Changelog

[0.2.0] - 2026-07-21

Cnpja.Company.offices is now [Cnpja.CompanyOffice.t()] (API subset without address)
Cnpja.Office.suframa is now [Cnpja.OfficeSuframa.t()] (embedded summary, not full Cnpja.Suframa)
Cnpja.Rfb no longer has registrations, links, suframa, or company fields (not returned by /rfb)
get_company/2 no longer documents or sends non-existent :simples / :simples_history query options

Lists API (saved CNPJ collections + async export):
- create_list/2, search_lists/1, get_list/2, update_list/3, delete_list/2
- create_list_export/3, search_list_exports/2, get_list_export/3
- Structs: Cnpja.CnpjList, Cnpja.CnpjListSummary, Cnpja.CnpjListPage, Cnpja.CnpjListExport, Cnpja.CnpjListExportPage, Cnpja.CnpjListExportId
Cnpja.CompanyOffice and Cnpja.OfficeSuframa response structs
Cnpja.Error.code from API error payloads
search_persons/1 filter :person_country_nin (country.id.nin)
Client support for POST, PATCH, DELETE (required by Lists API)
Configurable :receive_timeout (config and per-call option; default 30s)
Query params that are lists are joined as comma-separated values

get_simples/2 now sends history= (API param); :simples_history remains as a documented alias
Nested company offices and office-embedded SUFRAMA parse correctly against real OpenAPI DTOs

Office.suframa and Rfb.suframa type corrected from Suframa.t() | nil to [Suframa.t()] — the API returns an array
Office.links and Rfb.links type corrected from OfficeLinks.t() | nil to [Link.t()] — the API returns an array of {type, url} objects; Cnpja.OfficeLinks replaced by Cnpja.Link
Company.id and CompanyRef.id type corrected from integer() to String.t() — the API returns a string (e.g. "37335118")
RFB, Simples, CCC, and SUFRAMA endpoints corrected to use taxId as a query param instead of a path segment (e.g. /rfb?taxId=... not /rfb/37335118000180)
All search query param names corrected from camelCase (simplesOptant, stateIn) to API dot-notation (company.simples.optant.eq, address.state.in)
get_ccc/3 states argument now sent as states= query param instead of a path segment
Removed unreachable parse_error_body/1 fallback clause from the HTTP client

Cnpja.Link struct — public document link with type and url fields, replacing Cnpja.OfficeLinks
Extended search_offices/1 with full set of filter params: founded_gte/lte, status_date_gte/lte, special_date_gte/lte, street/district/number/details filters, phone/email field-level filters, activity_in/nin, simples_since_gte/lte, simei_since_gte/lte, and more

API paths corrected from plural to singular: /offices → /office, /companies → /company, /persons → /person. All previous versions returned 404 on these endpoints when called against the real CNPJá API.

Cnpja.CompanyRef — lightweight company reference embedded in Office and Rfb to avoid circular parsing
Cnpja.SizeLabel — company size struct with id, acronym, and text fields (distinct from Label)
Cnpja.PersonMembership and Cnpja.PersonMembershipCompany — typed person membership records
status_date, reason, special_date, special fields on Cnpja.Office and Cnpja.Rfb
name, equity, nature, size, jurisdiction fields on Cnpja.Rfb
jurisdiction field on Cnpja.Company
performed field on Cnpja.Activity
Full Cnpja.Suframa rewrite: since, head, approved (boolean), approval_date, address, phones, emails, incentives
Address parsing accepts both "zip" and "code" field names
Extended @camel_map covering all search, image, and certificate query parameters

Full coverage of all 17 CNPJá API endpoints
Typed structs for all API responses (Cnpja.Office, Cnpja.Company, Cnpja.Zip, etc.)
Tagged tuple error handling — {:ok, struct} / {:error, %Cnpja.Error{}}
Support for configuration via config.exs and per-call api_key: option
Bypass-based integration tests for all endpoints
ExDoc documentation published on HexDocs