The purpose of this repository is to help you understand how ESLint works, how can you create rules, parser and configs based on TypeScript language.
Before anything to do, run npm install
.
More about ESLint
ESLint is a tool for identifying and reporting on patterns found in ECMAScript/JavaScript code, with the goal of making code more consistent and avoiding bugs.
- 🧰 Parts of ESLint
- 💻 CLI
- ⚙️ CLI Engine
- 🎉 Linter
- 🔬 How dose Linter work in short
- 🧾 Rules
- 📌 Configs (In progress)
- 📌 Parsers (In progress)
- 📌 Write custom parser (In progress)
- 📌 TS-ESLint (In progress)
- 📌 Write and test TS-ESLint custom rule (In progress)
- 📌 Create (TS-)ESLint plugin (In progress)
flowchart TD
subgraph bin
A>eslint.js]
end
A -.-> C & D
subgraph lib
B>api.js] ---> E & F & G & I
C(cli.js) ---> E
G(linter) --> I & J
F(rule-tester) --> G
E(cli-engine) --> G
I(source-code)
J(rules)
end
D("Proxy call \nnpm init @eslint/config")
- CLI - This is the heart of the ESLint. It takes an array of arguments and then uses eslint to execute the commands.
- CLI Engine - This module is
CLIEngine
class that finds source code files and configuration files then does code verifying with theLinter
class. This includes the loading logic of configuration files, parsers, plugins, and formatters. - Linter - This module is the core
Linter
class that does code verifying based on configuration options. This file does not have I/O operation and does not interact with the console at all. For other Node.js programs that have JavaScript text to verify, they would be able to use this interface directly. - Rules - This contains built-in rules.
- Rule tester - This module is
RuleTester
class that is a wrapper around Mocha so that rules can be unit tested. - Source Code - This module is
SourceCode
class that is used to represent the parsed source code.
The cli
object is the API for the command line interface. Literally, the bin/eslint.js
file simply passes arguments
to the cli
object and then sets process.exitCode
to the returned exit code.
The main method is cli.execute()
, which accepts an array of strings that represent the command line options (as
if process.argv
were passed without the first two arguments). If you want to run ESLint from inside of another program
and have it act like the CLI, then cli
is the object to use.
This object’s responsibilities include:
- Interpreting command line arguments
- Reading from the file system
- Outputting to the console
- Outputting to the filesystem
- Use a formatter
- Returning the correct exit code
The CLIEngine
type represents the core functionality of the CLI except that it reads nothing from the command line and
doesn’t output anything by default. Instead, it accepts many (but not all) of the arguments that are passed into the
CLI. It reads both configuration and source files as well as managing the environment that is passed into the Linter
object.
The main method of the CLIEngine
is executeOnFiles()
, which accepts an array of file and directory names to run the
linter on.
This object’s responsibilities include:
- Managing the execution environment for
Linter
- Reading from the file system
- Reading configuration information from config files (including
.eslintrc
andpackage.json
)
The main method of the Linter
object is verify()
and accepts two arguments: the source text to verify and a
configuration object (the baked configuration of the given configuration file plus command line options). The method
first parses the given text with espree
(or whatever the configured parser is) and retrieves the AST. The AST is
produced with both line/column and range locations which are useful for reporting location of issues and retrieving the
source text related to an AST node, respectively.
Once the AST is available, estraverse
is used to traverse the AST from top to bottom. At each node, the Linter
object emits an event that has the same name as the node type (i.e., Identifier
, WithStatement
, etc.). On the way
back up the subtree, an event is emitted with the AST type name and suffixed with :exit
, such as Identifier:exit
-
this allows rules to take action both on the way down and on the way up in the traversal. Each event is emitted with the
appropriate AST node available.
This object’s responsibilities include:
- Inspecting JavaScript code strings
- Creating an AST for the code
- Executing rules on the AST
- Reporting back the results of the execution
flowchart LR
A([Source code]) --> ESPree
A -.-> B
B(Preprocessor) -.-> ESPree
subgraph ESPree
direction LR
C(Tokenization) --> D
D(AST)
end
ESPree --> E
E(Linting) --> G
E -.-> F
F(Postprocessor) -.-> G
G([Results])
Processors tell ESLint how to process files other than standard JavaScript.
Tokenization (also known as lexical analysis) is a step during source code parsing.
Side note: Before tokenization, line reconstruction is needed to convert the input character sequence to a canonical form ready for the tokenizer. Languages which (for example) allow arbitrary spaces within identifiers require this phase.
Tokens represent those units from which the code is built, like Keyword (const
), Identifier (foo
, bar
), Punctuator (=
, {
, }
).
These tokens help us to get some information about source or make changes and help parser to produce the AST object.
I have created a simple tokenizer which you can find src/ast-demo/convert-source-code-to-tokens.util.ts in this repository.
Run npm run ast-demo:start
to see what happens.
Abstract Syntax Tree is a tree representation of the abstract syntactics structure of text written in a formal
language (in our case the JavaScript code). Here is an online AST playground.
If you want to imagine how dose AST look like, here is a diagram about it.
I have created simple tokens-to-AST and AST-to-source parsers which you can find src/ast-demo/convert-tokens-to-ast.util.ts and src/ast-demo/convert-ast-to-source-code.util.ts
and also I have an exercise src/ast-demo/index.ts for you in this repository.
Run npm run ast-demo:start
to see what happens.
flowchart TD
A([Program])
A --> B
A --> C
subgraph Body
B(Variable declaration)
C(Function declaration) --> D
D(Return statement)
end
ESPree is a parser which creates
- Tokens (one dimensional array containing tokens)
- AST object based on ESTree specification
from source code.
In short the steps are the following
flowchart LR
A([Source code]) -- Tokenization --> B
B(Tokens) --> C
C(AST Object)
You can find src/espree-demo/index.ts in this repository.
Run npm run espree-demo:start
to see what happens.
Side note: ESPree uses AcornJS to tokenize and parse source code. Actually the AcornJS is just wrapped (extended with some ESPree feature).
Rules are the core building block of ESLint. A rule validates if your code meets a certain expectation, and what to do if it does not meet that expectation. Rules can also contain additional configuration options specific to that rule.
ESLint provides us with many built-in rules that you can use, on the other hand you have opportunity to write custom rules to which fit your idea.
I have produced a demo example which you can find in the src/js-rule-demo/prefer-strict-equal.rule.ts and src/js-rule-demo/index.ts files.
Run npm run js-rule-demo:start
to see what happens.
I have created a demo example which you can find in the src/js-rule-demo/prefer-strict-equal.rule.spec.ts file.
Run npm run js-rule-demo:test-watch
to see what happens.