kreuzberg 4.7.2 → 4.7.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c713d0a652ed4e1752ce0c3f6e044516e870349de149b2fdd5cb6c429c722d65
-  data.tar.gz: b519e5b70800b61bd6e5d6acef3a0fee0de8d53f0ad625e863753573f79caf4f
+  metadata.gz: 0f26fc086a0221056b94cd10141832240ef9563783835ef66555445e0d33442d
+  data.tar.gz: ed8488fcdd8bd12266ec2be8984ec5f5b60a32aaa562e64b442d9a4a641c8841
 SHA512:
-  metadata.gz: 0cec6f964c7975e905997422f296a129fe32974c221300f1d6f89d5f54b27ca48e7343522e961d044cc90d9c7f0f4ef20c42de0c95b20e6794614cd1243bd03b
-  data.tar.gz: 8b173be0cb820ade74c4e572465a28257911e9cd1d4ed510f93d26e65529c47f1e637f59a215171b8f931f4a45037bb9b9e5069fd4542b4f911fa36908c65364
+  metadata.gz: 485460f24ddbf58b82a41873a61c5d78408012b415e62a52daeed43df2d150b317115f3ca79bc029a9394188da70d6733f79ba8e61f589c58bee7d1a845b17d8
+  data.tar.gz: 5e80e6e34fe8125a934b141ad3ab058240b81f4b82d609e0e2aea822131cbd4954dd7112e152a90542fc56d494519182c62cc14b387f658944ebbc73bc5037a1

data/README.md

@@ -22,7 +22,7 @@
     <img src="https://img.shields.io/maven-central/v/dev.kreuzberg/kreuzberg?label=Java&color=007ec6" alt="Java">
   </a>
   <a href="https://github.com/kreuzberg-dev/kreuzberg/releases">
-    <img src="https://img.shields.io/github/v/tag/kreuzberg-dev/kreuzberg?label=Go&color=007ec6&filter=v4.7.2" alt="Go">
+    <img src="https://img.shields.io/github/v/tag/kreuzberg-dev/kreuzberg?label=Go&color=007ec6&filter=v4.7.3" alt="Go">
   </a>
   <a href="https://www.nuget.org/packages/Kreuzberg/">
     <img src="https://img.shields.io/nuget/v/Kreuzberg?label=C%23&color=007ec6" alt="C#">

data/ext/kreuzberg_rb/native/Cargo.lock

@@ -1565,9 +1565,9 @@ dependencies = [
 
 [[package]]
 name = "fastrand"
-version = "2.3.0"
+version = "2.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "37909eebbb50d72f9059c3b6d82c0463f2ff062c9e95845c43a6c9c0355411be"
+checksum = "a043dc74da1e37d6afe657061213aa6f425f855399a11d3463c6ecccc4dfda1f"
 
 [[package]]
 name = "fax"
@@ -2938,7 +2938,7 @@ dependencies = [
 
 [[package]]
 name = "kreuzberg-rb"
-version = "4.7.1"
+version = "4.7.2"
 dependencies = [
  "async-trait",
  "html-to-markdown-rs",

data/ext/kreuzberg_rb/native/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-rb"
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]

data/lib/kreuzberg/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Kreuzberg
-  VERSION = '4.7.2'
+  VERSION = '4.7.3'
 end

data/vendor/Cargo.toml

@@ -2,7 +2,7 @@
 members = ["kreuzberg", "kreuzberg-ffi", "kreuzberg-tesseract", "kreuzberg-paddle-ocr", "kreuzberg-pdfium-render"]
 
 [workspace.package]
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
@@ -32,8 +32,8 @@ html-to-markdown-rs = { version = "3.1.0", default-features = false }
 image = { version = "0.25.10", default-features = false }
 itertools = "0.14"
 js-sys = "0.3"
-kreuzberg = { path = "./crates/kreuzberg", version = "4.7.2", default-features = false }
-kreuzberg-ffi = { path = "./crates/kreuzberg-ffi", version = "4.7.2" }
+kreuzberg = { path = "./crates/kreuzberg", version = "4.7.3", default-features = false }
+kreuzberg-ffi = { path = "./crates/kreuzberg-ffi", version = "4.7.3" }
 lazy_static = "1.5.0"
 libc = "0.2.184"
 log = "0.4"
@@ -43,7 +43,7 @@ num_cpus = "1.17.0"
 once_cell = "1.21.4"
 ort = { version = "2.0.0-rc.12", features = ["std", "api-18"], default-features = false }
 parking_lot = "0.12.5"
-pdf_oxide = { version = "0.3.19", default-features = false }
+pdf_oxide = { version = "0.3.20", default-features = false }
 pdfium-render = { package = "kreuzberg-pdfium-render", path = "crates/kreuzberg-pdfium-render", version = "4.3" }
 rayon = "1.11.0"
 reqwest = { version = "0.13.2", default-features = false }

data/vendor/kreuzberg/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg"
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
@@ -307,7 +307,7 @@ ort = { version = "2.0.0-rc.12", default-features = false, features = [
 outlook-pst = { version = "1.2.0", optional = true }
 parking_lot = "0.12.5"
 pastey = "0.2"
-pdf_oxide = { version = "0.3.19", default-features = false, optional = true }
+pdf_oxide = { version = "0.3.20", default-features = false, optional = true }
 pdfium-render = { package = "kreuzberg-pdfium-render", path = "../kreuzberg-pdfium-render", features = ["thread_safe", "image_latest"], optional = true }
 pulldown-cmark = { version = "0.13" }
 quick-xml = { version = "0.39.2", features = ["serialize"], optional = true }
@@ -404,4 +404,4 @@ tempfile = "3.27.0"
 tokio = { version = "1.51.0", features = ["macros", "time"] }
 tokio-test = "0.4"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
-zip = { version = ">=7.0.0, <7.4.0", default-features = false, features = ["deflate-flate2"] }
+zip = { version = ">=7.0.0, <8.6.0", default-features = false, features = ["deflate-flate2"] }

data/vendor/kreuzberg/README.md

@@ -18,7 +18,7 @@ High-performance document intelligence library for Rust. Extract text, metadata,
 
 This is the core Rust library that powers the Python, TypeScript, and Ruby bindings.
 
-> **🚀 Version 4.7.2 Release**
+> **🚀 Version 4.7.3 Release**
 > This is a pre-release version. We invite you to test the library and [report any issues](https://github.com/kreuzberg-dev/kreuzberg/issues) you encounter.
 >
 > **Note**: The Rust crate is not currently published to crates.io for this RC. Use git dependencies or language bindings (Python, TypeScript, Ruby) instead.

data/vendor/kreuzberg/src/extractors/pdf/mod.rs

@@ -593,11 +593,21 @@ impl PdfExtractor {
                 );
                 (native_text, false)
             } else if decision.fallback || has_font_encoding_issues {
-                let (ocr_text, ocr_tbls, ocr_elems, ocr_doc) = run_ocr_with_layout(content, config, path).await?;
-                ocr_tables = ocr_tbls;
-                _ocr_elements_from_ocr = ocr_elems;
-                ocr_internal_doc = ocr_doc;
-                (ocr_text, true)
+                match run_ocr_with_layout(content, config, path).await {
+                    Ok((ocr_text, ocr_tbls, ocr_elems, ocr_doc)) => {
+                        ocr_tables = ocr_tbls;
+                        _ocr_elements_from_ocr = ocr_elems;
+                        ocr_internal_doc = ocr_doc;
+                        (ocr_text, true)
+                    }
+                    Err(e) => {
+                        tracing::warn!(
+                            error = %e,
+                            "OCR fallback failed; using native text extraction result"
+                        );
+                        (native_text, false)
+                    }
+                }
             } else {
                 (native_text, false)
             }

data/vendor/kreuzberg/src/ocr/tesseract_backend.rs

@@ -197,13 +197,20 @@ impl OcrBackend for TesseractBackend {
         let processor = Arc::clone(&self.processor);
         let image_bytes = image_bytes.to_vec();
 
-        let ocr_result = tokio::task::spawn_blocking(move || match output_format {
-            Some(fmt) => processor.process_image_with_format(&image_bytes, &tess_config_clone, fmt),
-            None => processor.process_image(&image_bytes, &tess_config_clone),
+        let ocr_result = tokio::task::spawn_blocking(move || {
+            std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| match output_format {
+                Some(fmt) => processor.process_image_with_format(&image_bytes, &tess_config_clone, fmt),
+                None => processor.process_image(&image_bytes, &tess_config_clone),
+            }))
+            .unwrap_or_else(|_| {
+                Err(crate::ocr::error::OcrError::ProcessingFailed(
+                    "Tesseract/Leptonica foreign exception caught".to_string(),
+                ))
+            })
         })
         .await
         .map_err(|e| crate::KreuzbergError::Plugin {
-            message: format!("Tesseract task panicked: {}", e),
+            message: format!("Tesseract task panicked or caught foreign exception: {}", e),
             plugin_name: "tesseract".to_string(),
         })?
         .map_err(|e| crate::KreuzbergError::Ocr {
@@ -302,13 +309,20 @@ impl OcrBackend for TesseractBackend {
         let processor = Arc::clone(&self.processor);
         let path_str = path.to_string_lossy().to_string();
 
-        let ocr_result = tokio::task::spawn_blocking(move || match output_format {
-            Some(fmt) => processor.process_image_file_with_format(&path_str, &tess_config_clone, fmt),
-            None => processor.process_image_file(&path_str, &tess_config_clone),
+        let ocr_result = tokio::task::spawn_blocking(move || {
+            std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| match output_format {
+                Some(fmt) => processor.process_image_file_with_format(&path_str, &tess_config_clone, fmt),
+                None => processor.process_image_file(&path_str, &tess_config_clone),
+            }))
+            .unwrap_or_else(|_| {
+                Err(crate::ocr::error::OcrError::ProcessingFailed(
+                    "Tesseract/Leptonica foreign exception caught".to_string(),
+                ))
+            })
         })
         .await
         .map_err(|e| crate::KreuzbergError::Plugin {
-            message: format!("Tesseract task panicked: {}", e),
+            message: format!("Tesseract task panicked or caught foreign exception: {}", e),
             plugin_name: "tesseract".to_string(),
         })?
         .map_err(|e| crate::KreuzbergError::Ocr {

data/vendor/kreuzberg/src/pdf/structure/adapters.rs

@@ -83,6 +83,7 @@ fn map_content_role(role: &ContentRole) -> (SemanticRole, Option<String>) {
 /// The resulting paragraphs feed into `apply_layout_overrides` and
 /// `assemble_internal_document`, matching the pdfium native text pipeline.
 #[cfg(feature = "ocr")]
+#[allow(dead_code)] // Called from extractors/pdf/ocr.rs only when layout-detection is also enabled
 pub(crate) fn ocr_doc_to_paragraphs(
     doc: &crate::types::internal::InternalDocument,
     page_height_px: u32,

data/vendor/kreuzberg/src/pdf/structure/bridge.rs

@@ -211,6 +211,15 @@ fn finalize_paragraph(
 
     // Join line texts with newlines (preserving full_text content exactly).
     let text: String = lines.iter().map(|l| l.text.as_str()).collect::<Vec<_>>().join("\n");
+
+    // Convert embedded HTML to markdown if detected (e.g., PDFs with HTML in text layer).
+    #[cfg(feature = "html")]
+    let text = if crate::pdf::text::contains_html_markup(&text) {
+        crate::pdf::text::convert_html_page_text(&text)
+    } else {
+        text
+    };
+
     let trimmed = text.trim();
     if trimmed.is_empty() {
         return None;
@@ -546,6 +555,13 @@ struct CharFontInfo {
 fn extract_page_blocks(page: &PdfPage) -> Option<(Vec<SegmentData>, String, Vec<f32>)> {
     let text_api = page.text().ok()?;
     let full_text = text_api.all();
+    // Convert embedded HTML to markdown if detected (PDFs with HTML in text layer).
+    #[cfg(feature = "html")]
+    let full_text = if crate::pdf::text::contains_html_markup(&full_text) {
+        crate::pdf::text::convert_html_page_text(&full_text)
+    } else {
+        full_text
+    };
     if full_text.trim().is_empty() {
         return None;
     }

data/vendor/kreuzberg/src/pdf/text.rs

@@ -53,7 +53,7 @@ fn fix_pdf_control_chars(text: &str) -> Cow<'_, str> {
 /// Some PDFs embed raw HTML in their text layer (e.g. from web-to-PDF converters).
 /// This function detects common HTML tags to determine if the text should be
 /// converted from HTML to markdown rather than used as-is.
-fn contains_html_markup(text: &str) -> bool {
+pub(crate) fn contains_html_markup(text: &str) -> bool {
     if !text.contains('<') {
         return false;
     }
@@ -72,7 +72,7 @@ fn contains_html_markup(text: &str) -> bool {
 /// Falls back to the original text if the `html` feature is not enabled
 /// or if conversion fails.
 #[cfg(feature = "html")]
-fn convert_html_page_text(text: &str) -> String {
+pub(crate) fn convert_html_page_text(text: &str) -> String {
     match crate::extraction::html::convert_html_to_markdown(text, None, None) {
         Ok(converted) => converted,
         Err(_) => text.to_owned(),

data/vendor/kreuzberg-ffi/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-ffi"
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
@@ -41,7 +41,7 @@ serde_json = { version = "1.0.149" }
 tokio = { version = "1.51.0", features = ["rt", "rt-multi-thread", "macros", "sync", "process", "fs", "time", "io-util"] }
 
 [target.'cfg(all(windows, target_env = "gnu"))'.dependencies]
-kreuzberg = { path = "../kreuzberg", version = "4.7.2", default-features = false, features = [
+kreuzberg = { path = "../kreuzberg", version = "4.7.3", default-features = false, features = [
     "pdf",
     "excel",
     "office",
@@ -64,7 +64,7 @@ kreuzberg = { path = "../kreuzberg", version = "4.7.2", default-features = false
 ] }
 
 [target.'cfg(not(all(windows, target_env = "gnu")))'.dependencies]
-kreuzberg = { path = "../kreuzberg", version = "4.7.2", default-features = false, features = ["bundled-pdfium", "full"] }
+kreuzberg = { path = "../kreuzberg", version = "4.7.3", default-features = false, features = ["bundled-pdfium", "full"] }
 
 [build-dependencies]
 cbindgen = "0.29"

data/vendor/kreuzberg-ffi/kreuzberg.h

@@ -9,8 +9,8 @@
 
 #define KREUZBERG_VERSION_MAJOR 4
 #define KREUZBERG_VERSION_MINOR 7
-#define KREUZBERG_VERSION_PATCH 2
-#define KREUZBERG_VERSION "4.7.2"
+#define KREUZBERG_VERSION_PATCH 3
+#define KREUZBERG_VERSION "4.7.3"
 
 
 #include <stdarg.h>

data/vendor/kreuzberg-paddle-ocr/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-paddle-ocr"
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]

data/vendor/kreuzberg-pdfium-render/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-pdfium-render"
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]

data/vendor/kreuzberg-tesseract/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "kreuzberg-tesseract"
-version = "4.7.2"
+version = "4.7.3"
 edition = "2024"
 rust-version = "1.91"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]

data/vendor/kreuzberg-tesseract/src/api.rs

@@ -2160,7 +2160,7 @@ impl Clone for TesseractAPI {
 }
 
 #[cfg(any(feature = "build-tesseract", feature = "build-tesseract-wasm"))]
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     fn TessBaseAPIMeanTextConf(handle: *mut c_void) -> c_int;
     fn TessBaseAPISetVariable(handle: *mut c_void, name: *const c_char, value: *const c_char) -> c_int;
     fn TessBaseAPIGetStringVariable(handle: *mut c_void, name: *const c_char) -> *const c_char;

data/vendor/kreuzberg-tesseract/src/choice_iterator.rs

@@ -69,7 +69,7 @@ impl Drop for ChoiceIterator {
     }
 }
 
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     fn TessChoiceIteratorDelete(handle: *mut c_void);
     fn TessChoiceIteratorNext(handle: *mut c_void) -> c_int;
     fn TessChoiceIteratorGetUTF8Text(handle: *mut c_void) -> *mut c_char;

data/vendor/kreuzberg-tesseract/src/leptonica.rs

@@ -29,7 +29,7 @@ use std::ffi::c_void;
 // ---------------------------------------------------------------------------
 
 #[cfg(any(feature = "build-tesseract", feature = "build-tesseract-wasm"))]
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     /// Allocates a new Pix with the given dimensions and bit depth.
     fn pixCreate(width: i32, height: i32, depth: i32) -> *mut c_void;
 

data/vendor/kreuzberg-tesseract/src/monitor.rs

@@ -60,7 +60,7 @@ impl Drop for TessMonitor {
     }
 }
 
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     pub fn TessMonitorCreate() -> *mut c_void;
     pub fn TessMonitorDelete(monitor: *mut c_void);
     pub fn TessMonitorSetDeadlineMSecs(monitor: *mut c_void, deadline: c_int);

data/vendor/kreuzberg-tesseract/src/mutable_iterator.rs

@@ -191,7 +191,7 @@ impl Drop for MutableIterator {
     }
 }
 
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     pub fn TessResultIteratorDelete(handle: *mut c_void);
     pub fn TessDeleteText(text: *mut c_char);
 }

data/vendor/kreuzberg-tesseract/src/page_iterator.rs

@@ -380,7 +380,7 @@ impl Drop for PageIterator {
     }
 }
 
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     pub fn TessPageIteratorDelete(handle: *mut c_void);
     pub fn TessPageIteratorBegin(handle: *mut c_void);
     pub fn TessPageIteratorNext(handle: *mut c_void, level: c_int) -> c_int;

data/vendor/kreuzberg-tesseract/src/result_iterator.rs

@@ -555,7 +555,7 @@ impl Drop for ResultIterator {
 }
 
 #[cfg(any(feature = "build-tesseract", feature = "build-tesseract-wasm"))]
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     pub fn TessResultIteratorDelete(handle: *mut c_void);
     pub fn TessPageIteratorBegin(handle: *mut c_void);
     pub fn TessResultIteratorGetUTF8Text(handle: *mut c_void, level: c_int) -> *mut c_char;

data/vendor/kreuzberg-tesseract/src/result_renderer.rs

@@ -198,7 +198,7 @@ impl Drop for TessResultRenderer {
     }
 }
 
-unsafe extern "C" {
+unsafe extern "C-unwind" {
     pub fn TessTextRendererCreate(outputbase: *const c_char) -> *mut c_void;
     pub fn TessHOcrRendererCreate(outputbase: *const c_char) -> *mut c_void;
     pub fn TessPDFRendererCreate(outputbase: *const c_char, datadir: *const c_char, textonly: c_int) -> *mut c_void;

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: kreuzberg
 version: !ruby/object:Gem::Version
-  version: 4.7.2
+  version: 4.7.3
 platform: ruby
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-04-04 00:00:00.000000000 Z
+date: 2026-04-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys