kreuzberg 4.0.8 → 4.1.0
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.
- checksums.yaml +4 -4
- data/Gemfile.lock +2 -2
- data/README.md +1 -1
- data/ext/kreuzberg_rb/native/Cargo.lock +94 -98
- data/ext/kreuzberg_rb/native/Cargo.toml +4 -2
- data/ext/kreuzberg_rb/native/src/batch.rs +139 -0
- data/ext/kreuzberg_rb/native/src/config/mod.rs +10 -0
- data/ext/kreuzberg_rb/native/src/config/types.rs +1058 -0
- data/ext/kreuzberg_rb/native/src/error_handling.rs +125 -0
- data/ext/kreuzberg_rb/native/src/extraction.rs +79 -0
- data/ext/kreuzberg_rb/native/src/gc_guarded_value.rs +35 -0
- data/ext/kreuzberg_rb/native/src/helpers.rs +176 -0
- data/ext/kreuzberg_rb/native/src/lib.rs +342 -3622
- data/ext/kreuzberg_rb/native/src/metadata.rs +34 -0
- data/ext/kreuzberg_rb/native/src/plugins/mod.rs +92 -0
- data/ext/kreuzberg_rb/native/src/plugins/ocr_backend.rs +159 -0
- data/ext/kreuzberg_rb/native/src/plugins/post_processor.rs +126 -0
- data/ext/kreuzberg_rb/native/src/plugins/validator.rs +99 -0
- data/ext/kreuzberg_rb/native/src/result.rs +326 -0
- data/ext/kreuzberg_rb/native/src/validation.rs +4 -0
- data/lib/kreuzberg/config.rb +66 -0
- data/lib/kreuzberg/result.rb +107 -2
- data/lib/kreuzberg/types.rb +104 -0
- data/lib/kreuzberg/version.rb +1 -1
- data/lib/kreuzberg.rb +0 -4
- data/sig/kreuzberg.rbs +105 -1
- data/vendor/Cargo.toml +3 -3
- data/vendor/kreuzberg/Cargo.toml +4 -3
- data/vendor/kreuzberg/README.md +1 -1
- data/vendor/kreuzberg/src/api/config.rs +69 -0
- data/vendor/kreuzberg/src/api/handlers.rs +99 -2
- data/vendor/kreuzberg/src/api/mod.rs +14 -7
- data/vendor/kreuzberg/src/api/router.rs +214 -0
- data/vendor/kreuzberg/src/api/startup.rs +243 -0
- data/vendor/kreuzberg/src/api/types.rs +78 -0
- data/vendor/kreuzberg/src/cache/cleanup.rs +277 -0
- data/vendor/kreuzberg/src/cache/core.rs +428 -0
- data/vendor/kreuzberg/src/cache/mod.rs +21 -843
- data/vendor/kreuzberg/src/cache/utilities.rs +156 -0
- data/vendor/kreuzberg/src/chunking/boundaries.rs +301 -0
- data/vendor/kreuzberg/src/chunking/builder.rs +294 -0
- data/vendor/kreuzberg/src/chunking/config.rs +52 -0
- data/vendor/kreuzberg/src/chunking/core.rs +1017 -0
- data/vendor/kreuzberg/src/chunking/mod.rs +14 -2211
- data/vendor/kreuzberg/src/chunking/processor.rs +10 -0
- data/vendor/kreuzberg/src/chunking/validation.rs +686 -0
- data/vendor/kreuzberg/src/core/config/extraction/core.rs +169 -0
- data/vendor/kreuzberg/src/core/config/extraction/env.rs +179 -0
- data/vendor/kreuzberg/src/core/config/extraction/loaders.rs +204 -0
- data/vendor/kreuzberg/src/core/config/extraction/mod.rs +42 -0
- data/vendor/kreuzberg/src/core/config/extraction/types.rs +93 -0
- data/vendor/kreuzberg/src/core/config/formats.rs +135 -0
- data/vendor/kreuzberg/src/core/config/mod.rs +20 -0
- data/vendor/kreuzberg/src/core/config/ocr.rs +73 -0
- data/vendor/kreuzberg/src/core/config/page.rs +57 -0
- data/vendor/kreuzberg/src/core/config/pdf.rs +111 -0
- data/vendor/kreuzberg/src/core/config/processing.rs +312 -0
- data/vendor/kreuzberg/src/core/config_validation/dependencies.rs +187 -0
- data/vendor/kreuzberg/src/core/config_validation/mod.rs +386 -0
- data/vendor/kreuzberg/src/core/config_validation/sections.rs +401 -0
- data/vendor/kreuzberg/src/core/extractor/batch.rs +246 -0
- data/vendor/kreuzberg/src/core/extractor/bytes.rs +116 -0
- data/vendor/kreuzberg/src/core/extractor/file.rs +240 -0
- data/vendor/kreuzberg/src/core/extractor/helpers.rs +71 -0
- data/vendor/kreuzberg/src/core/extractor/legacy.rs +62 -0
- data/vendor/kreuzberg/src/core/extractor/mod.rs +490 -0
- data/vendor/kreuzberg/src/core/extractor/sync.rs +208 -0
- data/vendor/kreuzberg/src/core/mod.rs +4 -1
- data/vendor/kreuzberg/src/core/pipeline/cache.rs +60 -0
- data/vendor/kreuzberg/src/core/pipeline/execution.rs +89 -0
- data/vendor/kreuzberg/src/core/pipeline/features.rs +108 -0
- data/vendor/kreuzberg/src/core/pipeline/format.rs +392 -0
- data/vendor/kreuzberg/src/core/pipeline/initialization.rs +67 -0
- data/vendor/kreuzberg/src/core/pipeline/mod.rs +135 -0
- data/vendor/kreuzberg/src/core/pipeline/tests.rs +975 -0
- data/vendor/kreuzberg/src/core/server_config/env.rs +90 -0
- data/vendor/kreuzberg/src/core/server_config/loader.rs +202 -0
- data/vendor/kreuzberg/src/core/server_config/mod.rs +380 -0
- data/vendor/kreuzberg/src/core/server_config/tests/basic_tests.rs +124 -0
- data/vendor/kreuzberg/src/core/server_config/tests/env_tests.rs +216 -0
- data/vendor/kreuzberg/src/core/server_config/tests/file_loading_tests.rs +341 -0
- data/vendor/kreuzberg/src/core/server_config/tests/mod.rs +5 -0
- data/vendor/kreuzberg/src/core/server_config/validation.rs +17 -0
- data/vendor/kreuzberg/src/embeddings.rs +136 -13
- data/vendor/kreuzberg/src/extraction/{archive.rs → archive/mod.rs} +45 -239
- data/vendor/kreuzberg/src/extraction/archive/sevenz.rs +98 -0
- data/vendor/kreuzberg/src/extraction/archive/tar.rs +118 -0
- data/vendor/kreuzberg/src/extraction/archive/zip.rs +101 -0
- data/vendor/kreuzberg/src/extraction/html/converter.rs +592 -0
- data/vendor/kreuzberg/src/extraction/html/image_handling.rs +95 -0
- data/vendor/kreuzberg/src/extraction/html/mod.rs +53 -0
- data/vendor/kreuzberg/src/extraction/html/processor.rs +659 -0
- data/vendor/kreuzberg/src/extraction/html/stack_management.rs +103 -0
- data/vendor/kreuzberg/src/extraction/html/types.rs +28 -0
- data/vendor/kreuzberg/src/extraction/mod.rs +6 -2
- data/vendor/kreuzberg/src/extraction/pptx/container.rs +159 -0
- data/vendor/kreuzberg/src/extraction/pptx/content_builder.rs +168 -0
- data/vendor/kreuzberg/src/extraction/pptx/elements.rs +132 -0
- data/vendor/kreuzberg/src/extraction/pptx/image_handling.rs +57 -0
- data/vendor/kreuzberg/src/extraction/pptx/metadata.rs +160 -0
- data/vendor/kreuzberg/src/extraction/pptx/mod.rs +558 -0
- data/vendor/kreuzberg/src/extraction/pptx/parser.rs +379 -0
- data/vendor/kreuzberg/src/extraction/transform/content.rs +205 -0
- data/vendor/kreuzberg/src/extraction/transform/elements.rs +211 -0
- data/vendor/kreuzberg/src/extraction/transform/mod.rs +480 -0
- data/vendor/kreuzberg/src/extraction/transform/types.rs +27 -0
- data/vendor/kreuzberg/src/extractors/archive.rs +2 -0
- data/vendor/kreuzberg/src/extractors/bibtex.rs +2 -0
- data/vendor/kreuzberg/src/extractors/djot_format/attributes.rs +134 -0
- data/vendor/kreuzberg/src/extractors/djot_format/conversion.rs +223 -0
- data/vendor/kreuzberg/src/extractors/djot_format/extractor.rs +172 -0
- data/vendor/kreuzberg/src/extractors/djot_format/mod.rs +24 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/block_handlers.rs +271 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/content_extraction.rs +257 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/event_handlers.rs +101 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/inline_handlers.rs +201 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/mod.rs +16 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/state.rs +78 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/table_extraction.rs +68 -0
- data/vendor/kreuzberg/src/extractors/djot_format/parsing/text_extraction.rs +61 -0
- data/vendor/kreuzberg/src/extractors/djot_format/rendering.rs +452 -0
- data/vendor/kreuzberg/src/extractors/docbook.rs +2 -0
- data/vendor/kreuzberg/src/extractors/docx.rs +12 -1
- data/vendor/kreuzberg/src/extractors/email.rs +2 -0
- data/vendor/kreuzberg/src/extractors/epub/content.rs +333 -0
- data/vendor/kreuzberg/src/extractors/epub/metadata.rs +137 -0
- data/vendor/kreuzberg/src/extractors/epub/mod.rs +186 -0
- data/vendor/kreuzberg/src/extractors/epub/parsing.rs +86 -0
- data/vendor/kreuzberg/src/extractors/excel.rs +4 -0
- data/vendor/kreuzberg/src/extractors/fictionbook.rs +2 -0
- data/vendor/kreuzberg/src/extractors/frontmatter_utils.rs +466 -0
- data/vendor/kreuzberg/src/extractors/html.rs +80 -8
- data/vendor/kreuzberg/src/extractors/image.rs +8 -1
- data/vendor/kreuzberg/src/extractors/jats/elements.rs +350 -0
- data/vendor/kreuzberg/src/extractors/jats/metadata.rs +21 -0
- data/vendor/kreuzberg/src/extractors/{jats.rs → jats/mod.rs} +10 -412
- data/vendor/kreuzberg/src/extractors/jats/parser.rs +52 -0
- data/vendor/kreuzberg/src/extractors/jupyter.rs +2 -0
- data/vendor/kreuzberg/src/extractors/latex/commands.rs +93 -0
- data/vendor/kreuzberg/src/extractors/latex/environments.rs +157 -0
- data/vendor/kreuzberg/src/extractors/latex/metadata.rs +27 -0
- data/vendor/kreuzberg/src/extractors/latex/mod.rs +146 -0
- data/vendor/kreuzberg/src/extractors/latex/parser.rs +231 -0
- data/vendor/kreuzberg/src/extractors/latex/utilities.rs +126 -0
- data/vendor/kreuzberg/src/extractors/markdown.rs +39 -162
- data/vendor/kreuzberg/src/extractors/mod.rs +9 -1
- data/vendor/kreuzberg/src/extractors/odt.rs +2 -0
- data/vendor/kreuzberg/src/extractors/opml/core.rs +165 -0
- data/vendor/kreuzberg/src/extractors/opml/mod.rs +31 -0
- data/vendor/kreuzberg/src/extractors/opml/parser.rs +479 -0
- data/vendor/kreuzberg/src/extractors/orgmode.rs +2 -0
- data/vendor/kreuzberg/src/extractors/pdf/extraction.rs +106 -0
- data/vendor/kreuzberg/src/extractors/{pdf.rs → pdf/mod.rs} +25 -324
- data/vendor/kreuzberg/src/extractors/pdf/ocr.rs +214 -0
- data/vendor/kreuzberg/src/extractors/pdf/pages.rs +51 -0
- data/vendor/kreuzberg/src/extractors/pptx.rs +9 -2
- data/vendor/kreuzberg/src/extractors/rst.rs +2 -0
- data/vendor/kreuzberg/src/extractors/rtf/encoding.rs +116 -0
- data/vendor/kreuzberg/src/extractors/rtf/formatting.rs +24 -0
- data/vendor/kreuzberg/src/extractors/rtf/images.rs +72 -0
- data/vendor/kreuzberg/src/extractors/rtf/metadata.rs +216 -0
- data/vendor/kreuzberg/src/extractors/rtf/mod.rs +142 -0
- data/vendor/kreuzberg/src/extractors/rtf/parser.rs +259 -0
- data/vendor/kreuzberg/src/extractors/rtf/tables.rs +83 -0
- data/vendor/kreuzberg/src/extractors/structured.rs +2 -0
- data/vendor/kreuzberg/src/extractors/text.rs +4 -0
- data/vendor/kreuzberg/src/extractors/typst.rs +2 -0
- data/vendor/kreuzberg/src/extractors/xml.rs +2 -0
- data/vendor/kreuzberg/src/keywords/processor.rs +14 -0
- data/vendor/kreuzberg/src/language_detection/processor.rs +10 -0
- data/vendor/kreuzberg/src/lib.rs +2 -2
- data/vendor/kreuzberg/src/mcp/errors.rs +312 -0
- data/vendor/kreuzberg/src/mcp/format.rs +211 -0
- data/vendor/kreuzberg/src/mcp/mod.rs +9 -3
- data/vendor/kreuzberg/src/mcp/params.rs +196 -0
- data/vendor/kreuzberg/src/mcp/server.rs +39 -1438
- data/vendor/kreuzberg/src/mcp/tools/cache.rs +179 -0
- data/vendor/kreuzberg/src/mcp/tools/extraction.rs +403 -0
- data/vendor/kreuzberg/src/mcp/tools/mime.rs +150 -0
- data/vendor/kreuzberg/src/mcp/tools/mod.rs +11 -0
- data/vendor/kreuzberg/src/ocr/backends/easyocr.rs +96 -0
- data/vendor/kreuzberg/src/ocr/backends/mod.rs +7 -0
- data/vendor/kreuzberg/src/ocr/backends/paddleocr.rs +27 -0
- data/vendor/kreuzberg/src/ocr/backends/tesseract.rs +134 -0
- data/vendor/kreuzberg/src/ocr/hocr.rs +60 -16
- data/vendor/kreuzberg/src/ocr/language_registry.rs +11 -235
- data/vendor/kreuzberg/src/ocr/mod.rs +1 -0
- data/vendor/kreuzberg/src/ocr/processor/config.rs +203 -0
- data/vendor/kreuzberg/src/ocr/processor/execution.rs +494 -0
- data/vendor/kreuzberg/src/ocr/processor/mod.rs +265 -0
- data/vendor/kreuzberg/src/ocr/processor/validation.rs +145 -0
- data/vendor/kreuzberg/src/ocr/tesseract_backend.rs +41 -24
- data/vendor/kreuzberg/src/pdf/bindings.rs +21 -8
- data/vendor/kreuzberg/src/pdf/hierarchy/bounding_box.rs +289 -0
- data/vendor/kreuzberg/src/pdf/hierarchy/clustering.rs +199 -0
- data/vendor/kreuzberg/src/pdf/{hierarchy.rs → hierarchy/extraction.rs} +6 -346
- data/vendor/kreuzberg/src/pdf/hierarchy/mod.rs +18 -0
- data/vendor/kreuzberg/src/plugins/extractor/mod.rs +319 -0
- data/vendor/kreuzberg/src/plugins/extractor/registry.rs +434 -0
- data/vendor/kreuzberg/src/plugins/extractor/trait.rs +391 -0
- data/vendor/kreuzberg/src/plugins/mod.rs +13 -0
- data/vendor/kreuzberg/src/plugins/ocr.rs +11 -0
- data/vendor/kreuzberg/src/plugins/processor/mod.rs +365 -0
- data/vendor/kreuzberg/src/plugins/processor/registry.rs +37 -0
- data/vendor/kreuzberg/src/plugins/processor/trait.rs +284 -0
- data/vendor/kreuzberg/src/plugins/registry/extractor.rs +416 -0
- data/vendor/kreuzberg/src/plugins/registry/mod.rs +116 -0
- data/vendor/kreuzberg/src/plugins/registry/ocr.rs +293 -0
- data/vendor/kreuzberg/src/plugins/registry/processor.rs +304 -0
- data/vendor/kreuzberg/src/plugins/registry/validator.rs +238 -0
- data/vendor/kreuzberg/src/plugins/validator/mod.rs +424 -0
- data/vendor/kreuzberg/src/plugins/validator/registry.rs +355 -0
- data/vendor/kreuzberg/src/plugins/validator/trait.rs +276 -0
- data/vendor/kreuzberg/src/stopwords/languages/asian.rs +40 -0
- data/vendor/kreuzberg/src/stopwords/languages/germanic.rs +36 -0
- data/vendor/kreuzberg/src/stopwords/languages/mod.rs +10 -0
- data/vendor/kreuzberg/src/stopwords/languages/other.rs +44 -0
- data/vendor/kreuzberg/src/stopwords/languages/romance.rs +36 -0
- data/vendor/kreuzberg/src/stopwords/languages/slavic.rs +36 -0
- data/vendor/kreuzberg/src/stopwords/mod.rs +7 -33
- data/vendor/kreuzberg/src/text/quality.rs +1 -1
- data/vendor/kreuzberg/src/text/quality_processor.rs +10 -0
- data/vendor/kreuzberg/src/text/token_reduction/core/analysis.rs +238 -0
- data/vendor/kreuzberg/src/text/token_reduction/core/mod.rs +8 -0
- data/vendor/kreuzberg/src/text/token_reduction/core/punctuation.rs +54 -0
- data/vendor/kreuzberg/src/text/token_reduction/core/reducer.rs +384 -0
- data/vendor/kreuzberg/src/text/token_reduction/core/sentence_selection.rs +68 -0
- data/vendor/kreuzberg/src/text/token_reduction/core/word_filtering.rs +156 -0
- data/vendor/kreuzberg/src/text/token_reduction/filters/general.rs +377 -0
- data/vendor/kreuzberg/src/text/token_reduction/filters/html.rs +51 -0
- data/vendor/kreuzberg/src/text/token_reduction/filters/markdown.rs +285 -0
- data/vendor/kreuzberg/src/text/token_reduction/filters.rs +131 -246
- data/vendor/kreuzberg/src/types/djot.rs +209 -0
- data/vendor/kreuzberg/src/types/extraction.rs +301 -0
- data/vendor/kreuzberg/src/types/formats.rs +443 -0
- data/vendor/kreuzberg/src/types/metadata.rs +560 -0
- data/vendor/kreuzberg/src/types/mod.rs +281 -0
- data/vendor/kreuzberg/src/types/page.rs +182 -0
- data/vendor/kreuzberg/src/types/serde_helpers.rs +132 -0
- data/vendor/kreuzberg/src/types/tables.rs +39 -0
- data/vendor/kreuzberg/src/utils/quality/heuristics.rs +58 -0
- data/vendor/kreuzberg/src/utils/{quality.rs → quality/mod.rs} +168 -489
- data/vendor/kreuzberg/src/utils/quality/patterns.rs +117 -0
- data/vendor/kreuzberg/src/utils/quality/scoring.rs +178 -0
- data/vendor/kreuzberg/src/utils/string_pool/buffer_pool.rs +325 -0
- data/vendor/kreuzberg/src/utils/string_pool/interned.rs +102 -0
- data/vendor/kreuzberg/src/utils/string_pool/language_pool.rs +119 -0
- data/vendor/kreuzberg/src/utils/string_pool/mime_pool.rs +235 -0
- data/vendor/kreuzberg/src/utils/string_pool/mod.rs +41 -0
- data/vendor/kreuzberg/tests/api_chunk.rs +313 -0
- data/vendor/kreuzberg/tests/api_embed.rs +6 -9
- data/vendor/kreuzberg/tests/batch_orchestration.rs +1 -0
- data/vendor/kreuzberg/tests/concurrency_stress.rs +7 -0
- data/vendor/kreuzberg/tests/core_integration.rs +1 -0
- data/vendor/kreuzberg/tests/docx_metadata_extraction_test.rs +130 -0
- data/vendor/kreuzberg/tests/epub_native_extractor_tests.rs +5 -14
- data/vendor/kreuzberg/tests/format_integration.rs +2 -0
- data/vendor/kreuzberg/tests/helpers/mod.rs +1 -0
- data/vendor/kreuzberg/tests/html_table_test.rs +11 -11
- data/vendor/kreuzberg/tests/ocr_configuration.rs +16 -0
- data/vendor/kreuzberg/tests/ocr_errors.rs +18 -0
- data/vendor/kreuzberg/tests/ocr_quality.rs +9 -0
- data/vendor/kreuzberg/tests/ocr_stress.rs +1 -0
- data/vendor/kreuzberg/tests/pipeline_integration.rs +50 -0
- data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs +13 -0
- data/vendor/kreuzberg/tests/plugin_system.rs +12 -0
- data/vendor/kreuzberg/tests/registry_integration_tests.rs +2 -0
- data/vendor/kreuzberg-ffi/Cargo.toml +2 -1
- data/vendor/kreuzberg-ffi/benches/result_view_benchmark.rs +2 -0
- data/vendor/kreuzberg-ffi/kreuzberg.h +347 -178
- data/vendor/kreuzberg-ffi/src/config/html.rs +318 -0
- data/vendor/kreuzberg-ffi/src/config/loader.rs +154 -0
- data/vendor/kreuzberg-ffi/src/config/merge.rs +104 -0
- data/vendor/kreuzberg-ffi/src/config/mod.rs +385 -0
- data/vendor/kreuzberg-ffi/src/config/parse.rs +91 -0
- data/vendor/kreuzberg-ffi/src/config/serialize.rs +118 -0
- data/vendor/kreuzberg-ffi/src/config_builder.rs +598 -0
- data/vendor/kreuzberg-ffi/src/error.rs +46 -14
- data/vendor/kreuzberg-ffi/src/helpers.rs +10 -0
- data/vendor/kreuzberg-ffi/src/html_options.rs +421 -0
- data/vendor/kreuzberg-ffi/src/lib.rs +16 -0
- data/vendor/kreuzberg-ffi/src/panic_shield.rs +11 -0
- data/vendor/kreuzberg-ffi/src/plugins/ocr_backend.rs +2 -0
- data/vendor/kreuzberg-ffi/src/result.rs +148 -122
- data/vendor/kreuzberg-ffi/src/result_view.rs +4 -0
- data/vendor/kreuzberg-tesseract/Cargo.toml +2 -2
- metadata +200 -28
- data/vendor/kreuzberg/src/api/server.rs +0 -518
- data/vendor/kreuzberg/src/core/config.rs +0 -1914
- data/vendor/kreuzberg/src/core/config_validation.rs +0 -949
- data/vendor/kreuzberg/src/core/extractor.rs +0 -1200
- data/vendor/kreuzberg/src/core/pipeline.rs +0 -1223
- data/vendor/kreuzberg/src/core/server_config.rs +0 -1220
- data/vendor/kreuzberg/src/extraction/html.rs +0 -1830
- data/vendor/kreuzberg/src/extraction/pptx.rs +0 -3102
- data/vendor/kreuzberg/src/extractors/epub.rs +0 -696
- data/vendor/kreuzberg/src/extractors/latex.rs +0 -653
- data/vendor/kreuzberg/src/extractors/opml.rs +0 -635
- data/vendor/kreuzberg/src/extractors/rtf.rs +0 -809
- data/vendor/kreuzberg/src/ocr/processor.rs +0 -858
- data/vendor/kreuzberg/src/plugins/extractor.rs +0 -1042
- data/vendor/kreuzberg/src/plugins/processor.rs +0 -650
- data/vendor/kreuzberg/src/plugins/registry.rs +0 -1339
- data/vendor/kreuzberg/src/plugins/validator.rs +0 -967
- data/vendor/kreuzberg/src/text/token_reduction/core.rs +0 -832
- data/vendor/kreuzberg/src/types.rs +0 -1713
- data/vendor/kreuzberg/src/utils/string_pool.rs +0 -762
- data/vendor/kreuzberg-ffi/src/config.rs +0 -1341
|
@@ -0,0 +1,293 @@
|
|
|
1
|
+
//! OCR backend registry.
|
|
2
|
+
|
|
3
|
+
use crate::plugins::OcrBackend;
|
|
4
|
+
use crate::{KreuzbergError, Result};
|
|
5
|
+
use std::collections::HashMap;
|
|
6
|
+
use std::sync::Arc;
|
|
7
|
+
|
|
8
|
+
/// Registry for OCR backend plugins.
|
|
9
|
+
///
|
|
10
|
+
/// Manages OCR backends with backend type and language-based selection.
|
|
11
|
+
///
|
|
12
|
+
/// # Thread Safety
|
|
13
|
+
///
|
|
14
|
+
/// The registry is thread-safe and can be accessed concurrently from multiple threads.
|
|
15
|
+
///
|
|
16
|
+
/// # Example
|
|
17
|
+
///
|
|
18
|
+
/// ```rust,no_run
|
|
19
|
+
/// use kreuzberg::plugins::registry::OcrBackendRegistry;
|
|
20
|
+
/// use std::sync::Arc;
|
|
21
|
+
///
|
|
22
|
+
/// let registry = OcrBackendRegistry::new();
|
|
23
|
+
/// // Register OCR backends
|
|
24
|
+
/// // registry.register(Arc::new(TesseractBackend::new()));
|
|
25
|
+
/// ```
|
|
26
|
+
pub struct OcrBackendRegistry {
|
|
27
|
+
pub(super) backends: HashMap<String, Arc<dyn OcrBackend>>,
|
|
28
|
+
}
|
|
29
|
+
|
|
30
|
+
impl OcrBackendRegistry {
|
|
31
|
+
/// Create a new OCR backend registry with default backends.
|
|
32
|
+
///
|
|
33
|
+
/// Registers the Tesseract backend by default if the "ocr" feature is enabled.
|
|
34
|
+
pub fn new() -> Self {
|
|
35
|
+
#[cfg(feature = "ocr")]
|
|
36
|
+
let mut registry = Self {
|
|
37
|
+
backends: HashMap::new(),
|
|
38
|
+
};
|
|
39
|
+
|
|
40
|
+
#[cfg(not(feature = "ocr"))]
|
|
41
|
+
let registry = Self {
|
|
42
|
+
backends: HashMap::new(),
|
|
43
|
+
};
|
|
44
|
+
|
|
45
|
+
#[cfg(feature = "ocr")]
|
|
46
|
+
{
|
|
47
|
+
use crate::ocr::tesseract_backend::TesseractBackend;
|
|
48
|
+
if let Ok(backend) = TesseractBackend::new() {
|
|
49
|
+
let _ = registry.register(Arc::new(backend));
|
|
50
|
+
}
|
|
51
|
+
}
|
|
52
|
+
|
|
53
|
+
registry
|
|
54
|
+
}
|
|
55
|
+
|
|
56
|
+
/// Create a new empty OCR backend registry without default backends.
|
|
57
|
+
///
|
|
58
|
+
/// This is useful for testing or when you want full control over backend registration.
|
|
59
|
+
pub fn new_empty() -> Self {
|
|
60
|
+
Self {
|
|
61
|
+
backends: HashMap::new(),
|
|
62
|
+
}
|
|
63
|
+
}
|
|
64
|
+
|
|
65
|
+
/// Register an OCR backend.
|
|
66
|
+
///
|
|
67
|
+
/// # Arguments
|
|
68
|
+
///
|
|
69
|
+
/// * `backend` - The OCR backend to register
|
|
70
|
+
///
|
|
71
|
+
/// # Returns
|
|
72
|
+
///
|
|
73
|
+
/// - `Ok(())` if registration succeeded
|
|
74
|
+
/// - `Err(...)` if initialization failed
|
|
75
|
+
///
|
|
76
|
+
/// # Example
|
|
77
|
+
///
|
|
78
|
+
/// ```rust,no_run
|
|
79
|
+
/// # use kreuzberg::plugins::registry::OcrBackendRegistry;
|
|
80
|
+
/// # use std::sync::Arc;
|
|
81
|
+
/// let mut registry = OcrBackendRegistry::new();
|
|
82
|
+
/// // let backend = Arc::new(MyOcrBackend::new());
|
|
83
|
+
/// // registry.register(backend)?;
|
|
84
|
+
/// # Ok::<(), kreuzberg::KreuzbergError>(())
|
|
85
|
+
/// ```
|
|
86
|
+
pub fn register(&mut self, backend: Arc<dyn OcrBackend>) -> Result<()> {
|
|
87
|
+
let name = backend.name().to_string();
|
|
88
|
+
|
|
89
|
+
super::validate_plugin_name(&name)?;
|
|
90
|
+
|
|
91
|
+
backend.initialize()?;
|
|
92
|
+
|
|
93
|
+
self.backends.insert(name, backend);
|
|
94
|
+
Ok(())
|
|
95
|
+
}
|
|
96
|
+
|
|
97
|
+
/// Get an OCR backend by name.
|
|
98
|
+
///
|
|
99
|
+
/// # Arguments
|
|
100
|
+
///
|
|
101
|
+
/// * `name` - Backend name
|
|
102
|
+
///
|
|
103
|
+
/// # Returns
|
|
104
|
+
///
|
|
105
|
+
/// The backend if found, or an error if not registered.
|
|
106
|
+
pub fn get(&self, name: &str) -> Result<Arc<dyn OcrBackend>> {
|
|
107
|
+
self.backends.get(name).cloned().ok_or_else(|| KreuzbergError::Plugin {
|
|
108
|
+
message: format!("OCR backend '{}' not registered", name),
|
|
109
|
+
plugin_name: name.to_string(),
|
|
110
|
+
})
|
|
111
|
+
}
|
|
112
|
+
|
|
113
|
+
/// Get an OCR backend that supports a specific language.
|
|
114
|
+
///
|
|
115
|
+
/// Returns the first backend that supports the language.
|
|
116
|
+
///
|
|
117
|
+
/// # Arguments
|
|
118
|
+
///
|
|
119
|
+
/// * `language` - Language code (e.g., "eng", "deu")
|
|
120
|
+
///
|
|
121
|
+
/// # Returns
|
|
122
|
+
///
|
|
123
|
+
/// The first backend supporting the language, or an error if none found.
|
|
124
|
+
pub fn get_for_language(&self, language: &str) -> Result<Arc<dyn OcrBackend>> {
|
|
125
|
+
self.backends
|
|
126
|
+
.values()
|
|
127
|
+
.find(|backend| backend.supports_language(language))
|
|
128
|
+
.cloned()
|
|
129
|
+
.ok_or_else(|| KreuzbergError::Plugin {
|
|
130
|
+
message: format!("No OCR backend supports language '{}'", language),
|
|
131
|
+
plugin_name: language.to_string(),
|
|
132
|
+
})
|
|
133
|
+
}
|
|
134
|
+
|
|
135
|
+
/// List all registered backend names.
|
|
136
|
+
pub fn list(&self) -> Vec<String> {
|
|
137
|
+
self.backends.keys().cloned().collect()
|
|
138
|
+
}
|
|
139
|
+
|
|
140
|
+
/// Remove a backend from the registry.
|
|
141
|
+
///
|
|
142
|
+
/// Calls `shutdown()` on the backend before removing.
|
|
143
|
+
pub fn remove(&mut self, name: &str) -> Result<()> {
|
|
144
|
+
if let Some(backend) = self.backends.remove(name) {
|
|
145
|
+
backend.shutdown()?;
|
|
146
|
+
}
|
|
147
|
+
Ok(())
|
|
148
|
+
}
|
|
149
|
+
|
|
150
|
+
/// Shutdown all backends and clear the registry.
|
|
151
|
+
pub fn shutdown_all(&mut self) -> Result<()> {
|
|
152
|
+
let names: Vec<_> = self.backends.keys().cloned().collect();
|
|
153
|
+
for name in names {
|
|
154
|
+
self.remove(&name)?;
|
|
155
|
+
}
|
|
156
|
+
Ok(())
|
|
157
|
+
}
|
|
158
|
+
}
|
|
159
|
+
|
|
160
|
+
impl Default for OcrBackendRegistry {
|
|
161
|
+
fn default() -> Self {
|
|
162
|
+
Self::new()
|
|
163
|
+
}
|
|
164
|
+
}
|
|
165
|
+
|
|
166
|
+
#[cfg(test)]
|
|
167
|
+
mod tests {
|
|
168
|
+
use super::*;
|
|
169
|
+
use crate::core::config::OcrConfig;
|
|
170
|
+
use crate::plugins::{OcrBackend, Plugin};
|
|
171
|
+
use crate::types::ExtractionResult;
|
|
172
|
+
use async_trait::async_trait;
|
|
173
|
+
|
|
174
|
+
struct MockOcrBackend {
|
|
175
|
+
name: String,
|
|
176
|
+
languages: Vec<String>,
|
|
177
|
+
}
|
|
178
|
+
|
|
179
|
+
impl Plugin for MockOcrBackend {
|
|
180
|
+
fn name(&self) -> &str {
|
|
181
|
+
&self.name
|
|
182
|
+
}
|
|
183
|
+
fn version(&self) -> String {
|
|
184
|
+
"1.0.0".to_string()
|
|
185
|
+
}
|
|
186
|
+
fn initialize(&self) -> Result<()> {
|
|
187
|
+
Ok(())
|
|
188
|
+
}
|
|
189
|
+
fn shutdown(&self) -> Result<()> {
|
|
190
|
+
Ok(())
|
|
191
|
+
}
|
|
192
|
+
}
|
|
193
|
+
|
|
194
|
+
#[async_trait]
|
|
195
|
+
impl OcrBackend for MockOcrBackend {
|
|
196
|
+
async fn process_image(&self, _: &[u8], _: &OcrConfig) -> Result<ExtractionResult> {
|
|
197
|
+
Ok(ExtractionResult {
|
|
198
|
+
content: "test".to_string(),
|
|
199
|
+
mime_type: "text/plain".to_string(),
|
|
200
|
+
metadata: crate::types::Metadata::default(),
|
|
201
|
+
tables: vec![],
|
|
202
|
+
detected_languages: None,
|
|
203
|
+
chunks: None,
|
|
204
|
+
images: None,
|
|
205
|
+
djot_content: None,
|
|
206
|
+
pages: None,
|
|
207
|
+
elements: None,
|
|
208
|
+
})
|
|
209
|
+
}
|
|
210
|
+
|
|
211
|
+
fn supports_language(&self, lang: &str) -> bool {
|
|
212
|
+
self.languages.iter().any(|l| l == lang)
|
|
213
|
+
}
|
|
214
|
+
|
|
215
|
+
fn backend_type(&self) -> crate::plugins::ocr::OcrBackendType {
|
|
216
|
+
crate::plugins::ocr::OcrBackendType::Custom
|
|
217
|
+
}
|
|
218
|
+
}
|
|
219
|
+
|
|
220
|
+
#[test]
|
|
221
|
+
fn test_ocr_backend_registry() {
|
|
222
|
+
let mut registry = OcrBackendRegistry::new_empty();
|
|
223
|
+
|
|
224
|
+
let backend = Arc::new(MockOcrBackend {
|
|
225
|
+
name: "test-ocr".to_string(),
|
|
226
|
+
languages: vec!["eng".to_string(), "deu".to_string()],
|
|
227
|
+
});
|
|
228
|
+
|
|
229
|
+
registry.register(backend).unwrap();
|
|
230
|
+
|
|
231
|
+
let retrieved = registry.get("test-ocr").unwrap();
|
|
232
|
+
assert_eq!(retrieved.name(), "test-ocr");
|
|
233
|
+
|
|
234
|
+
let eng_backend = registry.get_for_language("eng").unwrap();
|
|
235
|
+
assert_eq!(eng_backend.name(), "test-ocr");
|
|
236
|
+
|
|
237
|
+
let names = registry.list();
|
|
238
|
+
assert_eq!(names.len(), 1);
|
|
239
|
+
assert!(names.contains(&"test-ocr".to_string()));
|
|
240
|
+
}
|
|
241
|
+
|
|
242
|
+
#[test]
|
|
243
|
+
fn test_ocr_backend_registry_new_empty() {
|
|
244
|
+
let registry = OcrBackendRegistry::new_empty();
|
|
245
|
+
assert_eq!(registry.list().len(), 0);
|
|
246
|
+
}
|
|
247
|
+
|
|
248
|
+
#[test]
|
|
249
|
+
fn test_ocr_backend_get_missing() {
|
|
250
|
+
let registry = OcrBackendRegistry::new_empty();
|
|
251
|
+
let result = registry.get("nonexistent");
|
|
252
|
+
assert!(result.is_err());
|
|
253
|
+
}
|
|
254
|
+
|
|
255
|
+
#[test]
|
|
256
|
+
fn test_ocr_backend_get_for_language_missing() {
|
|
257
|
+
let registry = OcrBackendRegistry::new_empty();
|
|
258
|
+
let result = registry.get_for_language("fra");
|
|
259
|
+
assert!(result.is_err());
|
|
260
|
+
}
|
|
261
|
+
|
|
262
|
+
#[test]
|
|
263
|
+
fn test_ocr_backend_remove() {
|
|
264
|
+
let mut registry = OcrBackendRegistry::new_empty();
|
|
265
|
+
let backend = Arc::new(MockOcrBackend {
|
|
266
|
+
name: "test-backend".to_string(),
|
|
267
|
+
languages: vec!["eng".to_string()],
|
|
268
|
+
});
|
|
269
|
+
registry.register(backend).unwrap();
|
|
270
|
+
|
|
271
|
+
registry.remove("test-backend").unwrap();
|
|
272
|
+
assert_eq!(registry.list().len(), 0);
|
|
273
|
+
}
|
|
274
|
+
|
|
275
|
+
#[test]
|
|
276
|
+
fn test_ocr_backend_shutdown_all() {
|
|
277
|
+
let mut registry = OcrBackendRegistry::new_empty();
|
|
278
|
+
let backend1 = Arc::new(MockOcrBackend {
|
|
279
|
+
name: "backend1".to_string(),
|
|
280
|
+
languages: vec!["eng".to_string()],
|
|
281
|
+
});
|
|
282
|
+
let backend2 = Arc::new(MockOcrBackend {
|
|
283
|
+
name: "backend2".to_string(),
|
|
284
|
+
languages: vec!["deu".to_string()],
|
|
285
|
+
});
|
|
286
|
+
|
|
287
|
+
registry.register(backend1).unwrap();
|
|
288
|
+
registry.register(backend2).unwrap();
|
|
289
|
+
|
|
290
|
+
registry.shutdown_all().unwrap();
|
|
291
|
+
assert_eq!(registry.list().len(), 0);
|
|
292
|
+
}
|
|
293
|
+
}
|
|
@@ -0,0 +1,304 @@
|
|
|
1
|
+
//! Post-processor registry implementation.
|
|
2
|
+
|
|
3
|
+
use crate::Result;
|
|
4
|
+
use crate::plugins::{PostProcessor, ProcessingStage};
|
|
5
|
+
use std::collections::{BTreeMap, HashMap};
|
|
6
|
+
use std::sync::Arc;
|
|
7
|
+
|
|
8
|
+
/// Registry for post-processor plugins.
|
|
9
|
+
///
|
|
10
|
+
/// Manages post-processors organized by processing stage.
|
|
11
|
+
pub struct PostProcessorRegistry {
|
|
12
|
+
processors: HashMap<ProcessingStage, BTreeMap<i32, Vec<Arc<dyn PostProcessor>>>>,
|
|
13
|
+
name_index: HashMap<String, (ProcessingStage, i32)>,
|
|
14
|
+
}
|
|
15
|
+
|
|
16
|
+
impl PostProcessorRegistry {
|
|
17
|
+
/// Create a new empty post-processor registry.
|
|
18
|
+
pub fn new() -> Self {
|
|
19
|
+
Self {
|
|
20
|
+
processors: HashMap::new(),
|
|
21
|
+
name_index: HashMap::new(),
|
|
22
|
+
}
|
|
23
|
+
}
|
|
24
|
+
|
|
25
|
+
/// Register a post-processor.
|
|
26
|
+
///
|
|
27
|
+
/// # Arguments
|
|
28
|
+
///
|
|
29
|
+
/// * `processor` - The post-processor to register
|
|
30
|
+
/// * `priority` - Execution priority (higher = runs first within stage)
|
|
31
|
+
pub fn register(&mut self, processor: Arc<dyn PostProcessor>, priority: i32) -> Result<()> {
|
|
32
|
+
let name = processor.name().to_string();
|
|
33
|
+
let stage = processor.processing_stage();
|
|
34
|
+
|
|
35
|
+
super::validate_plugin_name(&name)?;
|
|
36
|
+
|
|
37
|
+
processor.initialize()?;
|
|
38
|
+
|
|
39
|
+
if self.name_index.contains_key(&name) {
|
|
40
|
+
self.remove(&name)?;
|
|
41
|
+
}
|
|
42
|
+
|
|
43
|
+
self.processors
|
|
44
|
+
.entry(stage)
|
|
45
|
+
.or_default()
|
|
46
|
+
.entry(priority)
|
|
47
|
+
.or_default()
|
|
48
|
+
.push(Arc::clone(&processor));
|
|
49
|
+
|
|
50
|
+
self.name_index.insert(name, (stage, priority));
|
|
51
|
+
|
|
52
|
+
Ok(())
|
|
53
|
+
}
|
|
54
|
+
|
|
55
|
+
/// Get all processors for a specific stage, in priority order.
|
|
56
|
+
///
|
|
57
|
+
/// # Arguments
|
|
58
|
+
///
|
|
59
|
+
/// * `stage` - The processing stage
|
|
60
|
+
///
|
|
61
|
+
/// # Returns
|
|
62
|
+
///
|
|
63
|
+
/// Vector of processors in priority order (highest first).
|
|
64
|
+
pub fn get_for_stage(&self, stage: ProcessingStage) -> Vec<Arc<dyn PostProcessor>> {
|
|
65
|
+
let mut result = Vec::new();
|
|
66
|
+
|
|
67
|
+
if let Some(priority_map) = self.processors.get(&stage) {
|
|
68
|
+
for (_priority, processors) in priority_map.iter().rev() {
|
|
69
|
+
for processor in processors {
|
|
70
|
+
result.push(Arc::clone(processor));
|
|
71
|
+
}
|
|
72
|
+
}
|
|
73
|
+
}
|
|
74
|
+
|
|
75
|
+
result
|
|
76
|
+
}
|
|
77
|
+
|
|
78
|
+
/// List all registered processor names.
|
|
79
|
+
pub fn list(&self) -> Vec<String> {
|
|
80
|
+
self.name_index.keys().cloned().collect()
|
|
81
|
+
}
|
|
82
|
+
|
|
83
|
+
/// Remove a processor from the registry.
|
|
84
|
+
pub fn remove(&mut self, name: &str) -> Result<()> {
|
|
85
|
+
let (stage, priority) = match self.name_index.remove(name) {
|
|
86
|
+
Some(location) => location,
|
|
87
|
+
None => return Ok(()),
|
|
88
|
+
};
|
|
89
|
+
|
|
90
|
+
let processor_to_shutdown = if let Some(priority_map) = self.processors.get_mut(&stage) {
|
|
91
|
+
let processor = priority_map.get_mut(&priority).and_then(|processors| {
|
|
92
|
+
processors
|
|
93
|
+
.iter()
|
|
94
|
+
.position(|p| p.name() == name)
|
|
95
|
+
.map(|pos| processors.remove(pos))
|
|
96
|
+
});
|
|
97
|
+
|
|
98
|
+
if let Some(processors) = priority_map.get(&priority)
|
|
99
|
+
&& processors.is_empty()
|
|
100
|
+
{
|
|
101
|
+
priority_map.remove(&priority);
|
|
102
|
+
}
|
|
103
|
+
|
|
104
|
+
if priority_map.is_empty() {
|
|
105
|
+
self.processors.remove(&stage);
|
|
106
|
+
}
|
|
107
|
+
processor
|
|
108
|
+
} else {
|
|
109
|
+
None
|
|
110
|
+
};
|
|
111
|
+
|
|
112
|
+
if let Some(processor) = processor_to_shutdown {
|
|
113
|
+
processor.shutdown()?;
|
|
114
|
+
}
|
|
115
|
+
|
|
116
|
+
Ok(())
|
|
117
|
+
}
|
|
118
|
+
|
|
119
|
+
/// Shutdown all processors and clear the registry.
|
|
120
|
+
pub fn shutdown_all(&mut self) -> Result<()> {
|
|
121
|
+
let names = self.list();
|
|
122
|
+
for name in names {
|
|
123
|
+
self.remove(&name)?;
|
|
124
|
+
}
|
|
125
|
+
Ok(())
|
|
126
|
+
}
|
|
127
|
+
}
|
|
128
|
+
|
|
129
|
+
impl Default for PostProcessorRegistry {
|
|
130
|
+
fn default() -> Self {
|
|
131
|
+
Self::new()
|
|
132
|
+
}
|
|
133
|
+
}
|
|
134
|
+
|
|
135
|
+
#[cfg(test)]
|
|
136
|
+
mod tests {
|
|
137
|
+
use super::*;
|
|
138
|
+
use crate::KreuzbergError;
|
|
139
|
+
use crate::core::config::ExtractionConfig;
|
|
140
|
+
use crate::plugins::Plugin;
|
|
141
|
+
use crate::types::ExtractionResult;
|
|
142
|
+
use async_trait::async_trait;
|
|
143
|
+
|
|
144
|
+
struct MockPostProcessor {
|
|
145
|
+
name: String,
|
|
146
|
+
stage: ProcessingStage,
|
|
147
|
+
}
|
|
148
|
+
|
|
149
|
+
impl Plugin for MockPostProcessor {
|
|
150
|
+
fn name(&self) -> &str {
|
|
151
|
+
&self.name
|
|
152
|
+
}
|
|
153
|
+
fn version(&self) -> String {
|
|
154
|
+
"1.0.0".to_string()
|
|
155
|
+
}
|
|
156
|
+
fn initialize(&self) -> Result<()> {
|
|
157
|
+
Ok(())
|
|
158
|
+
}
|
|
159
|
+
fn shutdown(&self) -> Result<()> {
|
|
160
|
+
Ok(())
|
|
161
|
+
}
|
|
162
|
+
}
|
|
163
|
+
|
|
164
|
+
#[async_trait]
|
|
165
|
+
impl PostProcessor for MockPostProcessor {
|
|
166
|
+
async fn process(&self, _result: &mut ExtractionResult, _: &ExtractionConfig) -> Result<()> {
|
|
167
|
+
Ok(())
|
|
168
|
+
}
|
|
169
|
+
|
|
170
|
+
fn processing_stage(&self) -> ProcessingStage {
|
|
171
|
+
self.stage
|
|
172
|
+
}
|
|
173
|
+
}
|
|
174
|
+
|
|
175
|
+
#[test]
|
|
176
|
+
fn test_post_processor_registry() {
|
|
177
|
+
let mut registry = PostProcessorRegistry::new();
|
|
178
|
+
|
|
179
|
+
let early = Arc::new(MockPostProcessor {
|
|
180
|
+
name: "early-processor".to_string(),
|
|
181
|
+
stage: ProcessingStage::Early,
|
|
182
|
+
});
|
|
183
|
+
|
|
184
|
+
let middle = Arc::new(MockPostProcessor {
|
|
185
|
+
name: "middle-processor".to_string(),
|
|
186
|
+
stage: ProcessingStage::Middle,
|
|
187
|
+
});
|
|
188
|
+
|
|
189
|
+
registry.register(early, 100).unwrap();
|
|
190
|
+
registry.register(middle, 50).unwrap();
|
|
191
|
+
|
|
192
|
+
let early_processors = registry.get_for_stage(ProcessingStage::Early);
|
|
193
|
+
assert_eq!(early_processors.len(), 1);
|
|
194
|
+
assert_eq!(early_processors[0].name(), "early-processor");
|
|
195
|
+
|
|
196
|
+
let middle_processors = registry.get_for_stage(ProcessingStage::Middle);
|
|
197
|
+
assert_eq!(middle_processors.len(), 1);
|
|
198
|
+
|
|
199
|
+
let names = registry.list();
|
|
200
|
+
assert_eq!(names.len(), 2);
|
|
201
|
+
}
|
|
202
|
+
|
|
203
|
+
#[test]
|
|
204
|
+
fn test_post_processor_registry_remove() {
|
|
205
|
+
let mut registry = PostProcessorRegistry::new();
|
|
206
|
+
|
|
207
|
+
let processor = Arc::new(MockPostProcessor {
|
|
208
|
+
name: "test-processor".to_string(),
|
|
209
|
+
stage: ProcessingStage::Early,
|
|
210
|
+
});
|
|
211
|
+
|
|
212
|
+
registry.register(processor, 50).unwrap();
|
|
213
|
+
assert_eq!(registry.get_for_stage(ProcessingStage::Early).len(), 1);
|
|
214
|
+
|
|
215
|
+
registry.remove("test-processor").unwrap();
|
|
216
|
+
assert_eq!(registry.get_for_stage(ProcessingStage::Early).len(), 0);
|
|
217
|
+
}
|
|
218
|
+
|
|
219
|
+
#[test]
|
|
220
|
+
fn test_post_processor_registry_default() {
|
|
221
|
+
let registry = PostProcessorRegistry::default();
|
|
222
|
+
assert_eq!(registry.list().len(), 0);
|
|
223
|
+
}
|
|
224
|
+
|
|
225
|
+
#[test]
|
|
226
|
+
fn test_post_processor_registry_invalid_name_empty() {
|
|
227
|
+
let mut registry = PostProcessorRegistry::new();
|
|
228
|
+
|
|
229
|
+
let processor = Arc::new(MockPostProcessor {
|
|
230
|
+
name: "".to_string(),
|
|
231
|
+
stage: ProcessingStage::Early,
|
|
232
|
+
});
|
|
233
|
+
|
|
234
|
+
let result = registry.register(processor, 50);
|
|
235
|
+
assert!(matches!(result, Err(KreuzbergError::Validation { .. })));
|
|
236
|
+
}
|
|
237
|
+
|
|
238
|
+
#[test]
|
|
239
|
+
fn test_post_processor_registry_invalid_name_whitespace() {
|
|
240
|
+
let mut registry = PostProcessorRegistry::new();
|
|
241
|
+
|
|
242
|
+
let processor = Arc::new(MockPostProcessor {
|
|
243
|
+
name: "my processor".to_string(),
|
|
244
|
+
stage: ProcessingStage::Early,
|
|
245
|
+
});
|
|
246
|
+
|
|
247
|
+
let result = registry.register(processor, 50);
|
|
248
|
+
assert!(matches!(result, Err(KreuzbergError::Validation { .. })));
|
|
249
|
+
}
|
|
250
|
+
|
|
251
|
+
#[test]
|
|
252
|
+
fn test_post_processor_registry_shutdown_all() {
|
|
253
|
+
let mut registry = PostProcessorRegistry::new();
|
|
254
|
+
|
|
255
|
+
let early = Arc::new(MockPostProcessor {
|
|
256
|
+
name: "early".to_string(),
|
|
257
|
+
stage: ProcessingStage::Early,
|
|
258
|
+
});
|
|
259
|
+
|
|
260
|
+
let late = Arc::new(MockPostProcessor {
|
|
261
|
+
name: "late".to_string(),
|
|
262
|
+
stage: ProcessingStage::Late,
|
|
263
|
+
});
|
|
264
|
+
|
|
265
|
+
registry.register(early, 100).unwrap();
|
|
266
|
+
registry.register(late, 50).unwrap();
|
|
267
|
+
|
|
268
|
+
assert_eq!(registry.list().len(), 2);
|
|
269
|
+
|
|
270
|
+
registry.shutdown_all().unwrap();
|
|
271
|
+
assert_eq!(registry.list().len(), 0);
|
|
272
|
+
}
|
|
273
|
+
|
|
274
|
+
#[test]
|
|
275
|
+
fn test_post_processor_registry_priority_order() {
|
|
276
|
+
let mut registry = PostProcessorRegistry::new();
|
|
277
|
+
|
|
278
|
+
let low = Arc::new(MockPostProcessor {
|
|
279
|
+
name: "low-priority".to_string(),
|
|
280
|
+
stage: ProcessingStage::Early,
|
|
281
|
+
});
|
|
282
|
+
|
|
283
|
+
let high = Arc::new(MockPostProcessor {
|
|
284
|
+
name: "high-priority".to_string(),
|
|
285
|
+
stage: ProcessingStage::Early,
|
|
286
|
+
});
|
|
287
|
+
|
|
288
|
+
registry.register(low, 10).unwrap();
|
|
289
|
+
registry.register(high, 100).unwrap();
|
|
290
|
+
|
|
291
|
+
let processors = registry.get_for_stage(ProcessingStage::Early);
|
|
292
|
+
assert_eq!(processors.len(), 2);
|
|
293
|
+
assert_eq!(processors[0].name(), "high-priority");
|
|
294
|
+
assert_eq!(processors[1].name(), "low-priority");
|
|
295
|
+
}
|
|
296
|
+
|
|
297
|
+
#[test]
|
|
298
|
+
fn test_post_processor_registry_empty_stage() {
|
|
299
|
+
let registry = PostProcessorRegistry::new();
|
|
300
|
+
|
|
301
|
+
let processors = registry.get_for_stage(ProcessingStage::Late);
|
|
302
|
+
assert_eq!(processors.len(), 0);
|
|
303
|
+
}
|
|
304
|
+
}
|