Skip to contents

The one door for data from any channel (the entry form, CSV files, DataFindR, scripts). The extraction is checked against the extraction form (types, required fields, allowed values, links) and, if valid, written to submissions/<study_id>/<extraction_id>/ with a gate report. Value checks from _rules.yml are recorded as warnings. Nothing is written to data/ until the submission is accepted with accept_submission() or reconcile().

Usage

submit_study(
  path = ".",
  study_id,
  metadata,
  data = data.frame(),
  person,
  channel = "script",
  tool = NULL,
  model = NULL,
  source_sha256 = NULL,
  reports = NULL,
  provenance = NULL,
  blind = FALSE
)

Arguments

path

Project root.

study_id

Study id, e.g. "hansen2019" (see suggest_study_id()).

metadata

Named list matching the form's metadata, or the path to a metadata.yml file.

data

Data frame of data points, or the path to a data.csv file.

person

Id of the person or tool (from people.yml) who made the extraction.

channel

How the data were produced, e.g. "entry_form", "csv_import", "datafindr", "script".

tool

Producing software as name@version. Defaults to metawoRld's version.

model

LLM model, if one was used.

source_sha256

Hash of the full text the values were read from (see hash_file()).

reports

Reports of the study (list of lists with doi, pmid, role). Defaults to metadata$doi/metadata$pmid.

provenance

Optional data frame with per-data-point provenance (measurement_id and any of the provenance columns), e.g. a model per value.

blind

Hide this extraction from other extractors until reconciled (used by the entry form for dual extraction)?

Value

The extraction id, invisibly. Validation errors abort with class metawoRld_validation_error; nothing is written then.