Recognize the Swift compiler in DW_AT_producer

[aprantl]

Summary

This patch adds support for Swift compiler producer strings to DWARFUnit.

[aprantl]

Sort variables in a more logical order.

[JDevlieghere]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

684

For consistency with the code below, should we first check if roducer.contains("swift") and then execute the regex? Otherwise the indentation on the line below is off.

[shafik]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

674

const?

[aprantl]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

684

I think it makes more sense to run the clang regex right away. Matching the regex is faster than first searching for clang and then running the regex.

[aprantl]

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

684

No, that does not preserve the semantics. We want to set the producer even if we can't parse the version. In any case, scanning the string twice in the Swift case is unnecessary.

[aprantl]

Const.

[JDevlieghere]

LGTM

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

684

Sounds reasonable to me. Should we do the same for clang? Then we just need to fix the indentation on line 685.

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.h

 
 typedef std::shared_ptr<DWARFUnit> DWARFUnitSP;
 
 enum DWARFProducer {
   eProducerInvalid = 0,
   eProducerClang,
   eProducerGCC,
   eProducerLLVMGCC,
+  eProducerSwift,
   eProducerOther
 };
 
 /// Base class describing the header of any kind of "unit."  Some information
 /// is specific to certain unit types.  We separate this class out so we can
 /// parse the header before deciding what specific kind of unit to construct.
 class DWARFUnitHeader {
   dw_offset_t m_offset = 0;

lldb/source/Plugins/SymbolFile/DWARF/DWARFUnit.cpp

void DWARFUnit::ParseProducerInfo() {
   if (!die)
     return;
 
   llvm::StringRef producer(
       die->GetAttributeValueAsString(this, DW_AT_producer, nullptr));
   if (producer.empty())
     return;
 
-  static RegularExpression g_llvm_gcc_regex(
+  static const RegularExpression g_swiftlang_version_regex(
+      llvm::StringRef(R"(swiftlang-([0-9]+\.[0-9]+\.[0-9]+(\.[0-9]+)?))"));
+  static const RegularExpression g_clang_version_regex(
+      llvm::StringRef(R"(clang-([0-9]+\.[0-9]+\.[0-9]+(\.[0-9]+)?))"));
+  static const RegularExpression g_llvm_gcc_regex(
       llvm::StringRef(R"(4\.[012]\.[01] )"
                       R"(\(Based on Apple Inc\. build [0-9]+\) )"
                       R"(\(LLVM build [\.0-9]+\)$)"));
-  static RegularExpression g_clang_version_regex(
-      llvm::StringRef(R"(clang-([0-9]+\.[0-9]+\.[0-9]+(\.[0-9]+)?))"));
 
-  if (g_llvm_gcc_regex.Execute(producer)) {
-    m_producer = eProducerLLVMGCC;
-  } else if (producer.contains("clang")) {
-    llvm::SmallVector<llvm::StringRef, 3> matches;
+  if (g_swiftlang_version_regex.Execute(producer, &matches)) {
+      m_producer_version.tryParse(matches[1]);
+    m_producer = eProducerSwift;
+  } else if (producer.contains("clang")) {
     if (g_clang_version_regex.Execute(producer, &matches))
       m_producer_version.tryParse(matches[1]);
     m_producer = eProducerClang;
-  } else if (producer.contains("GNU"))
+  } else if (producer.contains("GNU")) {
     m_producer = eProducerGCC;
+  } else if (g_llvm_gcc_regex.Execute(producer)) {
+    m_producer = eProducerLLVMGCC;
+  }
 }
 
 DWARFProducer DWARFUnit::GetProducer() {
   if (m_producer == eProducerInvalid)
     ParseProducerInfo();
   return m_producer;
 }
 

lldb/unittests/SymbolFile/DWARF/DWARFUnitTest.cpp

- Version:         4
         - AbbrCode:        0x0
 )";
 
   YAMLModuleTester t(yamldata);
   DWARFUnit *unit = t.GetDwarfUnit();
   ASSERT_TRUE((bool)unit);
   EXPECT_EQ(unit->GetProducer(), eProducerLLVMGCC);
 }
+
+TEST(DWARFUnitTest, SwiftProducer) {
+  const char *yamldata = R"(
+--- !ELF
+FileHeader:
+  Class:   ELFCLASS64
+  Data:    ELFDATA2LSB
+  Type:    ET_EXEC
+  Machine: EM_386
+DWARF:
+  debug_str:
+    - 'Apple Swift version 5.5 (swiftlang-1300.0.31.1 clang-1300.0.29.1)'
+  debug_abbrev:
+    - Table:
+        - Code:            0x00000001
+          Tag:             DW_TAG_compile_unit
+          Children:        DW_CHILDREN_yes
+          Attributes:
+            - Attribute:       DW_AT_producer
+              Form:            DW_FORM_strp
+  debug_info:
+    - Version:         4
+      AddrSize:        8
+      Entries:
+        - AbbrCode:        0x1
+          Values:
+            - Value:           0x0
+        - AbbrCode:        0x0
+)";
+
+  YAMLModuleTester t(yamldata);
+  DWARFUnit *unit = t.GetDwarfUnit();
+  ASSERT_TRUE((bool)unit);
+  EXPECT_EQ(unit->GetProducer(), eProducerSwift);
+  EXPECT_EQ(unit->GetProducerVersion(), llvm::VersionTuple(1300, 0, 31, 1));
+}