[NFC][SpecialCaseList] Precommit Version 4 tests

clang/docs/ReleaseNotes.rst

@@ -69,6 +69,8 @@ Potentially Breaking Changes
   call the member ``operator delete`` instead of the expected global
   delete operator. The old behavior is retained under ``-fclang-abi-compat=21``
   flag.
+- Clang warning suppressions file, ``--warning-suppression-mappings=``, now will
+  use the last matching entry instead of the longest one.
 - Trailing null statements in GNU statement expressions are no longer
   ignored by Clang; they now result in a void type. Clang previously
   matched GCC's behavior, which was recently clarified to be incorrect.

clang/docs/WarningSuppressionMappings.rst

@@ -63,7 +63,7 @@ Format
 Warning suppression mappings uses the same format as
 :doc:`SanitizerSpecialCaseList`.
 
-Sections describe which diagnostic group's behaviour to change, e.g.
+Sections describe which diagnostic group's behavior to change, e.g.
 ``[unused]``. When a diagnostic is matched by multiple sections, the latest
 section takes precedence.
 
@@ -76,7 +76,7 @@ Source files are matched against these globs either:
 - as paths relative to the current working directory
 - as absolute paths.
 
-When a source file matches multiple globs in a section, the longest one takes
+When a source file matches multiple globs in a section, the last one takes
 precedence.
 
 .. code-block:: bash

clang/include/clang/Basic/Diagnostic.h

@@ -971,7 +971,7 @@ class DiagnosticsEngine : public RefCountedBase<DiagnosticsEngine> {
   /// diagnostics in specific files.
   /// Mapping file is expected to be a special case list with sections denoting
   /// diagnostic groups and `src` entries for globs to suppress. `emit` category
-  /// can be used to disable suppression. Longest glob that matches a filepath
+  /// can be used to disable suppression. The last glob that matches a filepath
   /// takes precedence. For example:
   ///   [unused]
   ///   src:clang/*

clang/lib/Basic/Diagnostic.cpp

@@ -525,16 +525,15 @@ std::unique_ptr<WarningsSpecialCaseList>
 WarningsSpecialCaseList::create(const llvm::MemoryBuffer &Input,
                                 std::string &Err) {
   auto WarningSuppressionList = std::make_unique<WarningsSpecialCaseList>();
-  if (!WarningSuppressionList->createInternal(&Input, Err,
-                                              /*OrderBySize=*/true))
+  if (!WarningSuppressionList->createInternal(&Input, Err))
     return nullptr;
   return WarningSuppressionList;
 }
 
 void WarningsSpecialCaseList::processSections(DiagnosticsEngine &Diags) {
   static constexpr auto WarningFlavor = clang::diag::Flavor::WarningOrError;
   for (const auto &SectionEntry : sections()) {
-    StringRef DiagGroup = SectionEntry.SectionStr;
+    StringRef DiagGroup = SectionEntry.name();
     if (DiagGroup == "*") {
       // Drop the default section introduced by special case list, we only
       // support exact diagnostic group names.
@@ -588,15 +587,12 @@ bool WarningsSpecialCaseList::isDiagSuppressed(diag::kind DiagId,
 
   StringRef F = llvm::sys::path::remove_leading_dotslash(PLoc.getFilename());
 
-  StringRef LongestSup = DiagSection->getLongestMatch("src", F, "");
-  if (LongestSup.empty())
+  unsigned LastSup = DiagSection->getLastMatch("src", F, "");
+  if (LastSup == 0)
     return false;
 
-  StringRef LongestEmit = DiagSection->getLongestMatch("src", F, "emit");
-  if (LongestEmit.empty())
-    return true;
-
-  return LongestSup.size() > LongestEmit.size();
+  unsigned LastEmit = DiagSection->getLastMatch("src", F, "emit");
+  return LastSup > LastEmit;
 }
 
 bool DiagnosticsEngine::isSuppressedViaMapping(diag::kind DiagId,

clang/lib/Basic/ProfileList.cpp

@@ -36,7 +36,7 @@ class ProfileSpecialCaseList : public llvm::SpecialCaseList {
 
   bool hasPrefix(StringRef Prefix) const {
     for (const auto &It : sections())
-      if (It.Entries.count(Prefix) > 0)
+      if (It.hasPrefix(Prefix))
         return true;
     return false;
   }

clang/lib/Basic/SanitizerSpecialCaseList.cpp

@@ -42,7 +42,7 @@ void SanitizerSpecialCaseList::createSanitizerSections() {
     SanitizerMask Mask;
 
 #define SANITIZER(NAME, ID)                                                    \
-  if (S.SectionMatcher.matchAny(NAME))                                         \
+  if (S.matchName(NAME))                                                       \
     Mask |= SanitizerKind::ID;
 #define SANITIZER_GROUP(NAME, ID, ALIAS) SANITIZER(NAME, ID)
 
@@ -68,7 +68,7 @@ SanitizerSpecialCaseList::inSectionBlame(SanitizerMask Mask, StringRef Prefix,
     if (S.Mask & Mask) {
       unsigned LineNum = S.S.getLastMatch(Prefix, Query, Category);
       if (LineNum > 0)
-        return {S.S.FileIdx, LineNum};
+        return {S.S.fileIndex(), LineNum};
     }
   }
   return NotFound;

clang/unittests/Basic/DiagnosticTest.cpp

@@ -294,7 +294,7 @@ TEST_F(SuppressionMappingTest, EmitCategoryIsExcluded) {
                                             locForFile("foo.cpp")));
 }
 
-TEST_F(SuppressionMappingTest, LongestMatchWins) {
+TEST_F(SuppressionMappingTest, LastMatchWins) {
   llvm::StringLiteral SuppressionMappingFile = R"(
   [unused]
   src:*clang/*
@@ -327,10 +327,8 @@ TEST_F(SuppressionMappingTest, LongShortMatch) {
 
   EXPECT_TRUE(Diags.isSuppressedViaMapping(diag::warn_unused_function,
                                            locForFile("test/t1.cpp")));
-
-  // FIXME: This is confusing.
-  EXPECT_TRUE(Diags.isSuppressedViaMapping(diag::warn_unused_function,
-                                           locForFile("lld/test/t2.cpp")));
+  EXPECT_FALSE(Diags.isSuppressedViaMapping(diag::warn_unused_function,
+                                            locForFile("lld/test/t2.cpp")));
 }
 
 TEST_F(SuppressionMappingTest, ShortLongMatch) {

llvm/include/llvm/Support/SpecialCaseList.h

@@ -12,19 +12,11 @@
 #ifndef LLVM_SUPPORT_SPECIALCASELIST_H
 #define LLVM_SUPPORT_SPECIALCASELIST_H
 
-#include "llvm/ADT/ArrayRef.h"
-#include "llvm/ADT/RadixTree.h"
-#include "llvm/ADT/SmallVector.h"
-#include "llvm/ADT/StringMap.h"
-#include "llvm/ADT/iterator_range.h"
 #include "llvm/Support/Allocator.h"
-#include "llvm/Support/Compiler.h"
-#include "llvm/Support/GlobPattern.h"
-#include "llvm/Support/Regex.h"
+#include "llvm/Support/Error.h"
 #include <memory>
 #include <string>
 #include <utility>
-#include <variant>
 #include <vector>
 
 namespace llvm {
@@ -118,119 +110,45 @@ class SpecialCaseList {
   // classes.
   LLVM_ABI bool createInternal(const std::vector<std::string> &Paths,
                                vfs::FileSystem &VFS, std::string &Error);
-  LLVM_ABI bool createInternal(const MemoryBuffer *MB, std::string &Error,
-                               bool OrderBySize = false);
+  LLVM_ABI bool createInternal(const MemoryBuffer *MB, std::string &Error);
 
   SpecialCaseList() = default;
   SpecialCaseList(SpecialCaseList const &) = delete;
   SpecialCaseList &operator=(SpecialCaseList const &) = delete;
 
-private:
-  // Lagacy v1 matcher.
-  class RegexMatcher {
+  class Section {
   public:
-    LLVM_ABI Error insert(StringRef Pattern, unsigned LineNumber);
-    LLVM_ABI void preprocess(bool BySize);
-
-    LLVM_ABI void
-    match(StringRef Query,
-          llvm::function_ref<void(StringRef Rule, unsigned LineNo)> Cb) const;
-
-    struct Reg {
-      Reg(StringRef Name, unsigned LineNo, Regex &&Rg)
-          : Name(Name), LineNo(LineNo), Rg(std::move(Rg)) {}
-      StringRef Name;
-      unsigned LineNo;
-      Regex Rg;
-    };
-
-    std::vector<Reg> RegExes;
-  };
-
-  class GlobMatcher {
-  public:
-    LLVM_ABI Error insert(StringRef Pattern, unsigned LineNumber);
-    LLVM_ABI void preprocess(bool BySize);
-
-    LLVM_ABI void
-    match(StringRef Query,
-          llvm::function_ref<void(StringRef Rule, unsigned LineNo)> Cb) const;
-
-    struct Glob {
-      Glob(StringRef Name, unsigned LineNo, GlobPattern &&Pattern)
-          : Name(Name), LineNo(LineNo), Pattern(std::move(Pattern)) {}
-      StringRef Name;
-      unsigned LineNo;
-      GlobPattern Pattern;
-    };
-
-    std::vector<GlobMatcher::Glob> Globs;
-
-    RadixTree<iterator_range<StringRef::const_iterator>,
-              RadixTree<iterator_range<StringRef::const_reverse_iterator>,
-                        SmallVector<const GlobMatcher::Glob *, 1>>>
-        PrefixSuffixToGlob;
-
-    RadixTree<iterator_range<StringRef::const_iterator>,
-              SmallVector<const GlobMatcher::Glob *, 1>>
-        SubstrToGlob;
-  };
-
-  /// Represents a set of patterns and their line numbers
-  class Matcher {
-  public:
-    LLVM_ABI Matcher(bool UseGlobs, bool RemoveDotSlash);
-
-    LLVM_ABI Error insert(StringRef Pattern, unsigned LineNumber);
-    LLVM_ABI void preprocess(bool BySize);
-
-    LLVM_ABI void
-    match(StringRef Query,
-          llvm::function_ref<void(StringRef Rule, unsigned LineNo)> Cb) const;
+    LLVM_ABI Section(StringRef Name, unsigned FileIdx, bool UseGlobs);
+    LLVM_ABI Section(Section &&);
+    LLVM_ABI ~Section();
 
-    LLVM_ABI bool matchAny(StringRef Query) const {
-      bool R = false;
-      match(Query, [&](StringRef, unsigned) { R = true; });
-      return R;
-    }
+    // Return name of the section, it's entire string in [].
+    StringRef name() const { return Name; }
 
-    std::variant<RegexMatcher, GlobMatcher> M;
-    bool RemoveDotSlash;
-  };
-
-  using SectionEntries = StringMap<StringMap<Matcher>>;
+    // Returns true of string 'Name' matches section name interpreted as a glob.
+    LLVM_ABI bool matchName(StringRef Name) const;
 
-protected:
-  struct Section {
-    Section(StringRef Str, unsigned FileIdx, bool UseGlobs)
-        : SectionMatcher(UseGlobs, /*RemoveDotSlash=*/false), SectionStr(Str),
-          FileIdx(FileIdx) {}
-
-    Section(Section &&) = default;
-
-    Matcher SectionMatcher;
-    SectionEntries Entries;
-    std::string SectionStr;
-    unsigned FileIdx;
+    // Return sequence number of the file where this section is defined.
+    unsigned fileIndex() const { return FileIdx; }
 
     // Helper method to search by Prefix, Query, and Category. Returns
     // 1-based line number on which rule is defined, or 0 if there is no match.
     LLVM_ABI unsigned getLastMatch(StringRef Prefix, StringRef Query,
                                    StringRef Category) const;
 
-    // Helper method to search by Prefix, Query, and Category. Returns
-    // matching rule, or empty string if there is no match.
-    LLVM_ABI StringRef getLongestMatch(StringRef Prefix, StringRef Query,
-                                       StringRef Category) const;
+    /// Returns true if the section has any entries for the given prefix.
+    LLVM_ABI bool hasPrefix(StringRef Prefix) const;
 
   private:
     friend class SpecialCaseList;
-    LLVM_ABI void preprocess(bool OrderBySize);
-    LLVM_ABI const SpecialCaseList::Matcher *
-    findMatcher(StringRef Prefix, StringRef Category) const;
+    class SectionImpl;
+
+    StringRef Name;
+    unsigned FileIdx;
+    std::unique_ptr<SectionImpl> Impl;
   };
 
-  ArrayRef<const Section> sections() const { return Sections; }
+  const std::vector<Section> &sections() const;
 
 private:
   BumpPtrAllocator StrAlloc;
@@ -242,7 +160,7 @@ class SpecialCaseList {
 
   /// Parses just-constructed SpecialCaseList entries from a memory buffer.
   LLVM_ABI bool parse(unsigned FileIdx, const MemoryBuffer *MB,
-                      std::string &Error, bool OrderBySize);
+                      std::string &Error);
 };
 
 } // namespace llvm