automutate

automutate

[Rough Draft] Applies waves of mutations provided by other tools, such as linters.

There are various linters in the world and most are adding or have added ways to --fix rule failures automatically. This is great but hard to do for a couple of reasons:

• Overlapping rule failures - The possibility of mutations appling to overlapping sets of characters requires logic to handle applying one, then re-runing linting, and so on.
• Code bloat verses duplication - Most linters either provide hooks to apply fixes themselves (which can result in code bloat) or have an external project (which duplicates logic for finding rules).

automutate proposes that linters only propose how to fix rules, via a standardized JSON format. A linter-specific utility can request waves of these fixes to be passed into automutate.

Having a standardized source-agnostic project to apply mutations brings a couple of benefits:

• Reduced overhead - Projects no longer need to do this work themselves.
• Standardized base - Ramp-up time to switch between projects using automutate is reduced with common code.

In general, detecting rule failures is a separate concern from fixing them. Linters need to run quickly over a read-only set of files, often during built processes, while fixers typically run slowly and modify files on user request.

Mutations

A single mutation contains a unique type identifier, a range of character position(s) to apply to, and optionally other logic.

The following basic text manipulations are provided out of the box:

• multiple - Container for multiple mutations. This indicates to automutate that these must be applied all at once or not at all, which guarantees consistency with the built-in mutation overlap detection.
• text-delete - Deletes a range of characters.
• text-insert - Inserts a string at a point.
• text-replace - Replaces characters matching a string or regular expression within a range.
• text-swap - Swaps a range of characters with a new string.

For example:

{
    "ugly-file.txt": [
        {
            "range": {
                "begin": 7,
                "end": 14
            },
            "type": "text-delete"
        },
        {
            "insertion": "inconceivable!",
            "range": {
                "begin": 21
            },
            "type": "text-insert"
        }]
}

Linter-specific utilities may define their own mutations. For example, a language's linter may define a node-rename mutation rather than use a multiple mutation containing text-swap mutations.

Mutators

Each mutation is tied to an implementation of the abstract Mutator class by name. The default logic searches for these in user-provided directories under their camelCase name appended with "Mutator". text-insert, for example, would be matched to mutators/testInsertMutator.js.

Each mutator class is specific to a single type of mutation, and each mutator instance is specific to a file. Calls to mutate are given the current file contents as a string, along with the mutation to be applied, and return the file contents after the mutation.

Mutators are also given the original file contents at construction time, which allows for custom mutators to perform setup logic (for example, a language's linter creating an abstract syntax tree for the file).

Project Onboarding

In order to be compatible with automutate, a linter must allow for some rules to report proposed fixes using the standard format. Once that is possible, it's a matter of creating a mutations provider that continuously retrieves these proposed fixes using the linter.