Esc

Csv_Options

structextensionsince 0.36.1Source
  • macOS: Csv_Options works on macOS
  • Linux: Csv_Options works on Linux
  • Windows: Csv_Options works on Windows
  • Wasm: Csv_Options works on Wasm

Declaration

Csv_Options :: struct {
// an ASCII byte other than the quote, NUL, CR or LF    delimiter: u8 = #char ",";    quote: u8 = #char "\"";// treat the first record as column names; it is not returned    header: bool;// drop spaces around unquoted fields    trim: bool;// allow a quote inside an unquoted field    lax_quotes: bool;// fail with .WIDTH when a record's field count differs from the first    require_equal_width: bool;    max_field_bytes: int = 1 << 20;// per record    max_fields: int = 16384;    max_record_bytes: int = 16 << 20;// used when writing; reading accepts all three    line_ending := Csv_Line_Ending.CRLF;
// A typed row's strings (and the strings behind its optional pointers) are slices of the input when the field needed no unquoting, and copies when it had doubled quotes.    borrow: bool;}

How a CSV document is read and written.

Fields

borrow: bool
A typed row's strings (and the strings behind its optional pointers) are slices of the input when the field needed no unquoting, and copies when it had doubled quotes. The input must outlive the rows, and a row must not go through csv_row_free(*row): decode with context.allocator set to an arena or temp and reset that, or free through csv_row_free(*reader, *row), which leaves slices alone. Only for a slice input: a chunked source reuses its buffer, so csv_reader(.., source, ..) fails with .OPTIONS. Untyped records are always borrowed, so it changes nothing there. reader := csv_reader(User, input, .{ header = true, borrow = true },, temp); user, ok, error := csv_next(*reader,, temp); // user.name is a slice of `input` reset_temporary_storage();

See also