xberg 1.2.8 → 1.2.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a47ed0c309b4d9b5df23a9b396dfaeee17b3b7b835ceb40dc0bd830ded1f71d0
-  data.tar.gz: b540df1be17cd4be359093c60872761f4e96cd2d507da600a973a895fcf2383e
+  metadata.gz: 7529fe12878b0564beead15e3c2a88894059854701f51d3065fd4d3f084aeb9e
+  data.tar.gz: 585618a4f752a05ffcf1260b75d1c790542ccb8f87be94b0d87b9cd517e11d18
 SHA512:
-  metadata.gz: 52aeeecf622c4d746d6a14f7f6ef8d417ff0188adbbeefc0a120081e88cda967e8fe16ea9888c155c9917f50508cd8862ed839af0acf707f3667dd7381686575
-  data.tar.gz: 89d5d741965fa9aa7fba7831fc4932a9a96628bc1acb35fe90940b80a907f8622a7f9173ccbd1d264d1c3a4ab9ad0c81eefa1878188b8e5086d4836b946dccdd
+  metadata.gz: 6602574242548bf624c8b84cdc32eb6476ef03327aeaea911b74743f49dde20c1434d0cdfc695013de60924ac46e5d78622b76ba25c6360149201296953fa44a
+  data.tar.gz: adc7c03bb0a5d9349e7d9230e53027b5f7efa37869b8461797b32ded44215b22293e778af93310bb0ddb2fea66234ab6e5d0ef3f619421c9efad48495b894d84

data/ext/xberg_rb/native/Cargo.lock

@@ -3828,9 +3828,9 @@ checksum = "b6d2cec3eae94f9f509c767b45932f1ada8350c4bdb85af2fcab4a3c14807981"
 
 [[package]]
 name = "libredox"
-version = "0.1.25"
+version = "0.1.24"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "61ff90caf6077a803a240f62fdbe88645a890bbca49ef8174c3cb0404362171d"
+checksum = "6480ccc157a1389bb2e4891b24751b0f798ba640d22386f23143fbcc89da195a"
 dependencies = [
  "libc",
 ]
@@ -8998,9 +8998,9 @@ dependencies = [
 
 [[package]]
 name = "xberg"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "59f538f443f9448d9b0faa0988b70b228d03d77c1dc5bc5b234e1ae0f9ad7b11"
+checksum = "373e4a5be0fd41625c4243832d04f51997d4313e5dea0d781a7eb39205765b00"
 dependencies = [
  "aes-gcm",
  "ahash",
@@ -9136,9 +9136,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-candle-ocr"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98bb62acd6fbaa55a6318fed4422df7fd5bd9becd3edc0a654b680b8379ec960"
+checksum = "e050a875c61d507f27145be51a4a1d87922bedb307a0303732d264626fe8c3fc"
 dependencies = [
  "base64 0.23.1",
  "candle-core",
@@ -9158,9 +9158,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-gliner"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "27c0245ea3e0574ef895df526d0e4776655306d971ba39f45038d7aff25db253"
+checksum = "416f7924058670553eb858bfcf2d88ce29a9a7c42f2cdcdce77c12cafae30de2"
 dependencies = [
  "getrandom 0.2.17",
  "getrandom 0.3.4",
@@ -9174,9 +9174,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-libheif"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f5babb457f33b03fc2875a06be3488934bb0563f2bc03fcb555403e68ff2da0a"
+checksum = "60d71b81ecf779d67ec02fa6d6369ebf89ec5ac567acf995f9f50dd823450a23"
 dependencies = [
  "cfg-if",
  "enumn",
@@ -9187,9 +9187,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-libwpd"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c90652e58d056de1277392cbcbda2d06698167d01836a5ef4486af17c5a74757"
+checksum = "38d720cc8c033dd014b3055fd687545549166cce916273aba87391ffbcae7e45"
 dependencies = [
  "cc",
  "flate2",
@@ -9202,9 +9202,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-native-pdf"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "04c2aa99aa1b367505df8571f053271e77edd50ea5e6d4c49bbd25b19b9b2c93"
+checksum = "ddc05f380b448050b95361f59a4155291cc850286ceee2ebf98b83c5ab6e5dff"
 dependencies = [
  "aes 0.9.3",
  "base64 0.23.1",
@@ -9249,9 +9249,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-paddle-ocr"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a74e3b99b5ec7c44720161d2c63d60ee349350785da9952121a27f0c4160aa84"
+checksum = "95b1f2a67b2ce2051456a7136ff0c7a186445249acf27d8e6972dca1173fdbc2"
 dependencies = [
  "geo-clipper",
  "geo-types",
@@ -9266,7 +9266,7 @@ dependencies = [
 
 [[package]]
 name = "xberg-rb"
-version = "1.2.8"
+version = "1.2.9"
 dependencies = [
  "async-trait",
  "crawlberg",
@@ -9282,9 +9282,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-tesseract"
-version = "1.2.8"
+version = "1.2.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "92bf1e8a02849cf27c0ce988bb665b3c8206c19f5f6e08a065c1793fbbd32c94"
+checksum = "b6d6714013f52d238cf8e14614056c5d2fffb3662f40ae0434b79ec0cd945635"
 dependencies = [
  "cc",
  "cmake",

data/ext/xberg_rb/native/Cargo.toml

@@ -1,11 +1,11 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:440428bf83c731f3abfb9c84471777c863d1e31fe5499be1d6701d23053cb560
+# alef:hash:8ba9facc6695d874a10b55c3de0e9a1c523b385cc9865bdab6cf64cf399696d6
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
 [package]
 name = "xberg-rb"
-version = "1.2.8"
+version = "1.2.9"
 edition = "2024"
 license = "MIT"
 description = "High-performance document intelligence library"
@@ -82,13 +82,13 @@ tokio = { version = "1", features = ["rt-multi-thread"] }
 tracing = "0.1"
 
 [target.'cfg(all(target_os = "macos", target_arch = "x86_64"))'.dependencies]
-xberg = { version = "1.2.8", default-features = false, features = ["macos-intel-target", "simd-utf8"] }
+xberg = { version = "1.2.9", default-features = false, features = ["macos-intel-target", "simd-utf8"] }
 
 [target.'cfg(not(any(target_os = "windows", all(target_os = "macos", target_arch = "x86_64"))))'.dependencies]
-xberg = { version = "1.2.8", features = ["full", "pdf", "ocr", "paddle-ocr", "paddle-ocr-types", "layout-detection", "layout-types", "embeddings", "embedding-presets", "reranker", "reranker-presets", "sparse-embeddings", "sparse-embedding-presets", "late-interaction", "late-interaction-presets", "chunking", "keywords-yake", "keywords-rake", "language-detection", "html", "tree-sitter", "office", "email", "archives", "stopwords", "auto-rotate", "auto-rotate-types", "tokio-runtime", "api", "api-types", "mcp", "liter-llm", "quality", "svg", "transcription", "transcription-types", "classification", "captioning", "ner-onnx", "ner-llm", "diff", "markdown-footnotes", "redaction", "enrichment", "heuristics", "presets", "structured", "url-ingestion"] }
+xberg = { version = "1.2.9", features = ["full", "pdf", "ocr", "paddle-ocr", "paddle-ocr-types", "layout-detection", "layout-types", "embeddings", "embedding-presets", "reranker", "reranker-presets", "sparse-embeddings", "sparse-embedding-presets", "late-interaction", "late-interaction-presets", "chunking", "keywords-yake", "keywords-rake", "language-detection", "html", "tree-sitter", "office", "email", "archives", "stopwords", "auto-rotate", "auto-rotate-types", "tokio-runtime", "api", "api-types", "mcp", "liter-llm", "quality", "svg", "transcription", "transcription-types", "classification", "captioning", "ner-onnx", "ner-llm", "diff", "markdown-footnotes", "redaction", "enrichment", "heuristics", "presets", "structured", "url-ingestion"] }
 
 [target.'cfg(target_os = "windows")'.dependencies]
-xberg = { version = "1.2.8", default-features = false, features = ["windows-gnu-target", "simd-utf8"] }
+xberg = { version = "1.2.9", default-features = false, features = ["windows-gnu-target", "simd-utf8"] }
 
 # This crate deliberately does not use `[lints]` / `workspace = true`: its C-ABI /
 # PyO3 / napi / ext-php-rs / NIF boundary requires `unsafe` code, and the workspace's

data/ext/xberg_rb/src/lib.rs

@@ -1,5 +1,5 @@
 // This file is auto-generated by alef. DO NOT EDIT.
-// alef:hash:f1a359a24f246b68ebe2717a124008b9358925ba194cc8cb713eb90df7dc417d
+// alef:hash:9fd93c9044f1bedf7b944452cf36003a6468e2f40f9fe5a1d4e78bea797a60a3
 // Re-generate with: alef generate
 #![allow(dead_code, unused_imports, unused_variables)]
 #![allow(
@@ -11754,6 +11754,9 @@ pub struct ExtractionConfig {
     ocr_strategy: OcrStrategy,
     force_ocr_pages: Option<Vec<u32>>,
     disable_ocr: bool,
+    ocr_near_empty_fallback: Option<bool>,
+    ocr_scanned_page_quality_gate: Option<bool>,
+    ocr_embedded_images: Option<bool>,
     chunking: Option<ChunkingConfig>,
     content_filter: Option<ContentFilterConfig>,
     images: Option<ImageExtractionConfig>,
@@ -11920,6 +11923,42 @@ impl ExtractionConfig {
                 })?,
                 None => false,
             },
+            ocr_near_empty_fallback: match kwargs
+                .get(ruby.to_symbol("ocr_near_empty_fallback"))
+                .filter(|v| !v.is_nil())
+            {
+                Some(v) => Some(bool::try_convert(v).map_err(|e| {
+                    magnus::Error::new(
+                        unsafe { magnus::Ruby::get_unchecked() }.exception_type_error(),
+                        format!("invalid value for `ocr_near_empty_fallback`: {}", e),
+                    )
+                })?),
+                None => None,
+            },
+            ocr_scanned_page_quality_gate: match kwargs
+                .get(ruby.to_symbol("ocr_scanned_page_quality_gate"))
+                .filter(|v| !v.is_nil())
+            {
+                Some(v) => Some(bool::try_convert(v).map_err(|e| {
+                    magnus::Error::new(
+                        unsafe { magnus::Ruby::get_unchecked() }.exception_type_error(),
+                        format!("invalid value for `ocr_scanned_page_quality_gate`: {}", e),
+                    )
+                })?),
+                None => None,
+            },
+            ocr_embedded_images: match kwargs
+                .get(ruby.to_symbol("ocr_embedded_images"))
+                .filter(|v| !v.is_nil())
+            {
+                Some(v) => Some(bool::try_convert(v).map_err(|e| {
+                    magnus::Error::new(
+                        unsafe { magnus::Ruby::get_unchecked() }.exception_type_error(),
+                        format!("invalid value for `ocr_embedded_images`: {}", e),
+                    )
+                })?),
+                None => None,
+            },
             chunking: match kwargs.get(ruby.to_symbol("chunking")).filter(|v| !v.is_nil()) {
                 Some(v) => Some(ChunkingConfig::try_convert(v).map_err(|e| {
                     magnus::Error::new(
@@ -12376,6 +12415,18 @@ impl ExtractionConfig {
         self.disable_ocr
     }
 
+    fn ocr_near_empty_fallback(&self) -> Option<bool> {
+        self.ocr_near_empty_fallback
+    }
+
+    fn ocr_scanned_page_quality_gate(&self) -> Option<bool> {
+        self.ocr_scanned_page_quality_gate
+    }
+
+    fn ocr_embedded_images(&self) -> Option<bool> {
+        self.ocr_embedded_images
+    }
+
     fn chunking(&self) -> Option<ChunkingConfig> {
         self.chunking.clone()
     }
@@ -12578,6 +12629,12 @@ impl ExtractionConfig {
 
             disable_ocr: self.disable_ocr,
 
+            ocr_near_empty_fallback: self.ocr_near_empty_fallback,
+
+            ocr_scanned_page_quality_gate: self.ocr_scanned_page_quality_gate,
+
+            ocr_embedded_images: self.ocr_embedded_images,
+
             chunking: self.chunking.clone().map(Into::into),
 
             content_filter: self.content_filter.clone().map(Into::into),
@@ -12682,6 +12739,12 @@ impl ExtractionConfig {
 
             disable_ocr: self.disable_ocr,
 
+            ocr_near_empty_fallback: self.ocr_near_empty_fallback,
+
+            ocr_scanned_page_quality_gate: self.ocr_scanned_page_quality_gate,
+
+            ocr_embedded_images: self.ocr_embedded_images,
+
             chunking: self.chunking.clone().map(Into::into),
 
             content_filter: self.content_filter.clone().map(Into::into),
@@ -12780,6 +12843,12 @@ impl ExtractionConfig {
 
             disable_ocr: self.disable_ocr,
 
+            ocr_near_empty_fallback: self.ocr_near_empty_fallback,
+
+            ocr_scanned_page_quality_gate: self.ocr_scanned_page_quality_gate,
+
+            ocr_embedded_images: self.ocr_embedded_images,
+
             chunking: self.chunking.clone().map(Into::into),
 
             content_filter: self.content_filter.clone().map(Into::into),
@@ -12878,6 +12947,12 @@ impl ExtractionConfig {
 
             disable_ocr: self.disable_ocr,
 
+            ocr_near_empty_fallback: self.ocr_near_empty_fallback,
+
+            ocr_scanned_page_quality_gate: self.ocr_scanned_page_quality_gate,
+
+            ocr_embedded_images: self.ocr_embedded_images,
+
             chunking: self.chunking.clone().map(Into::into),
 
             content_filter: self.content_filter.clone().map(Into::into),
@@ -12976,6 +13051,12 @@ impl ExtractionConfig {
 
             disable_ocr: self.disable_ocr,
 
+            ocr_near_empty_fallback: self.ocr_near_empty_fallback,
+
+            ocr_scanned_page_quality_gate: self.ocr_scanned_page_quality_gate,
+
+            ocr_embedded_images: self.ocr_embedded_images,
+
             chunking: self.chunking.clone().map(Into::into),
 
             content_filter: self.content_filter.clone().map(Into::into),
@@ -49044,6 +49125,9 @@ impl From<ExtractionConfig> for xberg::ExtractionConfig {
             ocr_strategy: val.ocr_strategy.into(),
             force_ocr_pages: val.force_ocr_pages.map(|v| v.into_iter().collect()),
             disable_ocr: val.disable_ocr,
+            ocr_near_empty_fallback: val.ocr_near_empty_fallback,
+            ocr_scanned_page_quality_gate: val.ocr_scanned_page_quality_gate,
+            ocr_embedded_images: val.ocr_embedded_images,
             chunking: val.chunking.map(Into::into),
             content_filter: val.content_filter.map(Into::into),
             images: val.images.map(Into::into),
@@ -49112,6 +49196,9 @@ impl From<xberg::ExtractionConfig> for ExtractionConfig {
             ocr_strategy: val.ocr_strategy.into(),
             force_ocr_pages: val.force_ocr_pages.map(|v| v.into_iter().collect()),
             disable_ocr: val.disable_ocr,
+            ocr_near_empty_fallback: val.ocr_near_empty_fallback,
+            ocr_scanned_page_quality_gate: val.ocr_scanned_page_quality_gate,
+            ocr_embedded_images: val.ocr_embedded_images,
             chunking: val.chunking.map(Into::into),
             content_filter: val.content_filter.map(Into::into),
             images: val.images.map(Into::into),
@@ -57607,6 +57694,18 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("disable_ocr", method!(ExtractionConfig::disable_ocr, 0))?;
 
+    class.define_method(
+        "ocr_near_empty_fallback",
+        method!(ExtractionConfig::ocr_near_empty_fallback, 0),
+    )?;
+
+    class.define_method(
+        "ocr_scanned_page_quality_gate",
+        method!(ExtractionConfig::ocr_scanned_page_quality_gate, 0),
+    )?;
+
+    class.define_method("ocr_embedded_images", method!(ExtractionConfig::ocr_embedded_images, 0))?;
+
     class.define_method("chunking", method!(ExtractionConfig::chunking, 0))?;
 
     class.define_method("content_filter", method!(ExtractionConfig::content_filter, 0))?;

data/lib/xberg/version.rb

@@ -1,10 +1,10 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:c9baad720d0abc8473f6b123d41344001f49b707e150fe40599e7bad1ff6fc0d
+# alef:hash:af0361b6e9fb7d4844fed422cc5ad28ac7fdea52beb041fa1a3ab043aec2d450
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
 
 module Xberg
   ## The version string for this package.
-  VERSION = "1.2.8"
+  VERSION = "1.2.9"
 end

data/sig/types.rbs

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:4cc11f4292def27becadc457989d24ea976f6ddcaaa87e60c01e9796d7722c80
+# alef:hash:060fee672abbcdc0cf88ae2c4c85dc76420919f28c17b4cd7ccbd449f6d319b1
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
@@ -934,6 +934,9 @@ def initialize: (?data: String, ?format: String, ?image_index: Integer, ?page_nu
         attr_reader ocr_strategy: OcrStrategy
     attr_reader force_ocr_pages: Array[Integer]?
         attr_reader disable_ocr: bool
+        attr_reader ocr_near_empty_fallback: bool?
+        attr_reader ocr_scanned_page_quality_gate: bool?
+        attr_reader ocr_embedded_images: bool?
         attr_reader chunking: ChunkingConfig?
         attr_reader content_filter: ContentFilterConfig?
         attr_reader images: ImageExtractionConfig?
@@ -979,7 +982,7 @@ 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: 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 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, ?ocr_near_empty_fallback: bool, ?ocr_scanned_page_quality_gate: bool, ?ocr_embedded_images: 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

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: xberg
 version: !ruby/object:Gem::Version
-  version: 1.2.8
+  version: 1.2.9
 platform: ruby
 authors:
 - Na'aman Hirschfeld <naaman@xberg.io>
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-09-23 00:00:00.000000000 Z
+date: 2026-09-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys