Struct Index 

pub struct Index { /* private fields */ }

A Luci search index.

Provides a unified API for indexing and searching JSON documents. Owns the storage, schema, writer, and search state.

Implementations

impl Index

pub fn create(path: impl AsRef<Path>) -> Result<Self>

Create a new index at the given path with fully dynamic mappings.

All field types are inferred from the first document that contains each field. Mappings are persisted to disk on commit.

pub fn create_with_mapping( path: impl AsRef<Path>, mapping: Mapping, ) -> Result<Self>

Create a new index with explicit field mappings.

Unknown fields are handled according to the mapping’s dynamic mode (default: true, meaning auto-map and index).

pub fn create_with_settings( path: impl AsRef<Path>, mapping: Mapping, analysis_config: Option<AnalysisConfig>, ) -> Result<Self>

Create a new index with explicit mappings and analysis settings.

The analysis_config defines custom analyzers, tokenizers, char filters, and token filters. It is persisted in the index metadata so that open can rebuild the analyzer registry.

See [[feature-analysis-pipeline]].

pub fn open(path: impl AsRef<Path>) -> Result<Self>

Open an existing index at the given path.

Mappings are loaded from disk (persisted on each commit). If the index has no persisted mappings (pre-persistence format), an empty mapping with dynamic=true is used.

pub fn add(&self, doc: Value) -> Result<()>

Add a document. Auto-commits — the document is immediately searchable.

Blocks if a transaction is active (waits for it to complete). See [[architecture-concurrency-transactions]].

pub fn bulk(&self, docs: Vec<Value>) -> Result<Value>

Add multiple documents in a single call. Auto-commits at the end — all documents are searchable after this returns.

Blocks if a transaction is active (waits for it to complete). Reduces Python→Rust FFI overhead from one call per document to one call per batch. Fails fast on the first invalid document.

Returns {"took": <ms>, "count": <n>}.

See [[feature-bulk-api]].

pub fn get(&self, id: &str) -> Result<Option<Value>>

Get a document by its _id. Returns the source if found. See [[feature-document-crud]].

pub fn delete(&self, id: &str) -> Result<bool>

Delete a document by its _id. Returns true if found and deleted. See [[feature-document-crud]].

pub fn update(&self, id: &str, partial_doc: Value) -> Result<bool>

Update a document by its _id with a partial doc merge. Returns true if found and updated. See [[feature-document-crud]].

pub fn delete_by_query(&self, query: Value) -> Result<u64>

Delete all documents matching a query. Returns count deleted. See [[feature-document-crud]].

pub fn count(&self, query: Value) -> Result<u64>

Count documents matching a query. See [[feature-document-crud]].

pub fn force_merge(&self, max_segments: usize) -> Result<()>

Force-merge all segments down to at most max_segments.

Call after bulk indexing is complete to optimize for query performance. Invalidates the cached searcher.

pub fn search(&self, expr: &SearchExpression) -> Result<SearchResults>

Search the index with a native SearchExpression.

Thin entry point: refreshes the reader, delegates to Searcher::execute_query, then handles deletion filtering and total_hits capping.

pub fn set_memory_budget(&self, budget: usize)

Set the memory budget for auto-flush (bytes).

When the in-memory buffer exceeds this size, it is automatically flushed to a new segment. Smaller values produce more segments (better parallelism, more merge work). Default: 64 MB.

pub fn set_write_timeout(&self, timeout: Duration)

Set the timeout for acquiring the cross-process write lock.

Default: 5 seconds. If another process holds the write lock, retries until the timeout, then returns WriterLocked. Not persisted — applies only to this session.

pub fn schema(&self) -> &Mapping

Get the schema.

pub fn buffered_doc_count(&self) -> u32

Number of buffered (uncommitted) documents.

pub fn begin_transaction(&self) -> Result<()>

Begin a transaction. Blocks if another transaction is active.

While a transaction is active, add() and bulk() will block until the transaction completes.

pub fn end_transaction(&self)

End a transaction, waking any blocked writers.

Must only be called after a successful begin_transaction().

pub fn is_transaction_active(&self) -> bool

Whether a transaction is currently active.

pub fn txn_add(&self, doc: Value) -> Result<()>

Add a document without committing. For use inside a transaction.

The caller must have called begin_transaction() first.

pub fn txn_commit(&self) -> Result<()>

Commit all buffered documents. For use inside a transaction.

The caller must have called begin_transaction() first.

pub fn txn_rollback(&self)

Discard all buffered (uncommitted) documents. For transaction rollback.

Resets the writer’s in-memory buffer without flushing to storage.