querybara query

CLI
PostgreSQL
MySQL
MariaDB
MongoDB
Redis / Valkey
Elasticsearch

querybara query runs statements against a target and prints the results. The input comes from -e, from a file, or from stdin. On PostgreSQL, MySQL and MariaDB it runs SQL statements one by one; on MongoDB, Redis and Elasticsearch it runs that engine’s commands or requests. Use it for reports, checks in CI, and one-off statements.

Usage

Usage: querybara query [options] <target>

run SQL statements one by one from -e, a file, or stdin

Arguments:
  target                    profile name or id, or connection URI

Options:
  -e, --execute <sql>       SQL to run (several statements allowed)
  -f, --file <path>         a .sql file to run, streamed ("-" for stdin)
  --format <format>         result format (choices: "table", "csv", "tsv",
                            "json", "jsonl", default: "table")
  -p, --param <name=value>  bind a placeholder (:name, $1 or ?) to a value;
                            repeatable
  --stop-on-error           stop at the first failed statement (default)
  --continue                keep going after a failed statement
  --error-log <file>        write failed statements and their errors to a file
  --row-limit <n>           print at most n rows per result set (0: all)
                            (default: 0)
  --max-column-width <n>    table format: cut longer cells with …
  --database <name>         database to connect to
  --read-only               refuse statements that write
  -y, --yes                 run statements that need confirmation without asking
  --tls <mode>              TLS mode for this run: disable, require, verify-ca
                            or verify-full
  --ssh <user@host[:port]>  reach URI targets through this SSH server; repeat
                            for jump hosts, in order
  --ssh-key <path>          SSH private key file (OpenSSH, PEM or PuTTY .ppk)
  --ssh-password-env <VAR>  take the SSH password from this variable (default
                            QUERYBARA_SSH_PASSWORD, else a prompt)
  --ssh-agent               log in with the keys of ssh-agent (SSH_AUTH_SOCK) or
                            Pageant
  --proxy <url>             reach URI targets (or their first SSH server)
                            through socks5://host:port or http://host:port
  --ssh-accept-new          trust and remember an SSH host key not seen before
                            (a changed key is always refused)
  --known-hosts <path>      SSH known hosts file (default: the desktop app's)
  -h, --help                show help for a command

Statements run in order, each in its own auto-commit unless the script opens a transaction.
Files stream through the statement splitter (DELIMITER, dollar quoting and comments are
handled), so large dumps run in flat memory; progress shows on stderr in a terminal.

Results go to stdout; row counts, timings and notices to stderr. csv, tsv, json and jsonl
are exact and stream: CSV writes NULL as an empty field and '' as "", TSV follows PostgreSQL
COPY text (\N for NULL), JSON keeps bigints exact and writes binary as base64.

Safety: UPDATE/DELETE without WHERE, DROP and TRUNCATE ask for confirmation (or need
--yes); production profiles confirm every write; read-only profiles refuse writes.
Placeholders are bound only when --param is given. Ctrl+C cancels the running statement.

Examples:
  querybara query prod -e "select * from users where id = :id" --param id=42
  querybara query "postgres://app@localhost/app" -f migrate.sql --continue --error-log errors.log
  querybara query dev -f dump.sql --yes --quiet
  cat report.sql | querybara query dev --format csv > report.csv
  querybara query "mysql://[email protected]/app" --ssh ops@jump:22 --ssh ops@bastion --ssh-key ~/.ssh/id_ed25519 -e "select 1"

Redis targets (redis://, rediss:// or a Redis profile) run redis-cli command lines, one per
line, and print redis-cli's output (--format json or jsonl: the replies as JSON); DEL, FLUSHDB
and other destructive commands ask for confirmation (or need --yes):
  querybara query "redis://localhost:6379/0" -e 'SET greeting "hello world"'

Elasticsearch targets (http://, https:// or a saved profile) run Kibana console
requests: a method and path per request, then its JSON body (NDJSON lines for _bulk); each
response body prints as JSON with numbers exactly as sent (--format json or jsonl: objects with
the request, status and body). Deleting or closing indices, delete by query and other
destructive requests ask for confirmation (or need --yes); an error status fails the request:
  querybara query "http://elastic@localhost:9200" -e 'GET _cluster/health'
  querybara query search-prod -f requests.txt --format jsonl

The --tls and SSH options are described in Global options.

Run SQL

Statements run in order, each in its own auto-commit unless the script opens a transaction. A file streams through the statement splitter, which handles DELIMITER, dollar quoting and comments, so large dumps run in flat memory.

querybara query shop-prod -e "select id, total from shop.orders where customer_id = :id" --param id=42
querybara query "postgres://[email protected]/shop" -f migrate.sql --continue --error-log errors.log
cat report.sql | querybara query shop-dev --format csv > report.csv

Placeholders (:name, $1 or ?) are bound only when you pass --param.

If the script leaves a transaction open, the CLI rolls it back at the end and warns.

Output formats

Results go to stdout; row counts, timings and notices go to stderr.

`--format`	Output
`table`	A text table (default). `--max-column-width` cuts longer cells
`csv`	CSV. `NULL` is an empty field and an empty string is `""`
`tsv`	PostgreSQL `COPY` text: `\N` for `NULL`
`json`	JSON. Bigints stay exact; binary values are base64
`jsonl`	One JSON object per line

csv, tsv, json and jsonl are exact and stream as rows arrive. --row-limit prints at most that many rows per result set.

MongoDB, Redis and Elasticsearch

On a MongoDB target, each input is a command document, in Extended JSON or shell syntax. An array of documents runs in order. Documents print as relaxed Extended JSON; --format json prints canonical Extended JSON as an array, and jsonl one canonical document per line.

querybara query "mongodb://[email protected]/catalog" -e '{ find: "products", filter: { price: { $gt: 100 } } }'

On a Redis target (redis://, rediss:// or a Redis profile), each line is a redis-cli command line, and the output is redis-cli’s. --format json or jsonl prints the replies as JSON.

querybara query "redis://cache.example.com:6379/0" -e 'HGETALL cart:8812'

On an Elasticsearch target (http://, https:// or a saved profile), the input is console requests: a method and path per request, then its JSON body (NDJSON lines for _bulk). Each response body prints as JSON with numbers exactly as sent. An error status fails the request.

querybara query "https://[email protected]:9200" -e 'GET _cluster/health'
querybara query search-prod -f requests.txt --format jsonl

Safety

UPDATE or DELETE without WHERE, DROP and TRUNCATE ask for confirmation, or need --yes.
Profiles whose environment is production, or that confirm writes, ask before every write.
Read-only profiles refuse writes, and so does any target with --read-only.
On MongoDB, a read-only target refuses commands that write; drops, multi-document updates and deletes, and other destructive commands ask or need --yes.
On Redis, DEL, FLUSHDB and other destructive commands ask or need --yes.
On Elasticsearch, deleting or closing indices, delete by query and other destructive requests ask or need --yes.

In a terminal you answer [y]es, [N]o, [a]ll; all runs the remaining statements of the run without asking again. Without a terminal, and whenever the script comes from stdin, a statement that needs confirmation fails unless you pass --yes.

Ctrl + C cancels the running statement on the server. A second Ctrl + C quits at once.

Exit codes

Code	Meaning
0	Every statement ran
2	A statement failed or was refused (also with `--continue`), or another error
130	Interrupted

To run a file and get exit code 1 when some statements failed but the run went on, use querybara run-file with --continue.

Documents Querybara 0.1.1 · built frombc9f5aa