Save and Fetch (CSV)

A Table can be written to and read back from CSV. Writing is schema-free; reading requires you to declare the type of each column up front.

Writing — write_csv

write_csv(path) serializes the table to a CSV file at path.

use rayforce::{Runtime, Table, Value};

let _rt = Runtime::new()?;

let t = Table::new(
    &["sym", "price", "size"],
    &[
        Value::sym_vec(&["AAPL", "MSFT", "GOOG"]),
        Value::vec(&[101.5f64, 202.0, 303.25]),
        Value::vec(&[10i64, 20, 30]),
    ],
)?;

t.write_csv("/tmp/trades.csv")?;
# Ok::<(), rayforce::RayError>(())

Reading — Table::read_csv

Table::read_csv(&[type_tokens], path) reads a CSV, parsing each column according to the type token at the same position. There must be one token per column.

# use rayforce::{Runtime, Table, Value};
# let _rt = Runtime::new()?;
# let t = Table::new(&["sym","price","size"], &[Value::sym_vec(&["AAPL","MSFT","GOOG"]), Value::vec(&[101.5f64,202.0,303.25]), Value::vec(&[10i64,20,30])])?;
# t.write_csv("/tmp/trades.csv")?;
let loaded = Table::read_csv(&["SYMBOL", "F64", "I64"], "/tmp/trades.csv")?;

assert_eq!(loaded.shape(), (3, 3));
assert_eq!(loaded.column("price")?.as_slice::<f64>()?, &[101.5, 202.0, 303.25]);
# Ok::<(), rayforce::RayError>(())

Type tokens

Tokens are conventionally written in UPPERCASE, but matching is case-insensitive ("i64" works as well as "I64").

Token	Column type
B8	boolean
U8	byte
I16	16-bit integer
I32	32-bit integer
I64	64-bit integer
F32	32-bit float
F64	64-bit float
DATE	date
TIME	time
TIMESTAMP	timestamp
SYMBOL	symbol
STR	string
GUID	GUID

Aliases

Alias	Resolves to
SYM	SYMBOL
STRING	STR
BOOL, BOOLEAN	B8

# use rayforce::{Runtime, Table};
# let _rt = Runtime::new()?;
// these schemas are equivalent
# let _ = || -> rayforce::Result<()> {
let a = Table::read_csv(&["SYM", "F64", "I64"], "/tmp/trades.csv")?;
let b = Table::read_csv(&["symbol", "f64", "i64"], "/tmp/trades.csv")?;
# let _ = (a, b); Ok(()) };
# Ok::<(), rayforce::RayError>(())

Reading order

read_csv matches tokens to columns by position, so the token order must follow the CSV's column order. See Data Types for what each token represents.

For columnar on-disk storage, see Splayed and Parted.