xberg 1.0.0.pre.rc.19 → 1.0.0.pre.rc.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c6fbf86d3a69b56798c048b6a235d801bc89fece79979923e42c8efab543d615
-  data.tar.gz: 0defd8a51ae10bd5169d78abb11a63c3478e677cda49fc9f5b302d6bbca641ae
+  metadata.gz: f357179c1ad3a2d78fc33a5565eea6ad47070d2dc852c90f671f021e50f877e7
+  data.tar.gz: cde1eeb18c1d1448e1a5685f876cf318a4832ebcf711cfa4521d139dd9864678
 SHA512:
-  metadata.gz: a5cb3b42394c7cb55c8da00efc8d84156473b13dd48e65b6af7a134f07bbf06c3baba6381a40fc37708b917f796014f2c8998ec11e485ece5da2bffa5d2d77e3
-  data.tar.gz: bef2169dd05080bbc5b66545bcb83e529d50d113fefbb3ffb548636ba18e2370fceb905a7bd4f57b20897cb5c0f275a836766251a6af47be931b0942a5d7386d
+  metadata.gz: aa06b3f66bb2e4d97c27b4b8bce0c1de267d835441848be857aef18da451964cf81fb6488ed419f49322232d30d565cd89f35a2cd8cea8d1d8da1605132ed8ba
+  data.tar.gz: 41f4ebdc94c4e5d7ae3f3c10968f66612d25cf119edfdb9acbb435efef4a32b488e8c02992a601e47f60bd8298858617aace8faf34519c5083ed94baaf74e2b0

data/ext/xberg_rb/native/Cargo.lock

@@ -8233,9 +8233,9 @@ dependencies = [
 
 [[package]]
 name = "xberg"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a8ac7f4216cd6167b24ca0d5f98ba995f7db0458019930880a0208111a4c4c7d"
+checksum = "b8cf06b497fa85f72e34dac11316ae36551db41a59e93743fbdef4f0690de92c"
 dependencies = [
  "ahash 0.8.12",
  "async-trait",
@@ -8352,9 +8352,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-gliner"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "226ce0e7469b500ba3e565a7f52390fc9a5f38c0145053adebd8650992321754"
+checksum = "dcd5a0a6088686c158f92618a2639a3107794c4018f4f9633d9a733182287257"
 dependencies = [
  "ndarray 0.17.2",
  "ort",
@@ -8366,9 +8366,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-libheif"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "609582e69acd5e76e1810b23491bb4f25dd53082163f781a440e5e72b662035a"
+checksum = "cf76ec8a52f0f671435a8b0cc5f9ec687ff8741c94e04b82c85008f24c835b84"
 dependencies = [
  "cfg-if",
  "enumn",
@@ -8379,9 +8379,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-paddle-ocr"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a203923e55dadb758ca094c269cc31b683bac45854317674f01c8b64ccaa3d48"
+checksum = "4ff214bb615d895a47e3642e7af5043a1741457fa18639428696f7808d9587f6"
 dependencies = [
  "geo-clipper",
  "geo-types",
@@ -8396,7 +8396,7 @@ dependencies = [
 
 [[package]]
 name = "xberg-rb"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 dependencies = [
  "async-trait",
  "crawlberg",
@@ -8410,9 +8410,9 @@ dependencies = [
 
 [[package]]
 name = "xberg-tesseract"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c2d2d55bd70d7fb88dbdb8eae31d7cb46e77f09fa8deae2145463ffa802c5e26"
+checksum = "509d7350ecd7a08a54be7579020ee890a98a8dee545d3e6dc33003b6677daf9a"
 dependencies = [
  "cc",
  "cmake",

data/ext/xberg_rb/native/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "xberg-rb"
-version = "1.0.0-rc.19"
+version = "1.0.0-rc.20"
 edition = "2024"
 license = "MIT"
 description = "High-performance document intelligence library"
@@ -62,10 +62,10 @@ serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 tokio = { version = "1", features = ["rt-multi-thread"] }
 [target.'cfg(not(any(target_os = "windows", all(target_os = "macos", target_arch = "x86_64"))))'.dependencies]
-xberg = { version = "1.0.0-rc.19", 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 = "1.0.0-rc.20", 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 = "1.0.0-rc.19", features = ["windows-target"] }
+xberg = { version = "1.0.0-rc.20", features = ["windows-target"] }
 
 [target.'cfg(all(target_os = "macos", target_arch = "x86_64"))'.dependencies]
-xberg = { version = "1.0.0-rc.19", features = ["macos-intel-target"] }
+xberg = { version = "1.0.0-rc.20", features = ["macos-intel-target"] }

data/ext/xberg_rb/src/lib.rs

@@ -1,5 +1,5 @@
 // This file is auto-generated by alef. DO NOT EDIT.
-// alef:hash:a83cf6d4ae5b7f677ed5cf28d09d3872838586a6c81acc7eccf6c3b5d0507b75
+// alef:hash:597bd1ea631d4ac6d358d121dc5791450b5d9aaaabe53fda1e43bef2af72e4ac
 // Re-generate with: alef generate
 #![allow(dead_code, unused_imports, unused_variables)]
 #![allow(
@@ -558,6 +558,7 @@ pub struct ExtractionConfig {
     enable_quality_processing: bool,
     ocr: Option<OcrConfig>,
     force_ocr: bool,
+    ocr_strategy: OcrStrategy,
     force_ocr_pages: Option<Vec<u32>>,
     disable_ocr: bool,
     chunking: Option<ChunkingConfig>,
@@ -654,6 +655,10 @@ impl ExtractionConfig {
                 .get(ruby.to_symbol("force_ocr"))
                 .and_then(|v| bool::try_convert(v).ok())
                 .unwrap_or(false),
+            ocr_strategy: kwargs
+                .get(ruby.to_symbol("ocr_strategy"))
+                .and_then(|v| OcrStrategy::try_convert(v).ok())
+                .unwrap_or(OcrStrategy::Auto),
             force_ocr_pages: kwargs
                 .get(ruby.to_symbol("force_ocr_pages"))
                 .and_then(|v| <Vec<u32>>::try_convert(v).ok()),
@@ -795,6 +800,10 @@ impl ExtractionConfig {
         self.force_ocr
     }
 
+    fn ocr_strategy(&self) -> OcrStrategy {
+        self.ocr_strategy.clone()
+    }
+
     fn force_ocr_pages(&self) -> Option<Vec<u32>> {
         self.force_ocr_pages.clone()
     }
@@ -958,6 +967,8 @@ impl ExtractionConfig {
 
             force_ocr: self.force_ocr,
 
+            ocr_strategy: self.ocr_strategy.clone().into(),
+
             force_ocr_pages: self.force_ocr_pages.clone(),
 
             disable_ocr: self.disable_ocr,
@@ -1038,6 +1049,8 @@ impl ExtractionConfig {
 
             force_ocr: self.force_ocr,
 
+            ocr_strategy: self.ocr_strategy.clone().into(),
+
             force_ocr_pages: self.force_ocr_pages.clone(),
 
             disable_ocr: self.disable_ocr,
@@ -1115,6 +1128,7 @@ pub struct FileExtractionConfig {
     enable_quality_processing: Option<bool>,
     ocr: Option<OcrConfig>,
     force_ocr: Option<bool>,
+    ocr_strategy: Option<OcrStrategy>,
     force_ocr_pages: Option<Vec<u32>>,
     disable_ocr: Option<bool>,
     chunking: Option<ChunkingConfig>,
@@ -1195,6 +1209,9 @@ impl FileExtractionConfig {
             force_ocr: kwargs
                 .get(ruby.to_symbol("force_ocr"))
                 .and_then(|v| bool::try_convert(v).ok()),
+            ocr_strategy: kwargs
+                .get(ruby.to_symbol("ocr_strategy"))
+                .and_then(|v| OcrStrategy::try_convert(v).ok()),
             force_ocr_pages: kwargs
                 .get(ruby.to_symbol("force_ocr_pages"))
                 .and_then(|v| <Vec<u32>>::try_convert(v).ok()),
@@ -1294,6 +1311,10 @@ impl FileExtractionConfig {
         self.force_ocr
     }
 
+    fn ocr_strategy(&self) -> Option<OcrStrategy> {
+        self.ocr_strategy.clone()
+    }
+
     fn force_ocr_pages(&self) -> Option<Vec<u32>> {
         self.force_ocr_pages.clone()
     }
@@ -3786,7 +3807,6 @@ pub struct HierarchyConfig {
     enabled: bool,
     k_clusters: usize,
     include_bbox: bool,
-    ocr_coverage_threshold: Option<f32>,
 }
 
 unsafe impl IntoValueFromNative for HierarchyConfig {}
@@ -3842,9 +3862,6 @@ impl HierarchyConfig {
                 .get(ruby.to_symbol("include_bbox"))
                 .and_then(|v| bool::try_convert(v).ok())
                 .unwrap_or(true),
-            ocr_coverage_threshold: kwargs
-                .get(ruby.to_symbol("ocr_coverage_threshold"))
-                .and_then(|v| f32::try_convert(v).ok()),
         })
     }
 
@@ -3859,10 +3876,6 @@ impl HierarchyConfig {
     fn include_bbox(&self) -> bool {
         self.include_bbox
     }
-
-    fn ocr_coverage_threshold(&self) -> Option<f32> {
-        self.ocr_coverage_threshold
-    }
 }
 
 #[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
@@ -4894,7 +4907,7 @@ impl TranscriptionConfig {
             verify_hash: kwargs
                 .get(ruby.to_symbol("verify_hash"))
                 .and_then(|v| bool::try_convert(v).ok())
-                .unwrap_or(true),
+                .unwrap_or(false),
         })
     }
 
@@ -15711,6 +15724,75 @@ impl CellChange {
     }
 }
 
+#[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
+#[magnus::wrap(class = "Xberg::PropertyChange")]
+pub struct PropertyChange {
+    name: String,
+    from: Option<String>,
+    to: Option<String>,
+}
+
+unsafe impl IntoValueFromNative for PropertyChange {}
+
+impl magnus::TryConvert for PropertyChange {
+    fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
+        if let Ok(r) = <&PropertyChange as magnus::TryConvert>::try_convert(val) {
+            return Ok(r.clone());
+        }
+        let json_str: String = if let Ok(s) = <String as magnus::TryConvert>::try_convert(val) {
+            s
+        } else {
+            val.funcall::<_, _, String>("to_json", ()).map_err(|e| {
+                magnus::Error::new(
+                    unsafe { magnus::Ruby::get_unchecked() }.exception_type_error(),
+                    format!("no implicit conversion into PropertyChange: {}", e),
+                )
+            })?
+        };
+        serde_json::from_str::<PropertyChange>(&json_str).map_err(|e| {
+            magnus::Error::new(
+                unsafe { magnus::Ruby::get_unchecked() }.exception_type_error(),
+                format!("failed to deserialize PropertyChange: {}", e),
+            )
+        })
+    }
+}
+
+unsafe impl TryConvertOwned for PropertyChange {}
+
+impl PropertyChange {
+    fn new(args: &[magnus::Value]) -> Result<Self, magnus::Error> {
+        let ruby = unsafe { magnus::Ruby::get_unchecked() };
+        let args = magnus::scan_args::scan_args::<(), (Option<magnus::RHash>,), (), (), (), ()>(args)?;
+        let (kwargs_opt,) = args.optional;
+        let kwargs = kwargs_opt.unwrap_or_else(|| ruby.hash_new());
+        Ok(Self {
+            name: kwargs
+                .get(ruby.to_symbol("name"))
+                .and_then(|v| String::try_convert(v).ok())
+                .unwrap_or_default(),
+            from: kwargs
+                .get(ruby.to_symbol("from"))
+                .and_then(|v| String::try_convert(v).ok()),
+            to: kwargs
+                .get(ruby.to_symbol("to"))
+                .and_then(|v| String::try_convert(v).ok()),
+        })
+    }
+
+    fn name(&self) -> String {
+        self.name.clone()
+    }
+
+    fn from(&self) -> Option<String> {
+        self.from.clone()
+    }
+
+    fn to(&self) -> Option<String> {
+        self.to.clone()
+    }
+}
+
 #[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
 #[magnus::wrap(class = "Xberg::DocumentRevision")]
 pub struct DocumentRevision {
@@ -15822,6 +15904,7 @@ impl DocumentRevision {
 pub struct RevisionDelta {
     content: Vec<DiffLine>,
     table_changes: Vec<CellChange>,
+    property_changes: Vec<PropertyChange>,
 }
 
 unsafe impl IntoValueFromNative for RevisionDelta {}
@@ -15873,6 +15956,10 @@ impl RevisionDelta {
                 .get(ruby.to_symbol("table_changes"))
                 .and_then(|v| <Vec<CellChange>>::try_convert(v).ok())
                 .unwrap_or_default(),
+            property_changes: kwargs
+                .get(ruby.to_symbol("property_changes"))
+                .and_then(|v| <Vec<PropertyChange>>::try_convert(v).ok())
+                .unwrap_or_default(),
         })
     }
 
@@ -15883,6 +15970,10 @@ impl RevisionDelta {
     fn table_changes(&self) -> Vec<CellChange> {
         self.table_changes.clone()
     }
+
+    fn property_changes(&self) -> Vec<PropertyChange> {
+        self.property_changes.clone()
+    }
 }
 
 #[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
@@ -20319,6 +20410,8 @@ pub struct PdfMetadata {
     width: Option<i64>,
     height: Option<i64>,
     page_count: Option<u32>,
+    scanned_confidence: Option<f32>,
+    scanned_pages: Option<Vec<u32>>,
 }
 
 unsafe impl IntoValueFromNative for PdfMetadata {}
@@ -20380,6 +20473,12 @@ impl PdfMetadata {
             page_count: kwargs
                 .get(ruby.to_symbol("page_count"))
                 .and_then(|v| u32::try_convert(v).ok()),
+            scanned_confidence: kwargs
+                .get(ruby.to_symbol("scanned_confidence"))
+                .and_then(|v| f32::try_convert(v).ok()),
+            scanned_pages: kwargs
+                .get(ruby.to_symbol("scanned_pages"))
+                .and_then(|v| <Vec<u32>>::try_convert(v).ok()),
         })
     }
 
@@ -20406,6 +20505,14 @@ impl PdfMetadata {
     fn page_count(&self) -> Option<u32> {
         self.page_count
     }
+
+    fn scanned_confidence(&self) -> Option<f32> {
+        self.scanned_confidence
+    }
+
+    fn scanned_pages(&self) -> Option<Vec<u32>> {
+        self.scanned_pages.clone()
+    }
 }
 
 #[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
@@ -22126,6 +22233,65 @@ unsafe impl IntoValueFromNative for VlmFallbackPolicy {}
 unsafe impl TryConvertOwned for VlmFallbackPolicy {}
 impl VlmFallbackPolicy {}
 
+#[derive(Clone, Debug, serde::Serialize, serde::Deserialize)]
+#[serde(tag = "mode")]
+#[serde(rename_all = "snake_case")]
+pub enum OcrStrategy {
+    Auto,
+    ScannedPages { min_confidence: f64 },
+}
+
+impl Default for OcrStrategy {
+    fn default() -> Self {
+        Self::Auto
+    }
+}
+
+impl magnus::IntoValue for OcrStrategy {
+    fn into_value_with(self, handle: &Ruby) -> magnus::Value {
+        match serde_json::to_value(&self) {
+            Ok(v) => json_to_ruby(handle, v),
+            Err(_) => handle.qnil().into_value_with(handle),
+        }
+    }
+}
+
+impl magnus::TryConvert for OcrStrategy {
+    fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
+        // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
+        // For unit enums or when passed as a string, fall back to string-based conversion.
+        let json_str: String = if let Ok(s) = <String as magnus::TryConvert>::try_convert(val) {
+            s
+        } else {
+            val.funcall::<_, _, String>("to_json", ()).map_err(|e| {
+                magnus::Error::new(
+                    unsafe { Ruby::get_unchecked() }.exception_type_error(),
+                    format!("no implicit conversion into OcrStrategy: {}", e),
+                )
+            })?
+        };
+        // Try deserializing as JSON first (handles JSON strings like "\"markdown\"" or "{\"click\":{\"selector\":\"...\"}}\"")
+        // For internally-tagged enums, a bare variant string is wrapped as {"<tag>": value}.
+        // If that fails, try treating it as a plain string value and wrap in quotes
+        // If both fail, try as Custom variant (for untagged enum support)
+        serde_json::from_str(&json_str)
+            .or_else(|_| serde_json::from_value(serde_json::json!({ "mode": json_str })))
+            .or_else(|_| serde_json::from_str(&format!("\"{json_str}\"")))
+            .or_else(|_| {
+                // Try as a JSON string for Custom variant (untagged enums accept any remaining value)
+                match serde_json::to_value(&json_str) {
+                    Ok(val) => serde_json::from_value(val),
+                    Err(e) => Err(e),
+                }
+            })
+            .map_err(|e| magnus::Error::new(unsafe { Ruby::get_unchecked() }.exception_type_error(), e.to_string()))
+    }
+}
+
+unsafe impl IntoValueFromNative for OcrStrategy {}
+unsafe impl TryConvertOwned for OcrStrategy {}
+impl OcrStrategy {}
+
 #[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum TableChunkingMode {
@@ -29080,6 +29246,7 @@ impl From<ExtractionConfig> for xberg::ExtractionConfig {
             enable_quality_processing: val.enable_quality_processing,
             ocr: val.ocr.map(Into::into),
             force_ocr: val.force_ocr,
+            ocr_strategy: val.ocr_strategy.into(),
             force_ocr_pages: val.force_ocr_pages.map(|v| v.into_iter().collect()),
             disable_ocr: val.disable_ocr,
             chunking: val.chunking.map(Into::into),
@@ -29131,6 +29298,7 @@ impl From<xberg::ExtractionConfig> for ExtractionConfig {
             enable_quality_processing: val.enable_quality_processing,
             ocr: val.ocr.map(Into::into),
             force_ocr: val.force_ocr,
+            ocr_strategy: val.ocr_strategy.into(),
             force_ocr_pages: val.force_ocr_pages.map(|v| v.into_iter().collect()),
             disable_ocr: val.disable_ocr,
             chunking: val.chunking.map(Into::into),
@@ -29181,6 +29349,7 @@ impl From<FileExtractionConfig> for xberg::FileExtractionConfig {
             enable_quality_processing: val.enable_quality_processing,
             ocr: val.ocr.map(Into::into),
             force_ocr: val.force_ocr,
+            ocr_strategy: val.ocr_strategy.map(Into::into),
             force_ocr_pages: val.force_ocr_pages.map(|v| v.into_iter().collect()),
             disable_ocr: val.disable_ocr,
             chunking: val.chunking.map(Into::into),
@@ -29221,6 +29390,7 @@ impl From<xberg::FileExtractionConfig> for FileExtractionConfig {
             enable_quality_processing: val.enable_quality_processing,
             ocr: val.ocr.map(Into::into),
             force_ocr: val.force_ocr,
+            ocr_strategy: val.ocr_strategy.map(Into::into),
             force_ocr_pages: val.force_ocr_pages.map(|v| v.into_iter().collect()),
             disable_ocr: val.disable_ocr,
             chunking: val.chunking.map(Into::into),
@@ -29935,7 +30105,6 @@ impl From<HierarchyConfig> for xberg::HierarchyConfig {
             enabled: val.enabled,
             k_clusters: val.k_clusters,
             include_bbox: val.include_bbox,
-            ocr_coverage_threshold: val.ocr_coverage_threshold,
             ..Default::default()
         }
     }
@@ -29948,7 +30117,6 @@ impl From<xberg::HierarchyConfig> for HierarchyConfig {
             enabled: val.enabled,
             k_clusters: val.k_clusters,
             include_bbox: val.include_bbox,
-            ocr_coverage_threshold: val.ocr_coverage_threshold,
         }
     }
 }
@@ -33102,6 +33270,28 @@ impl From<xberg::CellChange> for CellChange {
     }
 }
 
+#[allow(clippy::redundant_closure, clippy::useless_conversion)]
+impl From<PropertyChange> for xberg::PropertyChange {
+    fn from(val: PropertyChange) -> Self {
+        Self {
+            name: val.name,
+            from: val.from,
+            to: val.to,
+        }
+    }
+}
+
+#[allow(clippy::redundant_closure, clippy::useless_conversion)]
+impl From<xberg::PropertyChange> for PropertyChange {
+    fn from(val: xberg::PropertyChange) -> Self {
+        Self {
+            name: val.name.to_string(),
+            from: val.from.map(|v| v.to_string()),
+            to: val.to.map(|v| v.to_string()),
+        }
+    }
+}
+
 #[allow(clippy::redundant_closure, clippy::useless_conversion)]
 impl From<DocumentRevision> for xberg::DocumentRevision {
     fn from(val: DocumentRevision) -> Self {
@@ -33137,6 +33327,7 @@ impl From<RevisionDelta> for xberg::RevisionDelta {
         Self {
             content: val.content.into_iter().map(Into::into).collect(),
             table_changes: val.table_changes.into_iter().map(Into::into).collect(),
+            property_changes: val.property_changes.into_iter().map(Into::into).collect(),
             ..Default::default()
         }
     }
@@ -33148,6 +33339,7 @@ impl From<xberg::RevisionDelta> for RevisionDelta {
         Self {
             content: val.content.into_iter().map(Into::into).collect(),
             table_changes: val.table_changes.into_iter().map(Into::into).collect(),
+            property_changes: val.property_changes.into_iter().map(Into::into).collect(),
         }
     }
 }
@@ -33984,6 +34176,8 @@ impl From<PdfMetadata> for xberg::pdf::metadata::PdfMetadata {
             width: val.width,
             height: val.height,
             page_count: val.page_count,
+            scanned_confidence: val.scanned_confidence,
+            scanned_pages: val.scanned_pages.map(|v| v.into_iter().collect()),
             ..Default::default()
         }
     }
@@ -33999,6 +34193,8 @@ impl From<xberg::pdf::metadata::PdfMetadata> for PdfMetadata {
             width: val.width,
             height: val.height,
             page_count: val.page_count,
+            scanned_confidence: val.scanned_confidence,
+            scanned_pages: val.scanned_pages.map(|v| v.into_iter().collect()),
         }
     }
 }
@@ -34613,6 +34809,28 @@ impl From<xberg::VlmFallbackPolicy> for VlmFallbackPolicy {
     }
 }
 
+impl From<OcrStrategy> for xberg::OcrStrategy {
+    fn from(val: OcrStrategy) -> Self {
+        match val {
+            OcrStrategy::Auto => Self::Auto,
+            OcrStrategy::ScannedPages { min_confidence } => Self::ScannedPages {
+                min_confidence: min_confidence,
+            },
+        }
+    }
+}
+
+impl From<xberg::OcrStrategy> for OcrStrategy {
+    fn from(val: xberg::OcrStrategy) -> Self {
+        match val {
+            xberg::OcrStrategy::Auto => Self::Auto,
+            xberg::OcrStrategy::ScannedPages { min_confidence } => Self::ScannedPages {
+                min_confidence: min_confidence,
+            },
+        }
+    }
+}
+
 impl From<TableChunkingMode> for xberg::TableChunkingMode {
     fn from(val: TableChunkingMode) -> Self {
         match val {
@@ -36370,6 +36588,8 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("force_ocr", method!(ExtractionConfig::force_ocr, 0))?;
 
+    class.define_method("ocr_strategy", method!(ExtractionConfig::ocr_strategy, 0))?;
+
     class.define_method("force_ocr_pages", method!(ExtractionConfig::force_ocr_pages, 0))?;
 
     class.define_method("disable_ocr", method!(ExtractionConfig::disable_ocr, 0))?;
@@ -36487,6 +36707,8 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("force_ocr", method!(FileExtractionConfig::force_ocr, 0))?;
 
+    class.define_method("ocr_strategy", method!(FileExtractionConfig::ocr_strategy, 0))?;
+
     class.define_method("force_ocr_pages", method!(FileExtractionConfig::force_ocr_pages, 0))?;
 
     class.define_method("disable_ocr", method!(FileExtractionConfig::disable_ocr, 0))?;
@@ -37045,11 +37267,6 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("include_bbox", method!(HierarchyConfig::include_bbox, 0))?;
 
-    class.define_method(
-        "ocr_coverage_threshold",
-        method!(HierarchyConfig::ocr_coverage_threshold, 0),
-    )?;
-
     let class = module.define_class("PostProcessorConfig", ruby.class_object())?;
 
     class.define_singleton_method("new", function!(PostProcessorConfig::new, -1))?;
@@ -39001,6 +39218,16 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("to", method!(CellChange::to, 0))?;
 
+    let class = module.define_class("PropertyChange", ruby.class_object())?;
+
+    class.define_singleton_method("new", function!(PropertyChange::new, -1))?;
+
+    class.define_method("name", method!(PropertyChange::name, 0))?;
+
+    class.define_method("from", method!(PropertyChange::from, 0))?;
+
+    class.define_method("to", method!(PropertyChange::to, 0))?;
+
     let class = module.define_class("DocumentRevision", ruby.class_object())?;
 
     class.define_singleton_method("new", function!(DocumentRevision::new, -1))?;
@@ -39025,6 +39252,8 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("table_changes", method!(RevisionDelta::table_changes, 0))?;
 
+    class.define_method("property_changes", method!(RevisionDelta::property_changes, 0))?;
+
     let class = module.define_class("DocumentSummary", ruby.class_object())?;
 
     class.define_singleton_method("new", function!(DocumentSummary::new, -1))?;
@@ -39706,6 +39935,10 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("page_count", method!(PdfMetadata::page_count, 0))?;
 
+    class.define_method("scanned_confidence", method!(PdfMetadata::scanned_confidence, 0))?;
+
+    class.define_method("scanned_pages", method!(PdfMetadata::scanned_pages, 0))?;
+
     let class = module.define_class("ProxyConfig", ruby.class_object())?;
 
     class.define_singleton_method("new", function!(ProxyConfig::new, -1))?;

data/lib/xberg/native.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:a83cf6d4ae5b7f677ed5cf28d09d3872838586a6c81acc7eccf6c3b5d0507b75
+# alef:hash:597bd1ea631d4ac6d358d121dc5791450b5d9aaaabe53fda1e43bef2af72e4ac
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 # frozen_string_literal: true
@@ -459,6 +459,81 @@ module Xberg
   end
 end
 
+module Xberg
+  # Which pages of a PDF get OCR'd when neither `force_ocr` nor `force_ocr_pages` applies.
+  module OcrStrategy
+    extend T::Helpers
+    extend T::Sig
+
+    interface!
+
+    # Dispatch from a Hash to the appropriate variant constructor.
+    # @param hash [Hash] with discriminator field and variant-specific fields
+    # @return [variant_class] an instance of the appropriate variant
+    sig { params(hash: T::Hash[T.untyped, T.untyped]).returns(T.untyped) }
+    def self.from_hash(hash)
+      discriminator = hash[:mode] || hash["mode"]
+      case discriminator
+      when "auto"
+        OcrStrategyAuto.from_hash(hash)
+      when "scanned_pages"
+        OcrStrategyScannedPages.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
+      end
+    end
+  end
+  ## OCR only when the native text layer fails a quality check (default).
+  ##
+  ## A scanner's invisible OCR sidecar passes that check, so scanned pages
+  ## carrying one are extracted natively. Use [`OcrStrategy::ScannedPages`]
+  ## to OCR them instead.
+  OcrStrategyAuto = Data.define do
+    include OcrStrategy
+    extend T::Sig
+
+    sig { returns(T::Boolean) }
+    def auto? = true
+    sig { returns(T::Boolean) }
+    def scanned_pages? = false
+    # @param hash [Hash] deserialized from the native extension
+    # @return [self]
+    sig { params(hash: T::Hash[T.untyped, T.untyped]).returns(T.attached_class) }
+    def self.from_hash(hash)
+      new
+    end
+  end
+  ## Additionally OCR every page that looks like a scan.
+  ##
+  ## Pages are graded on raster coverage, whether the text layer is invisible
+  ## or absent, the image codec, and the producer. Pages at or above
+  ## `min_confidence` are OCR'd; the rest keep native text and still go through
+  ## the `Auto` quality check.
+  ##
+  ## Detects that a text layer came from a scanner, not whether it is accurate,
+  ## so a page carrying a good sidecar is OCR'd too.
+  OcrStrategyScannedPages = Data.define(:min_confidence) do
+    include OcrStrategy
+    extend T::Sig
+
+    # Minimum scan confidence, in `[0.0, 1.0]`. Values outside the range are
+    # clamped. See [`DEFAULT_SCANNED_MIN_CONFIDENCE`] for how to pick one.
+    sig { returns(Float) }
+    # rubocop:disable Lint/UselessMethodDefinition
+    def min_confidence = super
+    sig { returns(T::Boolean) }
+    def auto? = false
+    sig { returns(T::Boolean) }
+    def scanned_pages? = true
+    # @param hash [Hash] deserialized from the native extension
+    # @return [self]
+    sig { params(hash: T::Hash[T.untyped, T.untyped]).returns(T.attached_class) }
+    def self.from_hash(hash)
+      new(min_confidence: hash[:min_confidence] || hash["min_confidence"])
+    end
+  end
+end
+
 module Xberg
   # How chunk size is measured.
   #

data/lib/xberg/version.rb

@@ -1,10 +1,10 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:a83cf6d4ae5b7f677ed5cf28d09d3872838586a6c81acc7eccf6c3b5d0507b75
+# alef:hash:597bd1ea631d4ac6d358d121dc5791450b5d9aaaabe53fda1e43bef2af72e4ac
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 # frozen_string_literal: true
 
 module Xberg
   ## The version string for this package.
-  VERSION = "1.0.0.pre.rc.19"
+  VERSION = "1.0.0.pre.rc.20"
 end

data/lib/xberg.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:a83cf6d4ae5b7f677ed5cf28d09d3872838586a6c81acc7eccf6c3b5d0507b75
+# alef:hash:597bd1ea631d4ac6d358d121dc5791450b5d9aaaabe53fda1e43bef2af72e4ac
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 # frozen_string_literal: true

data/sig/types.rbs

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:a83cf6d4ae5b7f677ed5cf28d09d3872838586a6c81acc7eccf6c3b5d0507b75
+# alef:hash:597bd1ea631d4ac6d358d121dc5791450b5d9aaaabe53fda1e43bef2af72e4ac
 # To regenerate: alef generate
 # To verify freshness: alef verify --exit-code
 
@@ -64,6 +64,7 @@ def initialize: (?prompt_template: String, labels: Array[String], multi_label: b
         attr_accessor enable_quality_processing: bool?
         attr_accessor ocr: OcrConfig?
         attr_accessor force_ocr: bool?
+        attr_accessor ocr_strategy: OcrStrategy?
     attr_accessor force_ocr_pages: Array[Integer]?
         attr_accessor disable_ocr: bool?
         attr_accessor chunking: ChunkingConfig?
@@ -103,7 +104,7 @@ def initialize: (?prompt_template: String, labels: Array[String], multi_label: b
         attr_accessor captioning: CaptioningConfig?
         attr_accessor qr_codes: bool?
 
-def initialize: (?use_cache: bool, ?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?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_output: HtmlOutputConfig, ?extraction_timeout_secs: Integer, ?max_concurrent_extractions: Integer, ?result_format: ResultFormat, ?security_limits: SecurityLimits, ?max_embedded_file_bytes: Integer, ?output_format: OutputFormat, ?jupyter_cell_rendering: JupyterCellRendering, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?use_layout_for_markdown: bool, ?include_document_structure: bool, ?acceleration: AccelerationConfig, ?cache_namespace: String, ?cache_ttl_secs: Integer, ?email: EmailConfig, ?url: UrlExtractionConfig, ?max_archive_depth: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
+def initialize: (?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_output: HtmlOutputConfig, ?extraction_timeout_secs: Integer, ?max_concurrent_extractions: Integer, ?result_format: ResultFormat, ?security_limits: SecurityLimits, ?max_embedded_file_bytes: Integer, ?output_format: OutputFormat, ?jupyter_cell_rendering: JupyterCellRendering, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?use_layout_for_markdown: bool, ?include_document_structure: bool, ?acceleration: AccelerationConfig, ?cache_namespace: String, ?cache_ttl_secs: Integer, ?email: EmailConfig, ?url: UrlExtractionConfig, ?max_archive_depth: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
     def needs_image_data: () -> bool
     def needs_image_processing: () -> bool
     def self.default: () -> ExtractionConfig
@@ -113,6 +114,7 @@ def initialize: (?use_cache: bool, ?enable_quality_processing: bool, ?ocr: OcrCo
         attr_accessor enable_quality_processing: bool?
         attr_accessor ocr: OcrConfig?
         attr_accessor force_ocr: bool?
+        attr_accessor ocr_strategy: OcrStrategy?
     attr_accessor force_ocr_pages: Array[Integer]?
         attr_accessor disable_ocr: bool?
         attr_accessor chunking: ChunkingConfig?
@@ -142,7 +144,7 @@ def initialize: (?use_cache: bool, ?enable_quality_processing: bool, ?ocr: OcrCo
         attr_accessor captioning: CaptioningConfig?
         attr_accessor qr_codes: bool?
 
-def initialize: (?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?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_output: HtmlOutputConfig, ?result_format: ResultFormat, ?output_format: OutputFormat, ?include_document_structure: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?timeout_secs: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?url: UrlExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
+def initialize: (?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_output: HtmlOutputConfig, ?result_format: ResultFormat, ?output_format: OutputFormat, ?include_document_structure: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?timeout_secs: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?url: UrlExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void
     end
 
     class SvgOptions
@@ -417,9 +419,8 @@ def initialize: (?extract_images: bool, ?extract_tables: bool, ?passwords: Array
         attr_accessor enabled: bool?
         attr_accessor k_clusters: Integer?
         attr_accessor include_bbox: bool?
-        attr_accessor ocr_coverage_threshold: Float?
 
-    def initialize: (?enabled: bool, ?k_clusters: Integer, ?include_bbox: bool, ?ocr_coverage_threshold: Float) -> void
+    def initialize: (?enabled: bool, ?k_clusters: Integer, ?include_bbox: bool) -> void
     def self.default: () -> HierarchyConfig
     end
 
@@ -1704,6 +1705,14 @@ def initialize: (findings: Array[RedactionFinding], total_redacted: Integer) ->
     def initialize: (row: Integer, col: Integer, from: String, to: String) -> void
     end
 
+    class PropertyChange
+        attr_reader name: String
+        attr_reader from: String
+        attr_reader to: String
+
+    def initialize: (name: String, ?from: String, ?to: String) -> void
+    end
+
     class DocumentRevision
         attr_reader revision_id: String
         attr_reader author: String
@@ -1718,8 +1727,9 @@ def initialize: (findings: Array[RedactionFinding], total_redacted: Integer) ->
     class RevisionDelta
     attr_accessor content: Array[DiffLine]?
     attr_accessor table_changes: Array[CellChange]?
+    attr_accessor property_changes: Array[PropertyChange]?
 
-def initialize: (?content: Array[DiffLine], ?table_changes: Array[CellChange]) -> void
+def initialize: (?content: Array[DiffLine], ?table_changes: Array[CellChange], ?property_changes: Array[PropertyChange]) -> void
     end
 
     class DocumentSummary
@@ -2182,8 +2192,10 @@ def initialize: (page_width: Integer, page_height: Integer, detections: Array[La
         attr_accessor width: Integer?
         attr_accessor height: Integer?
         attr_accessor page_count: Integer?
+        attr_accessor scanned_confidence: Float?
+    attr_accessor scanned_pages: Array[Integer]?
 
-    def initialize: (?pdf_version: String, ?producer: String, ?is_encrypted: bool, ?width: Integer, ?height: Integer, ?page_count: Integer) -> void
+def initialize: (?pdf_version: String, ?producer: String, ?is_encrypted: bool, ?width: Integer, ?height: Integer, ?page_count: Integer, ?scanned_confidence: Float, ?scanned_pages: Array[Integer]) -> void
     end
 
     class ProxyConfig
@@ -2347,6 +2359,9 @@ def initialize: (?urls: Array[SitemapUrl]) -> void
     class VlmFallbackPolicy
     end
 
+    class OcrStrategy
+    end
+
     class TableChunkingMode
         type value = :split | :repeat_header
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: xberg
 version: !ruby/object:Gem::Version
-  version: 1.0.0.pre.rc.19
+  version: 1.0.0.pre.rc.20
 platform: ruby
 authors:
 - Na'aman Hirschfeld <naaman@xberg.io>
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-07-09 00:00:00.000000000 Z
+date: 2026-07-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys