Output contracts

base_cli.output.render_records() supports text, csv, tsv, yaml, json, and ndjson formats. Install base-cli[yaml] before selecting yaml; the other formats are available from the core package. The requested text format is presentation-aware: it renders a table on a TTY and tab-delimited rows when stdout is redirected or piped.

Delimited output is intentionally automation-friendly:

• rows are streamed directly from the iterable, so CSV and TSV do not retain the complete result set in memory;
• the supplied columns sequence controls both column order and cell lookup;
• no column header or footer is emitted;
• values use the standard csv quoting rules, while ANSI escape sequences and other control characters are replaced with spaces.

ndjson is the bounded machine-output format for large or long-running results. It consumes the input iterable once and writes one flushed JSON object per record without first building a list. Each line has this stable shape:

{"schema_version":1,"schema":"base-cli.record","record":{"name":"base"}}

Use base_cli.NdjsonWriter when a consumer produces records incrementally. The StructuredRecord and StructuredResultWriter types describe the public producer boundary. Diagnostics remain on stderr; a consumer should not mix log lines into the NDJSON stream.

Terminal tables use Unicode display-cell width rather than Python string length. Long cells are bounded by max_cell_width (80 by default), and the complete table is fitted to the detected terminal width (120 columns as a safe fallback) using an ellipsis. Pass terminal_width and max_cell_width explicitly when a caller needs deterministic rendering in tests or a custom frontend.

For an optional polished human table, pass rich=True to render_records(). Rich is consulted only for interactive text; all redirected and structured formats retain the rules above and fall back to the built-in renderer if Rich is unavailable or fails.