pre-commit

pre-commit runs hooks on the files of each commit. The iglint repository provides three hooks:

Hook id	Runs	Files
`iglint`	`iglint check`	JSON and Python files
`iglint-fix`	`iglint check --fix` (safe fixes, then reports what is left)	JSON and Python files
`ign-lint`	`iglint ign-lint`, a drop-in for ign-lint's own hook	`view.json` files

iglint's hooks

repos:
  - repo: https://github.com/Mustry-Solutions/iglint
    rev: v0.1.0
    hooks:
      - id: iglint        # or iglint-fix to apply safe fixes

These hooks install iglint from the repository with pip, which builds the binary from source on first use and needs a Rust toolchain. To skip the build, install iglint yourself and declare a local hook that runs it from the PATH:

repos:
  - repo: local
    hooks:
      - id: iglint
        name: iglint
        entry: iglint check
        language: system
        types_or: [json, python]
        require_serial: true

With the PyPI wheel, pre-commit can install the prebuilt binary itself:

repos:
  - repo: local
    hooks:
      - id: iglint
        name: iglint
        entry: iglint check
        language: python
        additional_dependencies: ["iglint==0.1.0"]
        types_or: [json, python]
        require_serial: true

Only the changed files, but the whole workspace

pre-commit passes the changed files to the hook, and iglint reports and fixes only those files. It still analyses the whole workspace around them: every project and gateway configuration below the workspace root (the configuration file's directory, the directory that holds projects/, or the Git work tree). So cross-resource rules (an unknown view, an unused custom property read by another view) and the safety of fixes are the same as for a run over the whole repository. A warm single-file run takes a few tens of milliseconds on a workspace of 3,000 files.

A linked Git work tree (git worktree add, also one inside the repository such as .worktrees/feature) is a checkout of its own: the hook that runs in it analyses the work tree's projects as if it were a separate clone, and runs in the main checkout do not read the work tree's files. Submodules and nested repositories stay part of the repository around them. iglint reads the .git entries on disk, not GIT_DIR or GIT_WORK_TREE, so the hook and a run from a shell find the same workspace root.

Replacing ign-lint's hook

To switch from ign-lint's pre-commit hook, keep its id and change only the repository. The ign-lint hook has the same files pattern (view\.json$), file types, default args (--config=.ignition-lint-precommit.json) and file passing as ign-lint's hook, so your configuration, args overrides, output and exit status stay as they were:

repos:
  - repo: https://github.com/Mustry-Solutions/iglint   # was bw-design-group/ignition-lint
    rev: v0.1.0
    hooks:
      - id: ign-lint
        args: ["--config=rule_config.json", "--ignore-warnings"]   # optional, as before

As with ign-lint 0.6.4, the default --config=.ignition-lint-precommit.json must exist in the repository, or args must name another configuration. iglint ign-lint finds more than ign-lint (see Migrating from ign-lint), so expect some new findings.