Documentation
¶
Overview ¶
Package unittest builds the SQL that runs a Bruin unit test on the asset's configured connection. The tables the asset reads are replaced by inline fixture CTEs, so the test runs as a single read-only SELECT that issues no DDL and leaves no artifacts on the target.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func BuildWarehouseQuery ¶
func BuildWarehouseQuery(p Rewriter, dialect, renderedSQL string, test pipeline.UnitTest, schemas map[string][]pipeline.Column) (string, error)
BuildWarehouseQuery rewrites an asset's rendered SQL into a single read-only SELECT that reads from inline fixture CTEs instead of real tables, so a unit test can run on the configured warehouse with no DDL and no artifacts.
Every table the query reads must be accounted for, or it would hit real data: a mocked input becomes a fixture CTE built from its rows; an unmocked read whose asset declares columns becomes an empty (zero-row) typed CTE; an unmocked read with no declared columns is an error.
To assert an intermediate CTE instead of the final output, derive its query from this result with Rewriter.SelectFromCTE (WITH … SELECT * FROM <cte>) — the fixtures are already injected, so it needs no separate build.
func ResolveFixtures ¶
func ResolveFixtures(available []pipeline.Fixture, test pipeline.UnitTest) ([]pipeline.UnitTestInput, error)
ResolveFixtures expands the named fixtures a test references into mock inputs and merges them with the test's own inputs. A pipeline-level fixture supplies rows for one asset; a test pulls it in by name (UnitTest.Fixtures) so many tests can share a baseline set of rows (typically lookup or dimension tables) without repeating them.
The test's own inputs win: an explicit inputs entry for an asset fully replaces a referenced fixture for that same asset. It is an error to reference a fixture the pipeline does not define, or to reference two fixtures that both supply rows for the same asset (ambiguous).
Types ¶
type Rewriter ¶
type Rewriter interface {
ExtractSelect(sql string, dialect string) (string, error)
UsedTables(sql string, dialect string) ([]string, error)
RenameTables(sql string, dialect string, mapping map[string]string) (string, error)
PrependCTEs(sql string, dialect string, ctes []sqlparser.CTE) (string, error)
SelectFromCTE(sql string, dialect string, cteName string) (string, error)
FreezeTime(sql string, dialect string, executionTime string) (string, error)
}
Rewriter is the subset of the SQL parser the warehouse builder needs. *sqlparser.SQLParser satisfies it.
type UnitTestResult ¶
type UnitTestResult struct {
Passed bool
Message string // human-readable explanation when Passed is false
}
UnitTestResult is the outcome of running a single unit test.
func CompareCTEResult ¶
func CompareCTEResult(expected pipeline.UnitTestCTEExpected, columns []string, rows [][]interface{}) *UnitTestResult
CompareCTEResult compares the rows produced by a named intermediate CTE against its expectation, with the same semantics as the final output.
func CompareResult ¶
func CompareResult(expected pipeline.UnitTestExpected, columns []string, rows [][]interface{}) *UnitTestResult
CompareResult compares a query result (columns + rows, as returned by a connection via SelectWithSchema) against a unit test's expectation.
Source Files