CSV (serde/csv)

The serde/csv module provides blazing-fast tabular data parsing and serialization with full RFC 4180 compliance (quoted fields, embedded commas, newlines within quotes).

Modules

File	Purpose
parse.vx	CSV text → CsvValue row/field parser
value.vx	CsvValue type representing parsed rows and fields
encoder.vx	Struct → CSV row via Serde Serializer contract
decoder.vx	CSV row → Struct via Serde Deserializer contract

Parsing CSV

import { parse } from "serde/csv";

let input = "name,age,city\nAlice,30,Istanbul\nBob,25,Berlin";
let rows = parse(input);
// rows[0] = ["name", "age", "city"]
// rows[1] = ["Alice", "30", "Istanbul"]

Decoding into Structs

import { decode } from "serde/csv";

struct Person {
    name: string
    age: i32
    city: string
}

let csv = "name,age,city\nAlice,30,Istanbul";
let! person = Person { };
decode<Person>(csv, &person);
println("{person.name} lives in {person.city}");

Encoding from Structs

import { encode } from "serde/csv";

let person = Person { name: "Charlie", age: 28, city: "Tokyo" };
let csv_row = encode<Person>(&person);
// → "Charlie,28,Tokyo"

Performance

CSV parsing uses native C-string strtod conversion for numeric fields and direct byte scanning for delimiters. This bypasses the heavier JSON-style tokenizer, achieving benchmark results of ~806,302 ops/s for simple row parsing. Quoted field handling adds minimal overhead since the parser only branches on the " byte.