CsvParseStream: negative fieldsPerRecord doesn't work with skipFirstRow or columns

[mfulton26]

Describe the bug

According to fieldsPerRecord documentation, I should be able to parse CSV with variable length records. I can successfully do so but not when I use skipFirstRow and/or columns.

working example

import { CsvParseStream } from "jsr:@std/[email protected]/parse-stream";
import { assertEquals } from "jsr:@std/[email protected]/equals";
import { assertType, IsExact } from "jsr:@std/[email protected]/types";

const source = ReadableStream.from([
  "name,age\n",
  "Alice,34\n",
  "Bob\n", // incomplete record
]);
const stream = source.pipeThrough(new CsvParseStream({ fieldsPerRecord: -1 }));
const result = await Array.fromAsync(stream);

assertEquals(result, [
  ["name", "age"],
  ["Alice", "34"],
  ["Bob"],
]);
assertType<IsExact<typeof result, string[][]>>(true);

Steps to Reproduce

1. Create a script with the contents of any of the following examples

   not working with skipFirstRow

   import { CsvParseStream } from "jsr:@std/[email protected]/parse-stream";
   import { assertEquals } from "jsr:@std/[email protected]/equals";
   import { assertType, IsExact } from "jsr:@std/[email protected]/types";
   
   const source = ReadableStream.from([
     "name,age\n",
     "Alice,34\n",
     "Bob\n", // incomplete record
   ]);
   const stream = source.pipeThrough(
     new CsvParseStream({
       fieldsPerRecord: -1,
       skipFirstRow: true,
     }),
   );
   const result = await Array.fromAsync(stream);
   
   assertEquals(result, [
     { name: "Alice", age: "34" },
     { name: "Bob", age: undefined },
   ]);
   assertType<IsExact<typeof result, Record<string, string | undefined>[]>>(true);
   not working with columns

   import { CsvParseStream } from "jsr:@std/[email protected]/parse-stream";
   import { assertEquals } from "jsr:@std/[email protected]/equals";
   import { assertType, IsExact } from "jsr:@std/[email protected]/types";
   
   const source = ReadableStream.from([
     "Alice,34\n",
     "Bob\n", // incomplete record
   ]);
   const stream = source.pipeThrough(
     new CsvParseStream({
       fieldsPerRecord: -1,
       columns: ["name", "age"],
     }),
   );
   const result = await Array.fromAsync(stream);
   
   assertEquals(result, [
     { name: "Alice", age: "34" },
     { name: "Bob", age: undefined },
   ]);
   assertType<
     IsExact<typeof result, Record<"name" | "age", string | undefined>[]>
   >(true);
   not working with skipFirstRow and columns

   import { CsvParseStream } from "jsr:@std/[email protected]/parse-stream";
   import { assertEquals } from "jsr:@std/[email protected]/equals";
   import { assertType, IsExact } from "jsr:@std/[email protected]/types";
   
   const source = ReadableStream.from([
     "name,age\n",
     "Alice,34\n",
     "Bob\n", // incomplete record
   ]);
   const stream = source.pipeThrough(
     new CsvParseStream({
       fieldsPerRecord: -1,
       skipFirstRow: true,
       columns: ["name", "age"],
     }),
   );
   const result = await Array.fromAsync(stream);
   
   assertEquals(result, [
     { name: "Alice", age: "34" },
     { name: "Bob", age: undefined },
   ]);
   assertType<
     IsExact<typeof result, Record<"name" | "age", string | undefined>[]>
   >(true);

2. Execute the script with deno run

3. See error:

   error: Uncaught (in promise) Error: Syntax error on line 3: The record has 1 fields, but the header has 2 fields
       throw new Error(
             ^
       at convertRowToObject (https://jsr.io/@std/csv/1.0.5/_io.ts:233:11)
       at CsvParseStream.#pull (https://jsr.io/@std/csv/1.0.5/parse_stream.ts:460:28)

Expected behavior

1. The scripts should run successfully.
2. TypeScript type checking should also pass.
3. Record<T, string | undefined> or Partial<Record<T, string>> should be used instead of Record<T, string>

Environment

• OS: MacOS 15.3.1
• deno version: 2.2.0
• std version: jsr:@std/[email protected]