{{ summaryTitle }}
{{ summaryValue }}

{{ summaryLine }}

Dialect{{ dialectLabel }} Columns{{ resultsReady ? computation.values.column_count : '—' }} Delimiter{{ resultsReady ? computation.values.delimiter_label : '—' }}
One source row per line; quoted fields may contain delimiters or line breaks.
{{ sourceStatus || 'Drop CSV, TSV, or TXT onto the textarea.' }}
Auto resolves to a visible delimiter label in the summary and audits; pin a delimiter when the source is ambiguous.
Examples: public.users, dbo.import_queue, audit_event.
Choose the database that will receive the script.
Use one-per-row for review logs and multi-row for compact fixtures.
The active column order controls every generated value tuple.
Duplicate names are suffixed and flagged for review.
Use text mode for leading-zero IDs, SKUs, ZIP codes, and other exact text codes.
Confirm the destination column semantics before running the script.
Turn this option on when quote identifiers is required.
{{ quote_identifiers ? 'On' : 'Off' }}
Turn this option on when statement terminators is required.
{{ include_semicolons ? 'On' : 'Off' }}
The neutral default is off; enabling it changes presentation, not parsed values.
{{ add_review_header ? 'Included' : 'Off' }}
{{ sqlScript }}

The chart renderer is unavailable. Literal counts remain available in the column audit.

RowStatusCellsReview noteStatement previewCopy
{{ row.row }}{{ row.status }}{{ row.cells }}{{ row.note }}{{ row.preview }}
ColumnSQL identifierFilledBlankLiteral typesSampleCopy
{{ row.column }}{{ row.identifier }}{{ row.filled }}{{ row.blank }}{{ row.types }}{{ row.sample }}
SQL Formatter Format SQL in your browser with dialect-aware layout and an audit that flags fallback output, unbalanced parentheses, and risky reuse.
SQLite Manager Open a SQLite file in your browser and inspect schema or rows before running guarded SQL and saving deliberate edits as a new local copy.
.env File Generator Generate reviewed .env and .env.example files with key cleanup and secret-aware placeholders while flagging quote or duplicate risks locally.
Database Migration Risk Checker Check SQL migration risk before release with engine and table context, exposing destructive changes plus lock hazards, rewrites and backfills.

Turning a flat set of records into database rows is mainly an alignment problem. Every value must land under the intended column, every identifier must be valid for the destination database, and every literal must keep its original meaning. A script can be valid SQL and still put data in the wrong place.

An INSERT statement pairs an ordered column list with one or more ordered value tuples. The first value belongs to the first column, the second value to the second column, and so on. Headers, delimiters, embedded commas, quoted line breaks, and blank cells all affect that alignment before SQL syntax is produced.

Identifier
A schema, table, or column name. Its quoting marks depend on the SQL dialect.
Literal
A value written into SQL, such as a quoted string, number, Boolean value, or NULL.
Delimited record
A source row split into fields by a comma, tab, semicolon, or pipe while respecting quoted fields.

Text that resembles another data type needs special care. Product codes, postal codes, and account numbers may contain only digits but still belong in text columns. Blank cells also need an explicit policy because an empty string is a real string value, while NULL represents a missing or unknown value.

Generated statements do not know the destination schema, constraints, triggers, permissions, or current rows. Review the column order and literals, then test the script inside a transaction or a disposable database. Application code should still use parameterized queries for untrusted runtime values rather than building SQL strings.

How to Use This Tool:

Start with the destination table and the exact record shape that table expects.

  1. Paste Source rows or choose a local CSV or text file. Keep the source below 2 MB and select a delimiter when automatic detection would be ambiguous.
  2. Enter the Table name, choose the target SQL dialect, and either supply Columns in destination order or use the first source row as the column list.
  3. Choose Value typing, the Blank cells policy, and one statement per row or one multi-row statement. Use text mode when numeric-looking values must remain text.
  4. Read Row review and Column audit before copying the SQL. Fix shifted row widths, duplicate column renames, unsafe unquoted identifiers, or unexpected literal types before execution.

Interpreting Results:

The statement count describes the generated script, not rows already written to a database. In one-per-row mode it normally matches the source row count; multi-row mode produces one statement, with Oracle using an INSERT ALL block.

  • A Ready row has the same number of source cells as active columns. It does not prove that database types or constraints will accept those values.
  • A Review row identifies missing cells filled by the blank policy or extra cells omitted from the tuple.
  • The literal profile is a quick check for accidental type inference. Compare it with the destination schema, especially for identifiers that contain leading zeros.

Technical Details:

Delimited input must be parsed before it can be quoted as SQL. The transformation keeps quoted delimiters and line breaks inside their field, establishes one column order, maps every row to that order, and only then renders dialect-specific syntax.

Transformation Core:

StageExact behaviorReview consequence
Delimiter resolutionA fixed comma, tab, semicolon, or pipe is used, or the first 20 non-empty lines are scored for a consistent delimiter outside quoted fields.Pin the delimiter when several characters occur regularly.
Record parsingDouble-quoted fields may contain delimiters or line breaks; a doubled quote becomes one quote. An unclosed quoted field is retained and warned.Inspect the affected record before trusting later rows.
Column selectionColumns come from a comma-separated list or the first parsed row. Duplicate names receive _2, _3, and later suffixes.Confirm that renamed columns really exist at the destination.
Row alignmentMissing trailing cells become NULL or empty strings according to policy; extra cells are ignored.Any width difference becomes a row-review note.
Literal renderingValues become strings, numbers, Booleans, NULL, or empty strings under the rules below.Compare inferred types with column semantics.
Statement renderingEach tuple becomes its own INSERT, or tuples are combined into one multi-row statement. Oracle multi-row output uses INSERT ALL and SELECT 1 FROM DUAL.Run the final dialect output only in its intended database.

Identifiers and values use different quoting rules. Identifier quoting protects names such as reserved words or names containing spaces; string quoting protects literal text.

Selected dialectIdentifier formEscaped closing mark
MySQL and BigQuery`identifier`A backtick is doubled.
SQL Server[identifier]A closing bracket is doubled.
PostgreSQL, SQLite, Oracle, Snowflake, and ANSI SQL"identifier"A double quote is doubled.
Source valueInfer modeText mode
Case-insensitive NULLNULLNULL
Blank or missing cellNULL or '', according to the blank policy
true or falseTRUE/FALSE, except SQLite, SQL Server, and Oracle emit 1/0Quoted text
Recognized integer, decimal, or exponent formUnquoted numeric literalQuoted text
Other textSingle-quoted text with embedded apostrophes doubled

The generated text is deterministic for the same source and settings. It does not connect to the database, inspect column types, validate foreign keys, detect duplicate keys, or add conflict-handling clauses. Those checks belong in the destination database and deployment workflow.

Privacy Notes:

Rows and local files are processed in the browser for generation and are not uploaded by this tool. Database extracts can still contain personal or confidential data, so review what you paste, protect downloaded scripts, and clear shared clipboards or browser sessions when the work is finished.

Worked Examples:

A quoted comma stays in one field

With columns id, name, the source row 3,"Chen, Mei" produces two values rather than three. The name becomes 'Chen, Mei', and the row remains ready because its cell count matches the two columns.

A short row is held for review

If five columns are active but a row supplies four cells, the final value follows the blank policy. The script is still generated, but Row review reports the missing cell so the omission is not mistaken for a complete record.

References: