nexcore-dataframe

Sovereign columnar DataFrame engine for NexCore.

Zero unsafe code, zero external DataFrame dependency. Replaces polars across the workspace, eliminating CRITICAL (fast-float segfault) and HIGH (pyo3 buffer overflow) transitive vulnerabilities.

Core types

• [DataFrame] — columnar table with named, typed columns
• [Column] — named array of homogeneous nullable values
• [Scalar] — single typed value for comparisons and aggregation results
• [Counter] — optimized hash-based group-count (replaces polars group_by+count)
• [Schema] — column name → type mapping

Operations

• Filter: df.filter(&mask), df.filter_by("col", |v| pred)
• Sort: df.sort("col", descending), df.head(n), df.tail(n)
• Select: df.select(&["col1", "col2"]), df.drop_columns(&["col"])
• Aggregate: col.sum(), col.mean(), col.min(), col.max(), col.median()
• GroupBy: df.group_by(&["col"])?.agg(&[Agg::Sum("val".into())])
• Join: df.join(&other, &["key"], JoinType::Inner) (inner/left/right/outer/semi/anti)
• I/O: DataFrame::from_json(s), df.to_json(), df.to_json_file(path)