How CSV Headers and Data Types Affect JSON Output

Every value stays a string

CSV to JSON never infers numbers, booleans, dates, or null from cell content. A value such as 00123 stays "00123", a large numeric identifier is copied through exactly with no precision loss, and text like true or FALSE stays a literal string rather than becoming a JSON boolean.

Blank and duplicate headers get deterministic names

A blank header cell gets a positional fallback name such as column_2, based on its position in the row. Duplicate header names — including a real header that happens to collide with a generated fallback, such as a column literally named column_2 — are resolved with a single pass over the complete header list, so the real header keeps the bare name and the generated one is suffixed instead.

• name, name, name becomes name, name_2, name_3
• name, (blank), age becomes name, column_2, age
• name, (blank), column_2 becomes name, column_2, column_2_2

Short and long rows never lose data

A row with fewer values than the header row is padded with empty strings, so every JSON object has the same complete set of keys — no object is ever missing a key that others have. A row with more values than the header row gets additional positional keys, such as column_5, rather than having the extra values silently dropped.

Blank rows and the header-only case

A fully blank row in the middle of the file is preserved as an object whose values are all empty strings, rather than being silently skipped — except for one trailing blank line at the very end of the file, which is treated the same way spreadsheet software treats a final newline. A CSV with only a header row and no data rows is rejected when "First row contains headers" is enabled; turn the setting off to convert the single row as data instead.