xberg 1.2.9 → 1.3.2

data/lib/xberg/version.rb

@@ -1,10 +1,10 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:af0361b6e9fb7d4844fed422cc5ad28ac7fdea52beb041fa1a3ab043aec2d450
+# alef:hash:c7a2e4c0b7a9168d08ef7f09b8b967926ae2bdff61094ac93d54a86a9813924d
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
 
 module Xberg
   ## The version string for this package.
-  VERSION = "1.2.9"
+  VERSION = "1.3.2"
 end

data/sig/types.rbs

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:060fee672abbcdc0cf88ae2c4c85dc76420919f28c17b4cd7ccbd449f6d319b1
+# alef:hash:714d17cb31a0dbcf05ffbfd2931268de7747a7876060ed07e56d20da729e9edf
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
@@ -96,6 +96,8 @@ def initialize: (?entry_count: Integer, ?citation_keys: Array[String], ?authors:
         attr_reader backend: enum_BrowserBackend
         attr_reader endpoint: String?
         attr_reader timeout: Integer
+        attr_reader overall_timeout: Integer
+        attr_reader shutdown_timeout: Integer
         attr_reader wait: enum_BrowserWait
         attr_reader wait_selector: String?
         attr_reader extra_wait: Integer?
@@ -106,7 +108,7 @@ def initialize: (?entry_count: Integer, ?citation_keys: Array[String], ?authors:
         attr_reader capture_network_events: bool
         attr_reader session_affinity: bool
 
-def initialize: (?mode: 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
+def initialize: (?mode: enum_BrowserMode, ?backend: enum_BrowserBackend, ?endpoint: String, ?timeout: Integer, ?overall_timeout: Integer, ?shutdown_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
@@ -123,8 +125,9 @@ def initialize: (?mode: enum_BrowserMode, ?backend: enum_BrowserBackend, ?endpoi
         attr_reader llm: LlmConfig
         attr_reader prompt: String?
         attr_reader min_image_area: Integer
+        attr_reader alt_text: enum_CaptionAltTextMode
 
-    def initialize: (llm: LlmConfig, ?prompt: String, min_image_area: Integer) -> void
+    def initialize: (llm: LlmConfig, ?prompt: String, min_image_area: Integer, alt_text: enum_CaptionAltTextMode) -> void
     end
 
     class CellChange
@@ -297,8 +300,9 @@ def initialize: (?chunks: Array[CodeChunkInfo], ?data: CodeDataNode) -> void
         attr_reader wrap: bool
         attr_reader wrap_width: Integer
         attr_reader include_document_structure: bool
+        attr_reader extract_metadata: bool
 
-def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_navigation: bool, ?remove_forms: bool, ?strip_tags: Array[String], ?preserve_tags: Array[String], ?exclude_selectors: Array[String], ?skip_images: bool, ?max_depth: Integer, ?wrap: bool, ?wrap_width: Integer, ?include_document_structure: bool) -> void
+def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_navigation: bool, ?remove_forms: bool, ?strip_tags: Array[String], ?preserve_tags: Array[String], ?exclude_selectors: Array[String], ?skip_images: bool, ?max_depth: Integer, ?wrap: bool, ?wrap_width: Integer, ?include_document_structure: bool, ?extract_metadata: bool) -> void
     end
 
     class ContentFilterConfig
@@ -362,8 +366,9 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
         attr_reader max_depth: Integer?
     attr_reader exclude_selectors: Array[String]
         attr_reader tier_strategy: enum_TierStrategy
+        attr_reader base_url: String?
 
-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
+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, ?base_url: String) -> void
     end
 
     class CoreProperties
@@ -402,12 +407,19 @@ def initialize: (?heading_style: enum_HeadingStyle, ?list_indent_type: enum_List
         attr_reader allow_subdomains: bool
     attr_reader include_paths: Array[String]
     attr_reader exclude_paths: Array[String]
+        attr_reader path_patterns_match_query: bool
+        attr_reader dedup_include_query: bool
+        attr_reader strip_tracking_params: bool
+    attr_reader tracking_params: Array[String]
     attr_reader custom_headers: Hash[String, String]
         attr_reader request_timeout: Integer
         attr_reader rate_limit_ms: Integer?
         attr_reader max_redirects: Integer
         attr_reader retry_count: Integer
     attr_reader retry_codes: Array[Integer]
+        attr_reader retry_initial_delay_ms: Integer
+        attr_reader retry_max_delay_ms: Integer
+        attr_reader rate_limit_jitter_ratio: Float
         attr_reader cookies_enabled: bool
         attr_reader auth: AuthConfig?
         attr_reader max_body_size: Integer?
@@ -435,7 +447,7 @@ def initialize: (?heading_style: enum_HeadingStyle, ?list_indent_type: enum_List
         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: 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
+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], ?path_patterns_match_query: bool, ?dedup_include_query: bool, ?strip_tracking_params: bool, ?tracking_params: Array[String], ?custom_headers: Hash[String, String], ?request_timeout: Integer, ?rate_limit_ms: Integer, ?max_redirects: Integer, ?retry_count: Integer, ?retry_codes: Array[Integer], ?retry_initial_delay_ms: Integer, ?retry_max_delay_ms: Integer, ?rate_limit_jitter_ratio: Float, ?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
@@ -834,6 +846,16 @@ def initialize: (name: String, markdown: String, row_count: Integer, col_count:
 def initialize: (sheets: Array[ExcelSheet], metadata: Hash[String, String], ?revisions: Array[DocumentRevision]) -> void
     end
 
+    class ExternalRedactionFinding
+        attr_reader label: String
+        attr_reader text: String?
+        attr_reader start: Integer?
+        attr_reader end: Integer?
+        attr_reader score: Float?
+
+    def initialize: (?label: String, ?text: String, ?start: Integer, ?end: Integer, ?score: Float) -> void
+    end
+
     class ExtractInput
         attr_reader kind: enum_ExtractInputKind
         attr_reader bytes: String?
@@ -1319,8 +1341,9 @@ def initialize: (src: String, ?alt: String, ?title: String, ?dimensions: ImageDi
         attr_reader contrast_enhance: bool
         attr_reader binarization_method: String
         attr_reader invert_colors: bool
+        attr_reader normalize_shaded_rows: bool
 
-    def initialize: (?target_dpi: Integer, ?auto_rotate: bool, ?deskew: bool, ?denoise: bool, ?contrast_enhance: bool, ?binarization_method: String, ?invert_colors: bool) -> void
+    def initialize: (?target_dpi: Integer, ?auto_rotate: bool, ?deskew: bool, ?denoise: bool, ?contrast_enhance: bool, ?binarization_method: String, ?invert_colors: bool, ?normalize_shaded_rows: bool) -> void
     end
 
     class ImagePreprocessingMetadata
@@ -1680,6 +1703,7 @@ def initialize: (name: String, supported_languages: Array[String]) -> void
         attr_reader tesseract_config: TesseractConfig?
         attr_reader output_format: OutputFormat?
         attr_reader paddle_ocr_config: String?
+        attr_reader paddle_ocr_settings: PaddleOcrConfig?
         attr_reader backend_options: String?
         attr_reader element_config: OcrElementConfig?
         attr_reader quality_thresholds: OcrQualityThresholds?
@@ -1692,8 +1716,9 @@ def initialize: (name: String, supported_languages: Array[String]) -> void
         attr_reader security_limits: SecurityLimits?
     attr_reader tessdata_bytes: Hash[String, String]?
         attr_reader tessdata_path: String?
+        attr_reader numeric_repair: bool
 
-def initialize: (?enabled: bool, ?backend: String, ?language: Array[String], ?tesseract_config: TesseractConfig, ?output_format: OutputFormat, ?paddle_ocr_config: String, ?backend_options: String, ?element_config: OcrElementConfig, ?quality_thresholds: OcrQualityThresholds, ?pipeline: OcrPipelineConfig, ?auto_rotate: bool, ?vlm_fallback: VlmFallbackPolicy, ?vlm_config: LlmConfig, ?vlm_prompt: String, ?acceleration: AccelerationConfig, ?security_limits: SecurityLimits, ?tessdata_bytes: Hash[String, String], ?tessdata_path: String) -> void
+def initialize: (?enabled: bool, ?backend: String, ?language: Array[String], ?tesseract_config: TesseractConfig, ?output_format: OutputFormat, ?paddle_ocr_config: String, ?paddle_ocr_settings: PaddleOcrConfig, ?backend_options: String, ?element_config: OcrElementConfig, ?quality_thresholds: OcrQualityThresholds, ?pipeline: OcrPipelineConfig, ?auto_rotate: bool, ?vlm_fallback: VlmFallbackPolicy, ?vlm_config: LlmConfig, ?vlm_prompt: String, ?acceleration: AccelerationConfig, ?security_limits: SecurityLimits, ?tessdata_bytes: Hash[String, String], ?tessdata_path: String, ?numeric_repair: bool) -> void
     end
 
     class OcrElement
@@ -1752,10 +1777,11 @@ def initialize: (stages: Array[OcrPipelineStage], quality_thresholds: OcrQuality
     attr_reader language: Array[String]?
         attr_reader tesseract_config: TesseractConfig?
         attr_reader paddle_ocr_config: String?
+        attr_reader paddle_ocr_settings: PaddleOcrConfig?
         attr_reader vlm_config: LlmConfig?
         attr_reader backend_options: String?
 
-def initialize: (backend: String, priority: Integer, ?language: Array[String], ?tesseract_config: TesseractConfig, ?paddle_ocr_config: String, ?vlm_config: LlmConfig, ?backend_options: String) -> void
+def initialize: (backend: String, priority: Integer, ?language: Array[String], ?tesseract_config: TesseractConfig, ?paddle_ocr_config: String, ?paddle_ocr_settings: PaddleOcrConfig, ?vlm_config: LlmConfig, ?backend_options: String) -> void
     end
 
     class OcrPoint
@@ -2251,8 +2277,12 @@ def initialize: (detection_bbox: BBox, cells: Array[Array[String]], markdown: St
         attr_reader preserve_offsets: bool
     attr_reader custom_terms: Array[RedactionTerm]
     attr_reader custom_patterns: Array[RedactionPattern]
+    attr_reader findings: Array[ExternalRedactionFinding]
+        attr_reader min_score: Float?
+        attr_reader findings_path: String?
+        attr_reader findings_offset_encoding: enum_RedactionOffsetEncoding
 
-def initialize: (?categories: Array[PiiCategory], ?strategy: enum_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], ?findings: Array[ExternalRedactionFinding], ?min_score: Float, ?findings_path: String, ?findings_offset_encoding: enum_RedactionOffsetEncoding) -> void
     def validate: () -> void
     end
 
@@ -2538,9 +2568,9 @@ def initialize: (?rows: Integer, ?cols: Integer, ?cells: Array[GridCell]) -> voi
         attr_reader tessedit_char_blacklist: String
         attr_reader tessedit_use_primary_params_model: bool
         attr_reader textord_space_size_is_variable: bool
-        attr_reader thresholding_method: bool
+        attr_reader thresholding_method: Integer
 
-def initialize: (?language: Array[String], ?psm: Integer, ?output_format: String, ?oem: Integer, ?min_confidence: Float, ?preprocessing: ImagePreprocessingConfig, ?enable_table_detection: bool, ?table_min_confidence: Float, ?table_column_threshold: Integer, ?table_row_threshold_ratio: Float, ?use_cache: bool, ?classify_use_pre_adapted_templates: bool, ?language_model_ngram_on: bool, ?tessedit_dont_blkrej_good_wds: bool, ?tessedit_dont_rowrej_good_wds: bool, ?tessedit_enable_dict_correction: bool, ?tessedit_char_whitelist: String, ?tessedit_char_blacklist: String, ?tessedit_use_primary_params_model: bool, ?textord_space_size_is_variable: bool, ?thresholding_method: bool) -> void
+def initialize: (?language: Array[String], ?psm: Integer, ?output_format: String, ?oem: Integer, ?min_confidence: Float, ?preprocessing: ImagePreprocessingConfig, ?enable_table_detection: bool, ?table_min_confidence: Float, ?table_column_threshold: Integer, ?table_row_threshold_ratio: Float, ?use_cache: bool, ?classify_use_pre_adapted_templates: bool, ?language_model_ngram_on: bool, ?tessedit_dont_blkrej_good_wds: bool, ?tessedit_dont_rowrej_good_wds: bool, ?tessedit_enable_dict_correction: bool, ?tessedit_char_whitelist: String, ?tessedit_char_blacklist: String, ?tessedit_use_primary_params_model: bool, ?textord_space_size_is_variable: bool, ?thresholding_method: Integer) -> void
     end
 
     class TextAnnotation
@@ -2758,6 +2788,8 @@ def initialize: (?min: Integer, ?max: Integer, years: Array[Integer]) -> void
 
     type enum_CandleTrocrVariant = :"base-printed" | :"large-printed" | :"base-handwritten" | :"large-handwritten"
 
+    type enum_CaptionAltTextMode = :preserve | :combine | :replace
+
     class ChunkSizing
     end
 
@@ -2913,6 +2945,8 @@ def initialize: (?min: Integer, ?max: Integer, years: Array[Integer]) -> void
 
     type enum_ProcessingStage = :Early | :Middle | :Late
 
+    type enum_RedactionOffsetEncoding = :utf8_bytes | :unicode_code_points | :utf16_code_units
+
     type enum_RedactionStrategy = :mask | :hash | :token_replace | :drop
 
     type enum_ReductionLevel = :Off | :Light | :Moderate | :Aggressive | :Maximum
@@ -2989,6 +3023,8 @@ def self.extract: (ExtractInput input, ExtractionConfig config) -> ExtractionRes
 
 def self.extract_batch: (Array[ExtractInput] inputs, ExtractionConfig config) -> ExtractionResult
 
+def self.extract_with_external_redaction: (ExtractInput input, ExtractionConfig config, String findings_json, ?String offset_encoding, ?Integer max_findings) -> ExtractionResult
+
 def self.find_unmarked_claims: (String markdown) -> Array[String]
 
 def self.list_document_extractors: () -> Array[String]
@@ -2997,6 +3033,8 @@ def self.list_embedding_backends: () -> Array[String]
 
 def self.list_ocr_backend_capabilities: () -> Array[OcrBackendCapabilities]
 
+def self.list_ocr_backend_capabilities_for: (OcrConfig config) -> Array[OcrBackendCapabilities]
+
 def self.list_ocr_backends: () -> Array[String]
 
 def self.list_post_processors: () -> Array[String]
@@ -3019,6 +3057,12 @@ def self.max_sim_score: (json_value query, json_value doc) -> Float
 
 def self.ocr_backend_supports_language: (String backend, String language) -> bool
 
+def self.ocr_backend_supports_language_for: (String backend, String language, OcrConfig config) -> bool
+
+def self.pdf_page_count: (String pdf_bytes, ?String password) -> Integer
+
+def self.redact_external: (ExtractedDocument document, RedactionConfig config, String findings_json, ?String offset_encoding, ?Integer max_findings) -> ExtractedDocument
+
 def self.register_document_extractor: (_DocumentExtractor extractor) -> void
 
 def self.register_renderer: (_Renderer renderer) -> void
@@ -3031,7 +3075,7 @@ def self.verify_excerpt: (String excerpt, String source_text) -> bool
 
     interface _OcrBackend
         # Optional methods the bridge calls when the object defines them (the
-        # trait's Rust default behavior applies otherwise): process_image_file, supported_languages, supports_table_detection, supports_document_processing, emits_structured_markdown, confidence_semantics, page_orientation_handling, process_document.
+        # trait's Rust default behavior applies otherwise): process_image_file, supported_languages, supported_languages_for, supports_language_for, supports_table_detection, supports_document_processing, emits_structured_markdown, confidence_semantics, page_orientation_handling, process_document.
         # The lifecycle hooks `initialize`/`shutdown` are likewise optional (note:
         # Ruby constructors are private and never count as the lifecycle hook).
     def process_image: (String image_bytes, OcrConfig config) -> ExtractedDocument

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: xberg
 version: !ruby/object:Gem::Version
-  version: 1.2.9
+  version: 1.3.2
 platform: ruby
 authors:
 - Na'aman Hirschfeld <naaman@xberg.io>
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-09-24 00:00:00.000000000 Z
+date: 2026-10-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys