CMake: detect and refuse broken std::regex

Swift SVN r32256

Author: gribozavr

CMakeLists.txt

@@ -205,6 +205,7 @@ endif()
 # Include CMake modules
 #
 
+include(CheckCXXSourceRuns)
 include(CMakeParseArguments)
 include(SwiftTranslateFlag)
 include(SwiftHandleGybSources)
@@ -305,6 +306,34 @@ if(XCODE)
   swift_common_xcode_cxx_config()
 endif()
 
+function(check_working_std_regex result_var_name)
+  if("${CMAKE_SYSTEM_NAME}" STREQUAL "Darwin")
+    # Apple operating systems use libc++, which has a working std::regex.
+    set("${result_var_name}" TRUE PARENT_SCOPE)
+  else()
+    # libstdc++ 4.8 has an incomplete std::regex implementation, and crashes
+    # on many regexes.
+    # libstdc++ 4.9 works.
+    set(std_regex_test_source
+"
+#include <regex>
+const std::regex broken_regex{
+  \"([a]+)\",
+  std::regex::ECMAScript | std::regex::nosubs};
+
+int main() {}
+")
+
+    check_cxx_source_runs("${std_regex_test_source}" result)
+    if (result)
+      set("${result_var_name}" TRUE PARENT_SCOPE)
+    else()
+      set("${result_var_name}" FALSE PARENT_SCOPE)
+    endif()
+  endif()
+endfunction()
+check_working_std_regex(SWIFT_HAVE_WORKING_STD_REGEX)
+
 #
 # Enable additional warnings.
 #

include/swift/Config.h.in

@@ -6,5 +6,6 @@
 
 #cmakedefine SWIFT_ENABLE_OBJECT_LITERALS 1
 #cmakedefine SWIFT_ENABLE_TARGET_TVOS 1
+#cmakedefine SWIFT_HAVE_WORKING_STD_REGEX 1
 
 #endif // SWIFT_CONFIG_H

lib/IDE/SyntaxModel.cpp

@@ -22,6 +22,7 @@
 #include "swift/Basic/SourceManager.h"
 #include "swift/Parse/Lexer.h"
 #include "swift/Parse/Token.h"
+#include "swift/Config.h"
 #include "swift/Subsystems.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/Support/MemoryBuffer.h"
@@ -238,8 +239,8 @@ class ModelASTWalker : public ASTWalker {
   unsigned BufferID;
   std::vector<StructureElement> SubStructureStack;
   SourceLoc LastLoc;
-  static const std::regex URLRxs[3];
-  static const std::regex DocCommentRxs[3];
+  static const std::regex &getURLRegex(unsigned Index);
+  static const std::regex &getDocCommentRegex(unsigned Index);
 
   Optional<SyntaxNode> parseFieldNode(StringRef Text, StringRef OrigText,
                                       SourceLoc OrigLoc);
@@ -307,26 +308,32 @@ class ModelASTWalker : public ASTWalker {
   }
 };
 
-const std::regex ModelASTWalker::URLRxs[3] =  {
+const std::regex &ModelASTWalker::getURLRegex(unsigned Index) {
+  static const std::regex Regexes[3] =  {
     std::regex{ RegexStrURL, std::regex::ECMAScript | std::regex::nosubs },
     std::regex{ RegexStrMailURL, std::regex::ECMAScript | std::regex::nosubs },
     std::regex{ RegexStrRadarURL, std::regex::ECMAScript | std::regex::nosubs }
-};
+  };
+  return Regexes[Index];
+}
 
-const std::regex ModelASTWalker::DocCommentRxs[3] = {
-  std::regex {
-    RegexStrParameter,
-    std::regex::egrep | std::regex::icase | std::regex::optimize
-  },
-  std::regex {
-    RegexStrDocCommentParametersHeading,
-    std::regex::egrep | std::regex::icase | std::regex::optimize
-  },
-  std::regex {
-    RegexStrDocCommentField,
-    std::regex::egrep | std::regex::icase | std::regex::optimize
-  }
-};
+const std::regex &ModelASTWalker::getDocCommentRegex(unsigned Index) {
+  static const std::regex Regexes[3] =  {
+    std::regex {
+      RegexStrParameter,
+      std::regex::egrep | std::regex::icase | std::regex::optimize
+    },
+    std::regex {
+      RegexStrDocCommentParametersHeading,
+      std::regex::egrep | std::regex::icase | std::regex::optimize
+    },
+    std::regex {
+      RegexStrDocCommentField,
+      std::regex::egrep | std::regex::icase | std::regex::optimize
+    }
+  };
+  return Regexes[Index];
+}
 
 SyntaxStructureKind syntaxStructureKindFromNominalTypeDecl(NominalTypeDecl *N) {
   if (isa<ClassDecl>(N))
@@ -1291,6 +1298,10 @@ bool ModelASTWalker::processComment(CharSourceRange Range) {
 bool ModelASTWalker::findUrlStartingLoc(StringRef Text,
                                         unsigned &Start,
                                         std::regex &Regex) {
+#ifndef SWIFT_HAVE_WORKING_STD_REGEX
+  return false;
+#endif
+
   static const auto MailToPosition = std::find(URLProtocols.begin(),
                                                URLProtocols.end(),
                                                "mailto");
@@ -1306,11 +1317,11 @@ bool ModelASTWalker::findUrlStartingLoc(StringRef Text,
         Text.substr(Index - It->size(), It->size()) == *It) {
       Start = Index - It->size();
       if (It < MailToPosition)
-        Regex = URLRxs[0];
+        Regex = getURLRegex(0);
       else if (It < RadarPosition)
-        Regex = URLRxs[1];
+        Regex = getURLRegex(1);
       else
-        Regex = URLRxs[2];
+        Regex = getURLRegex(2);
       return true;
     }
   }
@@ -1350,8 +1361,13 @@ bool ModelASTWalker::searchForURL(CharSourceRange Range) {
 Optional<SyntaxNode> ModelASTWalker::parseFieldNode(StringRef Text,
                                                     StringRef OrigText,
                                                     SourceLoc OrigLoc) {
+#ifndef SWIFT_HAVE_WORKING_STD_REGEX
+  return None;
+#endif
+
   std::match_results<StringRef::iterator> Matches;
-  for (auto &Rx : DocCommentRxs) {
+  for (unsigned i = 0; i != 3; ++i) {
+    auto &Rx = getDocCommentRegex(i);
     bool HadMatch = std::regex_search(Text.begin(), Text.end(), Matches, Rx);
     if (HadMatch)
       break;

test/IDE/coloring.swift

@@ -1,5 +1,6 @@
 // RUN: %target-swift-ide-test -syntax-coloring -source-filename %s | FileCheck %s
 // RUN: %target-swift-ide-test -syntax-coloring -typecheck -source-filename %s | FileCheck %s
+// XFAIL: broken_std_regex
 
 #line 17 "abc.swift"
 // CHECK: <#kw>#line</#kw> <int>17</int> <str>"abc.swift"</str>

test/lit.site.cfg.in

@@ -37,6 +37,9 @@ if "@SWIFT_ENABLE_OBJECT_LITERALS@" == "TRUE":
 if "@SWIFT_OPTIMIZED@" == "TRUE":
     config.available_features.add("optimized_stdlib")
 
+if "@SWIFT_HAVE_WORKING_STD_REGEX@" == "FALSE":
+    config.available_features.add('broken_std_regex')
+
 # Let the main config do the real work.
 if config.test_exec_root is None:
     config.test_exec_root = os.path.dirname(os.path.realpath(__file__))