Contract: cultivares_registradas¶
The dataset reuses source contract rnc_registradas 1.0, constant RNC_REGISTRADAS_V1, internal name rnc.registradas. There is no contract alias using the dataset name. See the cultivar dataset API.
Schema¶
The ten columns retain the RNC API names and order. Text fields contain strings, in the installed pandas default dtype (str on pandas 3, object on 2). Civil dates use datetime64[ns], without timezone or time of day.
| Column | pandas type | Nullable | Meaning |
|---|---|---|---|
cultivar |
str | No | Published name; legitimate empty strings are retained |
nome_comum |
str | No | Common species name, with repeated internal whitespace collapsed into 1 space |
nome_cientifico |
str | No | Published scientific name, with repeated internal whitespace collapsed into 1 space |
grupo |
str | No | Species group |
situacao |
str | No | Published registration status |
nr_formulario |
str | No | Form number; may repeat or be empty |
nr_registro |
str | No | Registration number, a nonblank textual key |
data_registro |
datetime64[ns] | Yes | Civil registration date |
data_validade |
datetime64[ns] | Yes | Published civil validity date |
mantenedor |
str | No | Maintainer text; empty values and compound names are retained |
Primary key: nr_registro, within the processed family and resource. Registration and form numbers are different identifiers. A form number is not a key or an automatic replacement for a missing registration number. Identifiers are not converted to numbers; leading zeros and punctuation remain textual.
Guarantees and limits¶
The source and dataset always validate the contract, including calls without metadata. The source validates all rows before filtering. Missing required columns, duplicate column labels, missing/blank/duplicate keys and incompatible types fail; cultivar or maintainer names are not deduplication keys.
A published empty text value remains "", distinct from null. Missing dates remain NaT; invalid date text is not silently coerced. Both dates allow absence without imputation. The contract requires ns civil dates and rejects timezone or time of day, including an all-null date column with the wrong physical type.
A filtered empty result retains all ten columns and their types. A physically empty or invalid CSV is not a legitimate empty search result. Status, validity and maintainer text are not interpreted as marketing authorization, ownership rights or agronomic advice. Multi-name fields are not split.
Metadata and selection¶
schema_version and contract_version are "1.0". At the dataset layer, selected_source="rnc_registradas" and attempted_sources=["rnc_registradas"]; caching does not switch the route to another family. Hash, size and collection time describe the complete CSV, while records_count and columns describe the selected result. Population diagnostics, comparison against the search total and filters are in source_details.
nr_registro and nr_formulario use full textual equality after trimming surrounding whitespace. Other filters use literal case-insensitive substrings; especie searches nome_comum. Blank or invalid parameters are rejected before cache or network access, without prohibiting published empty values in the table.
The export is current. Its hash identifies an acquisition rather than a selectable historical revision; the dataset rejects deterministic contexts before I/O. See the RNC/SNPC source and the separate protected-cultivar contract.
Export of 2026-09-18¶
The CSVs of the 2026-09-18 export contain 38,325 RNC records and 5,424 SNPC records. The HTML search count that precedes each CSV matches it; this agreement establishes neither a transactional snapshot nor historical completeness.
In RNC, the published 22,946 blank form numbers, 4,256 blank cultivar names and 4,271 blank maintainers remain empty text; both dates are populated in that export. There are 924 repeated form-number groups with 15,142 occurrences. In SNPC, protection end has 5,267 dates, 155 conditions and two blanks; the 155 conditions and two blanks produce null dates while retaining the text. There are 102 certificate numbers repeated across distinct processes, with 204 occurrences. These secondary-identifier repetitions are not deduplicated.
The original acquisition in fetched_at retains timezone-aware UTC through cache and dataset calls. At both source and dataset level, fetch_timestamp equals acquisition time. Table dates are civil dates independent of these timestamps.