Struct CsvDecoder 

pub struct CsvDecoder { /* private fields */ }

Decodes CSV byte payloads into Arrow RecordBatches.

Ring Placement

• Ring 1: decode_batch() — parse CSV, build columnar Arrow output
• Ring 2: Construction (new / with_config) — one-time setup

Implementations

impl CsvDecoder

pub fn new(schema: SchemaRef) -> Self

Creates a new CSV decoder for the given Arrow schema with default config.

pub fn with_config(schema: SchemaRef, config: CsvDecoderConfig) -> Self

Creates a new CSV decoder with custom configuration.

pub fn parse_error_count(&self) -> u64

Returns the cumulative parse error count.

Trait Implementations

impl Debug for CsvDecoder

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

Formats the value using the given formatter.

impl FormatDecoder for CsvDecoder

fn decode_batch(&self, records: &[RawRecord]) -> SchemaResult<RecordBatch>

Decodes a batch of raw CSV records into an Arrow RecordBatch.

Each RawRecord.value contains one or more CSV lines (typically one line per record for streaming sources; may contain multiple lines for file-based sources).

Algorithm

1. Initialize one Arrow ArrayBuilder per schema column.
2. Concatenate all raw record bytes into a single buffer.
3. Create a csv::Reader with pre-configured settings.
4. For each CSV row:
   • Skip rows per skip_rows config.
   • For each field: apply the pre-computed CsvCoercion.
   • Handle field count mismatches per config.
5. Finish all builders and assemble into RecordBatch.

fn output_schema(&self) -> SchemaRef

Returns the Arrow schema produced by this decoder.

fn format_name(&self) -> &str

Returns the name of the format this decoder handles (e.g., "json").

fn decode_one(&self, record: &RawRecord) -> SchemaResult<RecordBatch>

Decodes a single raw record into an Arrow RecordBatch with one row.