Doesn't work with custom parser

[so1ve]

I don't think a reproduction is needed for this issue :(

For example, in eslint config:

{
  extends: ["plugin:markdown/recommended"],
  overrides: [
    { files: ["*.md"], parser: "eslint-parser-plain", rules: {} }
  ]
}

The parser does not take effect. (Actually it doesn't run at all!)

Is this by design?

[sam3k]

Hi @so1ve, could you provide a reproducible demo? Taking a quick look at the plugin docs shows that the API has changed between v1 and v2. Version 2 is configured like this it seems:

module.exports = {
    extends: "plugin:markdown/recommended",
    plugins: ["markdown"],
    overrides: [
        {
            files: ["**/*.md"],
            processor: "markdown/markdown"
        },
    ]
};

[so1ve]

Sure, please checkout https://github.com/so1ve/eslint-plugin-markdown-repro

1. Install the dependencies and run npm run lint
2. Console logs the content of test.md
3. Comment out the line in .eslintrc.js
4. Run npm run lint again
5. Console logs nothing

[sam3k]

Thank you for the demo @so1ve. Just ran it on my machine and I see that the linter throws error if you use markdown/markdown as processor:

module.exports = {
  parserOptions: {
    ecmaVersion: "latest",
    sourceType: "module"
  },
  plugins: ["markdown"],
  overrides: [
    {
      files: ["**/*.md"],
      processor: "markdown/markdown",
    },
  ],
  rules: {
    semi: "error"
  }
};

It also works if you remove the processor:

module.exports = {
  parserOptions: {
    ecmaVersion: "latest",
    sourceType: "module"
  },
  plugins: ["markdown"],
  overrides: [
    {
      files: ["**/*.md"],
    },
  ],
  rules: {
    semi: "error"
  }
};

Now, if I understood you correctly, it seems like the markdown errors stop working once you add a custom parser.js. I am able to get it to still throw error if I have the following config:

module.exports = {
  parserOptions: {
    ecmaVersion: "latest",
    sourceType: "module"
  },
  plugins: ["markdown"],
  overrides: [
    {
      files: ["**/*.md"],
      processor: "markdown/markdown", // Comment out this line
      parser: require.resolve("./parser.js")
    },
  ],
  rules: {
    semi: "error"
  }
};

and then use the following format in parser.js which is based on the example given in the docs:

var espree = require("espree");

function parseForESLint(code, options) {
  console.log('parseForESLint');
  return {
    ast: espree.parse(code, options),
    services: {
      isPlain: true,
      foo: function() {
        console.log("foo");
      }
    },
    scopeManager: null,
    visitorKeys: null
  };
};

module.exports = { parseForESLint };

This calls the custom parser and throws the error still.

[nzakas]

@sam3k when you say "throws errors", are you saying that ESLint is reporting linting errors or that it is throwing an error (with a stack trace)?

[sam3k]

@nzakas I meant ESLint reports lint error:

[sam3k]

@so1ve were you able to try out the configuration I've suggested?

[so1ve]

@sam3k Ah, sorry. Before I was busy with exams and didn't have time to see github. I'll try your solution later

[so1ve]

Haven't tried it yet, but I think it won't work as my expectations. In your code, you call espree.parse to parse the code, which means the input is JavaScript code, not markdown. However, I need a custom parser to return a Program node that contains the markdown file's content to let prettier format it

[sam3k]

@so1ve it seems like it returns JS AST. I've seen the ast object being defined as a Program like the following but not sure if this can be of any help to you:

function parseForESLint(code, options) {
  return {
    ast: {
      // ast is JS ast. We don't have JS, so this AST is for empty JS file
      type: 'Program',
      start: 0,
      end: 0,
      loc: { start: { line: 1, column: 0 }, end: { line: 1, column: 0 } },
      range: [0, 0],
      body: [],
      tokens: [],
      comments: [],
      // Used only by eslint-plugin-markdown-language
      mdCode: code,
    },
    services: {
      isPlain: true,
      foo: function() {
        console.log("foo");
      }
    },
    scopeManager: null,
    visitorKeys: null
  };
};

module.exports = { parseForESLint };

Not sure if this is helpful; but could you give it a shot?

[so1ve]

Maybe I am doing something wrong?

function parseForESLint(code, options) {
  console.log(code) // This logs nothing :(
  return {
    ast: {
      // ast is JS ast. We don't have JS, so this AST is for empty JS file
      type: 'Program',
      start: 0,
      end: 0,
      loc: { start: { line: 1, column: 0 }, end: { line: 1, column: 0 } },
      range: [0, 0],
      body: [],
      tokens: [],
      comments: [],
      // Used only by eslint-plugin-markdown-language
      mdCode: code,
    },
    services: {
      isPlain: true,
      foo: function() {
        console.log("foo");
      }
    },
    scopeManager: null,
    visitorKeys: null
  };
};

module.exports = { parseForESLint };

[sam3k]

@so1ve everything looks good. Can you try this fork I've made?

[so1ve]

It doesn't work either... It logs nothing :(

[nzakas]

It looks like this isn't about a problem with the plugin, but rather, a question about how to configure ESLint, so I'm moving this to a discussion.