https://github.com/savetheclocktower/tree-sitter-hyperlink

Tree-sitter grammar for detecting URLs in arbitrary text
https://github.com/savetheclocktower/tree-sitter-hyperlink

Last synced: about 1 month ago
JSON representation

Tree-sitter grammar for detecting URLs in arbitrary text

• Host: GitHub
• URL: https://github.com/savetheclocktower/tree-sitter-hyperlink
• Owner: savetheclocktower
• License: mit
• Created: 2023-03-20T23:25:29.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2024-10-21T00:38:39.000Z (2 months ago)
• Last Synced: 2024-10-21T04:00:29.171Z (2 months ago)
• Language: JavaScript
• Size: 49.8 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# tree-sitter-hyperlink

[Tree-sitter](https://github.com/tree-sitter/tree-sitter) grammar for detecting URLs in prose.

Eventually designed to do all the things [TextMate’s hyperlink helper bundle](https://github.com/textmate/hyperlink-helper.tmbundle/blob/master/Syntaxes/Hyperlink.tmLanguage) can do. That bundle got converted to Atom as [language-hyperlink](https://github.com/pulsar-edit/language-hyperlink/), but a tree-sitter version needs to exist for injecting into tree-sitter grammars in [Pulsar](https://pulsar-edit.dev/).

## Syntax

Support is currently limited to URLs that

* begin with `http` or `https`,

* have sections of ordinary text with dots/slashes in between, and

* do not end with any of `,."']`, all of which are far more likely to be meant

  as a prose delimiter rather than part of the URL.

URLs that end in `)` will have that `)` included in the URL _if_ it was preceded by an earlier `(` in the URL; otherwise it’ll be treated as a delimiter like the characters listed above.

Validity of the URL, or of any TLDs, is _far_ beyond the ambitions of this parser.

## Examples

URLs that will be correctly identified and highlighted:

```

http://example.com/foo?bar=baz

You might find my web site at https://example.com. The period at the end of the last sentence is not part of the URL.

One example would be [this one](http://example.com/foo?bar=baz), like in Markdown.

Or this one [http://example.com]

This fragment will be ignored http://

As will this one https://bleh

CSS URL without quotes:

@import url(https://www.example.com/style.css);

CSS URL with quotes:

div {

  background-image: url("https://www.example.com/style.gif");

}

also https://example.net.

Good news, Elvis: https://en.wikipedia.org/wiki/Alison_(song) (because the closing parenthesis is mistakenly assumed not to be part of the URL)

http://example.com (it'll recognize both instances)

[A link to the Elvis Costello song in question](https://en.wikipedia.org/wiki/Alison_(song)) will correctly interpret the first ) as being part of the URL, but _not_ the second one.

```

There are surely some URLs out there in the wild that run afoul of these rules, so open an issue if you like.

## Tests

What I need to be able to test this properly is to inspect _the exact boundaries_ of a match, and [neither](https://tree-sitter.github.io/tree-sitter/creating-parsers#command-test) [kind](https://tree-sitter.github.io/tree-sitter/syntax-highlighting#unit-testing) of tree-sitter test does exactly what I need it to.

Until I get around to something more rigorous: if you’re contributing a change and want to guard against regressions, compare the output of `tree-sitter parse examples/example.txt` with the contents of `example_tree.txt`.

## TODO

* Fix the issue with parens in URLs — ideally by allowing one `)` for each `(` encountered in a URL, and the same with `[]`