Skip to content
Chelis docs

CSV and JSON

Coral.Io provides CSV, JSON, and Parquet read/write helpers. Read inference covers int, float, bool, and string columns.

module Coral.BookIo
import Coral.Frame (from_pairs, ncols, int_col_of_list)
import Coral.Io (write_csv_frame, write_json_frame)
export (main)
def main() -> int64 = {
  frame = from_pairs([("id", int_col_of_list([cast(1, int64), cast(2, int64)])), ("price", FloatCol(to_tensor([cast(10.0, f32), cast(20.5, f32)]))), ("flag", BoolCol(neq(to_tensor([cast(1, int64), cast(0, int64)]), to_tensor([cast(0, int64), cast(0, int64)])))), ("city", StringCol(["london", "paris"]))])
  csv_unit = write_csv_frame(frame, "book-io.csv")
  json_unit = write_json_frame(frame, "book-io.json")
  ncols(frame)
}

Parquet

Section titled “Parquet”

read_parquet_frame and write_parquet_frame are exported by Coral.Io but currently call fail(...) at runtime. Parquet support remains gated on upstream Chelis-Lang/chelis shipping the runtime backing for Std.Io.Parquet; the v0.7.6 probe still resolves the import at check time but does not provide a usable runtime path. See docs/upstream-bugs.md for the full probe log.