Export & Dump

BroadSQL writes query results to a local file in one of two ways:

• EXPORT <fileName> (synonyms EXP, EXTRACT, EXT): a toggle. Once turned on, every SQL statement you run afterwards writes its result to that file instead of the screen, until EXPORT is called again with no argument (turns it off) or with a new file name. See command reference.
• DUMP <tableName>: a one-shot export of an entire table's content, in a single step. See command reference.

Both share the same destination-folder rule, the same extension-to-format mapping, and the same DefaultFileFormat setting described below.

Recommended alternative for new scripts: PULL covers the same ground (PULL <source> TO <name> AS CSV/TXT, or AS XLSX/ODS for a multi-tab spreadsheet) with one unified syntax and an explicit destination name/format, instead of EXPORT's toggle or DUMP's automatic format choice. EXPORT/DUMP aren't going anywhere and keep working exactly as described on this page; PULL's CSV output just uses a separate CsvSeparator setting rather than FieldsSeparator/ SET SEPARATOR, and rejects a column it can't map instead of writing whatever the driver returns for it (see PULL for the full comparison).

Where files are written

• EXPORT: if fileName includes a directory, that directory must already exist or the command fails with an error. If it doesn't include one, the file is written to DefaultFolder (see Application settings).
• DUMP: always writes to DefaultFolder, named after the table exactly as typed (schema included, e.g. DUMP PUBLIC.CUSTOMER produces PUBLIC.CUSTOMER.<ext>). There is no option to choose the destination or file name. Unlike EXPORT, DUMP does not check that DefaultFolder exists first.

Choosing a format

The format is determined by the file's extension:

If EXPORT is given a file name with no extension, one is appended automatically based on DefaultFileFormat (see below): .xlsx, .ods, .csv, or .txt. DUMP never receives an extension from the user; it always picks one itself:

• At or below the MaxRowXLSX threshold (or when the row count can't be determined), it uses the extension matching DefaultFileFormat.
• Above MaxRowXLSX, it always writes a tab-separated .txt file, regardless of DefaultFileFormat or the configured FieldsSeparator.

DefaultFileFormat

INI setting ([General], see Application settings) controlling the default format above. Accepted values: XLSX, ODS, CSV, TXT (not case-sensitive). It is the only optional export-related setting:

• Missing from the file: BroadSQL prints an INFO message at startup and uses ODS.
• Present but not one of the four values: the value is ignored, BroadSQL prints an INFO message at startup explaining why, and uses ODS.
• Valid: used as given, no message.

This setting only supplies a default. Naming an extension explicitly always wins: `EXPORT totals.csv always produces a CSV file no matter what DefaultFileFormat` says.

CSV vs. TXT

Before this setting existed, .csv and .txt produced an identical file; only the extension differed. They are now distinct:

• .csv always uses a comma, regardless of FieldsSeparator or SET SEPARATOR.
• .txt (and any other/unrecognized extension) uses the configured separator (FieldsSeparator in the INI, or whatever SET SEPARATOR last set for the session, tab by default).

Both otherwise share the same writer and behavior (see "CSV / TXT" below).

Interrupting an export

Pressing CTRL+C during an .xlsx, .ods, .csv, or .txt export stops the write cleanly without closing the database connection: rows written so far stay in the file, the rest of the result set is discarded. (Not supported for .mdb.)

Per-format details

XLSX (Excel)

• A NULL value produces a blank cell (not 0/false/a placeholder date).
• DATE/TIME/TIMESTAMP columns get a real, typed Excel cell with a matching format (yyyy-MM-dd, hh:mm:ss, yyyy-MM-dd hh:mm:ss): the time-of-day on a TIMESTAMP is preserved, not truncated to midnight.
• BIGINT/DECIMAL/NUMERIC values are written as a real (sortable, calculable) number when that round-trips exactly through a double; otherwise the cell falls back to exact text, since Excel numbers cannot represent arbitrary precision.
• The sheet name is derived from the table/query and sanitized (Excel-forbidden characters removed, truncated to 31 characters). A second sheet named "Query" records the exact SQL text and when it ran.
• A sheet is capped at 1,048,576 data rows (Excel's own limit); beyond that, the export stops cleanly, the file remains valid, and a warning reports how many rows were and weren't exported.
• Binary columns (BINARY/VARBINARY/LONGVARBINARY/OTHER/DATALINK) are written as the text "Binary content, not exported to Excel"; known limitation: this placeholder is currently shown even when the binary value is actually NULL, instead of a blank cell.
• BLOB/CLOB/NCLOB are not handled explicitly; they fall into the generic text case.
• .xls (the legacy 97-2003 binary format) is not supported.

ODS (OpenDocument Spreadsheet)

• Same NULL-handling, typed DATE/TIMESTAMP cells, and 1,048,576-row safety cap as XLSX above (the cap here is a memory safeguard, not an ODF format limit; unlike XLSX's streaming writer, the whole sheet is held in memory before being saved).
• TIME columns have no dedicated cell type in the underlying library and are written as formatted text (HH:mm:ss).
• BIGINT/DECIMAL/NUMERIC values are written with their exact decimal value, with no precision loss to guard against (unlike XLSX, the format doesn't round-trip through a double).
• The header row is bold with a light blue background. There is no equivalent of Excel's "Query" recap sheet.
• Binary columns get the same "Binary content, not exported to ODS" placeholder as XLSX, but a NULL binary value correctly produces a blank cell.
• BLOB/CLOB/NCLOB are not handled explicitly; they fall into the generic text case.

CSV / TXT

• RFC 4180 quoting: a value containing the separator, a double quote, or a line break is wrapped in double quotes (internal quotes doubled).
• A true SQL NULL produces an empty field; the literal text "null" (or any other value) is written as-is; the two are never confused.
• Always written as UTF-8 with a leading BOM, independent of the console's own encoding, so tools like Excel recognize accented characters correctly when opening the file directly.
• Re-running EXPORT in append mode to an existing, non-empty file does not repeat the header row.

MDB (MS Access), legacy format, frozen

This format is frozen (28/08/2026): still fully supported, but not receiving further development. It is not part of PULL's unified formats (AS H2/AS XLSX/AS ODS/AS CSV/AS TXT/AS JSON/ AS MD/AS HTML; see PULL) and there is no plan to add it there. If your workflow depends on .mdb export and you'd like to see it continue evolving, let us know: real usage is exactly what would justify further investment.

• Written in the Access 2000 (.mdb) file format via Jackcess, not .accdb (the format Access itself has defaulted to since Access 2007).
• The table name is derived from the query/table the same way as the Excel sheet name (unsanitized).
• If the target .mdb file already exists, a new table is added into it; existing tables are left untouched. If it doesn't exist, a new database file is created.

Related pages

• Application settings: DefaultFolder, DefaultFileFormat, MaxRowXLSX, FieldsSeparator.
• EXPORT, DUMP, SET SEPARATOR command reference.