Struct StoreMerger 

pub struct StoreMerger<'a, W: Write> { /* private fields */ }

Store merger - concatenates compressed blocks from multiple stores without recompression

This is much faster than rebuilding stores since it avoids:

• Decompressing blocks from source stores
• Re-serializing documents
• Re-compressing blocks at level 22

Limitations:

• All source stores must NOT use dictionaries (or use the same dictionary)
• Doc IDs are remapped sequentially

Implementations

impl<'a, W: Write> StoreMerger<'a, W>

pub fn new(writer: &'a mut W) -> Self

pub async fn append_store<F: AsyncFileRead>( &mut self, data_slice: &F, blocks: &[RawStoreBlock], ) -> Result<()>

Append raw compressed blocks from a store file

data_slice should be the data portion of the store (before index/footer) blocks contains the block metadata from the source store

pub async fn append_store_recompressing( &mut self, store: &AsyncStoreReader, ) -> Result<()>

Append blocks from a dict-compressed store by decompressing and recompressing.

For stores that use dictionary compression, raw blocks can’t be stacked directly because the decompressor needs the original dictionary. This method decompresses each block with the source dict, then recompresses without a dictionary so the merged output is self-contained.

pub fn finish(self) -> Result<u32>

Finish writing the merged store