feat(linter): implement @typescript-eslint/triple-slash-reference

[kaykdm]

implement @typescript-eslint/triple-slash-reference
Related issue: #503

original

• doc: https://github.com/typescript-eslint/typescript-eslint/blob/main/packages/eslint-plugin/docs/rules/triple-slash-reference.md
• code: https://github.com/typescript-eslint/typescript-eslint/blob/main/packages/eslint-plugin/src/rules/triple-slash-reference.ts

[codspeed-hq]

CodSpeed Performance Report

Merging #1903 will not alter performance

_{Comparing kaykdm:triple-slash-reference (dc71ea4) with main (6e0bd52)}

Summary

✅ 14 untouched benchmarks

[Boshen]

I think we can do this, given TSImportEqualsDeclaration and ImportDeclaration only appear at top level:

for each statement i in program.body
  if the statement is an import
    get all comments from i and i-1
      check for triple slash

For getting comments, use the range API from BTreeMap: comments.range(span.start..span.end)

[kaykdm]

I think we can do this, given TSImportEqualsDeclaration and ImportDeclaration only appear at top level:

for each statement i in program.body
  if the statement is an import
    get all comments from i and i-1
      check for triple slash

For getting comments, use the range API from BTreeMap: comments.range(span.start..span.end)

@Boshen
Thank you!
I will fix that to use program.body instead of ctx.nodes().
However, even we use program.body to find import statement and iterate comment with range, we still need to check other comments because triple-slash directives are not always with import statement.

Also, it made me realize that I don't need to iterate all comments since Triple-slash directives are only valid at the top of their containing file.
I should find the first program node to get start value and use it to iterate comments with range.
It seems like typescript-eslint uses program node's range to get comments.
https://github.com/typescript-eslint/typescript-eslint/blob/main/packages/eslint-plugin/src/rules/triple-slash-reference.ts#L100

[kaykdm]

@Boshen
I have done the following

• use program.body instead of ctx.nodes() to find TSImportEqualsDeclaration and ImportDeclaration
• find the start value of the first statement and use it for comments iteration range
• iterate comments first and then program.body to find import statement if necessary

I am not sure why performance regression is happening 🤔
I think not that much processing is going on in both checker.ts and pdf.mjs