Add plotly run command

[T4rk1n]

Add a command to run dash app from the terminal.

The command can be run two methods:

• plotly run app:app --debug
• python -m dash run app:app --debug

The arguments follow Dash.run, run help:

$ plotly run --help
usage: plotly run [-h] [--host HOST] [--port PORT] [--proxy PROXY] [--debug | --no-debug | -d] [--dev-tools-ui | --no-dev-tools-ui] [--dev-tools-props-check | --no-dev-tools-props-check]
                  [--dev-tools-serve-dev-bundles | --no-dev-tools-serve-dev-bundles] [--dev-tools-hot-reload | --no-dev-tools-hot-reload] [--dev-tools-hot-reload-interval DEV_TOOLS_HOT_RELOAD_INTERVAL]
                  [--dev-tools-hot-reload-watch-interval DEV_TOOLS_HOT_RELOAD_WATCH_INTERVAL] [--dev-tools-hot-reload-max-retry DEV_TOOLS_HOT_RELOAD_MAX_RETRY]
                  [--dev-tools-silence-routes-logging | --no-dev-tools-silence-routes-logging] [--dev-tools-disable-version-check | --no-dev-tools-disable-version-check] [--dev-tools-prune-errors | --no-dev-tools-prune-errors]
                  app

Run a local development server for a Dash app.

positional arguments:
  app                   The Dash app to run, in the format "module:variable".

options:
  -h, --help            show this help message and exit
  --host HOST           Host IP used to serve the application (Default: "127.0.0.1").
  --port PORT, -p PORT  Port used to serve the application (Default: "8050").
  --proxy PROXY         Proxy configuration string, e.g., "http://0.0.0.0:8050::https://my.domain.com".
  --debug, --no-debug, -d
                        Enable/disable Flask debug mode and dev tools.

dev tools options:
  --dev-tools-ui, --no-dev-tools-ui
                        Enable/disable the dev tools UI.
  --dev-tools-props-check, --no-dev-tools-props-check
                        Enable/disable component prop validation.
  --dev-tools-serve-dev-bundles, --no-dev-tools-serve-dev-bundles
                        Enable/disable serving of dev bundles.
  --dev-tools-hot-reload, --no-dev-tools-hot-reload
                        Enable/disable hot reloading.
  --dev-tools-hot-reload-interval DEV_TOOLS_HOT_RELOAD_INTERVAL
                        Interval in seconds for hot reload polling (Default: 3).
  --dev-tools-hot-reload-watch-interval DEV_TOOLS_HOT_RELOAD_WATCH_INTERVAL
                        Interval in seconds for server-side file watch polling (Default: 0.5).
  --dev-tools-hot-reload-max-retry DEV_TOOLS_HOT_RELOAD_MAX_RETRY
                        Max number of failed hot reload requests before failing (Default: 8).
  --dev-tools-silence-routes-logging, --no-dev-tools-silence-routes-logging
                        Enable/disable silencing of Werkzeug's route logging.
  --dev-tools-disable-version-check, --no-dev-tools-disable-version-check
                        Enable/disable the Dash version upgrade check.
  --dev-tools-prune-errors, --no-dev-tools-prune-errors
                        Enable/disable pruning of tracebacks to user code only.

[syl-oh]

Nice, that'll make remembering all the debug flags easier

[syl-oh]

Do we want to use that in the CI for some test to avoid regression?

[emilykl]

@T4rk1n What do you think about using app as the default variable name if no variable name is provided? So e.g. if you had a file named my_dash_app.py containing an app variable called app, you could just run

plotly run my_dash_app --debug

[emilykl]

@T4rk1n For the flags, is there a way to make the helptext clearer to the user about which option is the default? I.e. if I can pass --debug or --no-debug, how do I know what is the default behavior if I pass neither?

For that matter (since I do know that no-debug is the default), what is the use case for the --no-debug flag?

[ndrezn]

We should be able to detect the app object automatically given a Python file. Here's a pattern to do this: https://github.com/plotly/de-client/blob/main/src/de_client/cli/root.py#L302-L306

[T4rk1n]

@T4rk1n What do you think about using app as the default variable name if no variable name is provided? So e.g. if you had a file named my_dash_app.py containing an app variable called app, you could just run

plotly run app --debug

Yes that might be more convenient 👍

@T4rk1n For the flags, is there a way to make the helptext clearer to the user about which option is the default? I.e. if I can pass --debug or --no-debug, how do I know what is the default behavior if I pass neither?

For that matter (since I do know that no-debug is the default), what is the use case for the --no-debug flag?

Oh, the --no-<option> is added by argparse.BooleanOptionalAction, I didn't notice them before but they don't make much sense since those are flags, I'll see how to remove them.

[gvwilson]

check that len(app_split) == 2 here?

[gvwilson]

OK, I see that further down you're allowing users to provide just the module name, then using the first Dash you find in that module.

[gvwilson]

if I run with just the module name, but the module doesn't define a Dash, I think app_instance will be uninitialized at the point of the isinstance() check.