std/csv

Strict CSV rows, following RFC 4180 and Go encoding/csv. CRLF is normalized to LF, including inside quoted fields. Blank lines are skipped; spaces are data. Errors use one-based scalar line/column. Streams consume chunks only on demand, retain at most one row plus the current chunk, and end after the first error. Pulls are replayable.

OptionsCopy Anchor Link

Options = (delimiter = Int, fields = Int, crlf = Bool)

CsvErrorCopy Anchor Link

CsvError = (line = Int, column = Int, reason = String)

optionsCopy Anchor Link

A single scalar delimiter, expected width (0 infer, negative variable), and output line ending. Quotes, NUL, CR, LF, U+FFFD are invalid delimiters. csv.options(";", -1, True).

options = (delimiter : String, fields : Int, crlf : Bool) -> Result(

defaultsCopy Anchor Link

Comma delimiter, infer row width, LF output.

defaults = () -> Options { Options(delimiter = 44, fields = 0, crlf = False) }

parseCopy Anchor Link

Parses comma-separated text, inferring the width from the first row. csv.parse("a,b\n1,2\n") gives Ok([["a", "b"], ["1", "2"]]).

parse = (input : String) -> Result(List(List(String)), CsvError) {

parse_withCopy Anchor Link

Parses with an explicit delimiter and row-width policy.

parse_with = (input : String, config : Options) -> Result(

parse_headerCopy Anchor Link

Header-mode convenience parser: csv.parse_header("name\ni\n", csv.defaults()).

parse_header = (input : String, config : Options) -> Result(

printCopy Anchor Link

Prints minimal quoting, with one LF after every row. csv.print([["a", "b,c"]]) gives "a,\"b,c\"\n".

print = (rows : List(List(String))) -> String { print_with(rows, defaults()) }

print_withCopy Anchor Link

Prints with the configured delimiter and LF/CRLF. Empty rows print as "".

print_with = (rows : List(List(String)), config : Options) -> String {

rowsCopy Anchor Link

Parses a stream of arbitrary text chunks, emitting one row per pull. csv.rows(stream.from_list(["a,", "b\n"]), csv.defaults()).

rows = <E, N>(

recordsCopy Anchor Link

Uses the first row as unique column names. Empty input yields no records.

records = <E, N>(

decodeCopy Anchor Link

Decodes a named row as a JSON object of strings using an ordinary schema. Compose schema.string with schema.map for domain conversions.

decode = <A, E>(