[MLIR] Generate cpp comments for TableGen summary and description #139606
Conversation
…from TableGen summary and description
|
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-tablegen @llvm/pr-subscribers-mlir-core Author: Peng Chen (pchen7e2) ChangesThis commit takes the The main motivation is to improve the developer experience. When people work on compilers from an IDE, they will be able to hover over the symbols (e.g. <img width="971" alt="Screenshot 2025-05-12 at 12 00 24 PM" src="https://github.com/user-attachments/assets/6404e4cd-385c-4da1-b155-8f93326459e9" /> Full diff: https://github.com/llvm/llvm-project/pull/139606.diff 7 Files Affected:
diff --git a/mlir/test/mlir-tblgen/cpp-class-comments.td b/mlir/test/mlir-tblgen/cpp-class-comments.td
new file mode 100644
index 0000000000000..66d7e692dee25
--- /dev/null
+++ b/mlir/test/mlir-tblgen/cpp-class-comments.td
@@ -0,0 +1,74 @@
+// RUN: mlir-tblgen -gen-dialect-decls -I %S/../../include %s | FileCheck %s --check-prefix=DIALECT
+// RUN: mlir-tblgen -gen-op-decls -I %S/../../include %s | FileCheck %s --check-prefix=OP
+// RUN: mlir-tblgen -gen-typedef-decls -I %S/../../include %s | FileCheck %s --check-prefix=TYPE
+
+include "mlir/IR/OpBase.td"
+
+// check dialect with summary and description
+def A_Dialect : Dialect {
+ let name = "a";
+ let cppNamespace = "";
+
+ let summary = "This is a summary";
+ let description = [{
+
+ This is a description, needs trimming
+
+ }];
+// DIALECT: /// This is a summary
+// DIALECT-NEXT: /// This is a description, needs trimming
+// DIALECT-NEXT: class ADialect : public ::mlir::Dialect {
+}
+
+def A_SomeOp1 : Op<A_Dialect, "some_op1", []>{
+ let summary = "Some Op1 summary line1 \nsummary line2";
+
+ let description = [{
+ Some Op1 description
+ }];
+
+ let cppNamespace = "OP1";
+// OP: namespace OP1
+// OP-NEXT: /// Some Op1 summary line1
+// OP-NEXT: /// summary line2
+// OP-NEXT: /// Some Op1 description
+// OP-NEXT: class SomeOp1;
+}
+
+// test weird characters in description
+def A_SomeOp2 : Op<A_Dialect, "some_op2", []>{
+ let summary = "";
+
+ let description = [{
+ $ptr (`,` $mask^)? (`,` $other^)?
+ oilist(
+ `a` `=` $1 | `b` `=` $2
+ )
+ }];
+// OP: /// $ptr (`,` $mask^)? (`,` $other^)?
+// OP-NEXT: /// oilist(
+// OP-NEXT: /// `a` `=` $1 | `b` `=` $2
+// OP-NEXT: /// )
+// OP-NEXT: class SomeOp2;
+}
+
+class AType<string name> : TypeDef<A_Dialect, name> { }
+
+
+def A_TensorType : AType<"Tensor"> {
+ let typeName = "a.simple_a_tensor";
+
+ let summary = "Tensor Type A summary";
+
+ let description = [{
+ Tensor Type A description
+ }];
+
+ let extraClassDeclaration = [{
+ void getSignlessBlockType() const {
+ }
+ }];
+// TYPE: /// Tensor Type A summary
+// TYPE-NEXT: /// Tensor Type A description
+// TYPE-NEXT: class TensorType;
+}
diff --git a/mlir/tools/mlir-tblgen/AttrOrTypeDefGen.cpp b/mlir/tools/mlir-tblgen/AttrOrTypeDefGen.cpp
index 05686c0539754..2a6071602fa49 100644
--- a/mlir/tools/mlir-tblgen/AttrOrTypeDefGen.cpp
+++ b/mlir/tools/mlir-tblgen/AttrOrTypeDefGen.cpp
@@ -7,6 +7,7 @@
//===----------------------------------------------------------------------===//
#include "AttrOrTypeFormatGen.h"
+#include "CppGenUtilities.h"
#include "mlir/TableGen/AttrOrTypeDef.h"
#include "mlir/TableGen/Class.h"
#include "mlir/TableGen/CodeGenHelpers.h"
@@ -813,8 +814,14 @@ bool DefGenerator::emitDecls(StringRef selectedDialect) {
NamespaceEmitter nsEmitter(os, defs.front().getDialect());
// Declare all the def classes first (in case they reference each other).
- for (const AttrOrTypeDef &def : defs)
+ for (const AttrOrTypeDef &def : defs) {
+ std::string comments = tblgen::emitSummaryAndDescComments(
+ def.getSummary(), def.getDescription());
+ if (!comments.empty()) {
+ os << comments << "\n";
+ }
os << "class " << def.getCppClassName() << ";\n";
+ }
// Emit the declarations.
for (const AttrOrTypeDef &def : defs)
diff --git a/mlir/tools/mlir-tblgen/CMakeLists.txt b/mlir/tools/mlir-tblgen/CMakeLists.txt
index 9431c59860522..2a7ef7e0576c8 100644
--- a/mlir/tools/mlir-tblgen/CMakeLists.txt
+++ b/mlir/tools/mlir-tblgen/CMakeLists.txt
@@ -33,6 +33,7 @@ add_tablegen(mlir-tblgen MLIR
RewriterGen.cpp
SPIRVUtilsGen.cpp
TosaUtilsGen.cpp
+ CppGenUtilities.cpp
)
target_link_libraries(mlir-tblgen
diff --git a/mlir/tools/mlir-tblgen/CppGenUtilities.cpp b/mlir/tools/mlir-tblgen/CppGenUtilities.cpp
new file mode 100644
index 0000000000000..ebca20cc685f4
--- /dev/null
+++ b/mlir/tools/mlir-tblgen/CppGenUtilities.cpp
@@ -0,0 +1,39 @@
+//===- CppGenUtilities.cpp - MLIR cpp gen utilities --------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// Defines common utilities for generating cpp files from tablegen
+// structures.
+//
+//===----------------------------------------------------------------------===//
+
+#include "CppGenUtilities.h"
+#include "mlir/Support/IndentedOstream.h"
+
+std::string
+mlir::tblgen::emitSummaryAndDescComments(llvm::StringRef summary,
+ llvm::StringRef description) {
+
+ std::string comments = "";
+ StringRef trimmedSummary = summary.trim();
+ StringRef trimmedDesc = description.trim();
+ llvm::raw_string_ostream os(comments);
+ raw_indented_ostream ros(os);
+
+ if (!trimmedSummary.empty()) {
+ ros.printReindented(trimmedSummary, "/// ");
+ }
+
+ if (!trimmedDesc.empty()) {
+ if (!trimmedSummary.empty()) {
+ // If there is a summary, add a newline after it.
+ ros << "\n";
+ }
+ ros.printReindented(trimmedDesc, "/// ");
+ }
+ return comments;
+}
diff --git a/mlir/tools/mlir-tblgen/CppGenUtilities.h b/mlir/tools/mlir-tblgen/CppGenUtilities.h
new file mode 100644
index 0000000000000..231c59a9e148f
--- /dev/null
+++ b/mlir/tools/mlir-tblgen/CppGenUtilities.h
@@ -0,0 +1,29 @@
+//===- CppGenUtilities.h - MLIR cpp gen utilities ---------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines common utilities for generating cpp files from tablegen
+// structures.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef MLIR_TOOLS_MLIRTBLGEN_CPPGENUTILITIES_H_
+#define MLIR_TOOLS_MLIRTBLGEN_CPPGENUTILITIES_H_
+
+#include "llvm/ADT/StringRef.h"
+
+namespace mlir {
+namespace tblgen {
+
+// Emit the summary and description as a C++ comment, perperly aligned placed
+// adjacent to the class declaration of generated classes.
+std::string emitSummaryAndDescComments(llvm::StringRef summary,
+ llvm::StringRef description);
+} // namespace tblgen
+} // namespace mlir
+
+#endif // MLIR_TOOLS_MLIRTBLGEN_CPPGENUTILITIES_H_
diff --git a/mlir/tools/mlir-tblgen/DialectGen.cpp b/mlir/tools/mlir-tblgen/DialectGen.cpp
index 6cf71d2bb0174..292ff0b9daa73 100644
--- a/mlir/tools/mlir-tblgen/DialectGen.cpp
+++ b/mlir/tools/mlir-tblgen/DialectGen.cpp
@@ -10,6 +10,7 @@
//
//===----------------------------------------------------------------------===//
+#include "CppGenUtilities.h"
#include "DialectGenUtilities.h"
#include "mlir/TableGen/Class.h"
#include "mlir/TableGen/CodeGenHelpers.h"
@@ -109,6 +110,7 @@ tblgen::findDialectToGenerate(ArrayRef<Dialect> dialects) {
/// {1}: The dialect namespace.
/// {2}: The dialect parent class.
static const char *const dialectDeclBeginStr = R"(
+{3}
class {0} : public ::mlir::{2} {
explicit {0}(::mlir::MLIRContext *context);
@@ -245,8 +247,11 @@ static void emitDialectDecl(Dialect &dialect, raw_ostream &os) {
std::string cppName = dialect.getCppClassName();
StringRef superClassName =
dialect.isExtensible() ? "ExtensibleDialect" : "Dialect";
+
+ std::string comments = tblgen::emitSummaryAndDescComments(
+ dialect.getSummary(), dialect.getDescription());
os << llvm::formatv(dialectDeclBeginStr, cppName, dialect.getName(),
- superClassName);
+ superClassName, comments);
// If the dialect requested the default attribute printer and parser, emit
// the declarations for the hooks.
diff --git a/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp b/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp
index 3f397f3a8e6fd..373d3762cbb1a 100644
--- a/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp
+++ b/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp
@@ -11,6 +11,7 @@
//
//===----------------------------------------------------------------------===//
+#include "CppGenUtilities.h"
#include "OpClass.h"
#include "OpFormatGen.h"
#include "OpGenHelpers.h"
@@ -2640,8 +2641,7 @@ void OpEmitter::genSeparateArgParamBuilder() {
// Avoid emitting "resultTypes.size() >= 0u" which is always true.
if (!hasVariadicResult || numNonVariadicResults != 0)
- body << " "
- << "assert(resultTypes.size() "
+ body << " " << "assert(resultTypes.size() "
<< (hasVariadicResult ? ">=" : "==") << " "
<< numNonVariadicResults
<< "u && \"mismatched number of results\");\n";
@@ -4749,6 +4749,11 @@ static void emitOpClassDecls(const RecordKeeper &records,
for (auto *def : defs) {
Operator op(*def);
NamespaceEmitter emitter(os, op.getCppNamespace());
+ std::string comments = tblgen::emitSummaryAndDescComments(
+ op.getSummary(), op.getDescription());
+ if (!comments.empty()) {
+ os << comments << "\n";
+ }
os << "class " << op.getCppClassName() << ";\n";
}
|
Also add test cases for Attr and Enum
|
@joker-eph Thanks for the review! I added one more commit to generate the same comments for |
|
Thanks for the contribution! |
|
@pchen7e2 Congratulations on having your first Pull Request (PR) merged into the LLVM Project! Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR. Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues. How to do this, and the rest of the post-merge process, is covered in detail here. If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again. If you don't get any reports, no action is required from you. Your changes are working as expected, well done! |
This commit takes the
summaryanddescriptionof TableGen files and generate a cpp comments on top of the declarations of generated cpp classes. It only affectsDialect,TypeandOpfor now. I'd like to get people's opinions on the direction before doing the same for other components such asInterface.The main motivation is to improve the developer experience. When people work on compilers from an IDE, they will be able to hover over the symbols (e.g.
"ADialect::BOp") in their cpp code and see the summary and descriptions without having to referring to the.tdfiles. I attach a screenshot as an example here to show what it looks like in VSCode with LSP properly set up: