Metadata-Version: 2.3
Name: cwtwb
Version: 0.7.0
Summary: Tableau Workbook (.twb) Generation MCP Server
Project-URL: Homepage, https://github.com/imgwho/cwtwb
Project-URL: Changelog, https://github.com/imgwho/cwtwb/blob/master/CHANGELOG.md
Keywords: mcp,tableau,twb,visualization,workbook
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: GNU Affero General Public License v3 or later (AGPLv3+)
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Scientific/Engineering :: Visualization
Requires-Python: >=3.10
Requires-Dist: lxml>=5.0
Requires-Dist: mcp[cli]>=1.0
Description-Content-Type: text/markdown

# cwtwb

> **Tableau Workbook (.twb) Generation MCP Server**  
> Programmatically create Tableau workbooks with calculated fields, multiple chart types, dashboard layouts, and built-in structural validation.

## Overview

**cwtwb** is a Model Context Protocol (MCP) server that generates Tableau Desktop workbook files (`.twb`) from AI-driven tool calls. It provides atomic operations for building visualizations step by step:

1. **Zero-Config Start**: Works out-of-the-box with a built-in clean template and a minimal `Sample - Superstore` dataset, or load your own TWB template.
2. Add calculated fields
3. Create worksheets with various chart types
4. Assemble dashboards with flexible layouts
5. Save valid `.twb` files that open directly in Tableau Desktop

## Installation

```bash
pip install cwtwb
```

### Requirements

- Python ≥ 3.10
- [lxml](https://lxml.de/) ≥ 5.0
- [mcp](https://pypi.org/project/mcp/) ≥ 1.0

## Quick Start

### As MCP Server (Recommended for AI Tools)

To allow your AI Assistant to build Tableau Dashboards automatically, you need to add `cwtwb` to your MCP client's configuration file.

The easiest and safest way to use `cwtwb` is via `uvx` (an isolated Python environment runner via [uv](https://docs.astral.sh/uv/)). This requires zero installation.

#### 1. Claude Desktop
Open `~/Library/Application Support/Claude/claude_desktop_config.json` (Mac) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows) and add:

```json
{
  "mcpServers": {
    "cwtwb": {
      "command": "uvx",
      "args": ["cwtwb"]
    }
  }
}
```

#### 2. Cursor IDE
1. Open **Cursor Settings** -> **Features** -> **MCP**.
2. Click **Add New MCP Server**.
3. Set **Type** to `command`.
4. Set **Name** to `cwtwb`.
5. Set **Command** to `uvx cwtwb`.

#### 3. Claude Code (CLI)
Run the following command in your terminal to permanently add the server to your Claude Code workspace:

```bash
claude mcp add cwtwb -- uvx cwtwb
```

#### 4. VSCode (via Cline / RooCode)
Open your MCP settings file (usually located at `~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json` or equivalent on Windows) and append:

```json
{
  "mcpServers": {
    "cwtwb": {
      "command": "uvx",
      "args": ["cwtwb"]
    }
  }
}
```

### As Python Library

```python
from cwtwb.twb_editor import TWBEditor

editor = TWBEditor("templates/twb/superstore.twb")
editor.clear_worksheets()

# Add a calculated field
editor.add_calculated_field("Profit Ratio", "SUM([Profit])/SUM([Sales])")

# Create a bar chart
editor.add_worksheet("Sales by Category")
editor.configure_chart(
    worksheet_name="Sales by Category",
    mark_type="Bar",
    rows=["Category"],
    columns=["SUM(Sales)"],
)

# Create a pie chart
editor.add_worksheet("Segment Pie")
editor.configure_chart(
    worksheet_name="Segment Pie",
    mark_type="Pie",
    color="Segment",
    wedge_size="SUM(Sales)",
)

# Build a dashboard
editor.add_dashboard(
    dashboard_name="Overview",
    worksheet_names=["Sales by Category", "Segment Pie"],
    layout="horizontal",
)

editor.save("output/my_workbook.twb")
```

## MCP Tools

| Tool | Description |
|---|---|
| `create_workbook` | Load a TWB template and initialize the workspace |
| `list_fields` | List all available dimensions and measures |
| `add_parameter` | Add an interactive parameter for What-if analysis |
| `add_calculated_field` | Add a calculated field with Tableau formula |
| `remove_calculated_field` | Remove a previously added calculated field |
| `add_worksheet` | Add a new blank worksheet |
| `configure_chart` | Configure chart type and field mappings |
| `add_dashboard` | Create a dashboard combining worksheets |
| `add_dashboard_action` | Add filter or highlight actions to a dashboard |
| `generate_layout_json` | Build an interactive structured dashboard flexbox layout |
| `set_mysql_connection` | Configure the datasource to use a local MySQL connection |
| `set_tableauserver_connection` | Configure connection to an online Tableau Server |
| `save_workbook` | Save the final TWB file |

## Supported Chart Types

- **Bar** — horizontal/vertical bar charts
- **Line** — line charts and trends
- **Pie** — pie charts with color and wedge-size encodings
- **Area** — area charts
- **Map** — geographic maps with `geographic_field` and optional `map_fields` for LOD
- **Circle** / **Square** — shape marks
- **Text** — text tables and KPI cards (via `measure_values`)
- **Automatic** — Tableau's automatic mark type

## Built-in Validation

`save()` automatically validates the TWB XML structure before writing:

- **Fatal errors** (missing `<workbook>`, `<datasources>`, etc.) raise `TWBValidationError` and block saving
- **Warnings** (missing `<view>`, `<panes>`, etc.) are logged but don't block saving
- Disable with `editor.save("output.twb", validate=False)`

## Dashboard Layouts

| Layout | Description |
|---|---|
| `vertical` | Stack worksheets top to bottom |
| `horizontal` | Place worksheets side by side |
| `grid-2x2` | 2×2 grid layout (up to 4 worksheets) |
| `dict` (JSON) | **Declarative Custom Layouts**: An infinitely nestable FlexBox-style JSON structure for enterprise dashboards. |

Custom layouts can be built programmatically using the `TWBEditor` API by passing a `layout` nested dictionary. See `examples/scripts/demo_declarative_layout.py` for a complete example of generating a side-bar executive dashboard.

## Project Structure

```
cwtwb/
├── src/cwtwb/
│   ├── __init__.py          # Package init + API exports
│   ├── config.py            # Shared constants and utilities
│   ├── field_registry.py    # Field name ↔ TWB reference mapping
│   ├── twb_editor.py        # Core TWBEditor class (Mixin-based)
│   ├── charts.py            # ChartsMixin — chart configuration
│   ├── dashboards.py        # DashboardsMixin — dashboard creation
│   ├── connections.py       # ConnectionsMixin — DB connections
│   ├── parameters.py        # ParametersMixin — parameter management
│   ├── validator.py         # Runtime TWB XML structural validator
│   ├── layout.py            # Dashboard layout engine
│   └── server.py            # MCP server with FastMCP
├── tests/
│   ├── twb_assert.py        # TWBAssert chainable assertion DSL
│   ├── conftest.py          # Shared pytest fixtures
│   ├── test_twb_structure.py # Structural validation tests
│   └── ...                  # Other integration tests
├── examples/                # Example scripts and prompts
│   ├── scripts/             # Python examples for SDK
│   ├── prompts/             # Natural language prompts for MCP LLM
│   └── layouts/             # JSON declarative layout definitions
├── docs/
│   └── ROADMAP.md           # Project roadmap and priorities
├── pyproject.toml           # Package configuration
└── README.md
```

## Development

```bash
# Install in editable mode
pip install -e .

# Run test suite
pytest

# Test end-to-end workflow demo
python examples/scripts/demo_e2e_mcp_workflow.py

# Start MCP server
cwtwb
```

## License

AGPL-3.0-or-later
