CatDat is based on a SQLite database. During runtime of the application, it is read-only.
The local copy of the database is located at /database/catdat.db. It contains three main tables:
structurespropertiesimplications
The structures table stores data that is common to all types of categorical structures. These types are currently supported: categories, functors, morphisms, and symmetric monoidal categories. They are stored in the following table:
structure_types
Properties (whether satisfied or not) are associated with categorical structures via the following table:
property_assignments
To mark properties as assumptions or conclusions of an implication, there are two tables:
assumptionsconclusions
These tables are abstracted through the implications_view view.
Functor implications may also depend on properties of the domain or codomain category. Likewise, morphism implications may depend on properties of the ambient category. Such dependencies are stored in the following table:
associated_assumptions
Additional tables are available. For a complete overview, see the diagram below.
The schema defines the structure of the database: tables, views, indexes, and triggers. It is specified in several SQL files located in the subfolder /database/schema. The command
pnpm db:setup
deletes the old database file (if it exists) and creates a new one using this schema. This is required when the schema changes.
Database entries are defined in YAML files located in the subfolder /database/data. The command
pnpm db:seed
rebuilds the database by clearing all existing data and then parsing and inserting the entries defined in these YAML files.
For text-only changes to existing structures, properties, and implications, use
pnpm db:text
This compares the YAML text with the existing database and updates changed texts (such as descriptions and proofs) without changing relations or running deductions. Pass --watch to keep the command running and update whenever a YAML file changes.
From the defined satisfied properties of a given categorical structure, new properties can be automatically deduced using the implications. (For example, when a category has equalizers and products, we can infer that it is complete.) The same applies to unsatisfied properties.
Additionally, suitable implications may be dualized, and a categorical structure inherits all dualized properties of its dual structure, if available. Note that the YAML files mentioned above do not contain any derived data.
The command
pnpm db:deduce
deduces implications, satisfied properties, and unsatisfied properties.
The command
pnpm db:test
executes some tests that verify that the data behaves as expected.
Use the command
pnpm db:update
to run all the commands in sequence: pnpm db:sort (which sorts the JSON files used by the tests), pnpm db:seed, pnpm db:deduce, and pnpm db:test. This also creates a copy of the local database in the /static folder.
Pass --watch to rerun the update whenever a file in the data folder changes:
pnpm db:update --watch
While pnpm dev is running, the browser reloads automatically after pnpm db:update or pnpm db:text has changed the database.
Use the command
pnpm db:structure <structure-id> [group]
to list the satisfied, unsatisfied, unknown, and undecidable properties of a structure in the terminal, grouped as on its detail page. Pass one of these groups as a second argument to list only that group.
This is the database schema as of 25.10.2026; changes may occur.
