Struct QvdStreamReader 

pub struct QvdStreamReader<R: Read + Seek + BufRead> {
    pub header: QvdTableHeader,
    pub symbols: Vec<Vec<QvdSymbol>>,
    /* private fields */
}

A streaming QVD reader that reads rows in chunks without loading the entire index table into memory at once.

Symbol tables are always loaded fully (they are needed for decoding), but the index table is read chunk by chunk.

Fields

header: QvdTableHeader

symbols: Vec<Vec<QvdSymbol>>

Implementations

impl<R: Read + Seek + BufRead> QvdStreamReader<R>

pub fn open(reader: R) -> QvdResult<Self>

Open a QVD file for streaming reading.

pub fn total_rows(&self) -> usize

Total number of rows in the file.

pub fn remaining_rows(&self) -> usize

Number of rows remaining to read.

pub fn next_chunk(&mut self, chunk_size: usize) -> QvdResult<Option<QvdChunk>>

Read the next chunk of rows. Returns None when all rows have been read.

pub fn next_chunk_indices( &mut self, chunk_size: usize, ) -> QvdResult<Option<(Vec<Vec<i64>>, usize, usize)>>

Read the next chunk as column indices (without resolving symbols). More efficient if you plan to do your own symbol resolution.

pub fn read_filtered( &mut self, filter_col: &str, exists_index: &ExistsIndex, select_cols: Option<&[&str]>, chunk_size: usize, ) -> QvdResult<QvdTable>

Read filtered: stream the entire file, decode only the filter column per row, and collect indices only for matching rows and selected columns.

• filter_col: column name to filter on
• exists_index: ExistsIndex with values to match
• select_cols: if Some, only include these columns in the output; if None, include all
• chunk_size: number of rows to read per I/O chunk

Memory: symbol tables (small) + one chunk buffer + only matching row indices. For 87M rows with 23% match rate, this holds ~20M rows instead of 87M.

pub fn reset(&mut self) -> QvdResult<()>

Reset reader to the beginning.

pub fn column_names(&self) -> Vec<&str>

Column names.