Last updated: 2026-07-22

Data Import

Load a CSV, Excel, or JSON file straight into a database table — no hand-written CREATE TABLE or INSERT statements required. The Data Import wizard walks you through picking a file, choosing a destination, mapping columns, and running the import, across all five writable engines.

Opening the Data Import wizard

There are several ways to start an import:

• Main toolbar — click More, then Import Data...
• Keyboard shortcut — Cmd+I (macOS) or Ctrl+I (Windows/Linux)
• Command palette — press Cmd/Ctrl+Shift+P and run Import Data...
• Drag and drop — drop a supported file onto the app window
• Object Explorer context menu — right-click a table and choose Import Data into Table... to pre-fill it as the existing-table target, or right-click a database and choose Import Data... to open the wizard bound to that database

Importing needs an active connection (a database isn't required up front — you can pick one while setting the destination). Azure Data Explorer (Kusto) connections are read-only and don't offer Import Data, since there's no writable table to import into.

Supported files

• CSV / TSV / TXT — delimiter is auto-detected
• Excel (.xlsx) — pick the sheet to import from a Sheet selector
• JSON / NDJSON — a JSON array of objects, or newline-delimited JSON

The destination is either a brand-new table or an existing one (append), across SQL Server, PostgreSQL, MySQL, Oracle, and SQLite.

The four steps

The wizard is a single workspace tab with four steps — Source · Target · Columns · Import:

1

Source

Pick a file and Jam SQL Studio parses it immediately: delimiter and column kinds for CSV/TSV/TXT, a sheet picker for Excel, or array/NDJSON detection for JSON. Encoding uses BOM auto-detection with a manual Encoding picker for legacy encodings (UTF-8, UTF-16 LE/BE, Latin-1, Windows-1250/1252, ISO-8859-2, GBK, Shift-JIS) — this isn't universal charset auto-detection, so if text looks garbled, re-pick the encoding manually.

A capped preview grid shows the parsed columns and sample rows, sortable for a quick scan. A Filter rows control lets you narrow the file down by condition, with a scope toggle for whether the filter applies to the preview only or to the import itself.

2

Target

Choose Create new table or Add rows to existing table. For a new table, Jam SQL Studio infers each column's native type from the file's data (types debut here, per engine), and a live schema preview shows the table about to be created — column names, engine-native types, and any generated primary key — before you continue. For an existing table, a searchable combobox finds the table by name, and a read-only preview shows its current schema so you know what you're appending to.

3

Columns

Every file column gets an include checkbox, a mapped target column, and an editable type. Unmapped target columns (ones with no matching file column) can take a fixed value instead — useful for a constant like a batch id or an import timestamp. The same row-filter control from Source stays available here, and when you change the target, Jam SQL Studio automatically re-runs the file-to-target column matching and shows a re-match summary.

Columns the database computes for you — generated/computed columns — aren't mappable targets; the wizard marks them so you don't try.

4

Import

Pick an error policy: stop at the first error and roll back, or skip invalid rows and keep going — skipped rows are written to a bounded error report with the row, column, value, and reason. If you'd rather not have Jam SQL Studio execute anything, toggle Generate script instead of executing to get a SQL script (the table DDL plus INSERT statements) you can review or run yourself.

When importing into an existing table, you can additionally choose to empty the table first before the new rows go in — a destructive confirmation calls out that, on MySQL and Oracle, that truncate can't be rolled back if the import then fails.

After a successful import, an optional “Make this table easier to browse” card offers loose foreign key, JSON, or enum declarations inferred from the imported data. These are stored locally in MetaInfo and never change the database schema — save the ones you want, or dismiss the card.

All five engines

Data Import works the same way across SQL Server, PostgreSQL, MySQL, Oracle, and SQLite — the wizard steps, column mapping, and error handling are identical; only the emitted DDL and driver-native writes differ per engine.

Limitations

• Encoding is BOM auto-detect, not universal charset detection. UTF-8 and UTF-16 (with a byte-order mark) are detected automatically; other encodings need the manual Encoding picker on the Source step.
• Derived/computed columns aren't supported as import targets. The wizard can insert into ordinary columns only — database-computed columns are marked as unavailable in the Columns step.
• New-table CREATE always runs outside the row-import transaction, on every engine. If you cancel or the import fails partway through, a freshly created table is left behind empty — there's no warning before you start, but a post-failure Delete partial table action cleans it up. TRUNCATE-first (existing-table mode) is transactional on PostgreSQL, MSSQL, and SQLite, rolling back along with the rest of a failed import, but non-transactional (implicit commit) on MySQL and Oracle — there, a destructive confirmation warns you before you confirm that the truncate can't be undone if the import then fails.

Frequently asked questions