kreuzberg 4.6.3-aarch64-linux → 4.7.2-aarch64-linux

data/sig/kreuzberg.rbs

@@ -91,6 +91,123 @@ module Kreuzberg
     def serialize: () -> Hash[Symbol, untyped]
   end
 
+  # CSV format metadata (T::Struct from types.rb)
+  class CsvMetadata
+    attr_reader row_count: Integer
+    attr_reader column_count: Integer
+    attr_reader delimiter: String?
+    attr_reader has_header: bool
+    attr_reader column_types: Array[String]?
+
+    def initialize: (row_count: Integer, column_count: Integer, delimiter: String?, has_header: bool, column_types: Array[String]?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Year range for citation metadata (T::Struct from types.rb)
+  class YearRange
+    attr_reader min: Integer?
+    attr_reader max: Integer?
+    attr_reader years: Array[Integer]
+
+    def initialize: (min: Integer?, max: Integer?, years: Array[Integer]) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # BibTeX format metadata (T::Struct from types.rb)
+  class BibtexMetadata
+    attr_reader entry_count: Integer
+    attr_reader citation_keys: Array[String]
+    attr_reader authors: Array[String]
+    attr_reader year_range: YearRange?
+    attr_reader entry_types: Hash[String, Integer]?
+
+    def initialize: (entry_count: Integer, citation_keys: Array[String], authors: Array[String], year_range: YearRange?, entry_types: Hash[String, Integer]?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Citation format metadata (T::Struct from types.rb)
+  class CitationMetadata
+    attr_reader citation_count: Integer
+    attr_reader format: String?
+    attr_reader authors: Array[String]
+    attr_reader year_range: YearRange?
+    attr_reader dois: Array[String]
+    attr_reader keywords: Array[String]
+
+    def initialize: (citation_count: Integer, format: String?, authors: Array[String], year_range: YearRange?, dois: Array[String], keywords: Array[String]) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # FictionBook format metadata (T::Struct from types.rb)
+  class FictionBookMetadata
+    attr_reader genres: Array[String]
+    attr_reader sequences: Array[String]
+    attr_reader annotation: String?
+
+    def initialize: (genres: Array[String], sequences: Array[String], annotation: String?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # DBF field information (T::Struct from types.rb)
+  class DbfFieldInfo
+    attr_reader name: String
+    attr_reader field_type: String
+
+    def initialize: (name: String, field_type: String) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # DBF format metadata (T::Struct from types.rb)
+  class DbfMetadata
+    attr_reader record_count: Integer
+    attr_reader field_count: Integer
+    attr_reader fields: Array[DbfFieldInfo]
+
+    def initialize: (record_count: Integer, field_count: Integer, fields: Array[DbfFieldInfo]) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Contributor with role (T::Struct from types.rb)
+  class ContributorRole
+    attr_reader name: String
+    attr_reader role: String?
+
+    def initialize: (name: String, role: String?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # JATS format metadata (T::Struct from types.rb)
+  class JatsMetadata
+    attr_reader copyright: String?
+    attr_reader license: String?
+    attr_reader history_dates: Hash[String, String]
+    attr_reader contributor_roles: Array[ContributorRole]
+
+    def initialize: (copyright: String?, license: String?, history_dates: Hash[String, String], contributor_roles: Array[ContributorRole]) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # EPUB format metadata (T::Struct from types.rb)
+  class EpubMetadata
+    attr_reader coverage: String?
+    attr_reader dc_format: String?
+    attr_reader relation: String?
+    attr_reader source: String?
+    attr_reader dc_type: String?
+    attr_reader cover_image: String?
+
+    def initialize: (coverage: String?, dc_format: String?, relation: String?, source: String?, dc_type: String?, cover_image: String?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # PST format metadata (T::Struct from types.rb)
+  class PstMetadata
+    attr_reader message_count: Integer
+
+    def initialize: (message_count: Integer) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
   # HTML metadata (T::Struct from types.rb)
   class HtmlMetadata     attr_reader title: String?
     attr_reader description: String?
@@ -210,6 +327,162 @@ module Kreuzberg
     def serialize: () -> Hash[Symbol, untyped]
   end
 
+  # Code span location (tree-sitter process result)
+  class CodeSpan
+    attr_reader start_byte: Integer
+    attr_reader end_byte: Integer
+    attr_reader start_line: Integer
+    attr_reader start_column: Integer
+    attr_reader end_line: Integer
+    attr_reader end_column: Integer
+
+    def initialize: (start_byte: Integer, end_byte: Integer, start_line: Integer, start_column: Integer, end_line: Integer, end_column: Integer) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # File-level code metrics (tree-sitter process result)
+  class CodeFileMetrics
+    attr_reader total_lines: Integer
+    attr_reader code_lines: Integer
+    attr_reader comment_lines: Integer
+    attr_reader blank_lines: Integer
+    attr_reader total_bytes: Integer
+    attr_reader node_count: Integer
+    attr_reader error_count: Integer
+    attr_reader max_depth: Integer
+
+    def initialize: (total_lines: Integer, code_lines: Integer, comment_lines: Integer, blank_lines: Integer, total_bytes: Integer, node_count: Integer, error_count: Integer, max_depth: Integer) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Code structure item (tree-sitter process result)
+  class CodeStructureItem
+    attr_reader kind: String
+    attr_reader name: String?
+    attr_reader visibility: String?
+    attr_reader span: CodeSpan
+    attr_reader children: Array[CodeStructureItem]
+    attr_reader decorators: Array[String]
+    attr_reader doc_comment: String?
+    attr_reader signature: String?
+    attr_reader body_span: CodeSpan?
+
+    def initialize: (kind: String, name: String?, visibility: String?, span: CodeSpan, children: Array[CodeStructureItem], decorators: Array[String], doc_comment: String?, signature: String?, body_span: CodeSpan?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Import information (tree-sitter process result)
+  class CodeImportInfo
+    attr_reader source: String
+    attr_reader items: Array[String]
+    attr_reader alias: String?
+    attr_reader is_wildcard: bool
+    attr_reader span: CodeSpan
+
+    def initialize: (source: String, items: Array[String], alias: String?, is_wildcard: bool, span: CodeSpan) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Export information (tree-sitter process result)
+  class CodeExportInfo
+    attr_reader name: String
+    attr_reader kind: String
+    attr_reader span: CodeSpan
+
+    def initialize: (name: String, kind: String, span: CodeSpan) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Symbol information (tree-sitter process result)
+  class CodeSymbolInfo
+    attr_reader name: String
+    attr_reader kind: String
+    attr_reader type_annotation: String?
+    attr_reader span: CodeSpan
+
+    def initialize: (name: String, kind: String, type_annotation: String?, span: CodeSpan) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Comment information (tree-sitter process result)
+  class CodeCommentInfo
+    attr_reader text: String
+    attr_reader kind: String
+    attr_reader span: CodeSpan
+
+    def initialize: (text: String, kind: String, span: CodeSpan) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Documentation section (tree-sitter process result)
+  class CodeDocSection
+    attr_reader kind: String
+    attr_reader name: String?
+    attr_reader content: String
+
+    def initialize: (kind: String, name: String?, content: String) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Docstring information (tree-sitter process result)
+  class CodeDocstringInfo
+    attr_reader text: String
+    attr_reader format: String
+    attr_reader associated_item: String?
+    attr_reader span: CodeSpan
+    attr_reader sections: Array[CodeDocSection]
+
+    def initialize: (text: String, format: String, associated_item: String?, span: CodeSpan, sections: Array[CodeDocSection]) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Diagnostic from code analysis (tree-sitter process result)
+  class CodeDiagnostic
+    attr_reader message: String
+    attr_reader severity: String
+    attr_reader span: CodeSpan
+
+    def initialize: (message: String, severity: String, span: CodeSpan) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Chunk context (tree-sitter process result)
+  class CodeChunkContext
+    attr_reader parent_name: String?
+    attr_reader parent_kind: String?
+
+    def initialize: (parent_name: String?, parent_kind: String?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Code chunk (tree-sitter process result)
+  class CodeChunk
+    attr_reader content: String
+    attr_reader language: String
+    attr_reader span: CodeSpan
+    attr_reader context: CodeChunkContext?
+
+    def initialize: (content: String, language: String, span: CodeSpan, context: CodeChunkContext?) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
+  # Full code processing result (tree-sitter process result)
+  class CodeProcessResult
+    attr_reader language: String
+    attr_reader metrics: CodeFileMetrics
+    attr_reader structure: Array[CodeStructureItem]
+    attr_reader imports: Array[CodeImportInfo]
+    attr_reader exports: Array[CodeExportInfo]
+    attr_reader comments: Array[CodeCommentInfo]
+    attr_reader docstrings: Array[CodeDocstringInfo]
+    attr_reader symbols: Array[CodeSymbolInfo]
+    attr_reader diagnostics: Array[CodeDiagnostic]
+    attr_reader chunks: Array[CodeChunk]
+
+    def initialize: (language: String, metrics: CodeFileMetrics, structure: Array[CodeStructureItem], imports: Array[CodeImportInfo], exports: Array[CodeExportInfo], comments: Array[CodeCommentInfo], docstrings: Array[CodeDocstringInfo], symbols: Array[CodeSymbolInfo], diagnostics: Array[CodeDiagnostic], chunks: Array[CodeChunk]) -> void
+    def to_h: () -> Hash[Symbol, untyped]
+  end
+
   # Config namespace (defined in lib/kreuzberg/config.rb)
   module Config
     class OCR
@@ -491,6 +764,32 @@ module Kreuzberg
       def to_h: () -> Hash[Symbol, untyped]
     end
 
+    class TreeSitterProcessConfig
+      attr_reader structure: bool
+      attr_reader imports: bool
+      attr_reader exports: bool
+      attr_reader comments: bool
+      attr_reader docstrings: bool
+      attr_reader symbols: bool
+      attr_reader diagnostics: bool
+      attr_reader chunk_max_size: Integer?
+      attr_reader content_mode: String?
+
+      def initialize: (?structure: bool, ?imports: bool, ?exports: bool, ?comments: bool, ?docstrings: bool, ?symbols: bool, ?diagnostics: bool, ?chunk_max_size: Integer?, ?content_mode: String?) -> void
+      def to_h: () -> Hash[Symbol, untyped]
+    end
+
+    class TreeSitterConfig
+      attr_reader enabled: bool?
+      attr_reader cache_dir: String?
+      attr_reader languages: Array[String]?
+      attr_reader groups: Array[String]?
+      attr_reader process: TreeSitterProcessConfig?
+
+      def initialize: (?enabled: bool?, ?cache_dir: String?, ?languages: Array[String]?, ?groups: Array[String]?, ?process: (TreeSitterProcessConfig | Hash[Symbol, untyped])?) -> void
+      def to_h: () -> Hash[Symbol, untyped]
+    end
+
     class Extraction
       attr_reader use_cache: bool
       attr_reader enable_quality_processing: bool
@@ -514,6 +813,7 @@ module Kreuzberg
       attr_reader concurrency: Concurrency?
       attr_reader acceleration: Acceleration?
       attr_reader email: Email?
+      attr_reader tree_sitter: TreeSitterConfig?
       attr_reader max_concurrent_extractions: Integer?
       attr_reader max_archive_depth: Integer
       attr_reader output_format: String?
@@ -544,6 +844,7 @@ module Kreuzberg
         ?concurrency: (Concurrency | Hash[Symbol, untyped])?,
         ?acceleration: (Acceleration | Hash[Symbol, untyped])?,
         ?email: (Email | Hash[Symbol, untyped])?,
+        ?tree_sitter: (TreeSitterConfig | Hash[Symbol, untyped])?,
         ?max_concurrent_extractions: Integer?,
         ?max_archive_depth: Integer,
         ?output_format: String?,
@@ -845,6 +1146,7 @@ module Kreuzberg
     # Text chunk (Struct from result.rb)
     class Chunk
       attr_reader content: String
+      attr_reader chunk_type: String
       attr_reader byte_start: Integer
       attr_reader byte_end: Integer
       attr_reader token_count: Integer?
@@ -857,6 +1159,7 @@ module Kreuzberg
 
       def initialize: (
         content: String,
+        chunk_type: String,
         byte_start: Integer,
         byte_end: Integer,
         token_count: Integer?,

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: kreuzberg
 version: !ruby/object:Gem::Version
-  version: 4.6.3
+  version: 4.7.2
 platform: aarch64-linux
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-03-27 00:00:00.000000000 Z
+date: 2026-04-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -205,7 +205,6 @@ files:
 - spec/binding/cache_spec.rb
 - spec/binding/cli_proxy_spec.rb
 - spec/binding/cli_spec.rb
-- spec/binding/config_result_spec.rb
 - spec/binding/config_spec.rb
 - spec/binding/config_validation_spec.rb
 - spec/binding/embeddings_spec.rb
@@ -215,32 +214,13 @@ files:
 - spec/binding/font_config_spec.rb
 - spec/binding/images_spec.rb
 - spec/binding/keywords_extraction_spec.rb
-- spec/binding/metadata_types_spec.rb
 - spec/binding/pages_extraction_spec.rb
 - spec/binding/plugins/ocr_backend_spec.rb
 - spec/binding/plugins/postprocessor_spec.rb
 - spec/binding/plugins/validator_spec.rb
 - spec/binding/render_spec.rb
 - spec/binding/tables_spec.rb
-- spec/serialization_spec.rb
-- spec/smoke/package_spec.rb
 - spec/spec_helper.rb
-- spec/unit/config/chunking_config_spec.rb
-- spec/unit/config/embedding_config_spec.rb
-- spec/unit/config/extraction_config_spec.rb
-- spec/unit/config/font_config_spec.rb
-- spec/unit/config/hierarchy_config_spec.rb
-- spec/unit/config/image_extraction_config_spec.rb
-- spec/unit/config/image_preprocessing_config_spec.rb
-- spec/unit/config/keyword_config_spec.rb
-- spec/unit/config/language_detection_config_spec.rb
-- spec/unit/config/ocr_config_spec.rb
-- spec/unit/config/output_format_spec.rb
-- spec/unit/config/page_config_spec.rb
-- spec/unit/config/pdf_config_spec.rb
-- spec/unit/config/postprocessor_config_spec.rb
-- spec/unit/config/tesseract_config_spec.rb
-- spec/unit/config/token_reduction_config_spec.rb
 homepage: https://github.com/kreuzberg-dev/kreuzberg
 licenses:
 - MIT