tfl-decompiler 1.0.0

Convert Tableau Prep Flow files to Python/pandas scripts

TFL Decompiler

Convert Tableau Prep Flow files (.tfl) to Python/pandas scripts.

Say goodbye to Tableau Prep's proprietary flows. Transform your data transformations into portable, version-controlled, scalable Python code.

---

Table of Contents

• Installation
• Quick Start
• Usage
  • Command Line
  • Python API
• Example Output
• Supported Transformations
• Architecture
• Why Replace Tableau Prep?
• Contributing
• License

---

Installation

pip install tfl-decompiler

With optional database connectors:

# PostgreSQL
pip install "tfl-decompiler[postgresql]"

# MySQL
pip install "tfl-decompiler[mysql]"

# SAP HANA
pip install "tfl-decompiler[saphana]"

# All database connectors
pip install "tfl-decompiler[all]"

Requirements: Python 3.9+

---

Quick Start

tfl-decompile my_flow.tfl -o my_flow.py
python my_flow.py

---

Usage

Command Line

After installation, the tfl-decompile command is available on your PATH.

# Convert a flow to Python (save to file)
tfl-decompile my_flow.tfl -o my_flow.py

# Print generated code to stdout
tfl-decompile my_flow.tfl

# Analyze flow structure without generating code
tfl-decompile my_flow.tfl --analyze

# Batch convert all flows in a directory
for f in *.tfl; do tfl-decompile "$f" -o "${f%.tfl}.py"; done

You can also run it as a module:

python -m tfl_decompiler my_flow.tfl -o my_flow.py

Python API

from tfl_decompiler import decompile

# Generate Python code string
code = decompile("my_flow.tfl")
print(code)

# Generate and save to file
decompile("my_flow.tfl", output_path="my_flow.py")

# Inspect the flow structure
from tfl_decompiler import TFLExtractor

extractor = TFLExtractor("my_flow.tfl")
content = extractor.extract()
summary = extractor.get_flow_summary(content)
print(summary)

# Fine-grained control
from tfl_decompiler import FlowParser, CodeGenerator

parser = FlowParser(content)
graph = parser.build_graph()
generator = CodeGenerator(graph)
code = generator.generate()

---

Example Output

Given a Tableau Prep flow that reads a CSV, renames columns, filters rows, aggregates, and writes output:

#!/usr/bin/env python3
"""
Auto-generated from: sales_flow.tfl
Generated by: TFL Decompiler
"""

import pandas as pd
import numpy as np
import os
from pathlib import Path


def main(input_sales: str = "data/sales.csv", output_results: str = "output/results.csv"):
    """Execute the data transformation flow."""

    # Step 1: Load data from CSV file
    df_input_sales = pd.read_csv(input_sales)

    # Step 2: Clean step (3 operations): rename, changetype, filter
    df_clean_transform = df_input_sales.copy()
    df_clean_transform = df_clean_transform.rename(columns={"old_name": "new_name"})
    df_clean_transform["amount"] = df_clean_transform["amount"].astype(float)
    df_clean_transform = df_clean_transform[df_clean_transform["status"] == "active"]

    # Step 3: Aggregate by [product_category]
    df_agg_summary = df_clean_transform.groupby("product_category").agg(
        {"amount": "sum", "quantity": "mean"}
    ).reset_index()

    # Step 4: Output to CSV
    os.makedirs(Path(output_results).parent, exist_ok=True)
    df_agg_summary.to_csv(output_results, index=False)

    return df_agg_summary


if __name__ == "__main__":
    import argparse

    parser = argparse.ArgumentParser(description="Decompiled Tableau Prep flow")
    parser.add_argument("--input_sales", default="data/sales.csv")
    parser.add_argument("--output_results", default="output/results.csv")
    args = parser.parse_args()
    result = main(input_sales=args.input_sales, output_results=args.output_results)
    print(f"Processing complete. Output shape: {result.shape}")

---

Supported Transformations

Node Types

Node	Description
Input	CSV, Excel, JSON, PostgreSQL, MySQL, SAP HANA
Clean / Transform	Rename, filter, calculate, type conversion, and more
Pivot	Columns-to-rows (melt) and rows-to-columns (pivot_table)
Aggregate	groupby with all common aggregations
Join	Inner, left, right, full outer
Union	Vertical concatenation
Filter	Boolean conditions
Output	CSV, Excel, databases

Input Sources

Tableau	Pandas
CSV File	pd.read_csv()
Excel File	pd.read_excel()
JSON File	pd.read_json()
PostgreSQL	pd.read_sql() with psycopg2
MySQL	pd.read_sql() with mysql-connector
SAP HANA	pd.read_sql() with hdbcli

Clean Operations

Tableau	Pandas
Rename	df.rename(columns={})
Remove Field	df.drop(columns=[])
Change Type	df.astype() / pd.to_datetime()
Filter	df[condition]
Replace Values	df.replace()
Split Field	df[col].str.split()
Merge Fields	String concatenation
Fill Nulls	df.fillna()
Remove Nulls	df.dropna()
Calculated Field	New column assignment
Trim	df[col].str.strip()
Case Conversion	str.upper() / str.lower()

Pivot Operations

Tableau	Pandas
Columns to Rows	pd.melt()
Rows to Columns	df.pivot_table()

Aggregations

Tableau	Pandas
SUM	sum
AVG	mean
COUNT	count
COUNTD	nunique
MIN	min
MAX	max
MEDIAN	median
STDEV	std

Joins

Tableau	Pandas
Inner	pd.merge(how='inner')
Left	pd.merge(how='left')
Right	pd.merge(how='right')
Full Outer	pd.merge(how='outer')

Tableau Expression Syntax

Tableau	Pandas
[Field Name]	df['Field Name']
IF...THEN...ELSE...END	np.where()
CONTAINS([Field], 'x')	df['Field'].str.contains('x')
LEFT([Field], n)	df['Field'].str[:n]
UPPER([Field])	df['Field'].str.upper()
ZN([Field])	df['Field'].fillna(0)
IFNULL([Field], val)	df['Field'].fillna(val)
YEAR([Date])	df['Date'].dt.year

---

Architecture

tfl_decompiler/
├── __init__.py           # Public API exports
├── __main__.py           # python -m tfl_decompiler entry point
├── main.py               # CLI entry point (tfl-decompile command)
├── extractor.py          # .tfl ZIP extraction and JSON parsing
├── flow_parser.py        # Flow graph construction + topological sort
├── node_registry.py      # Node type → handler mapping
├── code_generator.py     # Python code generation
├── expression_parser.py  # Tableau calc → pandas expression translator
├── db_connectors.py      # Database connection helpers
├── utils.py              # Utility functions
├── nodes/
│   ├── base.py           # Abstract base node
│   ├── input_node.py     # Data source handlers
│   ├── clean_node.py     # Transform handlers
│   ├── pivot_node.py     # Pivot handlers
│   ├── aggregate_node.py # Aggregation handlers
│   ├── join_node.py      # Join handlers
│   ├── union_node.py     # Union handlers
│   ├── filter_node.py    # Filter handlers
│   └── output_node.py    # Output handlers
└── builder/
    ├── flow_builder.py   # High-level flow assembly
    └── node_builders.py  # Per-node code builders

---

Why Replace Tableau Prep?

Tableau Prep	Python / pandas
Proprietary format	Plain text, version-controlled
GUI-only	Scriptable and automatable
License required	Free and open source
Limited scalability	Scale to any size
Black box	Full transparency
Desktop only	Runs anywhere

---

Contributing

Contributions are welcome. Areas of interest:

• Additional node types
• More Tableau functions in the expression parser
• Additional database connectors
• Test cases with real .tfl files

Please open an issue or pull request on GitHub.

---

License

MIT License