RubyGems - kreuzberg - Versions diffs - 4.3.5-aarch64-linux - Mend

kreuzberg 4.3.5-aarch64-linux

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (82) hide show

checksums.yaml +7 -0
data/.gitignore +14 -0
data/.rspec +3 -0
data/.rubocop.yaml +1 -0
data/.rubocop.yml +543 -0
data/Gemfile +8 -0
data/Gemfile.lock +260 -0
data/README.md +399 -0
data/Rakefile +34 -0
data/Steepfile +51 -0
data/examples/async_patterns.rb +283 -0
data/extconf.rb +60 -0
data/kreuzberg.gemspec +253 -0
data/lib/kreuzberg/api_proxy.rb +125 -0
data/lib/kreuzberg/cache_api.rb +67 -0
data/lib/kreuzberg/cli.rb +57 -0
data/lib/kreuzberg/cli_proxy.rb +118 -0
data/lib/kreuzberg/config.rb +1241 -0
data/lib/kreuzberg/djot_content.rb +225 -0
data/lib/kreuzberg/document_structure.rb +204 -0
data/lib/kreuzberg/error_context.rb +136 -0
data/lib/kreuzberg/errors.rb +116 -0
data/lib/kreuzberg/extraction_api.rb +329 -0
data/lib/kreuzberg/mcp_proxy.rb +176 -0
data/lib/kreuzberg/ocr_backend_protocol.rb +40 -0
data/lib/kreuzberg/post_processor_protocol.rb +15 -0
data/lib/kreuzberg/result.rb +712 -0
data/lib/kreuzberg/setup_lib_path.rb +99 -0
data/lib/kreuzberg/types.rb +414 -0
data/lib/kreuzberg/validator_protocol.rb +16 -0
data/lib/kreuzberg/version.rb +5 -0
data/lib/kreuzberg.rb +102 -0
data/lib/kreuzberg_rb.so +0 -0
data/lib/libpdfium.so +0 -0
data/sig/kreuzberg/internal.rbs +184 -0
data/sig/kreuzberg.rbs +1337 -0
data/spec/binding/async_operations_spec.rb +473 -0
data/spec/binding/batch_operations_spec.rb +677 -0
data/spec/binding/batch_spec.rb +360 -0
data/spec/binding/cache_spec.rb +227 -0
data/spec/binding/cli_proxy_spec.rb +85 -0
data/spec/binding/cli_spec.rb +55 -0
data/spec/binding/config_result_spec.rb +377 -0
data/spec/binding/config_spec.rb +419 -0
data/spec/binding/config_validation_spec.rb +377 -0
data/spec/binding/embeddings_spec.rb +816 -0
data/spec/binding/error_handling_spec.rb +399 -0
data/spec/binding/error_recovery_spec.rb +488 -0
data/spec/binding/errors_spec.rb +66 -0
data/spec/binding/font_config_spec.rb +220 -0
data/spec/binding/images_spec.rb +732 -0
data/spec/binding/keywords_extraction_spec.rb +600 -0
data/spec/binding/metadata_types_spec.rb +1253 -0
data/spec/binding/pages_extraction_spec.rb +550 -0
data/spec/binding/plugins/ocr_backend_spec.rb +307 -0
data/spec/binding/plugins/postprocessor_spec.rb +269 -0
data/spec/binding/plugins/validator_spec.rb +273 -0
data/spec/binding/tables_spec.rb +650 -0
data/spec/fixtures/config.toml +38 -0
data/spec/fixtures/config.yaml +41 -0
data/spec/fixtures/invalid_config.toml +3 -0
data/spec/serialization_spec.rb +134 -0
data/spec/smoke/package_spec.rb +177 -0
data/spec/spec_helper.rb +40 -0
data/spec/unit/config/chunking_config_spec.rb +213 -0
data/spec/unit/config/embedding_config_spec.rb +343 -0
data/spec/unit/config/extraction_config_spec.rb +434 -0
data/spec/unit/config/font_config_spec.rb +285 -0
data/spec/unit/config/hierarchy_config_spec.rb +314 -0
data/spec/unit/config/image_extraction_config_spec.rb +209 -0
data/spec/unit/config/image_preprocessing_config_spec.rb +230 -0
data/spec/unit/config/keyword_config_spec.rb +229 -0
data/spec/unit/config/language_detection_config_spec.rb +258 -0
data/spec/unit/config/ocr_config_spec.rb +171 -0
data/spec/unit/config/output_format_spec.rb +380 -0
data/spec/unit/config/page_config_spec.rb +221 -0
data/spec/unit/config/pdf_config_spec.rb +267 -0
data/spec/unit/config/postprocessor_config_spec.rb +290 -0
data/spec/unit/config/tesseract_config_spec.rb +181 -0
data/spec/unit/config/token_reduction_config_spec.rb +251 -0
data/test/metadata_types_test.rb +959 -0
metadata +292 -0

checksums.yaml ADDED Viewed

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: 4e7a8855affaf95c9e76424a8389a5c8b398022ee7fbfe0cd5f81acd42b4441f
+  data.tar.gz: 827920bb1c20a6885dd27c6f96a2144771309b97cde2c31b7c907eb279132584
+SHA512:
+  metadata.gz: 6047eb8aa9fd0937981cd300410fa8fd64b4d78938af56c5c66adcb4ef550662d4a407b1b6f5921fff9754cf9af648efb19b623d11110241c6de0d7d266a7890
+  data.tar.gz: 9f98e886089fe5994359f5e46185dbb90cb2fc2b4469a3fd0c9b3722fa20fdca75348efcef4bdcdf4c8e9f1d0072728470a84d6af795f68468afbb70b5f85865

data/.gitignore ADDED Viewed

@@ -0,0 +1,14 @@
+tmp/
+lib/*.bundle
+lib/*.dylib
+lib/*.so
+lib/*.dll
+lib/*.dylib
+lib/*.so
+lib/*.dll
+# Vendor directory for local development (symlink to ../../crates/kreuzberg)
+# In CI, this is replaced by the actual vendored crate files
+# Exception: vendor/rb-sys is patched and committed
+!vendor/rb-sys/
+vendor/

data/.rspec ADDED Viewed

@@ -0,0 +1,3 @@
+--require spec_helper
+--format documentation
+--color

data/.rubocop.yaml ADDED Viewed

	@@ -0,0 +1 @@
1	+ inherit_from: .rubocop.yml

data/.rubocop.yml ADDED Viewed

@@ -0,0 +1,543 @@
+plugins:
+  - rubocop-performance
+  - rubocop-rspec
+AllCops:
+  TargetRubyVersion: 3.2
+  NewCops: enable
+  SuggestExtensions: false
+  Exclude:
+    - 'vendor/**/*'
+    - 'tmp/**/*'
+    - 'lib/**/*.bundle'
+    - 'ext/**/*'
+Style/StringLiterals:
+  Enabled: true
+  EnforcedStyle: single_quotes
+Style/StringLiteralsInInterpolation:
+  Enabled: true
+  EnforcedStyle: single_quotes
+Style/FrozenStringLiteralComment:
+  Enabled: true
+  EnforcedStyle: always
+Style/Documentation:
+  Enabled: false
+Layout/LineLength:
+  Max: 120
+  AllowedPatterns:
+    - '\A\s*#'
+  Exclude:
+    - 'spec/**/*'
+Metrics/BlockLength:
+  Enabled: true
+  Max: 350
+  CountComments: false
+  CountAsOne: []
+Metrics/MethodLength:
+  Max: 18
+  Exclude:
+    - 'spec/**/*'
+    - 'examples/**/*'
+Metrics/AbcSize:
+  Max: 20
+  Exclude:
+    - 'spec/**/*'
+    - 'examples/**/*'
+    - 'lib/kreuzberg/config.rb'
+Naming/FileName:
+  Enabled: true
+  Exclude:
+    - 'Gemfile'
+    - 'Rakefile'
+RSpec/SpecFilePathFormat:
+  Enabled: false
+RSpec/SpecFilePathSuffix:
+  Enabled: false
+RSpec/PendingWithoutReason:
+  Enabled: false
+RSpec/LeakyConstantDeclaration:
+  Enabled: false
+RSpec/UnspecifiedException:
+  Enabled: false
+Lint/ConstantDefinitionInBlock:
+  Enabled: false
+  Exclude:
+    - 'spec/**/*'
+RSpec/InstanceVariable:
+  Enabled: false
+Gemspec/DevelopmentDependencies:
+  Enabled: false
+Metrics/ParameterLists:
+  Max: 14
+  Exclude:
+    - 'lib/kreuzberg/config.rb'
+Metrics/CyclomaticComplexity:
+  Max: 7
+  Exclude:
+    - 'lib/kreuzberg/config.rb'
+Metrics/PerceivedComplexity:
+  Max: 12
+  Exclude:
+    - 'lib/kreuzberg/config.rb'
+Metrics/ClassLength:
+  Exclude:
+    - 'lib/kreuzberg/config.rb'
+RSpec/RepeatedExampleGroupBody:
+  Enabled: false
+RSpec/DescribeClass:
+  Enabled: false
+RSpec/PredicateMatcher:
+  Enabled: false
+Gemspec/AddRuntimeDependency:
+  Enabled: true
+Gemspec/AttributeAssignment:
+  Enabled: true
+Gemspec/DeprecatedAttributeAssignment:
+  Enabled: true
+Gemspec/RequireMFA:
+  Enabled: false
+Layout/EmptyLinesAfterModuleInclusion:
+  Enabled: true
+Layout/LineContinuationLeadingSpace:
+  Enabled: true
+Layout/LineContinuationSpacing:
+  Enabled: true
+Layout/LineEndStringConcatenationIndentation:
+  Enabled: true
+Layout/SpaceBeforeBrackets:
+  Enabled: true
+Lint/AmbiguousAssignment:
+  Enabled: true
+Lint/AmbiguousOperatorPrecedence:
+  Enabled: true
+Lint/AmbiguousRange:
+  Enabled: true
+Lint/ArrayLiteralInRegexp:
+  Enabled: true
+Lint/ConstantOverwrittenInRescue:
+  Enabled: true
+Lint/ConstantReassignment:
+  Enabled: true
+Lint/CopDirectiveSyntax:
+  Enabled: true
+Lint/DeprecatedConstants:
+  Enabled: true
+Lint/DuplicateBranch:
+  Enabled: true
+Lint/DuplicateMagicComment:
+  Enabled: true
+Lint/DuplicateMatchPattern:
+  Enabled: true
+Lint/DuplicateRegexpCharacterClassElement:
+  Enabled: true
+Lint/DuplicateSetElement:
+  Enabled: true
+Lint/EmptyBlock:
+  Enabled: true
+Lint/EmptyClass:
+  Enabled: true
+Lint/EmptyInPattern:
+  Enabled: true
+Lint/HashNewWithKeywordArgumentsAsDefault:
+  Enabled: true
+Lint/IncompatibleIoSelectWithFiberScheduler:
+  Enabled: true
+Lint/ItWithoutArgumentsInBlock:
+  Enabled: true
+Lint/LambdaWithoutLiteralBlock:
+  Enabled: true
+Lint/LiteralAssignmentInCondition:
+  Enabled: true
+Lint/MixedCaseRange:
+  Enabled: true
+Lint/NoReturnInBeginEndBlocks:
+  Enabled: true
+Lint/NonAtomicFileOperation:
+  Enabled: true
+Lint/NumberedParameterAssignment:
+  Enabled: true
+Lint/NumericOperationWithConstantResult:
+  Enabled: true
+Lint/OrAssignmentToConstant:
+  Enabled: true
+Lint/RedundantDirGlobSort:
+  Enabled: true
+Lint/RedundantRegexpQuantifiers:
+  Enabled: true
+Lint/RedundantTypeConversion:
+  Enabled: true
+Lint/RefinementImportMethods:
+  Enabled: true
+Lint/RequireRangeParentheses:
+  Enabled: true
+Lint/RequireRelativeSelfPath:
+  Enabled: true
+Lint/SharedMutableDefault:
+  Enabled: true
+Lint/SuppressedExceptionInNumberConversion:
+  Enabled: true
+Lint/SymbolConversion:
+  Enabled: true
+Lint/ToEnumArguments:
+  Enabled: true
+Lint/TripleQuotes:
+  Enabled: true
+Lint/UnescapedBracketInRegexp:
+  Enabled: true
+Lint/UnexpectedBlockArity:
+  Enabled: true
+Lint/UnmodifiedReduceAccumulator:
+  Enabled: true
+Lint/UselessConstantScoping:
+  Enabled: true
+Lint/UselessDefaultValueArgument:
+  Enabled: true
+Lint/UselessDefined:
+  Enabled: true
+Lint/UselessNumericOperation:
+  Enabled: true
+Lint/UselessOr:
+  Enabled: true
+Lint/UselessRescue:
+  Enabled: true
+Lint/UselessRuby2Keywords:
+  Enabled: true
+Metrics/CollectionLiteralLength:
+  Enabled: true
+Naming/BlockForwarding:
+  Enabled: true
+Naming/PredicateMethod:
+  Enabled: true
+Security/CompoundHash:
+  Enabled: true
+Security/IoMethods:
+  Enabled: true
+Style/AmbiguousEndlessMethodDefinition:
+  Enabled: true
+Style/ArgumentsForwarding:
+  Enabled: true
+Style/ArrayIntersect:
+  Enabled: true
+Style/ArrayIntersectWithSingleElement:
+  Enabled: true
+Style/BitwisePredicate:
+  Enabled: true
+Style/CollectionCompact:
+  Enabled: true
+Style/CollectionQuerying:
+  Enabled: true
+Style/CombinableDefined:
+  Enabled: true
+Style/ComparableBetween:
+  Enabled: true
+Style/ComparableClamp:
+  Enabled: true
+Style/ConcatArrayLiterals:
+  Enabled: true
+Style/DataInheritance:
+  Enabled: true
+Style/DigChain:
+  Enabled: true
+Style/DirEmpty:
+  Enabled: true
+Style/DocumentDynamicEvalDefinition:
+  Enabled: true
+Style/EmptyHeredoc:
+  Enabled: true
+Style/EmptyStringInsideInterpolation:
+  Enabled: true
+Style/EndlessMethod:
+  Enabled: true
+Style/EnvHome:
+  Enabled: true
+Style/ExactRegexpMatch:
+  Enabled: true
+Style/FetchEnvVar:
+  Enabled: true
+Style/FileEmpty:
+  Enabled: true
+Style/FileNull:
+  Enabled: true
+Style/FileRead:
+  Enabled: true
+Style/FileTouch:
+  Enabled: true
+Style/FileWrite:
+  Enabled: true
+Style/HashConversion:
+  Enabled: true
+Style/HashExcept:
+  Enabled: true
+Style/HashFetchChain:
+  Enabled: true
+Style/HashSlice:
+  Enabled: true
+Style/IfWithBooleanLiteralBranches:
+  Enabled: true
+Style/InPatternThen:
+  Enabled: true
+Style/ItAssignment:
+  Enabled: true
+Style/ItBlockParameter:
+  Enabled: true
+Style/KeywordArgumentsMerging:
+  Enabled: true
+Style/MagicCommentFormat:
+  Enabled: true
+Style/MapCompactWithConditionalBlock:
+  Enabled: true
+Style/MapIntoArray:
+  Enabled: true
+Style/MapToHash:
+  Enabled: true
+Style/MapToSet:
+  Enabled: true
+Style/MinMaxComparison:
+  Enabled: true
+Style/MultilineInPatternThen:
+  Enabled: true
+Style/NegatedIfElseCondition:
+  Enabled: true
+Style/NestedFileDirname:
+  Enabled: true
+Style/NilLambda:
+  Enabled: true
+Style/NumberedParameters:
+  Enabled: true
+Style/NumberedParametersLimit:
+  Enabled: true
+Style/ObjectThen:
+  Enabled: true
+Style/OpenStructUse:
+  Enabled: true
+Style/OperatorMethodCall:
+  Enabled: true
+Style/QuotedSymbols:
+  Enabled: true
+Style/RedundantArgument:
+  Enabled: true
+Style/RedundantArrayConstructor:
+  Enabled: true
+Style/RedundantArrayFlatten:
+  Enabled: true
+Style/RedundantConstantBase:
+  Enabled: true
+Style/RedundantCurrentDirectoryInPath:
+  Enabled: true
+Style/RedundantDoubleSplatHashBraces:
+  Enabled: true
+Style/RedundantEach:
+  Enabled: true
+Style/RedundantFilterChain:
+  Enabled: true
+Style/RedundantFormat:
+  Enabled: true
+Style/RedundantHeredocDelimiterQuotes:
+  Enabled: true
+Style/RedundantInitialize:
+  Enabled: true
+Style/RedundantInterpolationUnfreeze:
+  Enabled: true
+Style/RedundantLineContinuation:
+  Enabled: true
+Style/RedundantRegexpArgument:
+  Enabled: true
+Style/RedundantRegexpConstructor:
+  Enabled: true
+Style/RedundantSelfAssignmentBranch:
+  Enabled: true
+Style/RedundantStringEscape:
+  Enabled: true
+Style/ReturnNilInPredicateMethodDefinition:
+  Enabled: true
+Style/SafeNavigationChainLength:
+  Enabled: true
+Style/SelectByRegexp:
+  Enabled: true
+Style/SendWithLiteralMethodName:
+  Enabled: true
+Style/SingleLineDoEndBlock:
+  Enabled: true
+Style/StringChars:
+  Enabled: true
+Style/SuperArguments:
+  Enabled: true
+Style/SuperWithArgsParentheses:
+  Enabled: true
+Style/SwapValues:
+  Enabled: true
+Style/YAMLFileRead:
+  Enabled: true
+RSpec/ExampleLength:
+  Max: 50
+RSpec/MultipleExpectations:
+  Max: 25
+RSpec/NestedGroups:
+  Max: 6

data/Gemfile ADDED Viewed

@@ -0,0 +1,8 @@
+# frozen_string_literal: true
+source 'https://rubygems.org'
+gemspec
+gem 'pry', '~> 0.14', group: :development
+gem 'pry-byebug', '~> 3.10', group: :development, platforms: %i[ruby mri]