-
Notifications
You must be signed in to change notification settings - Fork 148
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[BUGFIX] Fix comment parsing to support multiple comments #663
base: main
Are you sure you want to change the base?
Conversation
It would be nice to have this in an 8.6.1 release. |
dce07e5
to
b7de92c
Compare
Because of an eager consumption of whitespace, the rule parsing would swallow a trailing comment, meaning the comment for the next rule would be affected. This patch addresses this by only consuming real whitespace without comments after a rule. Fixes MyIntervals#173 Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
b7de92c
to
edaa39a
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks a lot for this PR, @ziegenberg! ❤️
Looks generally good to me. 👍 I've added some review comments.
Once this is merged, I'm all for backporting this to the 8.x branch and release the 8.x branch once support for PHP 8.4 is merged as well.
@@ -220,18 +220,26 @@ public function parseCharacter($bIsForIdentifier) | |||
} | |||
|
|||
/** | |||
* @return array<int, Comment>|void | |||
* Consumes whitespace and comments and returns the comments. | |||
* If $withoutComment is true, only whitespace is consumed. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Please put code in comments in backticks so it gets displayed as code.
* | ||
* @param bool $withoutComment Do not consume comments, only whitespace. | ||
* | ||
* @return array<int, Comment>|void List of comments. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We can be even more specific (and omit the void return type, which is incorrect for this method and also is not allowed in union types):
* @return array<int, Comment>|void List of comments. | |
* @return list<Comment> list of comments |
* | ||
* @throws UnexpectedEOFException | ||
* @throws UnexpectedTokenException | ||
*/ | ||
public function consumeWhiteSpace(): array | ||
public function consumeWhiteSpace(bool $withoutComment = false): array |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not a big fan of boolean function parameters that switch behavior. Instead, please let's have a separate, well-named method, e.g., comsumeWhitespaceWithoutComment
or something alike.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe we can then also move the new method to Rule
. (I'm not 100% sure about this and would welcome your thoughts on this, @ziegenberg @sabberworm @JakeQZ.)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would go even further and rename the current function comsumeWhitespace
to comsumeWhitespaceWithComment
and create a new function comsumeWhitespace
which precisely only consumes white space. I was baffled to see comsumeWhitespace
also consume comments and return them.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, this would probably be a breaking change. So, I introduced the boolean function parameter to stay within the current wording (which I deem wrong) and still make the change non-breaking.
I would consider having comsumeWhitespace
and a new comsumeWhitespaceWithoutComment
a not-so-good solution. That would cement the semantically wrong situation.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like the idea of renaming this method!
Would you be willing to do this as a (or more specific), two pre-patch PRs?
- one (to be backported) that renames the method, adds the specific type annotation, and re-adds the old method to redirect to the new method, and marks it as
@deprecated will be removed in version 9.0
(including a corresponding line in the changelog - another (not to be backported) that removes the old method (including a corresponding entry in the changelog)
I'd be willing to review these two pre-PRs quickly so we can rebase on them.
We also might want to mark the renamed method as @internal
as it needs to be public
for technical reasons, but is not expected to be called from outside this library. What do you think?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, exactly!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I am not a big fan of comsumeWhitespaceWithoutComment
, having a function name telling you what it does not do. But I'll create the two pre-patch PRs as asked.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Alle consume*
functions are in the class ParserState
, so why move it to Rule
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah, it seems I misread your comment and got confused. 🤯
Okay, so I'm fine with this plan:
I would go even further and rename the current function
comsumeWhitespace
tocomsumeWhitespaceWithComment
and create a new functioncomsumeWhitespace
which precisely only consumes white space.
We'll need to do this in a non-breaking way for the parts that we backport, though.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All
consume*
functions are in the classParserState
, so why move it toRule
?
I've thought about this some more and now think having it in ParserState
is perfectly fine. :-)
@@ -1161,13 +1161,16 @@ public function commentExtracting(): void | |||
*/ | |||
public function flatCommentExtracting(): void |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We're planning to split the tests and testcases into smaller, more focused tests. To make this easier for us, please don't expand the test. Instead, could you please duplicate it into two separate tests, one with a single comment and one with multiple? Thanks!
Battle plan as discussed in our call right now:
|
I don't think we should ever reintroduce a method with the same name as one that previously existed, with different default funcionality. It will catch out anyone doing a major update through several major versions, which happens even (and probably moreso) in sizeable companies as they defer the work required to support the update, deeming it low priority when updating requires some other changes. So I think we may as well come up with a new method name now; maybe |
@ziegenberg Do we still need this PR now that #671/#672 have been merged? |
I propose we move the discussion of the next steps here: #679 |
Because of an eager consumption of whitespace, the rule parsing would swallow a trailing comment, meaning the comment for the next rule would be affected. This patch addresses this by only consuming real whitespace without comments after a rule.
Fixes #173