Friday, April 24, 2020

New tricks in CockroachDB: Sql statement redirection from an external file

CockroachDB is the SQL database for building global, scalable cloud services that survive disasters. It is designed to be a Postgresql compatible database with distributed roots. The Postgresql compatibility is being built from scratch in Go. That said, product is undergoing fast pace of innovation and development. One convenience method of passing sql statements from an external file to CLI has been missing. It is not available in the current GA release and I opened a GitHub issue to implement it. In lieu of available convenience option, this post will discuss other methods of achieving the same.

Scenario


I have a table called population with the following schema:

CREATE TABLE population ( state CHAR(2) NULL, city STRING NULL, population INT8 NULL, FAMILY "primary" (state, city, population, rowid) )

The associated CSV file has the following structure:

IN,Evansville,7415326
TX,Fort Worth,9000748
IN,South Bend,2590270
KS,Shawnee Mission,9444064
GA,Macon,9804704
WA,Seattle,7928277

The associated IMPORT command:

IMPORT TABLE population ( state char(2), city TEXT, population INT8 ) CSV DATA ('https://api.mockaroo.com/api/25164a90?count=1000&key=02f7f490');

The associated IMPORT INTO command:

IMPORT INTO population (state, city, population) CSV DATA ('https://api.mockaroo.com/api/25164a90?count=5000&key=02f7f490');

The big difference between IMPORT and IMPORT INTO is the former will create the table for you and the latter expects the table to exist. That said, you can repeatedly issue IMPORT INTO command.

CockroachDB has ability to import CSV from cloud storage, http(s) endpoint, local filesystem, etc. You can read more about IMPORT command, here and the associated IMPORT INTO command, here.

Method 1a: Unix file redirection


Suppose I want to pass an import statement from a file. With Unix file redirection the following can be done with command below:

cockroach sql -e="$(<import_into_statement.sql)" --url 'postgresql://'

-e is functionally equivalent to --execute. This is pretty convoluted and can be further simplified by the next approach.

Method 1b: Unix file redirection


cockroach sql < import_into_statement.sql --url 'postgresql://'

Method 2: Using psql


Given CockroachDB is Postgresql wire compatible, you can use psql utility to achieve the same. I'm using Mac OSX and psql can be easily installed with:

brew install libpq
brew link --force libpq


Once installed, import from an external file can be easily done with built-in -f and --file= flags:

psql -f import_into_statement.sql 'postgresql://root@localhost:26257/defaultdb?sslcert=certs%2Fclient.root.crt&sslkey=certs%2Fclient.root.key&sslmode=verify-full&sslrootcert=certs%2Fca.crt'

Additional file redirection tricks are illustrated in our docs.

No comments: