Skip to content
DDCELL

Free early testing build · feedback wanted

Clean messy data from the command line.

DCEll is a free early tester build for cleaning tabular files, SQL databases, and NoSQL databases — MongoDB, Cassandra, and Neo4j.

Download the free tester build Explore on GitHub

Windows installer · via GitHub Releases

Installer is not code-signed

This Windows installer is not code-signed yet. Windows SmartScreen or security software may show an unknown publisher warning. That does not prove the installer is malicious, but it also does not certify it as safe. Only install if you trust this project; review the source code and release details first. A signed installer is planned for a future release.

What it does

One tool for files, SQL, and NoSQL.

Files

Ingest supported files into SQLite, clean the tables, and export cleaned results.

  • CSV
  • TSV/TXT
  • Excel
  • ODS
  • JSON
  • XML
  • HTML tables
  • Parquet
  • Feather
  • ORC
  • HDF
  • Stata
  • SPSS
  • SAS

Pickle is opt-in only — loading pickle can execute code.

Recursive folders

Clean supported files across nested folders and export organized results.

SQL

Built on SQLAlchemy, working with configured database dialects and drivers.

NoSQL

MongoDB, Cassandra, and Neo4j. Defaults to a dry-run preview; writing changes requires explicit opt-in.

Cleaning operations

Careful by default, explicit when it matters.

For files and SQL tables.

Trim & quote cleanup
Strip stray whitespace and wrapping quotes.
Missing values
Handle empty and missing entries.
Duplicate removal
Optional — only when you ask for it.
Case & whitespace
Case transforms and whitespace collapse.
Unicode normalization
Consistent text representation.
Explicit conversions
Email, boolean, numeric, and date — invalid values are surfaced, not silently coerced.
Fuzzy grouping
Opt-in and lossy, so it’s off unless enabled.
Identifier protection
ID-like values are protected by default.
Reports
See what changed after each run.

Safety first

Your originals are not an experiment.

DCEll is designed to avoid surprising, destructive changes.

  • Preserves source data where possible.
  • File and SQL workflows can write to a separate target.
  • NoSQL defaults to dry-run; writes need explicit opt-in.
  • Pickle loading is disabled by default.
  • Review output and back up before any in-place writes.

Before you install

This Windows installer is not code-signed yet. Windows SmartScreen or security software may show an unknown publisher warning. That does not prove the installer is malicious, but it also does not certify it as safe. Only install if you trust this project; review the source code and release details first. A signed installer is planned for a future release.

Review the source codeRead release details

Get started

Four steps to a cleaner dataset.

  1. 01

    Download

    Get the Windows installer from GitHub Releases.

  2. 02

    Run the installer

    Read the unsigned-installer note above first.

  3. 03

    Open a terminal

    Any terminal on your Windows machine.

  4. 04

    Start exploring

    Run DCEll --help, then DCEll init for the guided setup.

$ DCEll --help

$ DCEll init

Download the free tester build

This is an early version. Tell us what breaks.

DCEll is free right now so it can be tested on real-world messy data. Use the built-in feedback command, or open an issue on GitHub.

Open an issue Explore on GitHub

FAQ

Questions, answered plainly.