manon — MongoDB Anonymizer

manon is a command-line tool that anonymizes MongoDB collections.
It samples a collection to infer its schema and automatically detects sensitive fields, then produces a deterministic, referentially-consistent anonymized copy of the data that can be used safely in non-production environments.

---

Motivation

Development, testing, and analytics teams often need realistic data that mirrors production volumes and shapes. Sharing raw production dumps is risky: they contain names, addresses, contact details, financial identifiers, and other personal information protected by GDPR and similar regulations.

Static masking only. manon produces a new, permanently anonymized copy of your data. It does not act as a proxy or rewrite queries on the fly (unlike dynamic-masking tools such as pgAnonymizer). The source collection is never modified; the anonymized copy exists as a separate dataset and must be refreshed whenever the source changes.

manon solves this by:

• Automatically detecting sensitive fields through a curated dictionary of field-name patterns (identifier.csv) mapped to masking categories (identifier_category.csv).
• Applying deterministic masking — the same input always produces the same output, so foreign-key relationships and join patterns are preserved across collections.
• Keeping the schema intact — non-sensitive fields (numbers, dates, booleans, nested objects, arrays) pass through unchanged unless explicitly annotated.
• Working entirely from the command line with no external services or configuration servers required.

Inspiration

manon stands on the shoulders of two existing open-source tools:

• pgAnonymizer — the idea of driving anonymization from a plain-text dictionary of field-name patterns mapped to masking categories comes directly from pgAnonymizer. The bundled identifier.csv and identifier_category.csv files are adapted from its approach.
• mongo2pg — the schema statistics format (field frequencies, type distributions, sampled values, distinct counts) is inspired by the schema-analysis output produced by mongo2pg.

---

Installation

manon is a single Rust binary with no runtime dependencies.

---

Download a pre-built binary

Go to the Releases page and download the archive that matches your platform:

Platform	File to download
Linux x86_64	manon-linux-x86_64
Linux arm64	manon-linux-aarch64
macOS Intel	manon-macos-x86_64
macOS Apple Silicon	manon-macos-aarch64
Windows x86_64	manon-windows-x86_64.exe
Windows arm64	manon-windows-aarch64.exe

Linux / macOS

# Replace <version> and <platform> with your values, e.g. v0.1.0 and linux-x86_64
version="v0.1.0"
platform="linux-x86_64"
curl -fL "https://github.com/pmpetit/mongodbanonymizer/releases/download/${version}/manon-${platform}" \
  -o manon
chmod +x manon
sudo mv manon /usr/local/bin/

Features

Feature	Description
Schema inference	Samples a MongoDB collection and produces a YAML schema that describes field types, probabilities, and distinct-value counts
Automatic field detection	Maps field names (exact, suffix, and prefix) to masking categories using bundled CSV dictionaries
Deterministic masking	Same input → same output; referential integrity is preserved across collections
Nested document support	Masking rules are applied recursively into embedded objects and arrays of objects
Array-of-scalars support	Coordinate arrays and similar numeric arrays are noisy-masked element by element
Compound value samples	The values lists in the YAML schema (including compound documents) are fully anonymized
Manual rule editing	The generated YAML schema can be hand-edited to add, remove, or change masking rules before applying
Live collection apply	Streams documents from a source cluster, masks them in flight, and inserts into a target cluster
Batch insert	Documents are written in configurable batches (default 500) for efficient bulk loading
Project management	init creates a project directory structure and config file for repeatable runs

Masking methods

Method	Behaviour
PRESERVE_TOKEN	Replaces each character class-by-class (uppercase↔uppercase, digit↔digit, separator kept); deterministic
REDACT_ALPHANUMERIC	Replaces every alphanumeric character with X / 0; separators kept
MASK_CONTACT_URI	Anonymizes e-mail local parts, URL paths/queries, and phone digits
MASK_NETWORK_ID	Anonymizes IPv4, IPv6, and MAC addresses segment by segment
GENERALIZE_LOCATION	Truncates postal codes to the first 2–3 characters
NOISY_DATE	Shifts dates by a deterministic ±30-day noise
NOISY_POSITION	Adds deterministic ±0.009° (~1 km) noise to geographic coordinates
STATIC_MAPPING	Maps the value to one of five tokens (A–E) deterministically
STATIC_BLOB_REPLACEMENT	Replaces the value with the literal string [REDACTED]

---

CLI Usage

manon init — create a project

manon init --project-cluster <base-dir> \
           --project-dbname  <project-name> \
           [--source-uri     mongodb://localhost:27017] \
           [--namespace      mydb.mycollection]

Creates <base-dir>/<project-name>/ with a config/ sub-directory containing a .conf file that stores the URI, namespace, and sampling parameters for subsequent commands.

---

manon infer — sample a collection and infer its schema

manon infer --source-uri mongodb://localhost:27017 \
            --namespace  mydb.listings \
            --output-dir ./schema \
            [--number 1000 | --percent 10]

Or using a project config file:

manon infer -c ./myproject/config/myproject.conf

Writes ./schema/listings/listings.yaml containing the inferred schema with masking annotations for all recognised sensitive fields.

Tip — whole database at once. Pass only the database name (no .) to --namespace and manon infer will enumerate and infer every collection in that database:

manon infer -s mongodb://prod:27017 -n mydb -o ./schema

Key options

Flag	Description
-s, --source-uri	MongoDB connection URI
-n, --namespace	<db>.<collection> for one collection, or just <db> to infer all collections in the database
--number	Number of documents to sample (default 1000)
--percent	Percentage of the collection to sample
-o, --output-dir	Directory to write YAML/JSON output
-c, --config	Path to a .conf file created by manon init

---

manon mask — refresh anonymized values in an existing schema file

manon mask <schema.yaml> [--output <out.yaml>]

Re-reads the CSV dictionaries, re-annotates any newly-recognised fields (including fields that were missed in a previous run), and replaces all values lists with freshly-anonymized data. Useful after editing a schema file manually or after updating the identifier dictionaries.

Key options

Flag	Description
(positional)	Path to the YAML schema file to update
-o, --output	Write result to a new file instead of updating in-place

---

manon apply — anonymize a live collection

manon apply --source-uri  mongodb://prod:27017   \
            --namespace   mydb.listings          \
            --target-uri  mongodb://dev:27017    \
            --masking-rules ./schema/listings/listings.yaml \
            [--target-namespace mydb_anon.listings]

Streams every document from the source collection, applies the masking rules defined in the YAML schema file, and inserts the anonymized documents into the target collection in batches of 500.

Tip — whole database at once. Pass only the database name (no .) to --namespace and point --masking-rules to the directory written by manon infer. Every collection that has a matching YAML file is processed; collections with no YAML are skipped with a warning:

manon apply -s mongodb://prod:27017 -n mydb \
            -m ./schema \
            -t mongodb://dev:27017 \
            --target-namespace mydb_anon

Add --percent 10 to copy only 10 % of each collection (useful for ephemeral environments).

Key options

Flag	Description
-s, --source-uri	Source MongoDB URI
-n, --namespace	Source namespace — <db>.<collection> for one collection, or just <db> for all collections in the database
-m, --masking-rules	Path to a YAML schema file (single collection) or a directory of per-collection YAML files (DB-level)
-t, --target-uri	Target MongoDB URI
--target-namespace	Target namespace or DB name (defaults to the same as --namespace)
-p, --percent	Copy only this percentage of each collection (e.g. 10 for 10%)
-c, --config	Path to a .conf file created by manon init

---

Typical workflow

# 1. Infer schema and masking rules from production
manon infer -s mongodb://prod:27017 -n mydb.users -o ./schema

# 2. Review / adjust ./schema/users/users.yaml as needed

# 3. Refresh anonymized value samples in the schema file
manon mask ./schema/users/users.yaml

# 4. Apply masking rules to populate a dev cluster
manon apply -s mongodb://prod:27017 \
            -n mydb.users \
            -t mongodb://dev:27017 \
            -m ./schema/users/users.yaml