[libclang] Add clang_File_tryGetRealPathName

[MaskRay]

Summary

clang_getFileName() may return a path relative to WorkingDir.
On Arch Linux, during clang_indexTranslationUnit(), clang_getFileName() on
CXIdxIncludedIncludedFileInfo::file may return
"/../lib64/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../include/c++/7.3.0/string",
for #include <string>.

I presume WorkingDir is somehow changed to /usr/lib or /usr/include and
clang_getFileName() returns a path relative to WorkingDir.

clang_File_tryGetRealPathName() returns a better file name more useful for
the indexer in this case.

[MaskRay]

description

[MaskRay]

Increase CINDEX_VERSION_MINOR

[MaskRay]

description

[MaskRay]

On Arch Linux, ../lib64/gcc/x86_64-pc-linux-gnu/7.2.1/../../../../include/c++/7.2.1 (Name) resolves to a path that requires leading path components (/usr/include), this kind of resembles jailbreak.

[MaskRay]

ping

[MaskRay]

Ping

[jbcoe]

Can you add some simple tests?

[MaskRay]

In D42893#1053754, @jbcoe wrote:

Can you add some simple tests?

Where should I add tests? I cannot find tests for other clang_File_* functions.

[MaskRay]

There is no unittest but I have verified with a C++ language server on Arch Linux. Without using this function, cquery/ccls will not be able to index system header files because of excessive .. in the path returned by clang_getFileName.

[jbcoe]

I’ll see if I can find a suitable location for a test.

[jbcoe]

Tests for libclang are in clang/unittests/libclang/LibclangTest.cpp

Given the surgical nature of this change I hope it will be quick to add a test.

[MaskRay]

Rebase

[MaskRay]

Add unittests/libclang/LibclangTest.cpp test

[MaskRay]

Done.

ninja -C ~/Dev/llvm/debug unittests/libclang/libclangTests
~/Dev/llvm/debug/tools/clang/unittests/libclang/libclangTests

cfe/trunk/include/clang-c/Index.h

 
 /**
  * \brief Returns non-zero if the \c file1 and \c file2 point to the same file,
  * or they are both NULL.
  */
 CINDEX_LINKAGE int clang_File_isEqual(CXFile file1, CXFile file2);
 
 /**
+ * \brief Returns the real path name of \c file.
+ *
+ * An empty string may be returned. Use \c clang_getFileName() in that case.
+ */
+CINDEX_LINKAGE CXString clang_File_tryGetRealPathName(CXFile file);
+
+/**
  * @}
  */
 
 /**
  * \defgroup CINDEX_LOCATIONS Physical source locations
  *
  * Clang represents physical source locations in its abstract syntax tree in
  * great detail, with file, line, and column information for the majority of

cfe/trunk/tools/libclang/CIndex.cpp

int clang_File_isEqual(CXFile file1, CXFile file2) {
   if (!file1 || !file2)
     return false;
 
   FileEntry *FEnt1 = static_cast<FileEntry *>(file1);
   FileEntry *FEnt2 = static_cast<FileEntry *>(file2);
   return FEnt1->getUniqueID() == FEnt2->getUniqueID();
 }
 
+CXString clang_File_tryGetRealPathName(CXFile SFile) {
+  if (!SFile)
+    return cxstring::createNull();
+
+  FileEntry *FEnt = static_cast<FileEntry *>(SFile);
+  return cxstring::createRef(FEnt->tryGetRealPathName());
+}
+
 //===----------------------------------------------------------------------===//
 // CXCursor Operations.
 //===----------------------------------------------------------------------===//
 
 static const Decl *getDeclFromExpr(const Stmt *E) {
   if (const ImplicitCastExpr *CE = dyn_cast<ImplicitCastExpr>(E))
     return getDeclFromExpr(CE->getSubExpr());
 

cfe/trunk/tools/libclang/libclang.exports

 clang_Cursor_isDynamicCall
 clang_Cursor_isExternalSymbol
 clang_Cursor_isNull
 clang_Cursor_isObjCOptional
 clang_Cursor_isVariadic
 clang_Cursor_getModule
 clang_Cursor_getStorageClass
 clang_File_isEqual
+clang_File_tryGetRealPathName
 clang_Module_getASTFile
 clang_Module_getParent
 clang_Module_getName
 clang_Module_getFullName
 clang_Module_getNumTopLevelHeaders
 clang_Module_getTopLevelHeader
 clang_Module_isSystem
 clang_IndexAction_create

cfe/trunk/unittests/libclang/LibclangTest.cpp

if (clang_reparseTranslationUnit(ClangTU, num_unsaved_files, unsaved_files,
       DEBUG(llvm::dbgs() << "Reparse failed\n");
       return false;
     }
     DisplayDiagnostics();
     return true;
   }
 };
 
+TEST_F(LibclangReparseTest, FileName) {
+  std::string CppName = "main.cpp";
+  WriteFile(CppName, "int main() {}");
+  ClangTU = clang_parseTranslationUnit(Index, CppName.c_str(), nullptr, 0,
+                                       nullptr, 0, TUFlags);
+  CXFile cxf = clang_getFile(ClangTU, CppName.c_str());
+
+  CXString cxname = clang_getFileName(cxf);
+  ASSERT_TRUE(strstr(clang_getCString(cxname), CppName.c_str()));
+  clang_disposeString(cxname);
+
+  cxname = clang_File_tryGetRealPathName(cxf);
+  ASSERT_TRUE(strstr(clang_getCString(cxname), CppName.c_str()));
+  clang_disposeString(cxname);
+}
 
 TEST_F(LibclangReparseTest, Reparse) {
   const char *HeaderTop = "#ifndef H\n#define H\nstruct Foo { int bar;";
   const char *HeaderBottom = "\n};\n#endif\n";
   const char *CppFile = "#include \"HeaderFile.h\"\nint main() {"
                          " Foo foo; foo.bar = 7; foo.baz = 8; }\n";
   std::string HeaderName = "HeaderFile.h";
   std::string CppName = "CppFile.cpp";