Struct Seeker 

pub struct Seeker<R> { /* private fields */ }

A specialized CSV stream seeker.

Implementations

impl<R: Read + Seek> Seeker<R>

pub fn has_headers(&self) -> bool

Returns whether this seeker has been configured to interpret the first record as a header.

pub fn initial_position(&self) -> u64

Returns the position the seekable stream was in when instantiating the seeker.

pub fn first_record_position(&self) -> u64

Returns the absolute byte offset of the first record (excluding header) of the seekable stream.

pub fn stream_len(&self) -> u64

Returns the total number of bytes contained in the seekable stream.

pub fn lookahead_len(&self) -> u64

Returns the number of bytes that will be read when performing a lookahead in the seekable stream when using Seeker::find_record_after.

pub fn range(&self) -> Range<u64>

Returns the first_record_position..stream_len range of the seeker.

pub fn exact_count(&self) -> Option<u64>

Returns the exact number of records (header excluded) contained in the seekable stream, if the initial sample built when instantiating the seeker exhausted the whole stream.

pub fn approx_count(&self) -> u64

Either returns the exact number of records (header excluded) contained in the seekable stream or an approximation based on statistics sampled from the beginning of the stream and the total stream length.

pub fn find_record_after( &mut self, from_pos: u64, ) -> Result<Option<(u64, ByteRecord)>>

Attempt to find the position, in the seekable stream, of the beginning of the CSV record just after the one where from_pos would end in.

Beware: if from_pos is the exact first byte of a CSV record, this method will still return the position of next CSV record because it has no way of knowing whether the byte just before from_pos is a newline.

This method will return an error if given from_pos is out of bounds.

This method will return None if it did not succeed in finding the next CSV record starting position. This can typically happen when seeking too close to the end of the stream, since this method needs to read ahead of the stream to test its heuristics.

match seeker.find_record_after(1024) {
    Ok(Some((pos, record))) => {
        // Everything went fine
    },
    Ok(None) => {
        // Lookahead failed
    },
    Err(err) => {
        // Either `from_pos` was out-of-bounds, or some IO error occurred
    }
}

pub fn segments(&mut self, count: usize) -> Result<Vec<(u64, u64)>>

Split the seekable stream into a maximum of count segments.

This method might return less than count segments if the stream seems too small to safely return that many segments.

pub fn byte_headers(&self) -> &ByteRecord

Returns the headers of the seekable stream, or just the first record the seeker was configured thusly.

pub fn first_byte_record(&mut self) -> Result<Option<ByteRecord>>

Attempt to read the first record of the seekable stream.

pub fn last_byte_record(&mut self) -> Result<Option<ByteRecord>>

Attempt to read the last record of the seekable stream by reading it in reverse.

pub fn into_inner(self) -> R

Returns the underlying reader without unwinding its position.

pub fn into_zero_copy_reader(self) -> Result<ZeroCopyReader<R>>

Transform the seeker into a ZeroCopyReader. Underlying reader will be correctly reset to the stream initial position beforehand.

pub fn into_splitter(self) -> Result<Splitter<R>>

Transform the seeker into a Splitter. Underlying reader will be correctly reset to the stream initial position beforehand.