Documentation

Read a CSV file with the default options.

Example

ghci> D.fastReadCsv "./data/taxi.csv"

Alias for fastReadCsv.

Example

ghci> D.readCsvFast "./data/taxi.csv"

Read a TSV file with the default options.

Example

ghci> D.fastReadTsv "./data/taxi.tsv"

Alias for fastReadTsv.

Example

ghci> D.readTsvFast "./data/taxi.tsv"

Like fastReadCsv but takes a ReadOptions record. Use this when you need to tune ragged-row handling, unclosed-quote handling, the whitespace-trimming knob, or the column selection; otherwise stick with fastReadCsv. The separator comes from columnSeparator, so this is a CsvReader the lazy scan can drive.

Example

ghci> D.fastReadCsvWithOpts D.defaultReadOptions{D.readColumns = Just ["id", "name"]} "./data/customers.csv"

TSV counterpart to fastReadCsvWithOpts; pins columnSeparator to tab.

Example

ghci> D.fastReadTsvWithOpts D.defaultReadOptions{D.safeRead = D.MaybeRead} "./data/taxi.tsv"

Read a CSV and coerce each column to the type declared in the supplied Schema. Schema columns bypass inference: parsed straight from file bytes as the declared type, avoiding row-1 misclassification.

Example

ghci> schema = D.makeSchema [("id", D.schemaType @Int)]
ghci> D.fastReadCsvWithSchema schema "./data/customers.csv"

Read a CSV and return only the named columns, in the order given. Shorthand for readColumns. The SIMD scan and delimiter classification still run over the whole file, but unnamed columns are never extracted, parsed or allocated. Naming a column the file does not have raises a ColumnsNotFoundException.

Example

ghci> D.fastReadCsvProj ["id", "name"] "./data/customers.csv"

In-memory version of fastReadCsv.

Example

ghci> D.fastReadCsvFromBytes "id,name\n1,Ada\n"

In-memory version of fastReadCsvWithSchema.

Example

ghci> schema = D.makeSchema [("id", D.schemaType @Int)]
ghci> D.fastReadCsvFromBytesWithSchema schema "id,name\n1,Ada\n"

Reads via a private (WriteCopy) mmap so the buffer is never copied. The separator byte and columnSeparator in opts must agree; use fastReadCsvWithOpts unless you need to pass a delimiter byte directly.

Example

ghci> D.readSeparated 44 D.defaultReadOptions "./data/taxi.csv"  -- 44 = ','

Identical to readSeparated but reads from an already-loaded byte buffer (e.g. a slice of a memory-mapped file). Zero-copy: the bytes are only read, never modified or grown.

Example

ghci> D.readSeparatedFromBytes 44 D.defaultReadOptions "id,name\n1,Ada\n"  -- 44 = ','

readSeparatedFromBytes with a forced chunk count, bypassing the capability/size gate. Intended for tests and benchmarks that need to pin the parallel split; 1 forces the sequential path.

Example

ghci> D.readSeparatedFromBytesChunks 1 44 D.defaultReadOptions "id,name\n1,Ada\n"  -- 44 = ','

Locate delimiter byte positions in the first originalLen bytes of csvFile. Treats an unclosed quoted field at EOF as a hard error; callers that want to suppress the exception can use getDelimiterIndicesPolicy with BestEffort.

Exceptions raised by the fast CSV parser. Catchable with catch or try.