vwoptimize.py 文件源码

def _convert_any_to_vw(source, format, output, weights, preprocessor, columnspec, named_labels, remap_label, ignoreheader):
    if named_labels is not None:
        assert not isinstance(named_labels, basestring)
        named_labels = set(named_labels)

    rows_source = open_anything(source, format, ignoreheader=ignoreheader)
    output = open(output, 'wb')

    for row in rows_source:
        try:
            vw_line = convert_row_to_vw(row, columnspec, preprocessor=preprocessor, weights=weights, named_labels=named_labels, remap_label=remap_label)
        except Exception:
            log_always('Failed to parse: %r', row)
            raise
        output.write(vw_line)

    flush_and_close(output)