CSV to SQL

No. Your CSV is never sent to our servers - all parsing, conversion to SQL, escaping, and copy/download happens directly in your browser on your device using JavaScript. The only network request the tool ever makes is the one you trigger explicitly with URL Import.

Four: ANSI / PostgreSQL / SQLite / Oracle (double-quoted identifiers, the default), MySQL / MariaDB (backtick identifiers), SQL Server (square-bracket identifiers), and No quoting (bare identifier names). Pick the dialect that matches your target database from the Dialect dropdown in the output toolbar - the generated SQL switches identifier wrapping accordingly so it pastes cleanly into your database tool.

When Delimiter is set to Auto, the tool checks the first 5 non-empty lines of your CSV against five candidates (comma, tab, semicolon, pipe, colon) and picks the one that produces a consistent column count across those lines. Pure first-line frequency is not used, so a line containing URLs or ISO timestamps will not trick it into picking colon. You can always override the choice manually.

It only transforms cells in three predictable, opt-in or type-driven ways: empty or whitespace-only cells become NULL or '' (your choice via Empty cells setting); cells whose entire trimmed value is null/true/false (any case) become unquoted SQL NULL/TRUE/FALSE; cells matching -?\d+(\.\d+)? become unquoted numbers. Everything else is wrapped in single quotes with embedded apostrophes doubled. Leading and trailing whitespace inside string cells is preserved exactly unless you turn on the Trim cell whitespace checkbox.

They are converted to the SQL keywords NULL or TRUE/FALSE - this is intentional so common CSV exports work out of the box. If you actually want the literal string "NULL" or "True" inserted as text (for example, a person's last name), wrap that cell in single quotes inside the CSV (so the value the parser sees is 'NULL') or post-process the generated SQL.

The numeric detector matches a strict pattern: optional minus, digits, optional decimal point and more digits. Scientific notation (1e10), thousands separators (1,000.00), leading plus signs (+5), and hex literals (0xFF) are treated as strings and quoted. If you need them inserted as numbers, normalize them in your spreadsheet first or run the SQL through a search-and-replace before executing.

Plain text only. The Upload picker is scoped to .csv, .tsv, and .txt; drag-and-drop accepts the same; URL Import calls fetch() directly from your browser, so the target URL must allow cross-origin requests via CORS headers. You can also type or paste CSV directly into the textarea.

No. Once you click Generate SQL, your input CSV, generated output, and every setting (delimiter, header row, table name, schema, dialect, empty-cell handling, batch mode, trim whitespace) are saved to your browser's localStorage under the key "csv-to-sql-data" and restored automatically on the next visit. Click Clear to wipe the input, output, and persisted data immediately.

Supports comma, tab, semicolon, pipe, and colon delimiters. When Delimiter is set to Auto, the tool examines the first 5 non-empty lines of your CSV against each candidate and picks the one that produces a consistent column count across those lines - not just the most-frequent character on line 1. This means a CSV with URLs (https://) or ISO timestamps (12:34:56) in the data will not be mis-detected as colon-separated. You can always override the choice manually.

Pick the dialect that matches your target database from the Dialect dropdown in the output toolbar. ANSI / PostgreSQL / SQLite / Oracle wraps identifiers in double quotes (the SQL standard, also accepted by SQL Server with QUOTED_IDENTIFIER ON and by MySQL with ANSI_QUOTES). MySQL / MariaDB uses backticks. SQL Server uses square brackets. No quoting leaves identifiers bare. Embedded special characters are escaped per dialect (doubled double quotes, doubled backticks, doubled closing brackets).

Quoted fields are parsed correctly, including embedded delimiters and newlines inside double-quoted cells, and embedded double quotes escaped by doubling (""). CRLF and LF row terminators are both handled. The parser is hand-written, not regex-based, so edge cases like trailing commas and unbalanced quotes degrade gracefully.

When Header row is on (default), the first row of your CSV is used as the column names in the generated SQL. When off, columns are auto-named col1, col2, col3, ... and every row is treated as data.

If a data row has fewer cells than the header, the missing cells are padded with empty values (which become NULL or '' depending on your Empty cells setting). If a row has more cells than the header, the extra cells are dropped. This means a slightly ragged CSV still produces well-formed SQL with consistent column counts.

Each cell is classified before being escaped: a cell whose trimmed value is null/true/false (any case) becomes the SQL keyword NULL/TRUE/FALSE; a cell matching -?\d+(\.\d+)? becomes an unquoted number; everything else is wrapped in single quotes with embedded apostrophes doubled (' -> ''). Note that scientific notation, hex literals, and thousands-separator numbers are treated as strings - normalize them in your spreadsheet first if you need them inserted as numbers.

An Empty cells dropdown in the output toolbar lets you choose how blank or whitespace-only cells are written: NULL inserts the SQL NULL keyword (default), Empty string '' inserts a single-quoted empty string. Pick based on whether your column is nullable or NOT NULL DEFAULT ''.

By default, leading and trailing whitespace inside string cells is preserved exactly - so codes like " A123" or padded fixed-width fields are inserted as you typed them. Toggle the Trim cell whitespace checkbox above the Generate SQL button to strip whitespace before escaping. The setting persists in localStorage between visits.

Toggle Batch mode in the output toolbar to combine all rows into a single multi-row SQL statement using the (...), (...), (...) tuple syntax instead of one statement per row. Batch mode is significantly faster for large bulk loads in PostgreSQL, MySQL, and SQLite, and is required for some databases that limit transaction overhead.

Set the target table name (default my_table) and an optional schema or database prefix (e.g. public, dbo, mydb). The schema prefix is wrapped in the same dialect-specific quoting as the table name, producing fully qualified references like "public"."my_table" for ANSI or [dbo].[my_table] for SQL Server.

The generated SQL is rendered with color-coded tokens: SQL keywords in blue, identifiers (backtick / double-quote / bracket wrapped) in amber, single-quoted strings in green, and numbers in purple. The highlighter recognizes all four dialect quoting styles, not just MySQL backticks.

Click Copy SQL to copy the entire generated output to your clipboard via navigator.clipboard.writeText(). The button briefly changes label to "Copied!" for 1.5 seconds as confirmation, then reverts.

Click Download SQL to save the output as a .sql file named after your table (for example my_table.sql, or insert.sql when no table name is set). The blob is created locally and the object URL is revoked immediately after the download triggers, so there is no memory leak.

Four ways to load CSV data: type or paste straight into the textarea, click Upload to pick a .csv / .tsv / .txt file, drag a file onto the input area, or click URL Import to reveal a URL field that fetches the CSV directly from your browser to the target endpoint with no proxy. URL Import requires the target to allow cross-origin requests via CORS headers.

Once you click Generate SQL, your input CSV, the generated output, and every setting (delimiter, header row, table name, schema, dialect, empty-cell handling, batch mode, trim whitespace) are saved to your browser's localStorage under the key "csv-to-sql-data" and restored automatically on the next visit. The save only fires after a successful generate, so a partially typed CSV does not overwrite your previous output.

Click Clear to wipe the input textarea, the rendered SQL, the error message, the URL Import bar, and reset every setting back to defaults (Auto delimiter, Header row on, my_table, no schema, ANSI dialect, NULL empty cells, Batch off, Trim off) plus the persisted localStorage entry in a single action.