Path: Add RealPath()

2024-11-26 15:45:42 +00:00 · 2023-12-05 15:21:37 +10:00 · 2023-12-05 15:21:37 +10:00 · 524625269f
parent b166ec3403
commit 524625269f
3 changed files with 192 additions and 20 deletions
--- a/src/common-tests/path_tests.cpp
+++ b/src/common-tests/path_tests.cpp
@ -5,7 +5,7 @@
 #include "common/types.h"
 #include <gtest/gtest.h>

-TEST(FileSystem, ToNativePath)
+TEST(Path, ToNativePath)
 {
  ASSERT_EQ(Path::ToNativePath(""), "");

@ -29,7 +29,7 @@ TEST(FileSystem, ToNativePath)
 #endif
 }

-TEST(FileSystem, IsAbsolute)
+TEST(Path, IsAbsolute)
 {
  ASSERT_FALSE(Path::IsAbsolute(""));
  ASSERT_FALSE(Path::IsAbsolute("foo"));
@ -61,7 +61,7 @@ TEST(FileSystem, IsAbsolute)
 #endif
 }

-TEST(FileSystem, Canonicalize)
+TEST(Path, Canonicalize)
 {
  ASSERT_EQ(Path::Canonicalize(""), Path::ToNativePath(""));
  ASSERT_EQ(Path::Canonicalize("foo/bar/../baz"), Path::ToNativePath("foo/baz"));
@ -72,10 +72,8 @@ TEST(FileSystem, Canonicalize)
  ASSERT_EQ(Path::Canonicalize("./foo"), Path::ToNativePath("foo"));
  ASSERT_EQ(Path::Canonicalize("../foo"), Path::ToNativePath("../foo"));
  ASSERT_EQ(Path::Canonicalize("foo/b🙃ar/../b🙃az/./foo"), Path::ToNativePath("foo/b🙃az/foo"));
-  ASSERT_EQ(
-    Path::Canonicalize(
-      "ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/b🙃az/../foℹ︎o"),
-    Path::ToNativePath("ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/foℹ︎o"));
+  ASSERT_EQ(Path::Canonicalize("ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/b🙃az/../foℹ︎o"),
+            Path::ToNativePath("ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/foℹ︎o"));
 #ifdef _WIN32
  ASSERT_EQ(Path::Canonicalize("C:\\foo\\bar\\..\\baz\\.\\foo"), "C:\\foo\\baz\\foo");
  ASSERT_EQ(Path::Canonicalize("C:/foo\\bar\\..\\baz\\.\\foo"), "C:\\foo\\baz\\foo");
@ -87,7 +85,7 @@ TEST(FileSystem, Canonicalize)
 #endif
 }

-TEST(FileSystem, Combine)
+TEST(Path, Combine)
 {
  ASSERT_EQ(Path::Combine("", ""), Path::ToNativePath(""));
  ASSERT_EQ(Path::Combine("foo", "bar"), Path::ToNativePath("foo/bar"));
@ -108,7 +106,7 @@ TEST(FileSystem, Combine)
 #endif
 }

-TEST(FileSystem, AppendDirectory)
+TEST(Path, AppendDirectory)
 {
  ASSERT_EQ(Path::AppendDirectory("foo/bar", "baz"), Path::ToNativePath("foo/baz/bar"));
  ASSERT_EQ(Path::AppendDirectory("", "baz"), Path::ToNativePath("baz"));
@ -122,7 +120,7 @@ TEST(FileSystem, AppendDirectory)
 #endif
 }

-TEST(FileSystem, MakeRelative)
+TEST(Path, MakeRelative)
 {
  ASSERT_EQ(Path::MakeRelative("", ""), Path::ToNativePath(""));
  ASSERT_EQ(Path::MakeRelative("foo", ""), Path::ToNativePath("foo"));
@ -141,8 +139,7 @@ TEST(FileSystem, MakeRelative)
  ASSERT_EQ(Path::MakeRelative(A "foo/b🙃ar", A "foo/b🙃az"), Path::ToNativePath("../b🙃ar"));
  ASSERT_EQ(Path::MakeRelative(A "f🙃oo/b🙃ar", A "f🙃oo/b🙃az"), Path::ToNativePath("../b🙃ar"));
  ASSERT_EQ(
-    Path::MakeRelative(A "ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/b🙃ar",
-                       A "ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/b🙃az"),
+    Path::MakeRelative(A "ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/b🙃ar", A "ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱/b🙃az"),
    Path::ToNativePath("../b🙃ar"));

 #undef A
@ -154,7 +151,7 @@ TEST(FileSystem, MakeRelative)
 #endif
 }

-TEST(FileSystem, GetExtension)
+TEST(Path, GetExtension)
 {
  ASSERT_EQ(Path::GetExtension("foo"), "");
  ASSERT_EQ(Path::GetExtension("foo.txt"), "txt");
@ -164,7 +161,7 @@ TEST(FileSystem, GetExtension)
  ASSERT_EQ(Path::GetExtension("a/b/foo"), "");
 }

-TEST(FileSystem, GetFileName)
+TEST(Path, GetFileName)
 {
  ASSERT_EQ(Path::GetFileName(""), "");
  ASSERT_EQ(Path::GetFileName("foo"), "foo");
@ -179,7 +176,7 @@ TEST(FileSystem, GetFileName)
 #endif
 }

-TEST(FileSystem, GetFileTitle)
+TEST(Path, GetFileTitle)
 {
  ASSERT_EQ(Path::GetFileTitle(""), "");
  ASSERT_EQ(Path::GetFileTitle("foo"), "foo");
@ -193,7 +190,7 @@ TEST(FileSystem, GetFileTitle)
 #endif
 }

-TEST(FileSystem, GetDirectory)
+TEST(Path, GetDirectory)
 {
  ASSERT_EQ(Path::GetDirectory(""), "");
  ASSERT_EQ(Path::GetDirectory("foo"), "");
@ -207,7 +204,7 @@ TEST(FileSystem, GetDirectory)
 #endif
 }

-TEST(FileSystem, ChangeFileName)
+TEST(Path, ChangeFileName)
 {
  ASSERT_EQ(Path::ChangeFileName("", ""), Path::ToNativePath(""));
  ASSERT_EQ(Path::ChangeFileName("", "bar"), Path::ToNativePath("bar"));
@ -227,13 +224,14 @@ TEST(FileSystem, ChangeFileName)
 #endif
 }

-TEST(FileSystem, SanitizeFileName)
+TEST(Path, SanitizeFileName)
 {
  ASSERT_EQ(Path::SanitizeFileName("foo"), "foo");
  ASSERT_EQ(Path::SanitizeFileName("foo/bar"), "foo_bar");
  ASSERT_EQ(Path::SanitizeFileName("f🙃o"), "f🙃o");
  ASSERT_EQ(Path::SanitizeFileName("ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱"), "ŻąłóРстуぬねのはen🍪⟑η∏☉ⴤℹ︎∩₲ ₱⟑♰⫳🐱");
-  ASSERT_EQ(Path::SanitizeFileName("abcdefghijlkmnopqrstuvwxyz-0123456789+&=_[]{}"), "abcdefghijlkmnopqrstuvwxyz-0123456789+&=_[]{}");
+  ASSERT_EQ(Path::SanitizeFileName("abcdefghijlkmnopqrstuvwxyz-0123456789+&=_[]{}"),
+            "abcdefghijlkmnopqrstuvwxyz-0123456789+&=_[]{}");
  ASSERT_EQ(Path::SanitizeFileName("some*path**with*asterisks"), "some_path__with_asterisks");
 #ifdef _WIN32
  ASSERT_EQ(Path::SanitizeFileName("foo:"), "foo_");
@ -244,3 +242,17 @@ TEST(FileSystem, SanitizeFileName)
 #endif
  ASSERT_EQ(Path::SanitizeFileName("foo/bar", false), "foo/bar");
 }
+
+#if 0
+
+// Relies on presence of files.
+TEST(Path, RealPath)
+{
+#ifdef _WIN32
+  ASSERT_EQ(Path::RealPath("C:\\Users\\Me\\Desktop\\foo\\baz"), "C:\\Users\\Me\\Desktop\\foo\\bar\\baz");
+#else
+  ASSERT_EQ(Path::RealPath("/lib/foo/bar"), "/usr/lib/foo/bar");
+#endif
+}
+
+#endif
--- a/src/common/file_system.cpp
+++ b/src/common/file_system.cpp
@ -12,6 +12,7 @@
 #include <cstdlib>
 #include <cstring>
 #include <limits>
+#include <numeric>

 #ifdef __APPLE__
 #include <mach-o/dyld.h>
@ -192,6 +193,161 @@ bool Path::IsAbsolute(const std::string_view& path)
 #endif
 }

+std::string Path::RealPath(const std::string_view& path)
+{
+  // Resolve non-absolute paths first.
+  std::vector<std::string_view> components;
+  if (!IsAbsolute(path))
+    components = Path::SplitNativePath(Path::Combine(FileSystem::GetWorkingDirectory(), path));
+  else
+    components = Path::SplitNativePath(path);
+
+  std::string realpath;
+  if (components.empty())
+    return realpath;
+
+  // Different to path because relative.
+  realpath.reserve(std::accumulate(components.begin(), components.end(), static_cast<size_t>(0),
+                                   [](size_t l, const std::string_view& s) { return l + s.length(); }) +
+                   components.size() + 1);
+
+#ifdef _WIN32
+  std::wstring wrealpath;
+  std::vector<WCHAR> symlink_buf;
+  wrealpath.reserve(realpath.size());
+  symlink_buf.resize(path.size() + 1);
+
+  // Check for any symbolic links throughout the path while adding components.
+  bool test_symlink = true;
+  for (const std::string_view& comp : components)
+  {
+    if (!realpath.empty())
+      realpath.push_back(FS_OSPATH_SEPARATOR_CHARACTER);
+    realpath.append(comp);
+    if (test_symlink)
+    {
+      DWORD attribs;
+      if (StringUtil::UTF8StringToWideString(wrealpath, realpath) &&
+          (attribs = GetFileAttributesW(wrealpath.c_str())) != INVALID_FILE_ATTRIBUTES)
+      {
+        // if not a link, go to the next component
+        if (attribs & FILE_ATTRIBUTE_REPARSE_POINT)
+        {
+          const HANDLE hFile =
+            CreateFileW(wrealpath.c_str(), FILE_READ_ATTRIBUTES, FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE,
+                        nullptr, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, nullptr);
+          if (hFile != INVALID_HANDLE_VALUE)
+          {
+            // is a link! resolve it.
+            DWORD ret = GetFinalPathNameByHandleW(hFile, symlink_buf.data(), static_cast<DWORD>(symlink_buf.size()),
+                                                  FILE_NAME_NORMALIZED);
+            if (ret > symlink_buf.size())
+            {
+              symlink_buf.resize(ret);
+              ret = GetFinalPathNameByHandleW(hFile, symlink_buf.data(), static_cast<DWORD>(symlink_buf.size()),
+                                              FILE_NAME_NORMALIZED);
+            }
+            if (ret != 0)
+              StringUtil::WideStringToUTF8String(realpath, std::wstring_view(symlink_buf.data(), ret));
+            else
+              test_symlink = false;
+
+            CloseHandle(hFile);
+          }
+        }
+      }
+      else
+      {
+        // not a file or link
+        test_symlink = false;
+      }
+    }
+  }
+
+  // GetFinalPathNameByHandleW() adds a \\?\ prefix, so remove it.
+  if (realpath.starts_with("\\\\?\\") && IsAbsolute(std::string_view(realpath.data() + 4, realpath.size() - 4)))
+    realpath.erase(0, 4);
+
+#else
+  // Why this monstrosity instead of calling realpath()? realpath() only works on files that exist.
+  std::string basepath;
+  std::string symlink;
+
+  basepath.reserve(realpath.capacity());
+  symlink.resize(realpath.capacity());
+
+  // Check for any symbolic links throughout the path while adding components.
+  bool test_symlink = true;
+  for (const std::string_view& comp : components)
+  {
+    if (!test_symlink)
+    {
+      realpath.push_back(FS_OSPATH_SEPARATOR_CHARACTER);
+      realpath.append(comp);
+      continue;
+    }
+
+    basepath = realpath;
+    if (realpath.empty() || realpath.back() != FS_OSPATH_SEPARATOR_CHARACTER)
+      realpath.push_back(FS_OSPATH_SEPARATOR_CHARACTER);
+    realpath.append(comp);
+
+    // Check if the last component added is a symlink
+    struct stat sb;
+    if (lstat(realpath.c_str(), &sb) != 0)
+    {
+      // Don't bother checking any further components once we error out.
+      test_symlink = false;
+      continue;
+    }
+    else if (!S_ISLNK(sb.st_mode))
+    {
+      // Nope, keep going.
+      continue;
+    }
+
+    for (;;)
+    {
+      ssize_t sz = readlink(realpath.c_str(), symlink.data(), symlink.size());
+      if (sz < 0)
+      {
+        // shouldn't happen, due to the S_ISLNK check above.
+        test_symlink = false;
+        break;
+      }
+      else if (static_cast<size_t>(sz) == symlink.size())
+      {
+        // need a larger buffer
+        symlink.resize(symlink.size() * 2);
+        continue;
+      }
+      else
+      {
+        // is a link, and we resolved it. gotta check if the symlink itself is relative :(
+        symlink.resize(static_cast<size_t>(sz));
+        if (!Path::IsAbsolute(symlink))
+        {
+          // symlink is relative to the directory of the symlink
+          realpath = basepath;
+          if (realpath.empty() || realpath.back() != FS_OSPATH_SEPARATOR_CHARACTER)
+            realpath.push_back(FS_OSPATH_SEPARATOR_CHARACTER);
+          realpath.append(symlink);
+        }
+        else
+        {
+          // Use the new, symlinked path.
+          realpath = symlink;
+        }
+
+        break;
+      }
+    }
+  }
+#endif
+
+  return realpath;
+}
+
 std::string Path::ToNativePath(const std::string_view& path)
 {
  std::string ret;
@ -1382,6 +1538,7 @@ std::string FileSystem::GetProgramPath()
    break;
  }

+  // Windows symlinks don't behave silly like Linux, so no need to RealPath() it.
  return StringUtil::WideStringToUTF8String(buffer);
 }

--- a/src/common/path.h
+++ b/src/common/path.h
@ -31,6 +31,9 @@ void SanitizeFileName(std::string* str, bool strip_slashes = true);
 /// Returns true if the specified path is an absolute path (C:\Path on Windows or /path on Unix).
 bool IsAbsolute(const std::string_view& path);

+/// Resolves any symbolic links in the specified path.
+std::string RealPath(const std::string_view& path);
+
 /// Makes the specified path relative to another (e.g. /a/b/c, /a/b -> ../c).
 /// Both paths must be relative, otherwise this function will just return the input path.
 std::string MakeRelative(const std::string_view& path, const std::string_view& relative_to);