mince 0.2.3

mince is a toolkit for slicing up data into bite-sized dashboards

mince 🔪🧄🧅

mince is a toolkit for slicing up data into bite-sized dashboards

The goal of mince is to minimize the effort required to build and maintain dashboards

mince provides a single coherent interface to help with many different dashboard-related tasks:

• ETL: data exploration, design of queries, execution of queries, persistent data storage
• UI: data visualization, custom interactive buttons
• hosting: building images, deploying images, network configuration
• live updates: collecting, storing, and serving data diffs efficiently

Table Of Contents

1. Getting started
2. Data Collection
3. UI Specification
4. Command Line Interface

Getting started

1. Installation

pip install mince

2. Define a new Dashboard class

Every dashboard needs 1) a data collector function and 2) a specification for how to display data

from mince import Dashboard, CollectKwargs, UiSpec
import polars as pl


class SomeNewDash(Dashboard):
    @classmethod
    def collect_data(cls, **kwargs: Unpack[CollectKwargs]) -> dict[str, pl.DataFrame]:
        ...

    @classmethod
    def load_spec(cls) -> UiSpec:
        ...

3. (optional) Register dashboard

Adding dashboard to the local registry makes it possible to refer to dashboards by their name instead of their class

python:

mince.register(SomeNewDash)

cli:

mince register some_module_name.SomeNewDash

4. Collect dashboard data

The collector function should return a dict of polars dataframes, which mince will then store in its data directory.

python:

mince.collect_dashboard_data(SomeNewDash)
# or, if registered: mince.collect_dashboard_data('some_new_dash')

cli:

mince collect some_new_dash

5. Run dashboard

Running the dashboard loads the most recently collected data and serves a UI on a local port (such as http://localhost:8052)

python:

mince.run_dashboard(SomeNewDash)
# or, if registered: mince.run_dashboard('some_new_dash')

cli:

mince run some_new_dash

Environment

• MINCE_ROOT root directory where mince stores collected data, default ~/data/mince
• MINCE_REGISTRY path to mince registry, default MINCE_ROOT/mince_registry.json

Data Collection

[WIP]

Data storage

Design goals

• use same path structure for local filesystem vs web urls
• allow atomic updates for groups of files

UI Specification

[WIP]

Command Line Interface

The mince cli aims to automate as many of these tasks as possible

• run dashboard mince run ... --pdb
• run dashboard with interactive debugger mince run ... --pdb
• list info about all running dashboards
  • mince ls
• disk caching of loaded datasets for quick dashboard restarts
• standardized hooks for data collection mince collect <DASHBOARD>
• cli generator to create a management cli for each dashboard <DASHBOARD> <SUBCOMMAND> ...
• easily load a dashboard's data into an interactive python session mince data <DASHBOARD> -i

Custom CLI's

• bypasses the registry
• takes config settings using arguments
• is meant to be intended within each dashboard package