odooscan

A quality and security scanner for Odoo modules, written in Rust. It parses Python with ruff's parser, and XML and CSV natively. It then builds one index of every model, field, method, XML id, view and access rule across the modules it is given, so rules can check what a single-file linter cannot. For example, it can tell you that a view names a field its model lacks, or that an XML id belongs to a module missing from depends.

uv tool install odooscan    # or: pip install odooscan
oscn check path/to/custom-addons --addons-path ~/repos/odoo/odoo,~/repos/odoo/enterprise

The wheels hold a prebuilt binary for Linux (glibc and musl), macOS and Windows; cargo build --release builds target/release/oscn anywhere else.

The documentation covers the options, configuration, CI, series, baselines and suppression, with a page per rule. oscn explain OD203 prints a rule's page in the terminal.

Layout

• crates/odooscan_core/src/project.rs finds modules, parses every file in parallel, and keeps per-file facts.
  • python.rs, xml.rs, csvdata.rs, js.rs and manifest.rs extract those facts and raise the single-file rules.
• index.rs merges the facts. It resolves fields through _inherit, _inherits and delegate=True, related comodels, view inheritance, JS module paths and the scripts bundles load, and the dependency scope of each module.
• checks.rs holds the cross-file rules.
• rules.rs is the registry: code, series range, severity and category. docs/rules/ has a page per rule.
• crates/odooscan is the CLI.

Testing

cargo test runs a snapshot over crates/odooscan_core/tests/fixtures. That fixture is a small core (base, mail, web) plus custom modules written to trigger every rule; review changes with cargo insta review. The false-positive oracle is Odoo itself: a scan of core must report exactly the findings recorded in tests/oracle/, as CONTRIBUTING.md describes.