SQL column type inference
Inference examines all nonempty values in a column. Boolean columns use BOOLEAN or SQLite INTEGER, integer columns use BIGINT, and decimal columns use a dialect-specific numeric type. A leading-zero identifier such as 007 remains text. Mixed types become TEXT. Choose text-only columns when identifiers, exact formatting or downstream import rules matter more than inferred numeric types.
Quoted names and escaped SQL values
Identifiers use double quotes for PostgreSQL and SQLite, and backticks for MySQL. Embedded identifier quotes are doubled. Text apostrophes are doubled, so O'Brien becomes O''Brien inside a string literal; MySQL backslashes are escaped as well. Table names with dots are treated as separate quoted name components. The result is generated SQL text, not a connection to a database.
Empty cells, headers and CSV quoting
The first row normally supplies column names; without a header, names become column_1 and onward. Empty-as-NULL is optional. Turning it off preserves empty strings. Duplicate or empty headers receive distinct names. Rows with inconsistent column counts are rejected so a delimiter error cannot shift values into the wrong SQL columns.
Format references
PostgreSQL SQL syntax · SQLite data types · MySQL string literals
File content is read and converted in the browser. Conversion does not upload the source to a server.