Request Input

parseFilterRequest covers the common case. Three additional, composable helpers cover the richer Spatie / JSON:API surface, control where input is sourced, and normalize field-name casing. All are pure reshapes — no validation or allow-listing (that happens downstream in applyFilter).

parseSpatieRequest — the full Spatie/JSON:API shape

A superset of parseFilterRequest: it understands the same filter / sort / search / distinct shapes plus JSON:API relation includes, sparse fieldsets, and cursor pagination. It returns a SpatieInput (a FilterInput extended with select, include, and cursor params) — ready to hand to applyFilter (offset) or applyCursor (keyset).

include and select are parsed but never applied. No runner reads them, because which relations and columns a client may reach is a policy decision this library deliberately leaves to you. Treat them as a request you validate and then act on with Lucid's preload / select yourself.

import { parseSpatieRequest } from '@adonis-agora/filter'

const input = parseSpatieRequest(request.qs())

page[after] / page[before] route to cursor pagination (after/first, before/last); otherwise page[number] / page[size] route to offset. after wins if both cursor bounds are present.

resolveInputFromRequest — where input comes from

Reads the raw structured input from a request according to an InputSource — useful when filters arrive in the body (a complex POST /search) rather than the query string:

import { resolveInputFromRequest } from '@adonis-agora/filter'

const raw = resolveInputFromRequest(req, 'auto')

The request is treated structurally ({ method, query, body }), so an AdonisJS HttpContext.request or a plain object both work. The result is always a fresh shallow copy.

normalizeInput — key casing

Normalizes an input object's top-level keys — case-transforming them and optionally stripping Id suffixes and empty values — so a snake_case API can map onto camelCase model columns (or vice-versa):

import { normalizeInput } from '@adonis-agora/filter'

normalizeInput({ company_id: 5, first_name: 'a' }, { normalizer: 'camelCase' })
// → { companyId: 5, firstName: 'a' }

normalizeInput({ companyId: 5 }, { normalizer: 'snakeCase', dropId: true })
// → { company: 5 }   (snake_case + trailing Id stripped)

• normalizer — 'camelCase', 'snakeCase', or a (key) => string function.
• dropId — strip a trailing Id/_id (and drop a bare id). Default false.
• stripEmpty — drop null/undefined/'' values. Default true.