kreuzberg 4.2.6 → 4.2.7

data/vendor/kreuzberg/src/api/handlers.rs

@@ -1,14 +1,11 @@
 //! API request handlers.
 
-use axum::{
-    Json,
-    extract::{Multipart, State},
-};
+use axum::{Json, extract::State};
 
 use crate::{batch_extract_bytes, cache, extract_bytes};
 
 use super::{
-    error::{ApiError, JsonApi},
+    error::{ApiError, JsonApi, MultipartApi},
     types::{
         ApiState, CacheClearResponse, CacheStatsResponse, ChunkRequest, ChunkResponse, EmbedRequest, EmbedResponse,
         ExtractResponse, HealthResponse, InfoResponse,
@@ -84,19 +81,18 @@ pub async fn info_handler() -> Json<InfoResponse> {
 ///
 /// The server's default config (loaded from kreuzberg.toml/yaml/json via discovery)
 /// is used as the base, and any per-request config overrides those defaults.
-// TODO: Add utoipa::path annotation once ExtractionResult implements ToSchema
-// #[utoipa::path(
-//     post,
-//     path = "/extract",
-//     tag = "extraction",
-//     request_body(content_type = "multipart/form-data"),
-//     responses(
-//         (status = 200, description = "Extraction successful", body = ExtractResponse),
-//         (status = 400, description = "Bad request", body = crate::api::types::ErrorResponse),
-//         (status = 413, description = "Payload too large", body = crate::api::types::ErrorResponse),
-//         (status = 500, description = "Internal server error", body = crate::api::types::ErrorResponse),
-//     )
-// )]
+#[utoipa::path(
+    post,
+    path = "/extract",
+    tag = "extraction",
+    request_body(content_type = "multipart/form-data"),
+    responses(
+        (status = 200, description = "Extraction successful", body = ExtractResponse),
+        (status = 400, description = "Bad request", body = crate::api::types::ErrorResponse),
+        (status = 413, description = "Payload too large", body = crate::api::types::ErrorResponse),
+        (status = 500, description = "Internal server error", body = crate::api::types::ErrorResponse),
+    )
+)]
 #[cfg_attr(
     feature = "otel",
     tracing::instrument(
@@ -107,10 +103,10 @@ pub async fn info_handler() -> Json<InfoResponse> {
 )]
 pub async fn extract_handler(
     State(state): State<ApiState>,
-    mut multipart: Multipart,
+    MultipartApi(mut multipart): MultipartApi,
 ) -> Result<Json<ExtractResponse>, ApiError> {
     let mut files = Vec::new();
-    let mut config = (*state.default_config).clone();
+    let mut config: Option<crate::core::config::ExtractionConfig> = None;
 
     while let Some(field) = multipart
         .next_field()
@@ -138,12 +134,12 @@ pub async fn extract_handler(
                     .await
                     .map_err(|e| ApiError::validation(crate::error::KreuzbergError::validation(e.to_string())))?;
 
-                config = serde_json::from_str(&config_str).map_err(|e| {
+                config = Some(serde_json::from_str(&config_str).map_err(|e| {
                     ApiError::validation(crate::error::KreuzbergError::validation(format!(
                         "Invalid extraction configuration: {}",
                         e
                     )))
-                })?;
+                })?);
             }
             "output_format" => {
                 let format_str = field
@@ -151,7 +147,9 @@ pub async fn extract_handler(
                     .await
                     .map_err(|e| ApiError::validation(crate::error::KreuzbergError::validation(e.to_string())))?;
 
-                config.output_format = match format_str.to_lowercase().as_str() {
+                // Ensure config exists before modifying output_format
+                let cfg = config.get_or_insert_with(|| (*state.default_config).clone());
+                cfg.output_format = match format_str.to_lowercase().as_str() {
                     "plain" => crate::core::config::OutputFormat::Plain,
                     "markdown" => crate::core::config::OutputFormat::Markdown,
                     "djot" => crate::core::config::OutputFormat::Djot,
@@ -177,18 +175,21 @@ pub async fn extract_handler(
     #[cfg(feature = "otel")]
     tracing::Span::current().record("files_count", files.len());
 
+    // Use provided config or fall back to default from state
+    let final_config = config.as_ref().unwrap_or(&state.default_config);
+
     if files.len() == 1 {
         let (data, mime_type, _file_name) = files
             .into_iter()
             .next()
             .expect("files.len() == 1 guarantees one element exists");
-        let result = extract_bytes(&data, mime_type.as_str(), &config).await?;
+        let result = extract_bytes(&data, mime_type.as_str(), final_config).await?;
         return Ok(Json(vec![result]));
     }
 
     let files_data: Vec<(Vec<u8>, String)> = files.into_iter().map(|(data, mime, _name)| (data, mime)).collect();
 
-    let results = batch_extract_bytes(files_data, &config).await?;
+    let results = batch_extract_bytes(files_data, final_config).await?;
     Ok(Json(results))
 }
 
@@ -492,6 +493,8 @@ pub async fn chunk_handler(JsonApi(request): JsonApi<ChunkRequest>) -> Result<Js
         overlap,
         trim: cfg.trim.unwrap_or(true),
         chunker_type,
+        embedding: None,
+        preset: None,
     };
 
     // Perform chunking - convert any remaining errors to validation errors since they're likely config issues

data/vendor/kreuzberg/src/api/openapi.rs

@@ -32,7 +32,7 @@ use utoipa::OpenApi;
     paths(
         crate::api::handlers::health_handler,
         crate::api::handlers::info_handler,
-        // Note: extract_handler omitted - requires ExtractionResult ToSchema impl
+        crate::api::handlers::extract_handler,
         crate::api::handlers::cache_stats_handler,
         crate::api::handlers::cache_clear_handler,
         crate::api::handlers::embed_handler,
@@ -53,6 +53,19 @@ use utoipa::OpenApi;
             crate::api::types::ChunkItem,
             crate::api::types::ChunkingConfigRequest,
             crate::api::types::ChunkingConfigResponse,
+            crate::types::extraction::ExtractionResult,
+            crate::types::extraction::Chunk,
+            crate::types::extraction::ChunkMetadata,
+            crate::types::extraction::ExtractedImage,
+            crate::types::extraction::Element,
+            crate::types::extraction::ElementMetadata,
+            crate::types::extraction::ElementId,
+            crate::types::extraction::ElementType,
+            crate::types::extraction::BoundingBox,
+            crate::types::metadata::Metadata,
+            crate::types::tables::Table,
+            crate::types::page::PageContent,
+            crate::types::djot::DjotContent,
         )
     ),
     tags(

data/vendor/kreuzberg/src/chunking/config.rs

@@ -2,43 +2,8 @@
 
 use serde::{Deserialize, Serialize};
 
-/// Configuration options for text chunking operations.
-///
-/// # Fields
-///
-/// * `max_characters` - Maximum number of characters per chunk (default: 2000)
-/// * `overlap` - Number of characters to overlap between consecutive chunks (default: 100)
-/// * `trim` - Whether to trim whitespace from chunk boundaries (default: true)
-/// * `chunker_type` - Type of chunker to use (Text or Markdown) (default: Text)
-pub struct ChunkingConfig {
-    pub max_characters: usize,
-    pub overlap: usize,
-    pub trim: bool,
-    pub chunker_type: ChunkerType,
-}
-
-impl Default for ChunkingConfig {
-    fn default() -> Self {
-        Self {
-            max_characters: 2000,
-            overlap: 100,
-            trim: true,
-            chunker_type: ChunkerType::Text,
-        }
-    }
-}
-
-/// Type of text chunker to use.
-///
-/// # Variants
-///
-/// * `Text` - Generic text splitter, splits on whitespace and punctuation
-/// * `Markdown` - Markdown-aware splitter, preserves formatting and structure
-#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
-pub enum ChunkerType {
-    Text,
-    Markdown,
-}
+// Re-export ChunkingConfig and ChunkerType from core config (canonical location)
+pub use crate::core::config::processing::{ChunkerType, ChunkingConfig};
 
 /// Result of a text chunking operation.
 ///

data/vendor/kreuzberg/src/chunking/core.rs

@@ -118,6 +118,8 @@ pub fn chunk_text_with_type(
         overlap,
         trim,
         chunker_type,
+        embedding: None,
+        preset: None,
     };
     chunk_text(text, &config, None)
 }
@@ -177,6 +179,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "This is a short text.";
         let result = chunk_text(text, &config, None).unwrap();
@@ -192,6 +196,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ";
         let result = chunk_text(text, &config, None).unwrap();
@@ -207,6 +213,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "abcdefghijklmnopqrstuvwxyz0123456789";
         let result = chunk_text(text, &config, None).unwrap();
@@ -230,6 +238,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Markdown,
+            embedding: None,
+            preset: None,
         };
         let markdown = "# Title\n\nParagraph one.\n\n## Section\n\nParagraph two.";
         let result = chunk_text(markdown, &config, None).unwrap();
@@ -244,6 +254,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Markdown,
+            embedding: None,
+            preset: None,
         };
         let markdown = "# Code Example\n\n```python\nprint('hello')\n```\n\nSome text after code.";
         let result = chunk_text(markdown, &config, None).unwrap();
@@ -258,6 +270,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Markdown,
+            embedding: None,
+            preset: None,
         };
         let markdown = "Check out [this link](https://example.com) for more info.";
         let result = chunk_text(markdown, &config, None).unwrap();
@@ -272,6 +286,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "  Leading and trailing spaces  should be trimmed  ";
         let result = chunk_text(text, &config, None).unwrap();
@@ -286,6 +302,8 @@ mod tests {
             overlap: 5,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "  Text with spaces  ";
         let result = chunk_text(text, &config, None).unwrap();
@@ -300,6 +318,8 @@ mod tests {
             overlap: 20,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let result = chunk_text("Some text", &config, None);
         assert!(result.is_err());
@@ -337,6 +357,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let texts = vec!["First text", "Second text", "Third text"];
         let results = chunk_texts_batch(&texts, &config).unwrap();
@@ -351,6 +373,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let texts = vec![
             "Short",
@@ -371,6 +395,8 @@ mod tests {
             overlap: 20,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let texts = vec!["Text one", "Text two"];
         let result = chunk_texts_batch(&texts, &config);
@@ -380,8 +406,8 @@ mod tests {
     #[test]
     fn test_chunking_config_default() {
         let config = ChunkingConfig::default();
-        assert_eq!(config.max_characters, 2000);
-        assert_eq!(config.overlap, 100);
+        assert_eq!(config.max_characters, 1000);
+        assert_eq!(config.overlap, 200);
         assert!(config.trim);
         assert_eq!(config.chunker_type, ChunkerType::Text);
     }
@@ -393,6 +419,8 @@ mod tests {
             overlap: 20,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "a".repeat(1000);
         let result = chunk_text(&text, &config, None).unwrap();
@@ -407,6 +435,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Line one\nLine two\nLine three\nLine four\nLine five";
         let result = chunk_text(text, &config, None).unwrap();
@@ -420,6 +450,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Markdown,
+            embedding: None,
+            preset: None,
         };
         let markdown = "# List Example\n\n- Item 1\n- Item 2\n- Item 3\n\nMore text.";
         let result = chunk_text(markdown, &config, None).unwrap();
@@ -434,6 +466,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Markdown,
+            embedding: None,
+            preset: None,
         };
         let markdown = "# Table\n\n| Col1 | Col2 |\n|------|------|\n| A    | B    |\n| C    | D    |";
         let result = chunk_text(markdown, &config, None).unwrap();
@@ -448,6 +482,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Special chars: @#$%^&*()[]{}|\\<>?/~`";
         let result = chunk_text(text, &config, None).unwrap();
@@ -462,6 +498,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Unicode: 你好世界 🌍 café résumé";
         let result = chunk_text(text, &config, None).unwrap();
@@ -477,6 +515,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "日本語のテキストです。これは長い文章で、複数のチャンクに分割されるべきです。";
         let result = chunk_text(text, &config, None).unwrap();
@@ -490,6 +530,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "English text mixed with 中文文本 and some français";
         let result = chunk_text(text, &config, None).unwrap();
@@ -503,6 +545,8 @@ mod tests {
             overlap: 5,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "AAAAA BBBBB CCCCC DDDDD EEEEE FFFFF";
         let result = chunk_text(text, &config, None).unwrap();
@@ -555,6 +599,8 @@ mod tests {
             overlap: 0,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "AAAAA BBBBB CCCCC DDDDD EEEEE FFFFF";
         let result = chunk_text(text, &config, None).unwrap();
@@ -581,6 +627,8 @@ mod tests {
             overlap: 3,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "0123456789 ABCDEFGHIJ KLMNOPQRST UVWXYZ";
         let result = chunk_text(text, &config, None).unwrap();
@@ -615,6 +663,8 @@ mod tests {
                 overlap,
                 trim: false,
                 chunker_type: ChunkerType::Text,
+                embedding: None,
+                preset: None,
             };
             let text = "Word ".repeat(30);
             let result = chunk_text(&text, &config, None).unwrap();
@@ -647,6 +697,8 @@ mod tests {
             overlap: 5,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "AAAAA BBBBB CCCCC DDDDD EEEEE";
         let result = chunk_text(text, &config, None).unwrap();
@@ -674,6 +726,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Page one content here. Page two starts here and continues.";
 
@@ -706,6 +760,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "This is some test content that should be split into multiple chunks.";
 
@@ -725,6 +781,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Some text content here.";
         let boundaries: Vec<PageBoundary> = vec![];
@@ -743,6 +801,8 @@ mod tests {
             overlap: 5,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "0123456789 AAAAAAAAAA 1111111111 BBBBBBBBBB 2222222222";
 
@@ -779,6 +839,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Page one content here. Page two content.";
 
@@ -802,6 +864,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Page one content here. Page two content.";
 
@@ -832,6 +896,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Page one content here. Page two content.";
 
@@ -862,6 +928,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "First page content here.Second page content here.Third page.";
 
@@ -897,6 +965,8 @@ mod tests {
             overlap: 10,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "All content on single page fits in one chunk.";
 
@@ -919,6 +989,8 @@ mod tests {
             overlap: 0,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "AAAAA BBBBB CCCCC DDDDD";
 
@@ -952,6 +1024,8 @@ mod tests {
             overlap: 5,
             trim: true,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "Page One Content Here.Page Two.";
 
@@ -982,6 +1056,8 @@ mod tests {
             overlap: 2,
             trim: false,
             chunker_type: ChunkerType::Text,
+            embedding: None,
+            preset: None,
         };
         let text = "0123456789ABCDEFGHIJ";
 

data/vendor/kreuzberg/src/chunking/mod.rs

@@ -60,7 +60,7 @@ pub mod validation;
 
 // Re-export submodule types and functions
 pub use boundaries::{calculate_page_range, validate_page_boundaries};
-pub use config::{ChunkerType, ChunkingConfig, ChunkingResult};
+pub use config::{ChunkerType, ChunkingConfig, ChunkingResult}; // ChunkingConfig re-exported from core::config::processing
 pub use core::{chunk_text, chunk_text_with_type, chunk_texts_batch};
 pub use processor::ChunkingProcessor;
 pub use validation::{ADAPTIVE_VALIDATION_THRESHOLD, precompute_utf8_boundaries, validate_utf8_boundaries};

data/vendor/kreuzberg/src/chunking/processor.rs

@@ -54,14 +54,7 @@ impl PostProcessor for ChunkingProcessor {
             None => return Ok(()),
         };
 
-        let chunk_config = crate::chunking::ChunkingConfig {
-            max_characters: chunking_config.max_chars,
-            overlap: chunking_config.max_overlap,
-            trim: true,
-            chunker_type: crate::chunking::ChunkerType::Text,
-        };
-
-        let chunking_result = crate::chunking::chunk_text(&result.content, &chunk_config, None)
+        let chunking_result = crate::chunking::chunk_text(&result.content, chunking_config, None)
             .map_err(|e| KreuzbergError::Other(format!("Chunking failed: {}", e)))?;
         result.chunks = Some(chunking_result.chunks);
 
@@ -87,14 +80,17 @@ mod tests {
     use super::*;
     use crate::core::config::ChunkingConfig;
     use crate::types::Metadata;
+    use std::borrow::Cow;
 
     #[tokio::test]
     async fn test_chunking_processor() {
         let processor = ChunkingProcessor;
         let config = ExtractionConfig {
             chunking: Some(ChunkingConfig {
-                max_chars: 100,
-                max_overlap: 10,
+                max_characters: 100,
+                overlap: 10,
+                trim: true,
+                chunker_type: crate::chunking::ChunkerType::Text,
                 embedding: None,
                 preset: None,
             }),
@@ -103,7 +99,7 @@ mod tests {
 
         let mut result = ExtractionResult {
 	            content: "This is a longer text that should be split into multiple chunks to test the chunking processor functionality.".to_string(),
-	            mime_type: "text/plain".to_string(),
+	            mime_type: Cow::Borrowed("text/plain"),
 	            metadata: Metadata::default(),
 	            tables: vec![],
 	            detected_languages: None,
@@ -128,7 +124,7 @@ mod tests {
 
         let mut result = ExtractionResult {
             content: "Some text".to_string(),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata: Metadata::default(),
             tables: vec![],
             detected_languages: None,
@@ -165,7 +161,7 @@ mod tests {
 
         let result = ExtractionResult {
             content: "Sample text".to_string(),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata: Metadata::default(),
             tables: vec![],
             detected_languages: None,
@@ -178,8 +174,10 @@ mod tests {
 
         let config_with_chunking = ExtractionConfig {
             chunking: Some(crate::core::config::ChunkingConfig {
-                max_chars: 100,
-                max_overlap: 10,
+                max_characters: 100,
+                overlap: 10,
+                trim: true,
+                chunker_type: crate::chunking::ChunkerType::Text,
                 embedding: None,
                 preset: None,
             }),
@@ -197,7 +195,7 @@ mod tests {
 
         let short_result = ExtractionResult {
             content: "Short".to_string(),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata: Metadata::default(),
             tables: vec![],
             detected_languages: None,
@@ -210,7 +208,7 @@ mod tests {
 
         let long_result = ExtractionResult {
             content: "a".repeat(100000),
-            mime_type: "text/plain".to_string(),
+            mime_type: Cow::Borrowed("text/plain"),
             metadata: Metadata::default(),
             tables: vec![],
             detected_languages: None,

data/vendor/kreuzberg/src/core/config/extraction/env.rs

@@ -94,8 +94,10 @@ impl ExtractionConfig {
 
             if self.chunking.is_none() {
                 self.chunking = Some(ChunkingConfig {
-                    max_chars: 1000,
-                    max_overlap: 200,
+                    max_characters: 1000,
+                    overlap: 200,
+                    trim: true,
+                    chunker_type: super::super::processing::ChunkerType::Text,
                     embedding: None,
                     preset: None,
                 });
@@ -103,8 +105,8 @@ impl ExtractionConfig {
 
             if let Some(ref mut chunking) = self.chunking {
                 // Validate against current overlap before updating
-                validate_chunking_params(max_chars, chunking.max_overlap)?;
-                chunking.max_chars = max_chars;
+                validate_chunking_params(max_chars, chunking.overlap)?;
+                chunking.max_characters = max_chars;
             }
         }
 
@@ -120,17 +122,19 @@ impl ExtractionConfig {
 
             if self.chunking.is_none() {
                 self.chunking = Some(ChunkingConfig {
-                    max_chars: 1000,
-                    max_overlap: 200,
+                    max_characters: 1000,
+                    overlap: 200,
+                    trim: true,
+                    chunker_type: super::super::processing::ChunkerType::Text,
                     embedding: None,
                     preset: None,
                 });
             }
 
             if let Some(ref mut chunking) = self.chunking {
-                // Validate against current max_chars before updating
-                validate_chunking_params(chunking.max_chars, max_overlap)?;
-                chunking.max_overlap = max_overlap;
+                // Validate against current max_characters before updating
+                validate_chunking_params(chunking.max_characters, max_overlap)?;
+                chunking.overlap = max_overlap;
             }
         }