Free Online pg_dump ViewerOpen a PostgreSQL dump, browse its tables and take out the rows you need — without restoring a server.
Reading one table out of a dump usually costs a full restore somewhere first. Here it takes three steps:
Open the Diagram tab when a column name is not enough. Select a table and its relationships stand out, so you can see which records belong together before exporting a result.
Open the file, then use the database it contains:
| In your dump | In the viewer |
|---|---|
| PostgreSQL dump (.sql, .sql.gz or .tar) | Opens as a PostgreSQL database in this tab; browse its data without starting a server. |
| Schemas, tables and views | Browse rows, types, indexes and row counts. |
| Related tables | See declared foreign keys in the Diagram tab before filtering or exporting data. |
| A result you need to share | Run read-only PostgreSQL SQL, then export the result as CSV or JSON. |
| Extensions used by the dump | Supported PostgreSQL extensions, including PostGIS and pgvector, load when the dump needs them. |
The dump is read from your disk in this tab and replayed into a PostgreSQL that exists only in the tab's memory - never on a server, never on disk, gone when the tab closes. Queries are read-only, so the file you opened is the file you still have. Because a dump is a script, open dumps you trust: one from a stranger is a program from a stranger.
Plain (.sql), gzipped plain (.sql.gz) and tar (-Ft). Custom format (-Fc) does not: only pg_restore reads that archive and it has no WebAssembly build, so convert it once with pg_restore -f dump.sql yourfile.dump and open the result.
PostgreSQL 18, compiled to WebAssembly by PGlite. A dump taken from an older server replays forward into it, which covers the versions in use today; a dump from a server newer than 18 is not guaranteed.
The dump's own CREATE EXTENSION lines are read before the engine starts, and exactly those extensions are loaded - so a dump that needs none downloads none, and a PostGIS dump gets PostGIS. An extension nobody has built for WebAssembly, such as TimescaleDB, cannot be loaded, and then the objects that needed it are named rather than silently missing.
There is no fixed limit, but the whole database is held in the tab's memory: PostgreSQL itself reserves around 650 MB before your data, and a 21 MB dump of 200,000 rows opens in about two seconds. The machine sets the ceiling.
No. The viewer runs SELECT and WITH queries only, and the database it builds is a throwaway copy - your .sql file is never written to. To take data elsewhere, export a result as CSV or JSON.
Need to move the database rather than read a dump of it? DBConvert migrates PostgreSQL data to and from the other major databases, schema and data alike.
Need help with a file? Contact us