Struct CsvSource 

pub struct CsvSource { /* private fields */ }

Column-mapped CSV data source.

Reads all rows from a CSV file at construction and exposes them as DataRecords. Suitable for small-to-medium datasets that fit comfortably in memory.

Modes

Configure the source with either anchor/positive columns (role mode) or a single text column (text mode):

use triplets_core::source::{CsvSource, CsvSourceConfig};

// Role mode: explicit anchor + positive columns.
let config = CsvSourceConfig::new("my_qna", "data/qna.csv")
    .with_anchor_column("question")
    .with_positive_column("answer")
    .with_trust(0.9);
let source = CsvSource::new(config).unwrap();

// Text mode: single text column.
let config2 = CsvSourceConfig::new("my_corpus", "data/corpus.csv")
    .with_text_column("text");
let source2 = CsvSource::new(config2).unwrap();

Implementations

impl CsvSource

pub fn new(config: CsvSourceConfig) -> Result<Self, SamplerError>

Load a CSV source from the given configuration.

Returns a SamplerError::Configuration error if the config is invalid, or a SamplerError::SourceUnavailable error if the CSV file cannot be opened or parsed.

Trait Implementations

impl DataSource for CsvSource

fn id(&self) -> &str

Stable source identifier used in records, metrics, and persistence state.

fn refresh( &self, _config: &SamplerConfig, cursor: Option<&SourceCursor>, limit: Option<usize>, ) -> Result<SourceSnapshot, SamplerError>

Fetch up to limit records starting from cursor state.

fn reported_record_count( &self, _config: &SamplerConfig, ) -> Result<u128, SamplerError>

Exact metadata record count reported by the source.

fn default_triplet_recipes(&self) -> Vec<TripletRecipe>

Optional source-provided default triplet recipes.

impl Debug for CsvSource

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl IndexableSource for CsvSource

fn id(&self) -> &str

Stable source identifier.

fn len_hint(&self) -> Option<usize>

Current index domain size, typically Some(total_records).

fn record_at(&self, idx: usize) -> Result<Option<DataRecord>, SamplerError>

Return the record at index idx, or None for sparse/missing positions.