01From delimited text
The first row supplies unique, non-empty column names. Delimiter detection supports comma, tab and semicolon inputs. Quoted commas, escaped quotes and multiline cells are supported. Each data row must match the header width. Empty lines are skipped. Cells stay strings, so identifiers such as 0012 keep their zeros.
02From structured rows
Use an array of flat objects. Columns are the union of their property names, in first-seen order. Missing or null values become empty cells. Nested objects, arrays, duplicate property names and unsafe large numeric integers are rejected. Store large identifiers as quoted strings. This conversion is not type-reversible: a number and a numeric string can look identical in the exported table.
03Spreadsheet safety
When exporting delimited text, the default option prefixes cells or headers beginning with =, +, -, @, tab or carriage return. This changes their text, including negative numbers. The result summary reports the count. Turning the option off preserves that text but can allow spreadsheet formula execution; inspect untrusted input before opening it.