Documentation
¶
Overview ¶
Package metadata exposes CSV/JSON column type inference as SPL builtins - useful for profiling an unfamiliar CSV/JSON data source (e.g. "is this column really numeric, or does it have stray text rows?") before writing a schema/import pipeline by hand.
The inference logic below is a direct port of the pure functions in github.com/oarkflow/metadata's field.go (InferCSVColumnType, InferCSVFileTypes, InferJSONFieldType, InferJSONFileType and their helpers), reimplemented here rather than imported: that module's other files pull in a full SQL-datasource connector layer (mysql/postgres/ mssql/sqlite via squealx) that both overlaps with plugins/database and, at the versions this repo currently pins, fails to build (a squealx subpackage those connector files import was removed upstream). Porting just the ~100 lines of inference logic avoids both problems.
Source Files