Method and exact limits

Whole parsed data as text

The original UTF-8 CSV bytes are read directly into a local worker. One leading UTF-8 encoding marker is treated as a marker; later U+FEFF characters are data. Quotes, escaped quotes, explicit delimiters and record boundaries are parsed. Every parsed field position and value is emitted as a literal text cell on one worksheet named Data. The first row is retained, not inferred as a header.

The exporter does not restore digits, clean data, guess encoding or delimiters, sample rows, apply formulas, create macros or fetch external workbook relationships. Empty fields are retained. A completed download is a real XLSX ZIP package, not an HTML file renamed to XLSX.

Fixed bounds

Maximum1MiB input,2,000rows,64columns,32,767UTF-16 units per field and16MiB output. Core processing is limited to1.5seconds; a separate1.9-second worker watchdog stops stalled work. Browser scheduling is not a real-time guarantee. Over-limit or malformed input stops, without truncated success.

Dates and amounts remain text. Native numeric calculations or date sorting require explicit downstream conversion, which can change spelling or numeric precision. CSV line endings pasted into a textarea can be normalized by the browser; original file mode is the byte-fidelity route.

Verification boundaries

Actual bounded output values, positions, dimensions and text types were checked with independent workbook readers. Current Chromium input/download/reset/privacy/mobile behavior was checked with synthetic files. This is not certification of every spreadsheet version, operating system or browser. The input file and saved output remain under your control.