SandboxCase.Sandbox.Logger (sandbox_case v0.4.0)

Sandbox adapter that captures logs per test.

Each test gets its own log buffer via an ETS table. Logs from the test process and any process in its $callers chain are routed to that buffer — no interleaving from concurrent tests.

Configuration

# Default — fail on unconsumed :error logs
logger: true

# Stricter — fail on unconsumed :warning and above
logger: [fail_on: :warning]

# Capture only, never fail
logger: [fail_on: false]

Pop-based assertions

Reading logs consumes them. At checkin, only unconsumed logs above the fail_on threshold trigger failure.

# Pop the next log at a level (returns message string or nil)
assert pop_log(sandbox, :error) =~ "sync failed"

# Pop all logs at a level (returns joined string)
assert logs(sandbox, :warning) =~ "duplicate"

# Happy path — never read logs, any error fails at checkin
test "creates user", %{sandbox: sandbox} do
  User.create!(attrs)
  # checkin: no errors → passes
end

Summary

Functions

get_logs(tokens)

Get all logs without consuming them (non-destructive). Returns a list of %{level: atom, message: binary, metadata: map}.

logs(sandbox, level \\ :debug)

Pop all log entries at or above level. Returns the messages joined with newlines as a string. Consumes the entries.

pop_log(sandbox, level \\ :debug)

Pop the next log entry at or above level. Returns the message string, or nil if no matching entry exists. Consumes the entry.