From a typed CSV to a tensor

// Synthetic records only: these values do not represent patients or clinical rules.
// Run from the project root; CSV paths are relative to the working directory.
import aner.data;
import aner.tensor;

fn main() -> Unit {
    var schema: DataSchema = data.schema();
    schema = data.column(schema, "record_id", "String", false);
    schema = data.column(schema, "age", "Int64", true);
    schema = data.column(schema, "measurement", "Float64", true);
    schema = data.column(schema, "site", "String", false);
    schema = data.column(schema, "eligible", "Bool", false);

    // This declares a plan. No file is loaded until summarize or collect.
    // Only unquoted NA is missing; the reader does not guess or silently clean.
    let source: DataScan = data.scan_csv("examples/fixtures/visits.csv", schema, "NA");
    let adults = data.where_ge(source, "age", 18);
    let selected = data.include(data.select(adults, "measurement"), "age");
    print(data.explain(selected));

    // One MiB for managed execution buffers; two selected rows per batch.
    // All source columns are validated, including rows that do not pass filters.
    let budget = 1048576;
    let report: DataReport = data.summarize(selected, budget, 2);
    print(data.describe(report));
    print("Validated source rows"); print(data.scanned_rows(report));
    print("Selected rows"); print(data.rows(report));
    print("Missing measurements"); print(data.missing(report, "measurement"));
    print("Mean measurement among observed selected rows");
    print(data.mean(report, "measurement"));

    // Exclusion is explicit: missing measurements are not filled with zero.
    // Collection executes a new scan of the same path; it is not a snapshot.
    let complete = data.where_present(selected, "measurement");
    let table: DataTable = data.collect(complete, 100, budget, 2);
    print("Collected complete rows"); print(data.rows(table));
    print("First column"); print(data.column_name(table, 0));
    let values = data.to_tensor(table, budget);
    print("Tensor rows"); print(tensor.rows(values));
    print("Tensor columns"); print(tensor.cols(values));
    print("First measurement"); print(tensor.value(values, 0, 0));

    // Identifiers retain their original text, including leading zeros.
    let identifiers = data.collect(data.select(source, "record_id"), 100, budget, 2);
    print("First record ID"); print(data.text(identifiers, 0, "record_id"));
}