNode], byte_start: Integer, byte_end: Integer) -> void
+def initialize: (kind: enum_CodeDataNodeKind, ?key: String, ?value: String, attributes: Array[CodeDataAttribute], children: Array[CodeDataNode], byte_start: Integer, byte_end: Integer) -> void
     end
 
     class CodeMetadata
@@ -318,8 +318,8 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
     end
 
     class ConversionOptions
-        attr_reader heading_style: HeadingStyle
-        attr_reader list_indent_type: ListIndentType
+        attr_reader heading_style: enum_HeadingStyle
+        attr_reader list_indent_type: enum_ListIndentType
         attr_reader list_indent_width: Integer
         attr_reader bullets: String
         attr_reader strong_em_symbol: String
@@ -332,17 +332,17 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
         attr_reader default_title: bool
         attr_reader br_in_tables: bool
         attr_reader compact_tables: bool
-        attr_reader highlight_style: HighlightStyle
+        attr_reader highlight_style: enum_HighlightStyle
         attr_reader extract_metadata: bool
-        attr_reader whitespace_mode: WhitespaceMode
+        attr_reader whitespace_mode: enum_WhitespaceMode
         attr_reader strip_newlines: bool
         attr_reader wrap: bool
         attr_reader wrap_width: Integer
         attr_reader convert_as_inline: bool
         attr_reader sub_symbol: String
         attr_reader sup_symbol: String
-        attr_reader newline_style: NewlineStyle
-        attr_reader code_block_style: CodeBlockStyle
+        attr_reader newline_style: enum_NewlineStyle
+        attr_reader code_block_style: enum_CodeBlockStyle
     attr_reader keep_inline_images_in: Array[String]
         attr_reader preprocessing: PreprocessingOptions
         attr_reader encoding: String
@@ -350,9 +350,9 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
     attr_reader strip_tags: Array[String]
     attr_reader preserve_tags: Array[String]
         attr_reader skip_images: bool
-        attr_reader url_escape_style: UrlEscapeStyle
-        attr_reader link_style: LinkStyle
-        attr_reader output_format: HtmlToMarkdownRsOutputFormat
+        attr_reader url_escape_style: enum_UrlEscapeStyle
+        attr_reader link_style: enum_LinkStyle
+        attr_reader output_format: enum_HtmlToMarkdownRsOutputFormat
         attr_reader include_document_structure: bool
         attr_reader extract_images: bool
         attr_reader max_image_size: Integer
@@ -360,9 +360,9 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
         attr_reader infer_dimensions: bool
         attr_reader max_depth: Integer?
     attr_reader exclude_selectors: Array[String]
-        attr_reader tier_strategy: TierStrategy
+        attr_reader tier_strategy: enum_TierStrategy
 
-def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType, ?list_indent_width: Integer, ?bullets: String, ?strong_em_symbol: String, ?escape_asterisks: bool, ?escape_underscores: bool, ?escape_misc: bool, ?escape_ascii: bool, ?code_language: String, ?autolinks: bool, ?default_title: bool, ?br_in_tables: bool, ?compact_tables: bool, ?highlight_style: HighlightStyle, ?extract_metadata: bool, ?whitespace_mode: WhitespaceMode, ?strip_newlines: bool, ?wrap: bool, ?wrap_width: Integer, ?convert_as_inline: bool, ?sub_symbol: String, ?sup_symbol: String, ?newline_style: NewlineStyle, ?code_block_style: CodeBlockStyle, ?keep_inline_images_in: Array[String], ?preprocessing: PreprocessingOptions, ?encoding: String, ?debug: bool, ?strip_tags: Array[String], ?preserve_tags: Array[String], ?skip_images: bool, ?url_escape_style: UrlEscapeStyle, ?link_style: LinkStyle, ?output_format: HtmlToMarkdownRsOutputFormat, ?include_document_structure: bool, ?extract_images: bool, ?max_image_size: Integer, ?capture_svg: bool, ?infer_dimensions: bool, ?max_depth: Integer, ?exclude_selectors: Array[String], ?tier_strategy: TierStrategy) -> void
+def initialize: (?heading_style: enum_HeadingStyle, ?list_indent_type: enum_ListIndentType, ?list_indent_width: Integer, ?bullets: String, ?strong_em_symbol: String, ?escape_asterisks: bool, ?escape_underscores: bool, ?escape_misc: bool, ?escape_ascii: bool, ?code_language: String, ?autolinks: bool, ?default_title: bool, ?br_in_tables: bool, ?compact_tables: bool, ?highlight_style: enum_HighlightStyle, ?extract_metadata: bool, ?whitespace_mode: enum_WhitespaceMode, ?strip_newlines: bool, ?wrap: bool, ?wrap_width: Integer, ?convert_as_inline: bool, ?sub_symbol: String, ?sup_symbol: String, ?newline_style: enum_NewlineStyle, ?code_block_style: enum_CodeBlockStyle, ?keep_inline_images_in: Array[String], ?preprocessing: PreprocessingOptions, ?encoding: String, ?debug: bool, ?strip_tags: Array[String], ?preserve_tags: Array[String], ?skip_images: bool, ?url_escape_style: enum_UrlEscapeStyle, ?link_style: enum_LinkStyle, ?output_format: enum_HtmlToMarkdownRsOutputFormat, ?include_document_structure: bool, ?extract_images: bool, ?max_image_size: Integer, ?capture_svg: bool, ?infer_dimensions: bool, ?max_depth: Integer, ?exclude_selectors: Array[String], ?tier_strategy: enum_TierStrategy) -> void
     end
 
     class CoreProperties
@@ -390,8 +390,8 @@ def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType
         attr_reader max_pages: Integer?
         attr_reader max_links_per_page: Integer?
         attr_reader max_concurrent: Integer?
-        attr_reader crawl_strategy: CrawlStrategyKind
-        attr_reader content_filter: ContentFilterKind?
+        attr_reader crawl_strategy: enum_CrawlStrategyKind
+        attr_reader content_filter: enum_ContentFilterKind?
         attr_reader bm25_query: String?
         attr_reader bm25_threshold: Float?
         attr_reader respect_robots_txt: bool
@@ -415,7 +415,7 @@ def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType
         attr_reader map_limit: Integer?
         attr_reader map_search: String?
         attr_reader download_assets: bool
-    attr_reader asset_types: Array[AssetCategory]
+    attr_reader asset_types: Array[enum_AssetCategory]
         attr_reader max_asset_size: Integer?
         attr_reader browser: BrowserConfig
         attr_reader proxy: ProxyConfig?
@@ -427,14 +427,14 @@ def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType
         attr_reader document_max_size: Integer?
     attr_reader document_mime_types: Array[String]
         attr_reader document_output_dir: String?
-        attr_reader document_content_encoding: DocumentContentEncoding?
+        attr_reader document_content_encoding: enum_DocumentContentEncoding?
         attr_reader warc_output: String?
         attr_reader browser_profile: String?
         attr_reader save_browser_profile: bool
         attr_reader ssrf: SsrfPolicy
         attr_reader ssrf_deny_private_explicit: bool?
 
-def initialize: (?max_depth: Integer, ?max_pages: Integer, ?max_links_per_page: Integer, ?max_concurrent: Integer, ?crawl_strategy: CrawlStrategyKind, ?content_filter: ContentFilterKind, ?bm25_query: String, ?bm25_threshold: Float, ?respect_robots_txt: bool, ?soft_http_errors: bool, ?user_agent: String, ?stay_on_domain: bool, ?allow_subdomains: bool, ?include_paths: Array[String], ?exclude_paths: Array[String], ?custom_headers: Hash[String, String], ?request_timeout: Integer, ?rate_limit_ms: Integer, ?max_redirects: Integer, ?retry_count: Integer, ?retry_codes: Array[Integer], ?cookies_enabled: bool, ?auth: AuthConfig, ?max_body_size: Integer, ?remove_tags: Array[String], ?content: ContentConfig, ?map_limit: Integer, ?map_search: String, ?download_assets: bool, ?asset_types: Array[AssetCategory], ?max_asset_size: Integer, ?browser: BrowserConfig, ?proxy: ProxyConfig, ?user_agents: Array[String], ?capture_screenshot: bool, ?follow_document_urls: bool, ?document_url_depth: Integer, ?download_documents: bool, ?document_max_size: Integer, ?document_mime_types: Array[String], ?document_output_dir: String, ?document_content_encoding: DocumentContentEncoding, ?warc_output: String, ?browser_profile: String, ?save_browser_profile: bool, ?ssrf: SsrfPolicy, ?ssrf_deny_private_explicit: bool) -> void
+def initialize: (?max_depth: Integer, ?max_pages: Integer, ?max_links_per_page: Integer, ?max_concurrent: Integer, ?crawl_strategy: enum_CrawlStrategyKind, ?content_filter: enum_ContentFilterKind, ?bm25_query: String, ?bm25_threshold: Float, ?respect_robots_txt: bool, ?soft_http_errors: bool, ?user_agent: String, ?stay_on_domain: bool, ?allow_subdomains: bool, ?include_paths: Array[String], ?exclude_paths: Array[String], ?custom_headers: Hash[String, String], ?request_timeout: Integer, ?rate_limit_ms: Integer, ?max_redirects: Integer, ?retry_count: Integer, ?retry_codes: Array[Integer], ?cookies_enabled: bool, ?auth: AuthConfig, ?max_body_size: Integer, ?remove_tags: Array[String], ?content: ContentConfig, ?map_limit: Integer, ?map_search: String, ?download_assets: bool, ?asset_types: Array[enum_AssetCategory], ?max_asset_size: Integer, ?browser: BrowserConfig, ?proxy: ProxyConfig, ?user_agents: Array[String], ?capture_screenshot: bool, ?follow_document_urls: bool, ?document_url_depth: Integer, ?download_documents: bool, ?document_max_size: Integer, ?document_mime_types: Array[String], ?document_output_dir: String, ?document_content_encoding: enum_DocumentContentEncoding, ?warc_output: String, ?browser_profile: String, ?save_browser_profile: bool, ?ssrf: SsrfPolicy, ?ssrf_deny_private_explicit: bool) -> void
     end
 
     class CsvConfig
@@ -471,10 +471,14 @@ def initialize: (?record_count: Integer, ?field_count: Integer, ?fields: Array[D
 
     class DeepseekOcrBackendOptions
         attr_reader model_path: String?
-        attr_reader device: CandleDevicePreference?
+        attr_reader model_id: String?
+        attr_reader hf_revision: String?
+        attr_reader cache_dir: String?
+        attr_reader device: enum_CandleDevicePreference?
         attr_reader version: Integer?
+        attr_reader dtype: enum_CandleDeepseekOcrDtype?
 
-    def initialize: (?model_path: String, ?device: CandleDevicePreference, ?version: Integer) -> void
+    def initialize: (?model_path: String, ?model_id: String, ?hf_revision: String, ?cache_dir: String, ?device: enum_CandleDevicePreference, ?version: Integer, ?dtype: enum_CandleDeepseekOcrDtype) -> void
     end
 
     class DetectResponse
@@ -550,10 +554,10 @@ def initialize: (plain_text: String, blocks: Array[FormattedBlock], metadata: Me
 
     class DoctorCheck
         attr_reader name: String
-        attr_reader status: ProbeStatus
+        attr_reader status: enum_ProbeStatus
         attr_reader message: String
 
-    def initialize: (name: String, status: ProbeStatus, message: String) -> void
+    def initialize: (name: String, status: enum_ProbeStatus, message: String) -> void
     end
 
     class DoctorReport
@@ -567,9 +571,9 @@ def initialize: (?checks: Array[DoctorCheck]) -> void
         attr_reader start_page: Integer
         attr_reader end_page: Integer
         attr_reader confidence: Float
-        attr_reader reason: BoundaryReason
+        attr_reader reason: enum_BoundaryReason
 
-    def initialize: (start_page: Integer, end_page: Integer, confidence: Float, reason: BoundaryReason) -> void
+    def initialize: (start_page: Integer, end_page: Integer, confidence: Float, reason: enum_BoundaryReason) -> void
     end
 
     class DocumentCounts
@@ -596,33 +600,33 @@ def initialize: (?checks: Array[DoctorCheck]) -> void
         attr_reader content: NodeContent
         attr_reader parent: Integer?
     attr_reader children: Array[Integer]
-        attr_reader content_layer: ContentLayer
+        attr_reader content_layer: enum_ContentLayer
         attr_reader page: Integer?
         attr_reader page_end: Integer?
         attr_reader bbox: BoundingBox?
     attr_reader annotations: Array[TextAnnotation]
     attr_reader attributes: Hash[String, String]?
 
-def initialize: (id: String, content: NodeContent, ?parent: Integer, children: Array[Integer], content_layer: ContentLayer, ?page: Integer, ?page_end: Integer, ?bbox: BoundingBox, annotations: Array[TextAnnotation], ?attributes: Hash[String, String]) -> void
+def initialize: (id: String, content: NodeContent, ?parent: Integer, children: Array[Integer], content_layer: enum_ContentLayer, ?page: Integer, ?page_end: Integer, ?bbox: BoundingBox, annotations: Array[TextAnnotation], ?attributes: Hash[String, String]) -> void
     end
 
     class DocumentRelationship
         attr_reader source: Integer
         attr_reader target: Integer
-        attr_reader kind: RelationshipKind
+        attr_reader kind: enum_RelationshipKind
 
-    def initialize: (source: Integer, target: Integer, kind: RelationshipKind) -> void
+    def initialize: (source: Integer, target: Integer, kind: enum_RelationshipKind) -> void
     end
 
     class DocumentRevision
         attr_reader revision_id: String
         attr_reader author: String?
         attr_reader timestamp: String?
-        attr_reader kind: RevisionKind
+        attr_reader kind: enum_RevisionKind
         attr_reader anchor: RevisionAnchor?
         attr_reader delta: RevisionDelta
 
-    def initialize: (revision_id: String, ?author: String, ?timestamp: String, kind: RevisionKind, ?anchor: RevisionAnchor, delta: RevisionDelta) -> void
+    def initialize: (revision_id: String, ?author: String, ?timestamp: String, kind: enum_RevisionKind, ?anchor: RevisionAnchor, delta: RevisionDelta) -> void
     end
 
     class DocumentStructure
@@ -638,10 +642,10 @@ def initialize: (?nodes: Array[DocumentNode], ?source_format: String, ?relations
 
     class DocumentSummary
         attr_reader text: String
-        attr_reader strategy: SummaryStrategy
+        attr_reader strategy: enum_SummaryStrategy
         attr_reader token_count: Integer?
 
-    def initialize: (text: String, strategy: SummaryStrategy, ?token_count: Integer) -> void
+    def initialize: (text: String, strategy: enum_SummaryStrategy, ?token_count: Integer) -> void
     end
 
     class DocxAppProperties
@@ -675,11 +679,11 @@ def initialize: (?core_properties: CoreProperties, ?app_properties: DocxAppPrope
 
     class Element
         attr_reader element_id: String
-        attr_reader element_type: ElementType
+        attr_reader element_type: enum_ElementType
         attr_reader text: String
         attr_reader metadata: ElementMetadata
 
-    def initialize: (element_id: String, element_type: ElementType, text: String, metadata: ElementMetadata) -> void
+    def initialize: (element_id: String, element_type: enum_ElementType, text: String, metadata: ElementMetadata) -> void
     end
 
     class ElementMetadata
@@ -830,21 +834,21 @@ def initialize: (sheets: Array[ExcelSheet], metadata: Hash[String, String], ?rev
     end
 
     class ExtractInput
-        attr_reader kind: ExtractInputKind
+        attr_reader kind: enum_ExtractInputKind
         attr_reader bytes: String?
         attr_reader uri: String?
         attr_reader mime_type: String?
         attr_reader filename: String?
         attr_reader config: FileExtractionConfig?
 
-    def initialize: (?kind: ExtractInputKind, ?bytes: String, ?uri: String, ?mime_type: String, ?filename: String, ?config: FileExtractionConfig) -> void
+    def initialize: (?kind: enum_ExtractInputKind, ?bytes: String, ?uri: String, ?mime_type: String, ?filename: String, ?config: FileExtractionConfig) -> void
     end
 
     class ExtractedDocument
         attr_reader content: String
         attr_reader mime_type: String
         attr_reader metadata: Metadata
-        attr_reader extraction_method: ExtractionMethod?
+        attr_reader extraction_method: enum_ExtractionMethod?
     attr_reader tables: Array[Table]
         attr_reader counts: DocumentCounts
     attr_reader detected_languages: Array[String]?
@@ -875,7 +879,7 @@ def initialize: (sheets: Array[ExcelSheet], metadata: Hash[String, String], ?rev
     attr_reader formulas: Array[Formula]
     attr_reader form_fields: Array[PdfFormField]
 
-def initialize: (?content: String, ?mime_type: String, ?metadata: Metadata, ?extraction_method: ExtractionMethod, ?tables: Array[Table], ?counts: DocumentCounts, ?detected_languages: Array[String], ?detected_language_confidences: Array[LanguageConfidence], ?chunks: Array[Chunk], ?images: Array[ExtractedImage], ?pages: Array[PageContent], ?elements: Array[Element], ?djot_content: DjotContent, ?ocr_elements: Array[OcrElement], ?document: DocumentStructure, ?extracted_keywords: Array[Keyword], ?quality_score: Float, ?processing_warnings: Array[ProcessingWarning], ?annotations: Array[PdfAnnotation], ?children: Array[ArchiveEntry], ?uris: Array[ExtractedUri], ?revisions: Array[DocumentRevision], ?structured_output: String, ?code_intelligence: String, ?llm_usage: Array[LlmUsage], ?entities: Array[Entity], ?summary: DocumentSummary, ?extraction_confidence: ExtractionConfidence, ?translation: Translation, ?page_classifications: Array[PageClassification], ?redaction_report: RedactionReport, ?formulas: Array[Formula], ?form_fields: Array[PdfFormField]) -> void
+def initialize: (?content: String, ?mime_type: String, ?metadata: Metadata, ?extraction_method: enum_ExtractionMethod, ?tables: Array[Table], ?counts: DocumentCounts, ?detected_languages: Array[String], ?detected_language_confidences: Array[LanguageConfidence], ?chunks: Array[Chunk], ?images: Array[ExtractedImage], ?pages: Array[PageContent], ?elements: Array[Element], ?djot_content: DjotContent, ?ocr_elements: Array[OcrElement], ?document: DocumentStructure, ?extracted_keywords: Array[Keyword], ?quality_score: Float, ?processing_warnings: Array[ProcessingWarning], ?annotations: Array[PdfAnnotation], ?children: Array[ArchiveEntry], ?uris: Array[ExtractedUri], ?revisions: Array[DocumentRevision], ?structured_output: String, ?code_intelligence: String, ?llm_usage: Array[LlmUsage], ?entities: Array[Entity], ?summary: DocumentSummary, ?extraction_confidence: ExtractionConfidence, ?translation: Translation, ?page_classifications: Array[PageClassification], ?redaction_report: RedactionReport, ?formulas: Array[Formula], ?form_fields: Array[PdfFormField]) -> void
     end
 
     class ExtractedImage
@@ -892,36 +896,36 @@ def initialize: (?content: String, ?mime_type: String, ?metadata: Metadata, ?ext
         attr_reader ocr_result: ExtractedDocument?
         attr_reader bounding_box: BoundingBox?
         attr_reader source_path: String?
-        attr_reader image_kind: ImageKind?
+        attr_reader image_kind: enum_ImageKind?
         attr_reader kind_confidence: Float?
         attr_reader cluster_id: Integer?
         attr_reader caption: String?
     attr_reader qr_codes: Array[QrCode]?
         attr_reader data_base64: String?
 
-def initialize: (?data: String, ?format: String, ?image_index: Integer, ?page_number: Integer, ?width: Integer, ?height: Integer, ?colorspace: String, ?bits_per_component: Integer, ?is_mask: bool, ?description: String, ?ocr_result: ExtractedDocument, ?bounding_box: BoundingBox, ?source_path: String, ?image_kind: ImageKind, ?kind_confidence: Float, ?cluster_id: Integer, ?caption: String, ?qr_codes: Array[QrCode], ?data_base64: String) -> void
+def initialize: (?data: String, ?format: String, ?image_index: Integer, ?page_number: Integer, ?width: Integer, ?height: Integer, ?colorspace: String, ?bits_per_component: Integer, ?is_mask: bool, ?description: String, ?ocr_result: ExtractedDocument, ?bounding_box: BoundingBox, ?source_path: String, ?image_kind: enum_ImageKind, ?kind_confidence: Float, ?cluster_id: Integer, ?caption: String, ?qr_codes: Array[QrCode], ?data_base64: String) -> void
     end
 
     class ExtractedUri
         attr_reader url: String
         attr_reader label: String?
         attr_reader page: Integer?
-        attr_reader kind: UriKind
+        attr_reader kind: enum_UriKind
 
-    def initialize: (url: String, ?label: String, ?page: Integer, kind: UriKind) -> void
+    def initialize: (url: String, ?label: String, ?page: Integer, kind: enum_UriKind) -> void
     end
 
     class ExtractionConfidence
         attr_reader text_coverage: Float
         attr_reader ocr_aggregate: Float?
-        attr_reader schema_compliance: SchemaCompliance
+        attr_reader schema_compliance: enum_SchemaCompliance
         attr_reader combined: Float
 
-    def initialize: (text_coverage: Float, ?ocr_aggregate: Float, schema_compliance: SchemaCompliance, combined: Float) -> void
+    def initialize: (text_coverage: Float, ?ocr_aggregate: Float, schema_compliance: enum_SchemaCompliance, combined: Float) -> void
     end
 
     class ExtractionConfig
-        attr_reader mime_detection_policy: MimeDetectionPolicy
+        attr_reader mime_detection_policy: enum_MimeDetectionPolicy
         attr_reader use_cache: bool
         attr_reader enable_quality_processing: bool
         attr_reader ocr: OcrConfig?
@@ -942,13 +946,13 @@ def initialize: (?data: String, ?format: String, ?image_index: Integer, ?page_nu
         attr_reader html_output: HtmlOutputConfig?
         attr_reader extraction_timeout_secs: Integer?
         attr_reader max_concurrent_extractions: Integer?
-        attr_reader result_format: ResultFormat
+        attr_reader result_format: enum_ResultFormat
         attr_reader security_limits: SecurityLimits?
         attr_reader max_embedded_file_bytes: Integer?
         attr_reader output_format: OutputFormat
         attr_reader escape_markdown: bool
         attr_reader table_anchors: bool
-        attr_reader jupyter_cell_rendering: JupyterCellRendering
+        attr_reader jupyter_cell_rendering: enum_JupyterCellRendering
         attr_reader apply_notebook_cell_tags: bool
         attr_reader layout: LayoutDetectionConfig?
         attr_reader transcription: TranscriptionConfig?
@@ -974,9 +978,11 @@ def initialize: (?data: String, ?format: String, ?image_index: Integer, ?page_nu
         attr_reader captioning: CaptioningConfig?
         attr_reader qr_codes: bool?
 
-def initialize: (?mime_detection_policy: MimeDetectionPolicy, ?use_cache: bool, ?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?ocr_strategy: OcrStrategy, ?force_ocr_pages: Array[Integer], ?disable_ocr: bool, ?chunking: ChunkingConfig, ?content_filter: ContentFilterConfig, ?images: ImageExtractionConfig, ?pdf_options: PdfConfig, ?token_reduction: TokenReductionOptions, ?language_detection: LanguageDetectionConfig, ?pages: PageConfig, ?keywords: KeywordConfig, ?postprocessor: PostProcessorConfig, ?html_options: ConversionOptions, ?html_output: HtmlOutputConfig, ?extraction_timeout_secs: Integer, ?max_concurrent_extractions: Integer, ?result_format: ResultFormat, ?security_limits: SecurityLimits, ?max_embedded_file_bytes: Integer, ?output_format: OutputFormat, ?escape_markdown: bool, ?table_anchors: bool, ?jupyter_cell_rendering: JupyterCellRendering, ?apply_notebook_cell_tags: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?use_layout_for_markdown: bool, ?include_document_structure: bool, ?acceleration: AccelerationConfig, ?cache_namespace: String, ?cache_ttl_secs: Integer, ?email: EmailConfig, ?csv: CsvConfig, ?geojson: GeoJsonExtractionConfig, ?concurrency: ConcurrencyConfig, ?url: UrlExtractionConfig, ?max_archive_depth: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?chunk_classification: ChunkClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
+def initialize: (?mime_detection_policy: enum_MimeDetectionPolicy, ?use_cache: bool, ?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?ocr_strategy: OcrStrategy, ?force_ocr_pages: Array[Integer], ?disable_ocr: bool, ?chunking: ChunkingConfig, ?content_filter: ContentFilterConfig, ?images: ImageExtractionConfig, ?pdf_options: PdfConfig, ?token_reduction: TokenReductionOptions, ?language_detection: LanguageDetectionConfig, ?pages: PageConfig, ?keywords: KeywordConfig, ?postprocessor: PostProcessorConfig, ?html_options: ConversionOptions, ?html_output: HtmlOutputConfig, ?extraction_timeout_secs: Integer, ?max_concurrent_extractions: Integer, ?result_format: enum_ResultFormat, ?security_limits: SecurityLimits, ?max_embedded_file_bytes: Integer, ?output_format: OutputFormat, ?escape_markdown: bool, ?table_anchors: bool, ?jupyter_cell_rendering: enum_JupyterCellRendering, ?apply_notebook_cell_tags: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?use_layout_for_markdown: bool, ?include_document_structure: bool, ?acceleration: AccelerationConfig, ?cache_namespace: String, ?cache_ttl_secs: Integer, ?email: EmailConfig, ?csv: CsvConfig, ?geojson: GeoJsonExtractionConfig, ?concurrency: ConcurrencyConfig, ?url: UrlExtractionConfig, ?max_archive_depth: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?chunk_classification: ChunkClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
     def validate: () -> void
     def needs_image_data: () -> bool
+    def runs_ocr_on_embedded_images: () -> bool
+    def wants_own_bytes_in_result: () -> bool
     def needs_image_processing: () -> bool
     end
 
@@ -1032,7 +1038,7 @@ def initialize: (?genres: Array[String], ?sequences: Array[String], ?annotation:
     end
 
     class FileExtractionConfig
-        attr_reader mime_detection_policy: MimeDetectionPolicy?
+        attr_reader mime_detection_policy: enum_MimeDetectionPolicy?
         attr_reader enable_quality_processing: bool?
         attr_reader ocr: OcrConfig?
         attr_reader force_ocr: bool?
@@ -1050,7 +1056,7 @@ def initialize: (?genres: Array[String], ?sequences: Array[String], ?annotation:
         attr_reader postprocessor: PostProcessorConfig?
         attr_reader html_options: ConversionOptions?
         attr_reader html_output: HtmlOutputConfig?
-        attr_reader result_format: ResultFormat?
+        attr_reader result_format: enum_ResultFormat?
         attr_reader output_format: OutputFormat?
         attr_reader include_document_structure: bool?
         attr_reader layout: LayoutDetectionConfig?
@@ -1068,7 +1074,7 @@ def initialize: (?genres: Array[String], ?sequences: Array[String], ?annotation:
         attr_reader captioning: CaptioningConfig?
         attr_reader qr_codes: bool?
 
-def initialize: (?mime_detection_policy: MimeDetectionPolicy, ?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?ocr_strategy: OcrStrategy, ?force_ocr_pages: Array[Integer], ?disable_ocr: bool, ?chunking: ChunkingConfig, ?content_filter: ContentFilterConfig, ?images: ImageExtractionConfig, ?pdf_options: PdfConfig, ?token_reduction: TokenReductionOptions, ?language_detection: LanguageDetectionConfig, ?pages: PageConfig, ?keywords: KeywordConfig, ?postprocessor: PostProcessorConfig, ?html_options: ConversionOptions, ?html_output: HtmlOutputConfig, ?result_format: ResultFormat, ?output_format: OutputFormat, ?include_document_structure: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?timeout_secs: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?url: UrlExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?chunk_classification: ChunkClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
+def initialize: (?mime_detection_policy: enum_MimeDetectionPolicy, ?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?ocr_strategy: OcrStrategy, ?force_ocr_pages: Array[Integer], ?disable_ocr: bool, ?chunking: ChunkingConfig, ?content_filter: ContentFilterConfig, ?images: ImageExtractionConfig, ?pdf_options: PdfConfig, ?token_reduction: TokenReductionOptions, ?language_detection: LanguageDetectionConfig, ?pages: PageConfig, ?keywords: KeywordConfig, ?postprocessor: PostProcessorConfig, ?html_options: ConversionOptions, ?html_output: HtmlOutputConfig, ?result_format: enum_ResultFormat, ?output_format: OutputFormat, ?include_document_structure: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?timeout_secs: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?url: UrlExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?chunk_classification: ChunkClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
     end
 
     class Footnote
@@ -1101,7 +1107,7 @@ def initialize: (label: String, content: Array[FormattedBlock]) -> void
     end
 
     class FormattedBlock
-        attr_reader block_type: BlockType
+        attr_reader block_type: enum_BlockType
         attr_reader level: Integer?
     attr_reader inline_content: Array[InlineElement]
         attr_reader attributes: Attributes?
@@ -1109,7 +1115,7 @@ def initialize: (label: String, content: Array[FormattedBlock]) -> void
         attr_reader code: String?
     attr_reader children: Array[FormattedBlock]
 
-def initialize: (block_type: BlockType, ?level: Integer, inline_content: Array[InlineElement], ?attributes: Attributes, ?language: String, ?code: String, children: Array[FormattedBlock]) -> void
+def initialize: (block_type: enum_BlockType, ?level: Integer, inline_content: Array[InlineElement], ?attributes: Attributes, ?language: String, ?code: String, children: Array[FormattedBlock]) -> void
     end
 
     class Formula
@@ -1127,13 +1133,13 @@ def initialize: (block_type: BlockType, ?level: Integer, inline_content: Array[I
     end
 
     class GlmOcrBackendOptions
-        attr_reader task: GlmOcrTaskKind?
-        attr_reader device: CandleDevicePreference?
-        attr_reader layout_mode: GlmOcrLayoutMode?
+        attr_reader task: enum_GlmOcrTaskKind?
+        attr_reader device: enum_CandleDevicePreference?
+        attr_reader layout_mode: enum_GlmOcrLayoutMode?
         attr_reader enable_chart_understanding: bool?
         attr_reader cache_dir: String?
 
-    def initialize: (?task: GlmOcrTaskKind, ?device: CandleDevicePreference, ?layout_mode: GlmOcrLayoutMode, ?enable_chart_understanding: bool, ?cache_dir: String) -> void
+    def initialize: (?task: enum_GlmOcrTaskKind, ?device: enum_CandleDevicePreference, ?layout_mode: enum_GlmOcrLayoutMode, ?enable_chart_understanding: bool, ?cache_dir: String) -> void
     end
 
     class GridCell
@@ -1224,7 +1230,7 @@ def initialize: (headings: Array[HeadingLevel]) -> void
         attr_reader canonical_url: String?
         attr_reader base_href: String?
         attr_reader language: String?
-        attr_reader text_direction: TextDirection?
+        attr_reader text_direction: enum_TextDirection?
     attr_reader open_graph: Hash[String, String]
     attr_reader twitter_card: Hash[String, String]
     attr_reader meta_tags: Hash[String, String]
@@ -1233,17 +1239,17 @@ def initialize: (headings: Array[HeadingLevel]) -> void
     attr_reader images: Array[ImageMetadataType]
     attr_reader structured_data: Array[StructuredData]
 
-def initialize: (?title: String, ?description: String, ?keywords: Array[String], ?author: String, ?canonical_url: String, ?base_href: String, ?language: String, ?text_direction: TextDirection, ?open_graph: Hash[String, String], ?twitter_card: Hash[String, String], ?meta_tags: Hash[String, String], ?headers: Array[HeaderMetadata], ?links: Array[LinkMetadata], ?images: Array[ImageMetadataType], ?structured_data: Array[StructuredData]) -> void
+def initialize: (?title: String, ?description: String, ?keywords: Array[String], ?author: String, ?canonical_url: String, ?base_href: String, ?language: String, ?text_direction: enum_TextDirection, ?open_graph: Hash[String, String], ?twitter_card: Hash[String, String], ?meta_tags: Hash[String, String], ?headers: Array[HeaderMetadata], ?links: Array[LinkMetadata], ?images: Array[ImageMetadataType], ?structured_data: Array[StructuredData]) -> void
     end
 
     class HtmlOutputConfig
         attr_reader css: String?
         attr_reader css_file: String?
-        attr_reader theme: HtmlTheme
+        attr_reader theme: enum_HtmlTheme
         attr_reader class_prefix: String
         attr_reader embed_css: bool
 
-    def initialize: (?css: String, ?css_file: String, ?theme: HtmlTheme, ?class_prefix: String, ?embed_css: bool) -> void
+    def initialize: (?css: String, ?css_file: String, ?theme: enum_HtmlTheme, ?class_prefix: String, ?embed_css: bool) -> void
     end
 
     class ImageDimensions
@@ -1295,10 +1301,10 @@ def initialize: (?width: Integer, ?height: Integer, ?format: String, ?exif: Hash
         attr_reader alt: String?
         attr_reader title: String?
         attr_reader dimensions: ImageDimensions?
-        attr_reader image_type: ImageType
+        attr_reader image_type: enum_ImageType
     attr_reader attributes: Array[KeyValueAttribute]
 
-def initialize: (src: String, ?alt: String, ?title: String, ?dimensions: ImageDimensions, image_type: ImageType, attributes: Array[KeyValueAttribute]) -> void
+def initialize: (src: String, ?alt: String, ?title: String, ?dimensions: ImageDimensions, image_type: enum_ImageType, attributes: Array[KeyValueAttribute]) -> void
     end
 
     class ImagePreprocessingConfig
@@ -1331,12 +1337,12 @@ def initialize: (src: String, ?alt: String, ?title: String, ?dimensions: ImageDi
     end
 
     class InlineElement
-        attr_reader element_type: InlineType
+        attr_reader element_type: enum_InlineType
         attr_reader content: String
         attr_reader attributes: Attributes?
     attr_reader metadata: Hash[String, String]?
 
-def initialize: (element_type: InlineType, content: String, ?attributes: Attributes, ?metadata: Hash[String, String]) -> void
+def initialize: (element_type: enum_InlineType, content: String, ?attributes: Attributes, ?metadata: Hash[String, String]) -> void
     end
 
     class JatsMetadata
@@ -1358,14 +1364,14 @@ def initialize: (?copyright: String, ?license: String, ?history_dates: Hash[Stri
     class Keyword
         attr_reader text: String
         attr_reader score: Float
-        attr_reader algorithm: KeywordAlgorithm
+        attr_reader algorithm: enum_KeywordAlgorithm
     attr_reader positions: Array[Integer]?
 
-def initialize: (text: String, score: Float, algorithm: KeywordAlgorithm, ?positions: Array[Integer]) -> void
+def initialize: (text: String, score: Float, algorithm: enum_KeywordAlgorithm, ?positions: Array[Integer]) -> void
     end
 
     class KeywordConfig
-        attr_reader algorithm: KeywordAlgorithm
+        attr_reader algorithm: enum_KeywordAlgorithm
         attr_reader max_keywords: Integer
         attr_reader min_score: Float
         attr_reader ngram_range: NgramRange
@@ -1373,7 +1379,7 @@ def initialize: (text: String, score: Float, algorithm: KeywordAlgorithm, ?posit
         attr_reader yake_params: YakeParams?
         attr_reader rake_params: RakeParams?
 
-    def initialize: (?algorithm: KeywordAlgorithm, ?max_keywords: Integer, ?min_score: Float, ?ngram_range: NgramRange, ?language: String, ?yake_params: YakeParams, ?rake_params: RakeParams) -> void
+    def initialize: (?algorithm: enum_KeywordAlgorithm, ?max_keywords: Integer, ?min_score: Float, ?ngram_range: NgramRange, ?language: String, ?yake_params: YakeParams, ?rake_params: RakeParams) -> void
     end
 
     class LanguageConfidence
@@ -1428,24 +1434,24 @@ def initialize: (name: String, model_repo: String, model_file: String, additiona
     end
 
     class LayoutDetection
-        attr_reader class_name: LayoutClass
+        attr_reader class_name: enum_LayoutClass
         attr_reader confidence: Float
         attr_reader bbox: BBox
 
-    def initialize: (class_name: LayoutClass, confidence: Float, bbox: BBox) -> void
+    def initialize: (class_name: enum_LayoutClass, confidence: Float, bbox: BBox) -> void
     end
 
     class LayoutDetectionConfig
-        attr_reader strategy: LayoutStrategy
+        attr_reader strategy: enum_LayoutStrategy
         attr_reader confidence_threshold: Float?
         attr_reader apply_heuristics: bool
-        attr_reader table_model: TableModel
-        attr_reader formula_model: FormulaModel?
-        attr_reader table_overlap_preference: TableOverlapPreference
+        attr_reader table_model: enum_TableModel
+        attr_reader formula_model: enum_FormulaModel?
+        attr_reader table_overlap_preference: enum_TableOverlapPreference
         attr_reader acceleration: AccelerationConfig?
         attr_reader enable_chart_understanding: bool
 
-    def initialize: (?strategy: LayoutStrategy, ?confidence_threshold: Float, ?apply_heuristics: bool, ?table_model: TableModel, ?formula_model: FormulaModel, ?table_overlap_preference: TableOverlapPreference, ?acceleration: AccelerationConfig, ?enable_chart_understanding: bool) -> void
+    def initialize: (?strategy: enum_LayoutStrategy, ?confidence_threshold: Float, ?apply_heuristics: bool, ?table_model: enum_TableModel, ?formula_model: enum_FormulaModel, ?table_overlap_preference: enum_TableOverlapPreference, ?acceleration: AccelerationConfig, ?enable_chart_understanding: bool) -> void
     end
 
     class LayoutRegion
@@ -1461,11 +1467,11 @@ def initialize: (name: String, model_repo: String, model_file: String, additiona
         attr_reader href: String
         attr_reader text: String
         attr_reader title: String?
-        attr_reader link_type: LinkType
+        attr_reader link_type: enum_LinkType
     attr_reader rel: Array[String]
     attr_reader attributes: Array[KeyValueAttribute]
 
-def initialize: (href: String, text: String, ?title: String, link_type: LinkType, rel: Array[String], attributes: Array[KeyValueAttribute]) -> void
+def initialize: (href: String, text: String, ?title: String, link_type: enum_LinkType, rel: Array[String], attributes: Array[KeyValueAttribute]) -> void
     end
 
     class LlmBudgetConfig
@@ -1633,13 +1639,13 @@ def initialize: (page_count: Integer, pages: Array[PageSignals]) -> void
     end
 
     class NerConfig
-        attr_reader backend: NerBackendKind
+        attr_reader backend: enum_NerBackendKind
     attr_reader categories: Array[EntityCategory]
         attr_reader model: String?
         attr_reader llm: LlmConfig?
     attr_reader custom_labels: Array[String]
 
-def initialize: (?backend: NerBackendKind, ?categories: Array[EntityCategory], ?model: String, ?llm: LlmConfig, ?custom_labels: Array[String]) -> void
+def initialize: (?backend: enum_NerBackendKind, ?categories: Array[EntityCategory], ?model: String, ?llm: LlmConfig, ?custom_labels: Array[String]) -> void
     end
 
     class NgramRange
@@ -1690,22 +1696,22 @@ def initialize: (?enabled: bool, ?backend: String, ?language: Array[String], ?te
         attr_reader text: String
         attr_reader geometry: OcrBoundingGeometry
         attr_reader confidence: OcrConfidence
-        attr_reader level: OcrElementLevel
+        attr_reader level: enum_OcrElementLevel
         attr_reader rotation: OcrRotation?
         attr_reader page_number: Integer
         attr_reader parent_id: String?
     attr_reader backend_metadata: Hash[String, String]
 
-def initialize: (?text: String, ?geometry: OcrBoundingGeometry, ?confidence: OcrConfidence, ?level: OcrElementLevel, ?rotation: OcrRotation, ?page_number: Integer, ?parent_id: String, ?backend_metadata: Hash[String, String]) -> void
+def initialize: (?text: String, ?geometry: OcrBoundingGeometry, ?confidence: OcrConfidence, ?level: enum_OcrElementLevel, ?rotation: OcrRotation, ?page_number: Integer, ?parent_id: String, ?backend_metadata: Hash[String, String]) -> void
     end
 
     class OcrElementConfig
         attr_reader include_elements: bool
-        attr_reader min_level: OcrElementLevel
+        attr_reader min_level: enum_OcrElementLevel
         attr_reader min_confidence: Float
         attr_reader build_hierarchy: bool
 
-    def initialize: (?include_elements: bool, ?min_level: OcrElementLevel, ?min_confidence: Float, ?build_hierarchy: bool) -> void
+    def initialize: (?include_elements: bool, ?min_level: enum_OcrElementLevel, ?min_confidence: Float, ?build_hierarchy: bool) -> void
     end
 
     class OcrExtractionResult
@@ -1780,8 +1786,10 @@ def initialize: (backend: String, priority: Integer, ?language: Array[String], ?
         attr_reader min_undecodable_ratio: Float
         attr_reader enable_provenance_ocr_routing: bool
         attr_reader min_provenance_fallback_ratio: Float
+        attr_reader enable_plausibility_ocr_routing: bool
+        attr_reader min_reliable_language_chunk_ratio: Float
 
-    def initialize: (?min_total_non_whitespace: Integer, ?min_non_whitespace_per_page: Float, ?min_meaningful_word_len: Integer, ?min_meaningful_words: Integer, ?min_alnum_ratio: Float, ?min_garbage_chars: Integer, ?max_fragmented_word_ratio: Float, ?critical_fragmented_word_ratio: Float, ?max_ocr_output_fragmented_word_ratio: Float, ?min_ocr_mean_confidence: Float, ?min_words_for_ocr_output_check: Integer, ?max_ocr_output_dict_invalid_word_ratio: Float, ?discard_suspected_ocr_noise: bool, ?min_avg_word_length: Float, ?min_words_for_avg_length_check: Integer, ?min_consecutive_repeat_ratio: Float, ?min_words_for_repeat_check: Integer, ?substantive_min_chars: Integer, ?non_text_min_chars: Integer, ?alnum_ws_ratio_threshold: Float, ?pipeline_min_quality: Float, ?min_undecodable_ratio: Float, ?enable_provenance_ocr_routing: bool, ?min_provenance_fallback_ratio: Float) -> void
+    def initialize: (?min_total_non_whitespace: Integer, ?min_non_whitespace_per_page: Float, ?min_meaningful_word_len: Integer, ?min_meaningful_words: Integer, ?min_alnum_ratio: Float, ?min_garbage_chars: Integer, ?max_fragmented_word_ratio: Float, ?critical_fragmented_word_ratio: Float, ?max_ocr_output_fragmented_word_ratio: Float, ?min_ocr_mean_confidence: Float, ?min_words_for_ocr_output_check: Integer, ?max_ocr_output_dict_invalid_word_ratio: Float, ?discard_suspected_ocr_noise: bool, ?min_avg_word_length: Float, ?min_words_for_avg_length_check: Integer, ?min_consecutive_repeat_ratio: Float, ?min_words_for_repeat_check: Integer, ?substantive_min_chars: Integer, ?non_text_min_chars: Integer, ?alnum_ws_ratio_threshold: Float, ?pipeline_min_quality: Float, ?min_undecodable_ratio: Float, ?enable_provenance_ocr_routing: bool, ?min_provenance_fallback_ratio: Float, ?enable_plausibility_ocr_routing: bool, ?min_reliable_language_chunk_ratio: Float) -> void
     end
 
     class OcrRotation
@@ -1830,9 +1838,9 @@ def initialize: (cells: Array[Array[String]], markdown: String, page_number: Int
         attr_reader drop_score: Float
         attr_reader model_tier: String
         attr_reader model_version: String
-        attr_reader inference_backend: PaddleInferenceBackend?
+        attr_reader inference_backend: enum_PaddleInferenceBackend?
 
-    def initialize: (?language: String, ?cache_dir: String, ?use_angle_cls: bool, ?enable_table_detection: bool, ?det_db_thresh: Float, ?det_db_box_thresh: Float, ?det_db_unclip_ratio: Float, ?det_limit_side_len: Integer, ?rec_batch_num: Integer, ?padding: Integer, ?drop_score: Float, ?model_tier: String, ?model_version: String, ?inference_backend: PaddleInferenceBackend) -> void
+    def initialize: (?language: String, ?cache_dir: String, ?use_angle_cls: bool, ?enable_table_detection: bool, ?det_db_thresh: Float, ?det_db_box_thresh: Float, ?det_db_unclip_ratio: Float, ?det_limit_side_len: Integer, ?rec_batch_num: Integer, ?padding: Integer, ?drop_score: Float, ?model_tier: String, ?model_version: String, ?inference_backend: enum_PaddleInferenceBackend) -> void
     def with_cache_dir: (String path) -> PaddleOcrConfig
     def with_table_detection: (bool enable) -> PaddleOcrConfig
     def with_angle_cls: (bool enable) -> PaddleOcrConfig
@@ -1848,14 +1856,14 @@ def initialize: (cells: Array[Array[String]], markdown: String, page_number: Int
     end
 
     class PaddleOcrVlBackendOptions
-        attr_reader task: PaddleOcrVlTaskKind?
+        attr_reader task: enum_PaddleOcrVlTaskKind?
         attr_reader model_path: String?
         attr_reader model_id: String?
         attr_reader hf_revision: String?
         attr_reader cache_dir: String?
-        attr_reader device: CandleDevicePreference?
+        attr_reader device: enum_CandleDevicePreference?
 
-    def initialize: (?task: PaddleOcrVlTaskKind, ?model_path: String, ?model_id: String, ?hf_revision: String, ?cache_dir: String, ?device: CandleDevicePreference) -> void
+    def initialize: (?task: enum_PaddleOcrVlTaskKind, ?model_path: String, ?model_id: String, ?hf_revision: String, ?cache_dir: String, ?device: enum_CandleDevicePreference) -> void
     end
 
     class PageBoundary
@@ -1970,11 +1978,11 @@ def initialize: (block_count: Integer, blocks: Array[HierarchicalBlock]) -> void
 
     class PageStructure
         attr_reader total_count: Integer
-        attr_reader unit_type: PageUnitType
+        attr_reader unit_type: enum_PageUnitType
     attr_reader boundaries: Array[PageBoundary]?
     attr_reader pages: Array[PageInfo]?
 
-def initialize: (total_count: Integer, unit_type: PageUnitType, ?boundaries: Array[PageBoundary], ?pages: Array[PageInfo]) -> void
+def initialize: (total_count: Integer, unit_type: enum_PageUnitType, ?boundaries: Array[PageBoundary], ?pages: Array[PageInfo]) -> void
     end
 
     class PatternMatch
@@ -1987,7 +1995,7 @@ def initialize: (total_count: Integer, unit_type: PageUnitType, ?boundaries: Arr
     end
 
     class PdfAnnotation
-        attr_reader annotation_type: PdfAnnotationType
+        attr_reader annotation_type: enum_PdfAnnotationType
         attr_reader content: String?
         attr_reader page_number: Integer
         attr_reader bounding_box: BoundingBox?
@@ -1998,7 +2006,7 @@ def initialize: (total_count: Integer, unit_type: PageUnitType, ?boundaries: Arr
     attr_reader quad_points: Array[BoundingBox]?
         attr_reader marked_text: String?
 
-def initialize: (annotation_type: PdfAnnotationType, ?content: String, page_number: Integer, ?bounding_box: BoundingBox, ?author: String, ?modified: String, ?color: String, ?subject: String, ?quad_points: Array[BoundingBox], ?marked_text: String) -> void
+def initialize: (annotation_type: enum_PdfAnnotationType, ?content: String, page_number: Integer, ?bounding_box: BoundingBox, ?author: String, ?modified: String, ?color: String, ?subject: String, ?quad_points: Array[BoundingBox], ?marked_text: String) -> void
     end
 
     class PdfConfig
@@ -2014,16 +2022,16 @@ def initialize: (annotation_type: PdfAnnotationType, ?content: String, page_numb
         attr_reader ocr_inline_images: bool
         attr_reader extract_form_fields: bool
         attr_reader reading_order: bool
-        attr_reader backend: PdfBackend
+        attr_reader backend: enum_PdfBackend
 
-def initialize: (?extract_images: bool, ?extract_tables: bool, ?passwords: Array[String], ?extract_metadata: bool, ?hierarchy: HierarchyConfig, ?extract_annotations: bool, ?top_margin_fraction: Float, ?bottom_margin_fraction: Float, ?allow_single_column_tables: bool, ?ocr_inline_images: bool, ?extract_form_fields: bool, ?reading_order: bool, ?backend: PdfBackend) -> void
+def initialize: (?extract_images: bool, ?extract_tables: bool, ?passwords: Array[String], ?extract_metadata: bool, ?hierarchy: HierarchyConfig, ?extract_annotations: bool, ?top_margin_fraction: Float, ?bottom_margin_fraction: Float, ?allow_single_column_tables: bool, ?ocr_inline_images: bool, ?extract_form_fields: bool, ?reading_order: bool, ?backend: enum_PdfBackend) -> void
     def validate: () -> void
     end
 
     class PdfFormField
         attr_reader name: String
         attr_reader full_name: String
-        attr_reader field_type: FormFieldType
+        attr_reader field_type: enum_FormFieldType
         attr_reader value: String?
         attr_reader default_value: String?
         attr_reader flags: Integer
@@ -2032,7 +2040,7 @@ def initialize: (?extract_images: bool, ?extract_tables: bool, ?passwords: Array
         attr_reader max_length: Integer?
         attr_reader tooltip: String?
 
-    def initialize: (name: String, full_name: String, field_type: FormFieldType, ?value: String, ?default_value: String, flags: Integer, ?page: Integer, ?bbox: BoundingBox, ?max_length: Integer, ?tooltip: String) -> void
+    def initialize: (name: String, full_name: String, field_type: enum_FormFieldType, ?value: String, ?default_value: String, flags: Integer, ?page: Integer, ?bbox: BoundingBox, ?max_length: Integer, ?tooltip: String) -> void
     end
 
     class PdfMetadata
@@ -2044,10 +2052,12 @@ def initialize: (?extract_images: bool, ?extract_tables: bool, ?passwords: Array
         attr_reader page_count: Integer?
         attr_reader scanned_confidence: Float?
     attr_reader scanned_pages: Array[Integer]?
+    attr_reader fabricated_text_pages: Array[Integer]?
+    attr_reader implausible_text_pages: Array[Integer]?
     attr_reader layout_gated_pages: Array[Integer]?
     attr_reader layout_gate_reasons: Array[String]?
 
-def initialize: (?pdf_version: String, ?producer: String, ?is_encrypted: bool, ?width: Integer, ?height: Integer, ?page_count: Integer, ?scanned_confidence: Float, ?scanned_pages: Array[Integer], ?layout_gated_pages: Array[Integer], ?layout_gate_reasons: Array[String]) -> void
+def initialize: (?pdf_version: String, ?producer: String, ?is_encrypted: bool, ?width: Integer, ?height: Integer, ?page_count: Integer, ?scanned_confidence: Float, ?scanned_pages: Array[Integer], ?fabricated_text_pages: Array[Integer], ?implausible_text_pages: Array[Integer], ?layout_gated_pages: Array[Integer], ?layout_gate_reasons: Array[String]) -> void
     end
 
     class PixelDimensions
@@ -2115,11 +2125,11 @@ def initialize: (?slide_count: Integer, ?slide_names: Array[String], ?image_coun
 
     class PreprocessingOptions
         attr_reader enabled: bool
-        attr_reader preset: PreprocessingPreset
+        attr_reader preset: enum_PreprocessingPreset
         attr_reader remove_navigation: bool
         attr_reader remove_forms: bool
 
-    def initialize: (?enabled: bool, ?preset: PreprocessingPreset, ?remove_navigation: bool, ?remove_forms: bool) -> void
+    def initialize: (?enabled: bool, ?preset: enum_PreprocessingPreset, ?remove_navigation: bool, ?remove_forms: bool) -> void
     end
 
     class PresentationHyperlink
@@ -2134,18 +2144,18 @@ def initialize: (?slide_count: Integer, ?slide_names: Array[String], ?image_coun
         attr_reader version: String
         attr_reader schema_name: String
         attr_reader description: String
-        attr_reader category: PresetCategory
+        attr_reader category: enum_PresetCategory
     attr_reader tags: Array[String]
         attr_reader schema: String
         attr_reader system_prompt: String
         attr_reader context_template: String?
-        attr_reader merge_mode: MergeMode
-        attr_reader preferred_call_mode: CallMode
+        attr_reader merge_mode: enum_MergeMode
+        attr_reader preferred_call_mode: enum_CallMode
         attr_reader emit_citations: bool
         attr_reader sample: PresetSample?
         attr_reader fingerprint: String
 
-def initialize: (id: String, version: String, schema_name: String, description: String, category: PresetCategory, tags: Array[String], schema: String, system_prompt: String, ?context_template: String, merge_mode: MergeMode, preferred_call_mode: CallMode, emit_citations: bool, ?sample: PresetSample, fingerprint: String) -> void
+def initialize: (id: String, version: String, schema_name: String, description: String, category: enum_PresetCategory, tags: Array[String], schema: String, system_prompt: String, ?context_template: String, merge_mode: enum_MergeMode, preferred_call_mode: enum_CallMode, emit_citations: bool, ?sample: PresetSample, fingerprint: String) -> void
     end
 
     class PresetSample
@@ -2160,13 +2170,13 @@ def initialize: (id: String, version: String, schema_name: String, description:
         attr_reader version: String
         attr_reader schema_name: String
         attr_reader description: String
-        attr_reader category: PresetCategory
+        attr_reader category: enum_PresetCategory
     attr_reader tags: Array[String]
-        attr_reader preferred_call_mode: CallMode
+        attr_reader preferred_call_mode: enum_CallMode
         attr_reader emit_citations: bool
         attr_reader fingerprint: String
 
-def initialize: (id: String, version: String, schema_name: String, description: String, category: PresetCategory, tags: Array[String], preferred_call_mode: CallMode, emit_citations: bool, fingerprint: String) -> void
+def initialize: (id: String, version: String, schema_name: String, description: String, category: enum_PresetCategory, tags: Array[String], preferred_call_mode: enum_CallMode, emit_citations: bool, fingerprint: String) -> void
     end
 
     class ProcessingWarning
@@ -2232,13 +2242,13 @@ def initialize: (detection_bbox: BBox, cells: Array[Array[String]], markdown: St
 
     class RedactionConfig
     attr_reader categories: Array[PiiCategory]
-        attr_reader strategy: RedactionStrategy
+        attr_reader strategy: enum_RedactionStrategy
         attr_reader ner: NerConfig?
         attr_reader preserve_offsets: bool
     attr_reader custom_terms: Array[RedactionTerm]
     attr_reader custom_patterns: Array[RedactionPattern]
 
-def initialize: (?categories: Array[PiiCategory], ?strategy: RedactionStrategy, ?ner: NerConfig, ?preserve_offsets: bool, ?custom_terms: Array[RedactionTerm], ?custom_patterns: Array[RedactionPattern]) -> void
+def initialize: (?categories: Array[PiiCategory], ?strategy: enum_RedactionStrategy, ?ner: NerConfig, ?preserve_offsets: bool, ?custom_terms: Array[RedactionTerm], ?custom_patterns: Array[RedactionPattern]) -> void
     def validate: () -> void
     end
 
@@ -2246,10 +2256,10 @@ def initialize: (?categories: Array[PiiCategory], ?strategy: RedactionStrategy,
         attr_reader start: Integer
         attr_reader end: Integer
         attr_reader category: PiiCategory
-        attr_reader strategy: RedactionStrategy
+        attr_reader strategy: enum_RedactionStrategy
         attr_reader replacement_token: String
 
-    def initialize: (start: Integer, end: Integer, category: PiiCategory, strategy: RedactionStrategy, replacement_token: String) -> void
+    def initialize: (start: Integer, end: Integer, category: PiiCategory, strategy: enum_RedactionStrategy, replacement_token: String) -> void
     end
 
     class RedactionPattern
@@ -2311,11 +2321,11 @@ def summaries: () -> Array[PresetSummary]
         attr_reader schema_name: String
         attr_reader schema: String
         attr_reader system_prompt: String
-        attr_reader merge_mode: MergeMode
-        attr_reader preferred_call_mode: CallMode
+        attr_reader merge_mode: enum_MergeMode
+        attr_reader preferred_call_mode: enum_CallMode
         attr_reader emit_citations: bool
 
-    def initialize: (id: String, version: String, fingerprint: String, schema_name: String, schema: String, system_prompt: String, merge_mode: MergeMode, preferred_call_mode: CallMode, emit_citations: bool) -> void
+    def initialize: (id: String, version: String, fingerprint: String, schema_name: String, schema: String, system_prompt: String, merge_mode: enum_MergeMode, preferred_call_mode: enum_CallMode, emit_citations: bool) -> void
     end
 
     class RevisionDelta
@@ -2406,11 +2416,11 @@ def initialize: (?deny_private: bool, ?allowlist: Array[HostMatcher], ?max_redir
     end
 
     class StructuredData
-        attr_reader data_type: StructuredDataType
+        attr_reader data_type: enum_StructuredDataType
         attr_reader raw_json: String
         attr_reader schema_type: String?
 
-    def initialize: (data_type: StructuredDataType, raw_json: String, ?schema_type: String) -> void
+    def initialize: (data_type: enum_StructuredDataType, raw_json: String, ?schema_type: String) -> void
     end
 
     class StructuredDataResult
@@ -2436,11 +2446,11 @@ def initialize: (content: String, format: String, metadata: Hash[String, String]
     end
 
     class SummarizationConfig
-        attr_reader strategy: SummaryStrategy
+        attr_reader strategy: enum_SummaryStrategy
         attr_reader max_tokens: Integer?
         attr_reader llm: LlmConfig?
 
-    def initialize: (?strategy: SummaryStrategy, ?max_tokens: Integer, ?llm: LlmConfig) -> void
+    def initialize: (?strategy: enum_SummaryStrategy, ?max_tokens: Integer, ?llm: LlmConfig) -> void
     end
 
     class SupportedFormat
@@ -2564,7 +2574,7 @@ def initialize: (?line_count: Integer, ?word_count: Integer, ?character_count: I
     end
 
     class TokenReductionConfig
-        attr_reader level: ReductionLevel
+        attr_reader level: enum_ReductionLevel
         attr_reader language_hint: String?
         attr_reader preserve_markdown: bool
         attr_reader preserve_code: bool
@@ -2577,7 +2587,7 @@ attr_reader custom_stopwords: Hash[String, Array[String]]?
         attr_reader enable_semantic_clustering: bool
         attr_reader preserve_important_words: bool
 
-def initialize: (?level: ReductionLevel, ?language_hint: String, ?preserve_markdown: bool, ?preserve_code: bool, ?semantic_threshold: Float, ?enable_parallel: bool, ?use_simd: bool, ?custom_stopwords: Hash[String, Array[String]], ?preserve_patterns: Array[String], ?target_reduction: Float, ?enable_semantic_clustering: bool, ?preserve_important_words: bool) -> void
+def initialize: (?level: enum_ReductionLevel, ?language_hint: String, ?preserve_markdown: bool, ?preserve_code: bool, ?semantic_threshold: Float, ?enable_parallel: bool, ?use_simd: bool, ?custom_stopwords: Hash[String, Array[String]], ?preserve_patterns: Array[String], ?target_reduction: Float, ?enable_semantic_clustering: bool, ?preserve_important_words: bool) -> void
     end
 
     class TokenReductionOptions
@@ -2589,7 +2599,7 @@ def initialize: (?level: ReductionLevel, ?language_hint: String, ?preserve_markd
 
     class TranscriptionConfig
         attr_reader enabled: bool
-        attr_reader model: WhisperModel
+        attr_reader model: enum_WhisperModel
         attr_reader language: String?
         attr_reader timestamps: bool
         attr_reader max_duration_ms: Integer?
@@ -2599,7 +2609,7 @@ def initialize: (?level: ReductionLevel, ?language_hint: String, ?preserve_markd
         attr_reader allow_network: bool
         attr_reader verify_hash: bool
 
-    def initialize: (?enabled: bool, ?model: WhisperModel, ?language: String, ?timestamps: bool, ?max_duration_ms: Integer, ?max_bytes: Integer, ?timeout_ms: Integer, ?model_cache_dir: String, ?allow_network: bool, ?verify_hash: bool) -> void
+    def initialize: (?enabled: bool, ?model: enum_WhisperModel, ?language: String, ?timestamps: bool, ?max_duration_ms: Integer, ?max_bytes: Integer, ?timeout_ms: Integer, ?model_cache_dir: String, ?allow_network: bool, ?verify_hash: bool) -> void
     end
 
     class Translation
@@ -2640,22 +2650,22 @@ def initialize: (?enabled: bool, ?cache_dir: String, ?languages: Array[String],
         attr_reader diagnostics: bool
         attr_reader data_extraction: bool
         attr_reader chunk_max_size: Integer?
-        attr_reader content_mode: CodeContentMode
+        attr_reader content_mode: enum_CodeContentMode
 
-    def initialize: (?structure: bool, ?imports: bool, ?exports: bool, ?comments: bool, ?docstrings: bool, ?symbols: bool, ?diagnostics: bool, ?data_extraction: bool, ?chunk_max_size: Integer, ?content_mode: CodeContentMode) -> void
+    def initialize: (?structure: bool, ?imports: bool, ?exports: bool, ?comments: bool, ?docstrings: bool, ?symbols: bool, ?diagnostics: bool, ?data_extraction: bool, ?chunk_max_size: Integer, ?content_mode: enum_CodeContentMode) -> void
     end
 
     class TrocrBackendOptions
-        attr_reader variant: CandleTrocrVariant?
-        attr_reader device: CandleDevicePreference?
+        attr_reader variant: enum_CandleTrocrVariant?
+        attr_reader device: enum_CandleDevicePreference?
         attr_reader cache_dir: String?
         attr_reader hf_revision: String?
 
-    def initialize: (?variant: CandleTrocrVariant, ?device: CandleDevicePreference, ?cache_dir: String, ?hf_revision: String) -> void
+    def initialize: (?variant: enum_CandleTrocrVariant, ?device: enum_CandleDevicePreference, ?cache_dir: String, ?hf_revision: String) -> void
     end
 
     class UrlExtractionConfig
-        attr_reader mode: UrlExtractionMode
+        attr_reader mode: enum_UrlExtractionMode
         attr_reader crawl: CrawlConfig
         attr_reader document_url_pattern: String?
         attr_reader max_document_urls_per_result: Integer?
@@ -2663,7 +2673,7 @@ def initialize: (?enabled: bool, ?cache_dir: String, ?languages: Array[String],
         attr_reader allow_local_file_inputs: bool
         attr_reader allow_file_uris: bool
 
-    def initialize: (?mode: UrlExtractionMode, ?crawl: CrawlConfig, ?document_url_pattern: String, ?max_document_urls_per_result: Integer, ?max_total_urls: Integer, ?allow_local_file_inputs: bool, ?allow_file_uris: bool) -> void
+    def initialize: (?mode: enum_UrlExtractionMode, ?crawl: CrawlConfig, ?document_url_pattern: String, ?max_document_urls_per_result: Integer, ?max_total_urls: Integer, ?allow_local_file_inputs: bool, ?allow_file_uris: bool) -> void
     end
 
     class UserChunkConfig
@@ -2721,87 +2731,55 @@ def initialize: (?min: Integer, ?max: Integer, years: Array[Integer]) -> void
     class AnnotationKind
     end
 
-    class AssetCategory
-        type value = :document | :image | :audio | :video | :font | :stylesheet | :script | :archive | :data | :other
-    end
+    type enum_AssetCategory = :document | :image | :audio | :video | :font | :stylesheet | :script | :archive | :data | :other
 
     class AuthConfig
     end
 
-    class BlockType
-        type value = :paragraph | :heading | :blockquote | :code_block | :list_item | :ordered_list | :bullet_list | :task_list | :definition_list | :definition_term | :definition_description | :div | :section | :thematic_break | :raw_block | :math_display
-    end
+    type enum_BlockType = :paragraph | :heading | :blockquote | :code_block | :list_item | :ordered_list | :bullet_list | :task_list | :definition_list | :definition_term | :definition_description | :div | :section | :thematic_break | :raw_block | :math_display
 
-    class BoundaryReason
-        type value = :start | :page_one_marker | :letterhead_reset | :density_shift | :end
-    end
+    type enum_BoundaryReason = :start | :page_one_marker | :letterhead_reset | :density_shift | :end
 
-    class BrowserBackend
-        type value = :chromiumoxide | :native
-    end
+    type enum_BrowserBackend = :chromiumoxide | :native
 
-    class BrowserMode
-        type value = :auto | :always | :never | :stealth
-    end
+    type enum_BrowserMode = :auto | :always | :never | :stealth
 
-    class BrowserWait
-        type value = :network_idle | :selector | :fixed
-    end
+    type enum_BrowserWait = :network_idle | :selector | :fixed
 
-    class CallMode
-        type value = :text_only | :vision_only | :text_plus_vision
-    end
+    type enum_CallMode = :text_only | :vision_only | :text_plus_vision
 
-    class CandleDevicePreference
-        type value = :auto | :cpu | :cuda | :metal
-    end
+    type enum_CandleDeepseekOcrDtype = :auto | :f32 | :f16 | :bf16
 
-    class CandleTrocrVariant
-        type value = :"base-printed" | :"large-printed" | :"base-handwritten" | :"large-handwritten"
-    end
+    type enum_CandleDevicePreference = :auto | :cpu | :cuda | :metal
+
+    type enum_CandleTrocrVariant = :"base-printed" | :"large-printed" | :"base-handwritten" | :"large-handwritten"
 
     class ChunkSizing
     end
 
-    class ChunkType
-        type value = :heading | :party_list | :definitions | :operative_clause | :signature_block | :schedule | :table_like | :formula | :code_block | :function | :class | :module | :image | :org_chart | :diagram | :unknown
-    end
+    type enum_ChunkType = :heading | :party_list | :definitions | :operative_clause | :signature_block | :schedule | :table_like | :formula | :code_block | :function | :class | :module | :image | :org_chart | :diagram | :unknown
 
-    class ChunkerType
-        type value = :text | :markdown | :yaml | :semantic
-    end
+    type enum_ChunkerType = :text | :markdown | :yaml | :semantic
 
     class ChunkingReason
     end
 
-    class CodeBlockStyle
-        type value = :Indented | :Backticks | :Tildes
-    end
+    type enum_CodeBlockStyle = :Indented | :Backticks | :Tildes
 
-    class CodeContentMode
-        type value = :chunks | :raw | :structure
-    end
+    type enum_CodeContentMode = :chunks | :raw | :structure
 
-    class CodeDataNodeKind
-        type value = :key_value | :element | :sequence
-    end
+    type enum_CodeDataNodeKind = :key_value | :element | :sequence
 
     class ConfidenceSemantics
     def self.legibility: (Float scale_max) -> ConfidenceSemantics
 
     end
 
-    class ContentFilterKind
-        type value = :bm25
-    end
+    type enum_ContentFilterKind = :bm25
 
-    class ContentLayer
-        type value = :body | :header | :footer | :footnote
-    end
+    type enum_ContentLayer = :body | :header | :footer | :footnote
 
-    class CrawlStrategyKind
-        type value = :bfs | :dfs | :best_first | :adaptive
-    end
+    type enum_CrawlStrategyKind = :bfs | :dfs | :best_first | :adaptive
 
     class CredentialProviderConfig
     end
@@ -2809,139 +2787,81 @@ def initialize: (?min: Integer, ?max: Integer, years: Array[Integer]) -> void
     class DiffLine
     end
 
-    class DocumentContentEncoding
-        type value = :base64
-    end
+    type enum_DocumentContentEncoding = :base64
 
-    class ElementType
-        type value = :title | :narrative_text | :heading | :list_item | :table | :image | :page_break | :code_block | :formula | :block_quote | :footer | :header
-    end
+    type enum_ElementType = :title | :narrative_text | :heading | :list_item | :table | :image | :page_break | :code_block | :formula | :block_quote | :footer | :header
 
     class EmbeddingModelType
     end
 
-    class EmbeddingsEmbeddingBackend
-        type value = :onnx | :static
-    end
+    type enum_EmbeddingsEmbeddingBackend = :onnx | :static
 
     class EntityCategory
     end
 
-    class ExecutionProviderType
-        type value = :auto | :cpu | :coreml | :cuda | :tensorrt
-    end
+    type enum_ExecutionProviderType = :auto | :cpu | :coreml | :cuda | :tensorrt
 
-    class ExtractInputKind
-        type value = :bytes | :uri
-    end
+    type enum_ExtractInputKind = :bytes | :uri
 
-    class ExtractionMethod
-        type value = :native | :ocr | :mixed
-    end
+    type enum_ExtractionMethod = :native | :ocr | :mixed
 
-    class FormFieldType
-        type value = :text | :checkbox | :radio | :choice | :signature | :button | :unknown
-    end
+    type enum_FormFieldType = :text | :checkbox | :radio | :choice | :signature | :button | :unknown
 
     class FormatMetadata
     end
 
-    class FormulaModel
-        type value = :latex_ocr
-    end
+    type enum_FormulaModel = :latex_ocr
 
-    class GlmOcrLayoutMode
-        type value = :whole_page | :paired
-    end
+    type enum_GlmOcrLayoutMode = :whole_page | :paired
 
-    class GlmOcrTaskKind
-        type value = :ocr | :table | :formula | :chart | :caption
-    end
+    type enum_GlmOcrTaskKind = :ocr | :table | :formula | :chart | :caption
 
-    class HeadingStyle
-        type value = :Underlined | :Atx | :AtxClosed
-    end
+    type enum_HeadingStyle = :Underlined | :Atx | :AtxClosed
 
-    class HighlightStyle
-        type value = :DoubleEqual | :Html | :Bold | :None
-    end
+    type enum_HighlightStyle = :DoubleEqual | :Html | :Bold | :None
 
     class HostMatcher
     end
 
-    class HtmlTheme
-        type value = :default | :github | :dark | :light | :unstyled
-    end
+    type enum_HtmlTheme = :default | :github | :dark | :light | :unstyled
 
-    class HtmlToMarkdownRsOutputFormat
-        type value = :Markdown | :Djot | :Plain
-    end
+    type enum_HtmlToMarkdownRsOutputFormat = :Markdown | :Djot | :Plain
 
-    class ImageKind
-        type value = :photograph | :diagram | :chart | :drawing | :text_block | :decoration | :logo | :icon | :tile_fragment | :mask | :page_raster | :unknown
-    end
+    type enum_ImageKind = :photograph | :diagram | :chart | :drawing | :text_block | :decoration | :logo | :icon | :tile_fragment | :mask | :page_raster | :unknown
 
     class ImageOutputFormat
     end
 
-    class ImageType
-        type value = :"data-uri" | :"inline-svg" | :external | :relative
-    end
+    type enum_ImageType = :"data-uri" | :"inline-svg" | :external | :relative
 
-    class InlineType
-        type value = :text | :strong | :emphasis | :highlight | :subscript | :superscript | :insert | :delete | :code | :link | :image | :span | :math | :raw_inline | :footnote_ref | :symbol
-    end
+    type enum_InlineType = :text | :strong | :emphasis | :highlight | :subscript | :superscript | :insert | :delete | :code | :link | :image | :span | :math | :raw_inline | :footnote_ref | :symbol
 
-    class JupyterCellRendering
-        type value = :source | :outputs | :both
-    end
+    type enum_JupyterCellRendering = :source | :outputs | :both
 
-    class KeywordAlgorithm
-        type value = :yake | :rake
-    end
+    type enum_KeywordAlgorithm = :yake | :rake
 
     class LateInteractionModelType
     end
 
-    class LayoutClass
-        type value = :caption | :chart | :footnote | :formula | :list_item | :page_footer | :page_header | :picture | :section_header | :table | :text | :title | :document_index | :code | :checkbox_selected | :checkbox_unselected | :form | :key_value_region
-    end
+    type enum_LayoutClass = :caption | :chart | :footnote | :formula | :list_item | :page_footer | :page_header | :picture | :section_header | :table | :text | :title | :document_index | :code | :checkbox_selected | :checkbox_unselected | :form | :key_value_region
 
-    class LayoutStrategy
-        type value = :always | :auto
-    end
+    type enum_LayoutStrategy = :always | :auto
 
-    class LinkStyle
-        type value = :Inline | :Reference
-    end
+    type enum_LinkStyle = :Inline | :Reference
 
-    class LinkType
-        type value = :anchor | :internal | :external | :email | :phone | :other
-    end
+    type enum_LinkType = :anchor | :internal | :external | :email | :phone | :other
 
-    class ListIndentType
-        type value = :Spaces | :Tabs
-    end
+    type enum_ListIndentType = :Spaces | :Tabs
 
-    class ListType
-        type value = :Bullet | :Numbered | :Lettered | :Indented
-    end
+    type enum_ListType = :Bullet | :Numbered | :Lettered | :Indented
 
-    class MergeMode
-        type value = :object_merge | :array_concat | :object_first
-    end
+    type enum_MergeMode = :object_merge | :array_concat | :object_first
 
-    class MimeDetectionPolicy
-        type value = :prefer_content | :trust_extension | :content_only
-    end
+    type enum_MimeDetectionPolicy = :prefer_content | :trust_extension | :content_only
 
-    class NerBackendKind
-        type value = :onnx | :llm
-    end
+    type enum_NerBackendKind = :onnx | :llm
 
-    class NewlineStyle
-        type value = :Spaces | :Backslash
-    end
+    type enum_NewlineStyle = :Spaces | :Backslash
 
     class NoChunkingReason
     end
@@ -2949,16 +2869,12 @@ def initialize: (?min: Integer, ?max: Integer, years: Array[Integer]) -> void
     class NodeContent
     end
 
-    class OcrBackendType
-        type value = :Tesseract | :PaddleOCR | :Candle | :Custom
-    end
+    type enum_OcrBackendType = :Tesseract | :PaddleOCR | :Candle | :Custom
 
     class OcrBoundingGeometry
     end
 
-    class OcrElementLevel
-        type value = :word | :line | :block | :page
-    end
+    type enum_OcrElementLevel = :word | :line | :block | :page
 
     class OcrStrategy
     end
@@ -2966,148 +2882,84 @@ def initialize: (?min: Integer, ?max: Integer, years: Array[Integer]) -> void
     class OutputFormat
     end
 
-    class PSMMode
-        type value = :OsdOnly | :AutoOsd | :AutoOnly | :Auto | :SingleColumn | :SingleBlockVertical | :SingleBlock | :SingleLine | :SingleWord | :CircleWord | :SingleChar
-    end
+    type enum_PSMMode = :OsdOnly | :AutoOsd | :AutoOnly | :Auto | :SingleColumn | :SingleBlockVertical | :SingleBlock | :SingleLine | :SingleWord | :CircleWord | :SingleChar
 
-    class PaddleInferenceBackend
-        type value = :ort | :tract
-    end
+    type enum_PaddleInferenceBackend = :ort | :tract
 
-    class PaddleLanguage
-        type value = :English | :Chinese | :Japanese | :Korean | :German | :French | :Latin | :Cyrillic | :TraditionalChinese | :Thai | :Greek | :EastSlavic | :Arabic | :Devanagari | :Tamil | :Telugu
-    end
+    type enum_PaddleLanguage = :English | :Chinese | :Japanese | :Korean | :German | :French | :Latin | :Cyrillic | :TraditionalChinese | :Thai | :Greek | :EastSlavic | :Arabic | :Devanagari | :Tamil | :Telugu
 
-    class PaddleOcrVlTaskKind
-        type value = :ocr | :table | :formula | :chart
-    end
+    type enum_PaddleOcrVlTaskKind = :ocr | :table | :formula | :chart
 
-    class PageOrientationHandling
-        type value = :SelfCorrecting | :RecognisesRotatedText | :RequiresUpright
-    end
+    type enum_PageOrientationHandling = :SelfCorrecting | :RecognisesRotatedText | :RequiresUpright
 
-    class PageUnitType
-        type value = :page | :slide | :sheet
-    end
+    type enum_PageUnitType = :page | :slide | :sheet
 
-    class PdfAnnotationType
-        type value = :text | :highlight | :link | :stamp | :underline | :strike_out | :squiggly | :ink | :square | :circle | :polygon | :poly_line | :line | :caret | :file_attachment | :sound | :movie | :other
-    end
+    type enum_PdfAnnotationType = :text | :highlight | :link | :stamp | :underline | :strike_out | :squiggly | :ink | :square | :circle | :polygon | :poly_line | :line | :caret | :file_attachment | :sound | :movie | :other
 
-    class PdfBackend
-        type value = :native | :pdfium
-    end
+    type enum_PdfBackend = :native | :pdfium
 
     class PiiCategory
     end
 
-    class PreprocessingPreset
-        type value = :Minimal | :Standard | :Aggressive
-    end
+    type enum_PreprocessingPreset = :Minimal | :Standard | :Aggressive
 
-    class PresetCategory
-        type value = :finance | :identity | :legal | :logistics | :medical | :hr | :other
-    end
+    type enum_PresetCategory = :finance | :identity | :legal | :logistics | :medical | :hr | :other
 
-    class ProbeStatus
-        type value = :pass | :warn | :fail | :skip
-    end
+    type enum_ProbeStatus = :pass | :warn | :fail | :skip
 
-    class ProcessingStage
-        type value = :Early | :Middle | :Late
-    end
+    type enum_ProcessingStage = :Early | :Middle | :Late
 
-    class RedactionStrategy
-        type value = :mask | :hash | :token_replace | :drop
-    end
+    type enum_RedactionStrategy = :mask | :hash | :token_replace | :drop
 
-    class ReductionLevel
-        type value = :Off | :Light | :Moderate | :Aggressive | :Maximum
-    end
+    type enum_ReductionLevel = :Off | :Light | :Moderate | :Aggressive | :Maximum
 
-    class RegionKind
-        type value = :Figure | :DenseTable | :ComplexLayout | :Caption
-    end
+    type enum_RegionKind = :Figure | :DenseTable | :ComplexLayout | :Caption
 
-    class RelationshipKind
-        type value = :footnote_reference | :citation_reference | :internal_link | :caption | :label | :toc_entry | :cross_reference
-    end
+    type enum_RelationshipKind = :footnote_reference | :citation_reference | :internal_link | :caption | :label | :toc_entry | :cross_reference
 
-    class RerankerHead
-        type value = :cross_encoder | :qwen3_generative
-    end
+    type enum_RerankerHead = :cross_encoder | :qwen3_generative
 
     class RerankerModelType
     end
 
-    class ResultFormat
-        type value = :unified | :element_based
-    end
+    type enum_ResultFormat = :unified | :element_based
 
     class RevisionAnchor
     end
 
-    class RevisionKind
-        type value = :insertion | :deletion | :format_change | :comment
-    end
+    type enum_RevisionKind = :insertion | :deletion | :format_change | :comment
 
-    class SchemaCompliance
-        type value = :all_valid | :partial_valid | :all_invalid
-    end
+    type enum_SchemaCompliance = :all_valid | :partial_valid | :all_invalid
 
     class SparseEmbeddingModelType
     end
 
-    class StructuredDataType
-        type value = :"json-ld" | :microdata | :rdfa
-    end
+    type enum_StructuredDataType = :"json-ld" | :microdata | :rdfa
 
-    class SummaryStrategy
-        type value = :extractive | :abstractive
-    end
+    type enum_SummaryStrategy = :extractive | :abstractive
 
-    class TableChunkingMode
-        type value = :split | :repeat_header
-    end
+    type enum_TableChunkingMode = :split | :repeat_header
 
-    class TableModel
-        type value = :tatr | :slanet_wired | :slanet_wireless | :slanet_plus | :slanet_auto | :disabled
-    end
+    type enum_TableModel = :tatr | :slanet_wired | :slanet_wireless | :slanet_plus | :slanet_auto | :disabled
 
-    class TableOverlapPreference
-        type value = :content | :native | :layout
-    end
+    type enum_TableOverlapPreference = :content | :native | :layout
 
-    class TextDirection
-        type value = :ltr | :rtl | :auto
-    end
+    type enum_TextDirection = :ltr | :rtl | :auto
 
-    class TierStrategy
-        type value = :auto | :tier2 | :tier1
-    end
+    type enum_TierStrategy = :auto | :tier2 | :tier1
 
-    class UriKind
-        type value = :hyperlink | :image | :anchor | :citation | :reference | :email
-    end
+    type enum_UriKind = :hyperlink | :image | :anchor | :citation | :reference | :email
 
-    class UrlEscapeStyle
-        type value = :Angle | :Percent
-    end
+    type enum_UrlEscapeStyle = :Angle | :Percent
 
-    class UrlExtractionMode
-        type value = :auto | :document | :crawl
-    end
+    type enum_UrlExtractionMode = :auto | :document | :crawl
 
     class VlmFallbackPolicy
     end
 
-    class WhisperModel
-        type value = :tiny | :base | :small | :medium | :large_v3
-    end
+    type enum_WhisperModel = :tiny | :base | :small | :medium | :large_v3
 
-    class WhitespaceMode
-        type value = :Normalized | :Strict
-    end
+    type enum_WhitespaceMode = :Normalized | :Strict
 
 def self.classify_chunks_owned: (ExtractedDocument result, ChunkClassificationConfig config) -> ExtractedDocument
 
@@ -3180,7 +3032,7 @@ def self.verify_excerpt: (String excerpt, String source_text) -> bool
         # Ruby constructors are private and never count as the lifecycle hook).
     def process_image: (String image_bytes, OcrConfig config) -> ExtractedDocument
     def supports_language: (String lang) -> bool
-    def backend_type: () -> OcrBackendType
+    def backend_type: () -> enum_OcrBackendType
     end
 
     interface _PostProcessor
@@ -3189,7 +3041,7 @@ def self.verify_excerpt: (String excerpt, String source_text) -> bool
         # The lifecycle hooks `initialize`/`shutdown` are likewise optional (note:
         # Ruby constructors are private and never count as the lifecycle hook).
     def process: (ExtractedDocument result, ExtractionConfig config) -> void
-    def processing_stage: () -> ProcessingStage
+    def processing_stage: () -> enum_ProcessingStage
     end
 
     interface _Validator