Skip to contents

Translate SQL between dialects

Source: R/transpile.R
sql_transpile.Rd

Parses sql with the from dialect and regenerates it in the to dialect, rewriting functions, quoting, and constructs as needed (e.g. MySQL IFNULL() becomes PostgreSQL COALESCE()).

Usage

sql_transpile(
  sql,
  from,
  to,
  pretty = FALSE,
  unsupported = c("raise", "warn", "ignore")
)

Arguments

sql

A single character string with one or more SQL statements (separated by ;).

from

Source dialect name (see sql_dialects()).

to

Target dialect name (see sql_dialects()).

pretty

If TRUE, pretty-print the output.

unsupported

How to handle constructs that cannot be represented in the target dialect: "raise" (default) throws a polyglot_transpile_error; "warn" and "ignore" continue and return the closest supported translation (with "warn", upstream collects diagnostics but still returns a result).

Value

A character vector with one element per input statement.

Semantic limitations

Transpilation is syntactic and best-effort: identical syntax can still behave differently across engines (implicit casts, collations, NULL ordering, integer division, time zone handling...). Always test the translated SQL against the target database before using it in production.

See also

sql_format(), sql_parse(), sql_validate()

Examples

sql_transpile("SELECT IFNULL(a, b) FROM t", from = "mysql", to = "postgres")
#> [1] "SELECT COALESCE(a, b) FROM t"
sql_transpile(
  "SELECT DATE_TRUNC('month', created_at) FROM events",
  from = "postgres", to = "duckdb"
)
#> [1] "SELECT DATE_TRUNC('month', created_at) FROM events"