file_exchange.md

File Exchange

Upload

Upload CSV or Parquet files to an Exasol table. If the table does not exist, exapump auto-creates it by inferring the schema from the file.

exapump upload data.csv --table schema.my_table
exapump upload *.parquet --table schema.my_table

Options

Flag	Default	Description
--table	required	Target table name (e.g. schema.table)
--dry-run	—	Preview inferred schema without loading data
--delimiter	,	CSV field delimiter
--quote	"	CSV quoting character
--escape	—	CSV escape character
--no-header	—	Treat the first row as data, not a header
--null-value	""	String to interpret as NULL

Examples

# Upload a CSV with a custom delimiter
exapump upload data.tsv --table my_schema.events --delimiter $'\t'

# Upload Parquet files
exapump upload part-*.parquet --table my_schema.events

# Dry run — preview the inferred schema
exapump upload data.csv --table my_schema.events --dry-run

---

Export

Export an Exasol table or query result to a CSV or Parquet file.

exapump export --table schema.my_table --output data.csv --format csv
exapump export --query 'SELECT * FROM t WHERE id > 100' --output result.parquet --format parquet

Options

Flag	Default	Description
--table	—	Table to export (mutually exclusive with --query)
--query	—	SQL query to export results from (mutually exclusive with --table)
--output	required	Output file path
--format	required	Export format: csv or parquet
--delimiter	,	CSV field delimiter
--quote	"	CSV quoting character
--no-header	—	Exclude header row from output
--null-value	""	String to represent NULL values
--timeout	—	Client-side export deadline in seconds; CSV only
--compression	—	Compression codec for Parquet: snappy, gzip, lz4, zstd, none
--max-rows-per-file	—	Maximum rows per output file (enables splitting)
--max-file-size	—	Maximum file size per output file, e.g. 500KB, 1MB, 2GB (enables splitting)

Three different timeouts can bound a CSV export, and only one of them is the flag above. --timeout sets a client-side export deadline, in seconds, for CSV exports only. ?query_timeout=<seconds> in the DSN sets a server-enforced query bound that applies to every format. ?timeout=<seconds> in the DSN sets only the connection's connect deadline; it does not bound an export. If --timeout elapses, exapump deletes the output files the export had written. On a split CSV export (--max-rows-per-file or --max-file-size), --timeout bounds only the download phase; the file-writing phase that follows runs unbounded. The single-file path stays bounded through both phases.

A Parquet export has no client-side deadline and cannot take one; --timeout is rejected for --format parquet. exapump versions before 0.12.0 bounded it implicitly at 300 seconds, whether or not it was split. ?query_timeout=<seconds> in the DSN is the only bound available for a Parquet export.

Examples

# Export a table to CSV
exapump export --table my_schema.events --output events.csv --format csv

# Export a CSV with a 300-second client-side deadline
exapump export --table my_schema.events --output events.csv --format csv --timeout 300

# Export a query result to compressed Parquet
exapump export --query 'SELECT * FROM t' --output out.parquet --format parquet --compression zstd

# Split output into files of at most 100MB
exapump export --table my_schema.big_table --output chunks.parquet --format parquet --max-file-size 100MB