[clangd] Implement simple folding for preprocessor branches #140959
Conversation
Extract directive branches information from DirectiveTree, fold branches that don't end with eof. Fixes clangd/clangd#1661
|
Thank you for submitting a Pull Request (PR) to the LLVM Project! This PR will be automatically labeled and the relevant teams will be notified. If you wish to, you can add reviewers by using the "Reviewers" section on this page. If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers. If you have further questions, they may be answered by the LLVM GitHub User Guide. You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums. |
|
@llvm/pr-subscribers-clangd @llvm/pr-subscribers-clang-tools-extra Author: None (aketchum15) ChangesThis pull request is based on #121449 by sr-tream. I added a unit test for the functionality developed by them. Full diff: https://github.com/llvm/llvm-project/pull/140959.diff 4 Files Affected:
diff --git a/clang-tools-extra/clangd/SemanticSelection.cpp b/clang-tools-extra/clangd/SemanticSelection.cpp
index dd7116e619e6d..57e7ff8b97c65 100644
--- a/clang-tools-extra/clangd/SemanticSelection.cpp
+++ b/clang-tools-extra/clangd/SemanticSelection.cpp
@@ -220,6 +220,24 @@ getFoldingRanges(const std::string &Code, bool LineFoldingOnly) {
auto EndPosition = [&](const Token &T) {
return offsetToPosition(Code, EndOffset(T));
};
+
+ // Preprocessor directives
+ auto PPRanges = pairDirectiveRanges(DirectiveStructure, OrigStream);
+ for (const auto &R : PPRanges) {
+ auto BTok = OrigStream.tokens()[R.Begin];
+ auto ETok = OrigStream.tokens()[R.End];
+ if (ETok.Kind == tok::eof)
+ continue;
+ if (BTok.Line >= ETok.Line)
+ continue;
+
+ Position Start = EndPosition(BTok);
+ Position End = StartPosition(ETok);
+ if (LineFoldingOnly)
+ End.line--;
+ AddFoldingRange(Start, End, FoldingRange::REGION_KIND);
+ }
+
auto Tokens = ParseableStream.tokens();
// Brackets.
for (const auto &Tok : Tokens) {
diff --git a/clang-tools-extra/clangd/support/DirectiveTree.cpp b/clang-tools-extra/clangd/support/DirectiveTree.cpp
index 7ea08add7a107..25ef1dcffd750 100644
--- a/clang-tools-extra/clangd/support/DirectiveTree.cpp
+++ b/clang-tools-extra/clangd/support/DirectiveTree.cpp
@@ -356,5 +356,59 @@ TokenStream DirectiveTree::stripDirectives(const TokenStream &In) const {
return Out;
}
+namespace {
+class RangePairer {
+ std::vector<Token::Range> &Ranges;
+
+public:
+ RangePairer(std::vector<Token::Range> &Ranges) : Ranges(Ranges) {}
+
+ void walk(const DirectiveTree &T) {
+ for (const auto &C : T.Chunks)
+ std::visit(*this, C);
+ }
+
+ void operator()(const DirectiveTree::Code &C) {}
+
+ void operator()(const DirectiveTree::Directive &) {}
+
+ void operator()(const DirectiveTree::Conditional &C) {
+ Token::Range Range;
+ Token::Index Last;
+ auto First = true;
+ for (const auto &B : C.Branches) {
+ if (First) {
+ First = false;
+ } else {
+ Range = {Last, B.first.Tokens.Begin};
+ Ranges.push_back(Range);
+ }
+ Last = B.first.Tokens.Begin;
+ }
+ Range = {Last, C.End.Tokens.Begin};
+ Ranges.push_back(Range);
+
+ for (const auto &B : C.Branches)
+ walk(B.second);
+ }
+};
+} // namespace
+
+std::vector<Token::Range> pairDirectiveRanges(const DirectiveTree &Tree,
+ const TokenStream &Code) {
+ std::vector<Token::Range> Ranges;
+ RangePairer(Ranges).walk(Tree);
+
+ // Transform paired ranges to start with last token in its logical line
+ for (auto &R : Ranges) {
+ const Token *Tok = &Code.tokens()[R.Begin + 1];
+ while (Tok->Kind != tok::eof && !Tok->flag(LexFlags::StartsPPLine))
+ ++Tok;
+ Tok = Tok - 1;
+ R.Begin = Tok->OriginalIndex;
+ }
+ return Ranges;
+}
+
} // namespace clangd
} // namespace clang
diff --git a/clang-tools-extra/clangd/support/DirectiveTree.h b/clang-tools-extra/clangd/support/DirectiveTree.h
index 34f5a888863f2..373af322bca0c 100644
--- a/clang-tools-extra/clangd/support/DirectiveTree.h
+++ b/clang-tools-extra/clangd/support/DirectiveTree.h
@@ -124,6 +124,10 @@ llvm::raw_ostream &operator<<(llvm::raw_ostream &,
/// The choices are stored in Conditional::Taken nodes.
void chooseConditionalBranches(DirectiveTree &, const TokenStream &Code);
+/// Pairs preprocessor conditional directives and computes their token ranges.
+std::vector<Token::Range> pairDirectiveRanges(const DirectiveTree &Tree,
+ const TokenStream &Code);
+
} // namespace clangd
} // namespace clang
diff --git a/clang-tools-extra/clangd/unittests/SemanticSelectionTests.cpp b/clang-tools-extra/clangd/unittests/SemanticSelectionTests.cpp
index 7faef6f95d8f9..b4249590c7f66 100644
--- a/clang-tools-extra/clangd/unittests/SemanticSelectionTests.cpp
+++ b/clang-tools-extra/clangd/unittests/SemanticSelectionTests.cpp
@@ -370,6 +370,26 @@ TEST(FoldingRanges, PseudoParserWithoutLineFoldings) {
//[[ foo
/* bar */]]
)cpp",
+ R"cpp(
+ //Ignore non-conditional directives
+ #define A 1
+
+ void func() {[[
+ int Variable = 100;
+
+ #ifdef FOO[[
+ Variable = 1;
+ ]]#else[[
+ Variable = 2;
+ //handle nested directives
+ #if 1[[
+ Variable = 3;
+ ]]#endif
+ ]]#endif
+
+
+ ]]}
+ )cpp",
};
for (const char *Test : Tests) {
auto T = Annotations(Test);
|
|
Ping |
There was a problem hiding this comment.
Thanks for adding some unit tests, and sorry I didn't get a chance to look at this sooner.
Unfortunately, it looks like the patch needs a bit more work, as I'm seeing it crash on incomplete code, for example:
#ifdef WALDO
#This crashes, at least on an assertions-enabled build, with:
include/llvm/ADT/ArrayRef.h:454: T &llvm::MutableArrayRefclang::clangd::Token::operator const [T = clang::clangd::Token]: Assertion `Index < this->size() && "Invalid index!"' failed.
That will need to be investigated and fixed before this patch can land. (Note that operations like folding ranges need to be robust and not crash on invalid code, especially on invalid code that is a plausible intermediate product of writing valid code, as in the above example, because the editor can potentially ask clangd to compute folding ranges after every keystroke.)
(As part of fixing this assertion failure, please include a unit test for the crashing testcase as well.)
| @@ -220,6 +220,24 @@ getFoldingRanges(const std::string &Code, bool LineFoldingOnly) { | |||
| auto EndPosition = [&](const Token &T) { | |||
| return offsetToPosition(Code, EndOffset(T)); | |||
| }; | |||
|
|
|||
| // Preprocessor directives | |||
There was a problem hiding this comment.
A couple of minor points here:
- Since this code block does not use
PreprocessedorParseableStream, while the code blocks for producing the other kinds of regions do, I think it would make sense for code organization to move this block above the declarations of those variables. - At the top of this function, please revise the comment
// FIXME( usaxena95): Collect PP conditional regions, includes and other code regions ... "to// FIXME( usaxena95): Collect includes and other code regions ...
| @@ -220,6 +220,24 @@ getFoldingRanges(const std::string &Code, bool LineFoldingOnly) { | |||
| auto EndPosition = [&](const Token &T) { | |||
| return offsetToPosition(Code, EndOffset(T)); | |||
| }; | |||
|
|
|||
| // Preprocessor directives | |||
| auto PPRanges = pairDirectiveRanges(DirectiveStructure, OrigStream); | |||
There was a problem hiding this comment.
Could you add a testcase that has a nested conditional in both branches of an outer conditional, and checks that both nested conditionals get a folding range? (It would be easy to get this wrong by passing ParseableStream rather than OrigStream here, which would have the effect of only processing the branch of the outer conditional which we heuristically decided to treat as active.)
This pull request is based on #121449 by sr-tream. I added a unit test for the functionality developed by them.