Data formats

Complete specification for CSV, JSON, and JSONL/NDJSON input.

CSV / TSV (.csv, .tsv)

RFC 4180 compliant parser with extensions.

Delimiter auto-detection

First line scanned for delimiter: comma (,), semicolon (;), tab (\t). No config needed.

Force delimiter:

// lakreet.config.json
{ "delimiter": ";" }

Quoting (RFC 4180)

Case	Input	Parsed value
Basic	hello	hello
With comma	"hello, world"	hello, world
With newline	"line1\nline2"	line1\nline2
Escaped quote	"He said ""Hi"""	He said "Hi"
Leading/trailing space	" padded "	padded

Quotes preserved verbatim (not stripped).

BOM handling

UTF-8 BOM (\uFEFF) stripped automatically. Works with Excel "UTF-8 with BOM".

Line numbers

Errors reference 1-based line numbers (header = line 1):

Line 5, column "Opacity": "abc" is not a number.

Column count

All rows must match header column count. Mismatch = error:

Line 8: expected 3 columns, found 4.

Comment columns

Headers starting with # ignored:

Source Text,Opacity,#internal_id
Budi,100,row-001

JSON (.json)

Array of objects

[
  { "Source Text": "Budi", "Opacity": 100 },
  { "Source Text": "Sari", "Opacity": 60 }
]

Object with rows key

{
  "rows": [
    { "Source Text": "Budi", "Opacity": 100 },
    { "Source Text": "Sari", "Opacity": 60 }
  ]
}

Type preservation

JSON type	Control type	Example
string	text	"Budi"
number	slider	100, 75.5
boolean	checkbox	true, false
null	text	"" (empty)

Missing keys = validation error (unless --empty-as-default for text).

JSONL / NDJSON (.jsonl, .ndjson)

One JSON object per line. Ideal for large datasets.

{"Source Text": "Budi", "Opacity": 100}
{"Source Text": "Sari", "Opacity": 60}
{"Source Text": "Andi", "Opacity": 25}

• Empty lines skipped
• Each line = valid JSON
• Line numbers in errors = JSONL line number

Validation rules (all formats)

Rule	Error example
Column matches control	Column "Foo" is not a control
Slider = finite number	"abc" is not a number
Checkbox = boolean	"yes" is not a boolean
Media file exists	file not found: /path/missing.png
Required columns present	column "Source Text" is required

Next steps

• Prepare a CSV — practical guide
• Use JSON or JSONL — how-to
• Map columns — header → control mapping
• Control types — input formats per type