disambiguate fold vs parenthesized assignment #239
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jdrouhard
reviewed
Jan 26, 2024
|
Thanks! |
This was referenced Jan 26, 2024
|
Hey this looks great, this was actually my original idea to fix this but I felt there could've been a better way, since this feels more like a bodge that just happens to work (in my head), but if it's the only solution then it's totally okay. Though I do have a few nitpicks to point out (and wish we got resolved before merging) - namely not generating with master and the naming/exposing of the lhs rule, I'll just push a fix for that though so no worries. |
brandonspark
added a commit
to semgrep/semgrep
that referenced
this pull request
Feb 3, 2024
## What: This PR updates both the `semgrep-c` and `semgrep-cpp` modules with the latest changes from `tree-sitter-c` and `tree-sitter-cpp`. ## Why: I recently fixed an [upstream tree-sitter issue](tree-sitter/tree-sitter-cpp#239) in the C++ grammar. Unfortunately, this means we need to take all of the new `tree-sitter-cpp` changes to pull it in. ## How: Lots of grammar hacking. In several places, I augmented the C AST to be more in line with the CPP AST. This makes it noticeably less simple, but is required, as the `tree-sitter-c` grammar becomes more complex. This leads to a gross amount of code duplication between the C and C++ translation code, but since the typed CSTs are different at the moment, there isn't a good way to prevent that. This PR also makes use of the groundwork I laid down in #9681, which introduces a `preproc_if_poly` (and friends) types, which are polymorphic types that encode the structure of the preprocessor statements that occur in C and C++. Here, we see that we save massive amounts of logic duplication with minimal boilerplate. ## Test plan: `make test` Added a parsing test for the parenthesized assignment thing that all of this was originally for, too.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, there is an ambiguity in parsing fold expressions and parenthesized assignment expressions. This is described in issues #201 and #212.
This arises in the following trace:
The crux of the issue is that we do not know whether to shift the
=forward, or to reduce the<id>to anexpression_not_binary. If we do the former, we can no longer parse a fold expression, which expects<exp> <fold_operator> '...'. If we do the latter, we can no longer parse a parenthesized assignment expression, as an assignment expression does not permit arbitrary expressions to appear on the LHS.The current behavior is that
tree-sitter-cppwill reduce the<id>, and thus fail to parse any parenthesized assignment expression. Unfortunately, these are reasonably common, so this is actually disastrous.This PR makes it so that parenthesized expressions specifically allow assignment expressions which feature expressions on the LHS, so that it can still parse even if we reduce the identifier. This causes the grammar to be a little more permissive than it needs to be, but it does not disallow valid programs, as another proposed solution to this problem does.
I believe it is better for us to be more permissive, in this case, so that we no longer disallow valid programs, with the understanding that we can restore stringency at a later date.