@spec dump_uuid!(Ecto.UUID.t()) :: binary()

Dumps a UUID string into the binary form a raw SQL parameter expects.

Raises ArgumentError when uuid is not a valid UUID.

@spec escape_like(String.t()) :: String.t()

Escapes the LIKE/ILIKE metacharacters (\, %, _) in term so a user-supplied search string matches literally instead of as a wildcard.

Every metacharacter is escaped in a single pass, so the result is the same regardless of which character appears first. The caller wraps the result in %...% (or similar) and, for raw SQL, appends ESCAPE '\\' to the condition.

Examples

iex> GoodAnalytics.SQL.escape_like("50%off")
"50\\%off"

iex> GoodAnalytics.SQL.escape_like("a_b")
"a\\_b"

@spec integer_value(Decimal.t() | integer() | nil) :: integer()

Coerces a numeric aggregate result to an integer.

Postgres sum/count can come back as a Decimal, a plain integer, or nil (empty set). Returns the integer value, treating nil as 0.

Ecto query fragment that normalizes a URL column for page-level grouping: strips the query string, collapses duplicate slashes (keeping ://), and drops a trailing slash, defaulting empty/null to "/".

Single source of truth so the analytics "Top Pages" breakdown grouping and any drill-down filter on the same dimension stay in lockstep. Import the module and use it inside a query:

import GoodAnalytics.SQL
from(e in Event, group_by: normalized_url(e.url), select: normalized_url(e.url))

@spec not_set() :: String.t()

The label for the null/empty bucket in a dimension breakdown.

Rows whose dimension value is NULL or absent are grouped under this label via coalesce(column, not_set()). Query builders and any consumer that renders or joins on the bucket value must reference this single source so the label stays consistent across grouping, joins, and display.