nicegui-builder 0.0.0.dev3

Plugin-first NiceGUI builders for declarative layouts, forms, and tables

nicegui-builder

Home

nicegui-builder is a Python library for building NiceGUI interfaces from declarative layouts and schema-aware plugins.

[!WARNING] This is an early pre-release published to invite discovery and feedback. The project is not presented as production-ready or fully stabilized yet. Expect API changes, rough edges, and evolving behavior while the library is still taking shape.

It currently provides three main entry points:

• builder(layout) for declarative NiceGUI rendering
• form(source, flavor="") for plugin-driven forms
• table(source, variant="std") for plugin-driven tabular views

Install

Base install:

pip install nicegui-builder

Optional pandas support:

pip install "nicegui-builder[pandas]"

CLI entry point:

nicegui-builder --help

Stable API

Top-level imports from nicegui_builder are the supported stable public API. See docs/public-api.md for the current stable surface.

Quick Start

builder(layout)

Use builder(...) when you already have a declarative layout.

import yaml
from pathlib import Path

from nicegui import ui
from nicegui_builder import builder

layout = yaml.safe_load(
    Path("src/nicegui_builder/examples/demo_basic_builder.yml").read_text(encoding="utf-8")
)

builder(layout)
ui.run()

form(source, flavor="")

Use form(...) when you want a plugin to inspect a supported source and render a form.

from pydantic import BaseModel
from nicegui import ui

from nicegui_builder import form


class Contact(BaseModel):
    firstname: str
    lastname: str
    email: str


handle = form(Contact)
ui.run()

If a matching YAML layout is not found, the plugin can fall back to an automatic layout.

Built-in automatic pydantic flavors:

• std: responsive editable grid
• compact: simple stacked editable layout
• detail: read-only detail view
• filters: search-oriented filter form
• actionable: editable layout with status, error, and action areas

Examples:

form(Contact, flavor="compact")
form(Contact, flavor="detail")
form(Contact, flavor="filters")
handle = form(Contact, flavor="actionable")

For datetime fields, the built-in pydantic plugin uses a split date_input + time_input widget by default. That split stays grouped as one logical field, but the layout can choose its wrapper container.

Example:

- field__starts_at:
    container: grid
    params:
      columns: 2
    classes: col-span-12 gap-2

That lets you keep the date/time pair together while placing it inside a row, column, grid, or another declarative container.

table(source, variant="std")

Use table(...) when you want a plugin to inspect a collection-like source and render a table.

import pandas as pd
from nicegui import ui

from nicegui_builder import table


df = pd.DataFrame(
    [
        {"name": "Ada", "country": "UK"},
        {"name": "Grace", "country": "US"},
    ]
)

handle = table(df)
ui.run()

For the pandas plugin, a richer filtered table variant is also available:

handle = table(df, variant="filters")

Working With Form Handles

form(...) returns a FormHandle.

Read values:

handle.get_values()

Rebuild a model:

handle.to_model()
handle.submit(lambda model: print(model), as_model=True)

Validate:

result = handle.validate(as_model=True)
print(result.valid)
print(result.errors_by_field())
print(result.error_messages())

Reset:

handle.reset_to_source()
handle.reset_to_defaults()
handle.reset_to_empty()

Track changes:

handle.changed_fields()
handle.is_dirty()

Live helpers:

handle.live_dirty_badge()
handle.live_reset_button()
handle.live_submit_button(
    "Save",
    lambda model: print(model),
    as_model=True,
    strategy="dirty_and_valid",
)
handle.live_error_panel(as_model=True, strategy="invalid")

Validation helpers:

handle.apply_errors(as_model=True)
handle.clear_errors()
handle.validate_field("email", as_model=True)
handle.field_error_message("email", as_model=True)
handle.apply_field_errors("email", as_model=True)
handle.live_validation(as_model=True, mode="change")

Action helpers:

handle.submit_button("Save", lambda model: print(model), as_model=True)
handle.reset_button()

handle.create_button(lambda model: save_new(model))
handle.update_button(lambda model: save_existing(model))
handle.delete_button(lambda instance: delete_instance(instance))

handle.crud_bar(
    on_create=lambda model: save_new(model),
    on_update=lambda model: save_existing(model),
    on_delete=lambda instance: delete_instance(instance),
)

handle.action_bar(
    "Save",
    lambda model: print(model),
    as_model=True,
    live_changed_badge=True,
    live_reset_button=True,
    live_submit_button=True,
    submit_strategy="dirty_and_valid",
)

Working With Table Handles

table(...) returns a TableHandle.

Inspect the table:

handle.spec
handle.component
handle.get_rows()

Sort, paginate, select, and export:

handle.sort_rows("name")
handle.set_pagination(rows_per_page=25, sort_by="score", descending=True)
handle.set_selected_rows([{"name": "Ada"}])
csv_text = handle.export_csv()

Filter:

handle.set_filter("name", "ada", op="contains")
handle.set_filter("score", [10, 20], op="between")
handle.apply_filters()
handle.clear_filters().apply_filters()

CRUD-style actions:

handle.crud_bar(
    on_create=lambda table_handle: open_create_dialog(),
    on_delete_selected=lambda rows: delete_rows(rows),
    on_export=lambda csv_text: print(csv_text),
)

CLI

The project also exposes a small CLI for quick experimentation.

List bundled examples:

nicegui-builder examples list

Run a bundled example:

nicegui-builder examples run 01_basic_builder --port 8080
nicegui-builder examples run 03_pydantic_form_basic --port 8080

Render a YAML layout directly:

nicegui-builder layout run path/to/layout.yml --port 8080

Render a form from a Python object path:

nicegui-builder form run my_module:MyModel --flavor actionable --port 8080

This is especially useful to:

• launch bundled examples quickly
• test a declarative layout without writing a dedicated script
• try a form source interactively while tuning layouts and flavors

Use Ctrl+C as the normal cross-platform way to stop a running CLI session.

Pydantic Field Mapping

The pydantic plugin resolves field__... nodes using:

• field type
• field constraints and metadata
• plugin-local widget mapping in src/nicegui_builder/plugins/pydantic/pydantic-nicegui.yml

Example:

- card.tight:
    classes: p-3 gap-3
    children:
    - label:
        params:
          text: Contact info form
    - grid:
        params:
          columns: 12
        classes: w-full
        children:
        - field__firstname:
            classes: col-span-6
        - field__lastname:
            classes: col-span-6
        - field__email:
            methods: email
            classes: w-full
        - field__starts_at:
            container: grid
            params:
              columns: 2
            classes: col-span-12 gap-2

More Docs

• docs/public-api.md: stable public API
• docs/plugin.md: how to build a new plugin
• docs/architecture.md: architecture and class diagram
• docs/publishing.md: TestPyPI and PyPI publishing guide
• docs/product-roadmap.md: roadmap history