CSV to SQL INSERT

A CSV to SQL INSERT converter reads CSV or TSV data and produces SQL INSERT statements you can paste into a database client, migration script, or seed file. ToolsSonic's converter parses headers from the first row, quotes identifiers according to your chosen dialect, escapes apostrophes in string values, infers numbers, booleans, and NULL tokens, groups rows into batches, and lets you download the result as a .sql file — all in your browser.

Six dialects are available: Generic SQL (standard double-quote identifiers), PostgreSQL (double-quote identifiers), MySQL (backtick identifiers), SQLite (double-quote identifiers), SQL Server (square-bracket identifiers), and Oracle (double-quote identifiers). Selecting a dialect sets the default identifier quoting mode, which you can override manually.

Identifier quoting wraps column and table names in dialect-appropriate delimiters to prevent conflicts with reserved words and names containing spaces or special characters. ToolsSonic supports four modes: Dialect default (backticks for MySQL, square brackets for SQL Server, double-quotes for everything else), Double quotes, Backticks, Square brackets, and None (review) — which emits bare identifiers and shows a warning for unsafe names.

String values are wrapped in single quotes, and any apostrophe inside a value is doubled (replaced with two consecutive single quotes, per the SQL standard). For example, the value O'Reilly becomes 'O''Reilly' in the generated SQL. This is correct for all six supported dialects. The converter does not use backslash escaping.

When the Infer numbers, booleans, NULL tokens option is enabled, the converter checks each cell value before wrapping it in single quotes. Cells that look like integers or decimals are emitted as bare numeric literals. Cells with the value true or false (case-insensitive) are emitted as TRUE or FALSE. Cells with the value NULL or \N are emitted as SQL NULL. You can disable inference to emit all non-empty values as quoted strings.

Empty cells have two policy options. SQL NULL emits the SQL keyword NULL for any cell that is empty after trimming. Empty string emits '' (two consecutive single quotes) to insert an empty string literal. Choose NULL for optional fields and columns that distinguish between missing and empty. Choose Empty string for required text columns that should never be NULL.

Batch size controls how many data rows appear in a single INSERT statement. One INSERT per row is the most portable and easiest to debug. Up to 25 rows and Up to 100 rows produce multi-row INSERT statements that are faster for bulk inserts in most databases. All rows puts every data row into one statement. Batch INSERT is supported by all six dialects but some databases impose limits on statement size.

Yes. Enter a schema name in the Schema field and a table name in the Table name field. ToolsSonic quotes each segment of the qualified path separately, producing output like "public"."users" for PostgreSQL or [dbo].[orders] for SQL Server. Leave the Schema field empty to use an unqualified table name.

Yes. The parser is quote-aware and follows RFC 4180 conventions. It handles commas inside double-quoted fields, embedded newlines inside quoted fields, and escaped double-quotes written as two consecutive double-quotes. It reports unclosed quoted fields and inconsistent row widths as warnings in the generation review panel.

The Auto-detect mode scans the first 20 lines of the input and scores four candidates: comma, tab, semicolon, and pipe. Whichever delimiter appears most frequently (outside quoted fields) is selected. You can also choose a delimiter manually: Comma (standard CSV), Tab (TSV), Semicolon (European CSV locales), or Pipe (log files).

No data is uploaded. CSV parsing and SQL generation run entirely in browser memory using vanilla JavaScript. There is no fetch() call, no XMLHttpRequest, no sendBeacon, no localStorage write, and no sessionStorage write. Closing the tab clears every byte of your data. This makes ToolsSonic safe for internal customer data, financial exports, and confidential database seeds.

No. The tool generates SQL text only. It never connects to a database, never executes SQL, never validates your schema or constraints, and never checks for duplicate rows or constraint violations. Generated SQL must be reviewed and run by you in your own database environment using a database client, migration tool, or script runner.

The Parsed data preview renders up to 20 data rows (excluding the header) as an HTML table so you can visually confirm that the parser correctly identified delimiters, quoted fields, and column alignment before generating SQL. It is not a database query result — it shows what the CSV parser read from your input.

The generation review panel shows a green Ready status and a summary of rows and columns when generation succeeds without warnings. When warnings occur, it shows a yellow Review status and a list of issues: unclosed quoted fields, inconsistent row widths, unsafe identifier names in None quoting mode, missing table name, and non-standard table name characters.

Paste TSV data (tab-separated) into the input panel. Set Input delimiter to Auto-detect or Tab (TSV) manually. The rest of the workflow is identical to CSV — set the table name, dialect, and batch size, then click Generate SQL. Tab characters inside quoted TSV fields are handled correctly by the parser.

Yes. Select MySQL from the SQL dialect dropdown. The converter automatically uses backtick identifier quoting for column names and table names. For example, a column named order becomes `order` in the generated SQL. You can also set the identifier quoting mode manually to Backticks on any other dialect.

Select PostgreSQL from the SQL dialect dropdown. Set the table and column names, then set Batch size to Up to 25 rows, Up to 100 rows, or All rows. PostgreSQL supports multi-row INSERT with a single INSERT INTO ... VALUES (...), (...), (...) syntax, which is what ToolsSonic generates.

The statistics strip shows seven values: Data rows (rows after the header), Columns (header field count), INSERT statements (number of generated statements), Values (rows × columns), SQL bytes (size of the generated SQL), Inferred literals (how many values were emitted as numbers, booleans, or NULL rather than strings), and Warnings (total diagnostic count).

Yes. Click Download .sql to save the generated SQL as a file named after your table name, for example users.sql. The file uses UTF-8 encoding. Copy SQL copies the output to the clipboard. Both actions happen in the browser with no server involved.