kjarni::indexer

IndexInfo

pub struct IndexInfo

Information about an existing index

Fields

• pub path: String,: Path to the index
• pub document_count: usize,: Number of documents
• pub segment_count: usize,: Number of segments
• pub dimension: usize,: Embedding dimension
• pub size_bytes: u64,: Total size in bytes
• pub embedding_model: Option<String>,: Embedding model used (if recorded)
• pub created_at: Option<u64>,: When the index was created

IndexStats

pub struct IndexStats

Statistics returned after indexing

Fields

• pub documents_indexed: usize,: Number of documents indexed
• pub chunks_created: usize,: Number of chunks created
• pub dimension: usize,: Embedding dimension used
• pub size_bytes: u64,: Total index size in bytes
• pub files_processed: usize,: Number of files processed
• pub files_skipped: usize,: Number of files skipped (errors, unsupported)
• pub elapsed_ms: u64,: Time taken in milliseconds

Indexer

pub struct Indexer

builder

pub fn builder(model: &str) -> IndexerBuilder

new

pub async fn new(model: &str) -> IndexerResult<Self>

Create with default settings

info

pub fn info(index_path: &str) -> IndexerResult<IndexInfo>

Get index info (static - doesn't need embedder)

delete

pub fn delete(index_path: &str) -> IndexerResult<()>

Delete an index

create

pub async fn create(&self, index_path: &str, inputs: &[&str]) -> IndexerResult<IndexStats>

Create a new index from files/directories

create_with_options

pub async fn create_with_options(&self, index_path: &str, inputs: &[&str], force: bool) -> IndexerResult<IndexStats>

Create with force overwrite option

create_with_callback

pub async fn create_with_callback<F, C>(&self, index_path: &str, inputs: &[&str], force: bool, on_progress: Option<F>, is_cancelled: Option<C>) -> IndexerResult<IndexStats> where F: Fn(ProgressStage, usize, usize, Option<&str>), C: Fn() -> bool,

Create with callback and cancellation support (for FFI)

add

pub async fn add(&self, index_path: &str, inputs: &[&str]) -> IndexerResult<usize>

Add documents to existing index

add_with_callback

pub async fn add_with_callback<F, C>(&self, index_path: &str, inputs: &[&str], on_progress: Option<F>, is_cancelled: Option<C>) -> IndexerResult<usize> where F: Fn(ProgressStage, usize, usize, Option<&str>), C: Fn() -> bool,

callback support (for FFI)

model_name

pub fn model_name(&self) -> &str

dimension

pub fn dimension(&self) -> usize

chunk_size

pub fn chunk_size(&self) -> usize

batch_size

pub fn batch_size(&self) -> usize

is_quiet

pub fn is_quiet(&self) -> bool

has_progress_callback

pub fn has_progress_callback(&self) -> bool

IndexerBuilder

pub struct IndexerBuilder

new

pub fn new(model: &str) -> Self

cpu

pub fn cpu(mut self) -> Self

gpu

pub fn gpu(mut self) -> Self

device

pub fn device(mut self, d: &str) -> Self

cache_dir

pub fn cache_dir(mut self, dir: impl Into<PathBuf>) -> Self

chunk_size

pub fn chunk_size(mut self, size: usize) -> Self

chunk_overlap

pub fn chunk_overlap(mut self, overlap: usize) -> Self

extension

pub fn extension(mut self, ext: &str) -> Self

extensions

pub fn extensions(mut self, exts: &[&str]) -> Self

exclude

pub fn exclude(mut self, pattern: &str) -> Self

recursive

pub fn recursive(mut self, r: bool) -> Self

include_hidden

pub fn include_hidden(mut self, h: bool) -> Self

max_file_size

pub fn max_file_size(mut self, bytes: usize) -> Self

max_docs_per_segment

pub fn max_docs_per_segment(mut self, n: usize) -> Self

batch_size

pub fn batch_size(mut self, n: usize) -> Self

quiet

pub fn quiet(mut self, q: bool) -> Self

on_progress

pub fn on_progress<F>(mut self, callback: F) -> Self where F: Fn(&Progress, Option<&str>) + Send + Sync + 'static,

cancel_token

pub fn cancel_token(mut self, token: watch::Receiver<bool>) -> Self

build

pub async fn build(self) -> IndexerResult<Indexer>

IndexerError

pub enum IndexerError

Error types for indexing operations

Variants

• NoInputs,
• PathNotFound(String),
• IndexExists(String),
• IndexNotFound(String),
• EmbedderError(#[from] crate::embedder::EmbedderError),
• IndexingFailed(#[from] anyhow::Error),
• DimensionMismatch
• Cancelled,

IndexerResult

pub type IndexerResult<T> = Result<T, IndexerError>