pyspark-data-toolkit 0.3.0

Modular toolkit for Data Engineering with PySpark and Delta Lake — schema management, auditing, profiling, normalization, JSON handling, window functions, and Delta Lake operations.

pyspark-data-toolkit — Utilities Library for PySpark & Delta Lake

A production-ready utility library to accelerate Data Engineering workflows with PySpark and Delta Lake.

Stop rewriting boilerplate code. pyspark-data-toolkit delivers robust, modular functions for schema management, audit trails, string normalization, profiling, union operations, windowing, and Delta Lake management — all with logging support and data governance best practices built-in.

---

📦 Installation

pip install pyspark-data-toolkit

For local development:

git clone https://github.com/<your-org>/pyspark-data-toolkit.git
cd pyspark-data-toolkit
pip install -e ".[dev]"

---

✨ Key Features

• 📦 Modular design — Import only what you need (audit, schema, delta, json, profiling, etc.)
• ⚡ Production-grade — Exception handling and structured logging
• 🧪 Profiling & validation — Null analysis, stats, outliers, schema diffs
• 🧼 Normalization utilities — Clean strings and standardize column names
• 🔄 Schema enforcement — Apply, validate, and cast schemas safely
• 🧱 Delta Lake utilities — Merge, replace, optimize, vacuum, z-order
• 🪟 Window-based deduplication — Keep the latest records per key
• 📊 Audit metadata — Control columns, row hashes, batch IDs
• 🔗 JSON manipulation — Nested structure extraction, flattening
• 🧠 Diff utilities — Compare DataFrames and tag changes

---

📋 Modules Overview

Module	Description
audit_utils	Add control/audit columns and validate ingestion metadata.
dataframe_utils	Common transformations for DataFrames.
datetime_utils	Date/time parsing, formatting, and timezone conversion.
delta_table_utils	Delta Lake management: merges, optimizations, partitions.
diff_utils	Compare DataFrames and schemas, summarize differences.
json_utils	Extract and flatten JSON/nested structures in columns.
normalization_utils	Normalize strings and column names, safe numeric conversion.
profiling_utils	Null analysis, stats, outliers, cardinality, skew, correlations.
schema_utils	Apply, validate, and cast schemas from specs.
union_utils	Schema-aligned DataFrame unions or JSON merges.
window_utils	Latest-record selection and deduplication by window.

---

🚀 Quick Start Example

from pyspark.sql import SparkSession, Row
from pyspark_data_toolkit.audit_utils import add_control_columns
from pyspark_data_toolkit.profiling_utils import profile_nulls

spark = SparkSession.builder.getOrCreate()

df = spark.createDataFrame([
    Row(id=1, value="A"),
    Row(id=2, value=None)
])

# Add control columns (e.g., ingestion timestamp, batch ID)
df = add_control_columns(df)

# Profile null values
null_report = profile_nulls(df)
print(null_report)

---

📖 Module Usage Examples

💡 Below are short, illustrative snippets. Full pipelines and advanced use cases are available in EXAMPLES.md.cd

Audit

from pyspark_data_toolkit.audit_utils import *
df = add_control_columns(df, add_hash=True, version='v2')
df = check_row_duplicates(df)
df = add_audit_trail_columns(df)

Delta Lake

from pyspark_data_toolkit.delta_table_utils import write_delta_table, merge_delta_table
write_delta_table(spark, df, "db.table", "/path", arq_format="delta", mode="overwrite", partition_cols=("part",))
merge_delta_table(spark, df_updates, "db.table", "/path", merge_cols=("id",))

Profiling

from pyspark_data_toolkit.profiling_utils import profile_nulls, profile_numeric_stats
nulls = profile_nulls(df)
stats = profile_numeric_stats(df)

Normalization

from pyspark_data_toolkit.normalization_utils import normalize_strings, normalize_column_names
df = normalize_strings(df, columns=["name"])
df = normalize_column_names(df)

Union

from pyspark_data_toolkit.union_utils import union_all_with_schema
df_union = union_all_with_schema([df1, df2])

Window

from pyspark_data_toolkit.window_utils import drop_duplicates_keep_latest
df_latest = drop_duplicates_keep_latest(df, keys=["id"], order_col="timestamp")

---

🏆 Best Practices

• Be modular — Import only the needed functions for clarity and performance.
• Profile early — Run profiling after ingestion to catch anomalies quickly.
• Validate schemas — Apply and validate schemas before transformations.
• Governance first — Use audit utilities to ensure traceability.
• Delta Lake safety — When overwriting, always set replace_where to avoid unwanted partition drops.

---

🔧 Dependencies

• Python >= 3.8
• PySpark >= 3.0
• Delta Lake (optional, required only for Delta modules)

---

🧪 Testing

# Install dev dependencies
pip install -e ".[dev]"

# Run all tests
make test

---

📝 Changelog

v0.1.0

• Initial release
• Schema management utilities
• Profiling (nulls, stats, outliers, diffs)
• Normalization functions
• Delta Lake operations (merge, optimize, vacuum)
• JSON extraction and flattening

---

🤝 Contributing

Contributions are welcome!

1. Fork the project
2. Create your feature branch (git checkout -b feature/pyspark-data-toolkit)
3. Commit your changes (git commit -m 'Add new feature')
4. Push to your branch (git push origin feature/pyspark-data-toolkit)
5. Open a Pull Request

---

📜 License

MIT License — see LICENSE for details.