Console output
AppConsole is a thin facade over rich.console.Console with four output
modes:
from rn_forge.commons import console, OutputMode
console.info("Starting {}", "job")
console.success("Done in {}s", 1.2)
console.table("name", "status", rows=[("alpha", "ok"), ("beta", "failed")])
The quiet/JSON tri-mode pattern
A CLI typically needs three output personalities: normal (styled,
human-readable), --quiet (suppress everything but an optional one-line
summary), and --json (machine-readable only). Model that by switching
console's mode instead of writing separate code paths:
from rn_forge.commons import console, OutputMode
def run(*, quiet: bool = False, json_output: bool = False) -> None:
if json_output:
console.set_mode(OutputMode.JSON)
elif quiet:
console.set_mode(OutputMode.QUIET)
result = do_work()
# One call, three behaviours:
# - RICH: prints `result` as Rich would (a Table, a str, ...)
# - QUIET: prints only quiet_text, or nothing
# - JSON: serializes `result` via JsonUtils.serialize
console.emit(result, quiet_text="ok")
table() follows the same pattern — it renders a Rich table normally, and emits
[{column: cell, ...}, ...] in JSON mode, so call sites never need an
if json: ... branch:
warning()/error() always go to stderr; fail() calls error() then raises
SystemExit — no Typer dependency, so it works the same from a plain script as
from a Typer command. The Typer-specific layer built on top of this one is
rn-forge-cli.
The underlying rich.console.Console is always reachable via console.rich for
anything this facade doesn't cover — a Progress bar, a Live display, or a
custom renderable.
Where the mode comes from in a CLI
Nothing in this package sets the mode for you — it is a plain facade, usable
from a script, a Django management command, a worker or a test. In an rn-forge
command-line application, rn-forge-cli drives it: its --quiet and --json
flags call set_mode before the command body runs, so a command just calls
console.emit(...) and gets the right personality.