Report the metadata a package still needs, with the call that fills it

Lists every field that still blocks validate_salmon_datapackage(path, require_iris = TRUE), and prints the exact set_sdp_dataset() / set_sdp_table() / set_sdp_column() / set_sdp_code() call that fills it. Replace the <...> placeholder in the printed call with the real value and paste it – the paste is the audit trail, just as it is for accept_suggestion().

Usage

review_metadata(path)

Arguments

path: Path to the package directory.

Value

An ms_metadata_review tibble subclass, one row per unfilled field, with the package path attached as the review_path attribute. Empty when nothing is outstanding.

Details

This is the companion to review_semantics(), and it sees something that review structurally cannot: a slot with no candidates at all. review_semantics() builds its queue from retrieved suggestions, so a field nothing was found for never appears there. review_metadata() builds its list from the package's own required-field rules, so an empty shortlist and a full one look the same to it.

What it reports:

unresolved MISSING DESCRIPTION: / MISSING METADATA: / REVIEW REQUIRED: placeholders in any metadata field;
schema-required fields (constraints.required) that are blank – a column the file does not have counts as blank in every row;
draft IRIs still carrying the REVIEW: prefix wherever strict validation refuses one: any schema-declared *_iri field of tables.csv, and the six semantic IRI fields of column_dictionary.csv (term_iri, property_iri, entity_iri, unit_iri, constraint_iri and statistical_modifier_iri). A marker in codes.csv or dataset.csv is not listed, because strict validation does not refuse it there. Where retrieval found candidates for one, review_semantics() shows them;
measurement columns missing term_iri, property_iri, entity_iri or unit_iri;
tables.csv rows with a blank observation_unit_iri.

It never contacts an LLM, and under the default options it never contacts a network: the SDP schema it reads the required fields from is the copy bundled with metasalmon. When the metasalmon.sdp_schema_source or metasalmon.sdp_schema_base_url option selects a different schema, it reads that one, as the package writers do: from this session's cache once they have loaded it, and otherwise by fetching it as they would.

Examples

pkg <- create_sdp(
  data.frame(spawner_count = c(12L, 19L)),
  path = tempfile("sdp-review-"),
  seed_semantics = FALSE,
  seed_verbose = FALSE,
  check_updates = FALSE
)
#> Some measurement semantic IRI fields are still blank in this review-ready
#> package.
#> ℹ That does not block review-ready creation, but those gaps must be filled
#>   before final validation or publication.
#>   term_iri: spawner_count (rows 1) property_iri: spawner_count (rows 1)
#>   entity_iri: spawner_count (rows 1) unit_iri: spawner_count (rows 1)
#> ℹ Review metadata/column_dictionary.csv first (and metadata/tables.csv if
#>   present), then fill the remaining gaps there.
#> ℹ If you want candidate IRIs later, run `suggest_semantics()` or recreate the
#>   package with `seed_semantics = TRUE` before final validation.
#> ✔ Dictionary validation passed
#> ✔ Created Salmon Data Package at /var/folders/pm/twz8_z1j6_zb996w0b17bz2r0000gn/T//RtmpP4K6i1/sdp-review-1101157331d76
#> Created review-ready one-shot package with `create_sdp()`.
#> ℹ Prefilled semantic values were written directly into the metadata CSVs only
#>   where target fields were blank. Compatible table observation-unit drafts can
#>   be auto-applied using observation-unit/description first and otherwise table
#>   label/id fallback. Any "REVIEW:" entries already live in the metadata CSVs
#>   and must be confirmed or edited there.
#> ℹ No shortlist was seeded, so set any IRI you already know with
#>   `set_sdp_column()`.
#> ℹ Then run `review_metadata(pkg_path)` for the free-text fields and any
#>   required IRI nothing was suggested for; it prints the `set_sdp_dataset()` /
#>   `set_sdp_table()` / `set_sdp_column()` call that fills each one.
#> ℹ Finish with `validate_salmon_datapackage(pkg_path, require_iris = TRUE)`,
#>   rebuilding the EDH XML first if you need it. README-review.txt has the same
#>   checklist.
review_metadata(pkg)
#> ── dataset.csv ───────────────────────────────────────────────────────────────
#>    description: placeholder text, refused by strict validation
#>       MISSING DESCRIPTION: describe the contents and purpose of dataset
#>       'dataset-1'.
#>    creator: placeholder text, refused by strict validation
#>       MISSING METADATA: add creator, team, or originating program.
#>    contact_name: placeholder text, refused by strict validation
#>       MISSING METADATA: add primary contact name or team.
#>    contact_email: placeholder text, refused by strict validation
#>       MISSING METADATA: add primary contact email.
#> 
#>    set_sdp_dataset("/var/folders/pm/twz8_z1j6_zb996w0b17bz2r0000gn/T//RtmpP4K6i1/sdp-review-1101157331d76",
#>      description = "<describe the contents and purpose of dataset 'dataset-1'>",
#>      creator = "<add creator, team, or originating program>",
#>      contact_name = "<add primary contact name or team>",
#>      contact_email = "<add primary contact email>"
#>    )
#> 
#> ── tables.csv · table_1 ──────────────────────────────────────────────────────
#>    description: placeholder text, refused by strict validation
#>       MISSING DESCRIPTION: describe what each row in table 'table_1'
#>       represents.
#>    observation_unit: placeholder text, refused by strict validation
#>       MISSING METADATA: describe the observation unit for table 'table_1'.
#>    observation_unit_iri: required IRI and not yet decided
#> 
#>    set_sdp_table("/var/folders/pm/twz8_z1j6_zb996w0b17bz2r0000gn/T//RtmpP4K6i1/sdp-review-1101157331d76", "table_1",
#>      description = "<describe what each row in table 'table_1' represents>",
#>      observation_unit = "<describe the observation unit for table 'table_1'>",
#>      observation_unit_iri = "<IRI for what one row represents>"
#>    )
#> 
#> ── column_dictionary.csv · table_1 · spawner_count ───────────────────────────
#>    column_description: placeholder text, refused by strict validation
#>       MISSING DESCRIPTION: define what 'spawner_count' means in table
#>       'table_1'.
#>    term_iri: required IRI and not yet decided
#>    property_iri: required IRI and not yet decided
#>    entity_iri: required IRI and not yet decided
#>    unit_iri: required IRI and not yet decided
#> 
#>    set_sdp_column("/var/folders/pm/twz8_z1j6_zb996w0b17bz2r0000gn/T//RtmpP4K6i1/sdp-review-1101157331d76", "spawner_count", table = "table_1",
#>      column_description = "<define what 'spawner_count' means in table 'table_1'>",
#>      term_iri = "<IRI for term_iri>",
#>      property_iri = "<IRI for property_iri>",
#>      entity_iri = "<IRI for entity_iri>",
#>      unit_iri = "<IRI for unit_iri>"
#>    )
#> 
#> ── next ──────────────────────────────────────────────────────────────────────
#>    12 fields still block strict validation.
#>    Replace each <...> with the real value, then paste the calls above.
#>    5 of them are IRIs -- review_semantics() shows candidates for any that have them.
#>    Then: validate_salmon_datapackage("/var/folders/pm/twz8_z1j6_zb996w0b17bz2r0000gn/T//RtmpP4K6i1/sdp-review-1101157331d76", require_iris = TRUE)
#>