Function infer_schema

nodedb/src/engine/timeseries/ilp_schema.rs:21–60

Infers a columnar schema from a batch of ILP lines. Scans all lines to discover tag keys and field keys, then builds a schema: timestamp + tag columns (Symbol) + field columns (typed).

(lines: &[IlpLine<'_>])

/// Scans all lines to discover tag keys and field keys, then builds
/// a schema: timestamp + tag columns (Symbol) + field columns (typed).
pub fn infer_schema(lines: &[IlpLine<'_>]) -> ColumnarSchema {
    let mut tag_keys: Vec<String> = Vec::new();
    let mut field_keys: Vec<(String, ColumnType)> = Vec::new();
    let mut seen_tags: std::collections::HashSet<String> = std::collections::HashSet::new();
    let mut seen_fields: std::collections::HashSet<String> = std::collections::HashSet::new();

    for line in lines {
        for &(key, _) in &line.tags {
            if seen_tags.insert(key.to_string()) {
                tag_keys.push(key.to_string());
            }
        }
        for &(key, ref val) in &line.fields {
            if seen_fields.insert(key.to_string()) {
                let col_type = match val {
                    FieldValue::Float(_) => ColumnType::Float64,
                    FieldValue::Int(_) | FieldValue::UInt(_) => ColumnType::Int64,
                    FieldValue::Str(_) => ColumnType::Symbol,
                    FieldValue::Bool(_) => ColumnType::Float64,
                };
                field_keys.push((key.to_string(), col_type));
            }
        }
    }

    let mut columns = Vec::with_capacity(1 + tag_keys.len() + field_keys.len());
    columns.push(("timestamp".to_string(), ColumnType::Timestamp));
    for tag in &tag_keys {
        columns.push((tag.clone(), ColumnType::Symbol));
    }
    for (field, ty) in &field_keys {
        columns.push((field.clone(), *ty));
    }

    ColumnarSchema {
        timestamp_idx: 0,
        codecs: vec![nodedb_codec::ColumnCodec::Auto; columns.len()],
        columns,
    }
}

/// Inject the three reserved bitemporal Int64 columns (`_ts_system`,
/// `_ts_valid_from`, `_ts_valid_until`) at the end of the schema when