conventional-commits / parser Goto Github PK

It would make it much easier for upstream tooling to use this parser, if we provide a helper that converted the tree into conventional-changelog-parser format.

Most of the existing unist syntax trees have all nodes documented, like nlcst. Would be good to add this once we have the first version of the parser.

TypeScript is widely used within the JS/node ecosystem, would be good to add definitions when we have our first version. I can help with that 😄

To provide actionable debug information and errors, we should implement Position tracking, as defined in uninst.

Refs: #1

Compare

<footer>          ::= <token>, <separator>, <whitespace>*, <value>
/* "!" should be added to the AST as a <breaking-change> node with the value "!" */
<token>           ::= <breaking-change>
                   |  <type>, "(" <scope> ")", ["!"]
                   |  <type>, ["!"]
<separator>       ::= ":" | " #"
<value>           ::= <text>, <continuation>+
                   |  <text>
<continuation>    ::= <newline>, <whitespace>+, <text>

with https://git-scm.com/docs/git-interpret-trailers

• "When reading trailers, there can be whitespaces after the token, the separator and the value"
• "There can also be whitespaces inside the token and the value"
  • This one seems like it makes parsing trailers a lot harder and doesn't seem common

@byCedric says:

Looks awesome! I'll test run it this weekend too. One thing that could help during development/debugging is adding debug messages. With that, the code path should be visible. (e.g. this)

Refs: #1

The amount of ritual to setup a Node, update its value, and record its source position is becoming burdensome.

Let's introduce a Node builder that:

1. provides a helper for rolling back go a prior source position, and or returning the Node.
2. provides shortcuts for tracking source positions.

CC: @byCedric
Refs #5

From: #15 (comment)

I feel like there will be tooling value in having this as as well { line: <line>, column: <column> } on the error. Maybe before line 80 we could add error.position = this.pos?

Another helpful pattern I like to follow (used in node, which is good precedence I think) is to have distinct .code properties for these errors. If we were tooling around errors, checking err.code === 'EOF_ERROR is much better than err.message.startsWith('unexpected token EOF at').

👋 Hi! I just was shown this repository - it looks great, thanks for making it. Really nice when tools publish reference implementations ❤️.

I don't see any meaningful maintenance activity since 2021. There are some issues or PRs that have been awaiting triage for a year or more, such as #42. So I'm hesitant to build on top of the published package or contribute if this repo is in a "frozen" state.

Is there >=1 maintainer who's still got time to work on this repo?

Seen on both GitHub and npm:

https://github.com/conventional-commits/parser/workflows/ci/badge.svg is not found.

I'm guessing this is because there hasn't been a meaningful action-causing commit since 2021?

So that there's no lost information in the parse tree, we should create <whitespace> and <newline> tokens for spacing characters that fall outside of <text> nodes.

Currently, the grammar only prevents the sequence !:. So someone could have a type like He!!o.

::= <any UTF8-octets except newline or parens or ":" or "!:" or whitespace>+

Personally, I would find this confusing to see but it also would greatly simplify this grammar rule to only be looking at characters rather than sequence of characters.

As per conventional commits specification v1.0.0 §8: “Each footer MUST consist of a word token, followed by either a :<space> or <space># separator, followed by a string value.”

However, the reference grammar currently specifies:

<separator>       ::= ":" | " #"
/* missing <space>     ^^ */

This is similar to but distinct from #41. The inconsintency was pointed out in crate-ci/git-conventional#54.

I'm assuming this is not intended but I'm unsure if its worth specifying this out of scope.

To simplify parsing logic, we create multiple text nodes when parsing body, let's collapse these together into one Node:

Refs #11

Hello,
Not sure I am asking in the right place, I have juste created a rust parser crate.

I was wondering if there was a generic test suite to ensure my parser is compliant with the spec ?