Merge "Add (partial) support for Windows long paths"

base/include/android-base/utf8.h

@@ -22,6 +22,8 @@
 #else
 // Bring in prototypes for standard APIs so that we can import them into the utf8 namespace.
 #include <fcntl.h>      // open
+#include <stdio.h>      // fopen
+#include <sys/stat.h>   // mkdir
 #include <unistd.h>     // unlink
 #endif
 
@@ -53,6 +55,19 @@ bool UTF8ToWide(const char* utf8, std::wstring* utf16);
 // Convert a UTF-8 std::string (including any embedded NULL characters) to
 // UTF-16. Returns whether the conversion was done successfully.
 bool UTF8ToWide(const std::string& utf8, std::wstring* utf16);
+
+// Convert a file system path, represented as a NULL-terminated string of
+// UTF-8 characters, to a UTF-16 string representing the same file system
+// path using the Windows extended-lengh path representation.
+//
+// See https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx#MAXPATH:
+//   ```The Windows API has many functions that also have Unicode versions to
+//   permit an extended-length path for a maximum total path length of 32,767
+//   characters. To specify an extended-length path, use the "\\?\" prefix.
+//   For example, "\\?\D:\very long path".```
+//
+// Returns whether the conversion was done successfully.
+bool UTF8PathToWindowsLongPath(const char* utf8, std::wstring* utf16);
 #endif
 
 // The functions in the utf8 namespace take UTF-8 strings. For Windows, these
@@ -73,9 +88,13 @@ bool UTF8ToWide(const std::string& utf8, std::wstring* utf16);
 namespace utf8 {
 
 #ifdef _WIN32
+FILE* fopen(const char* name, const char* mode);
+int mkdir(const char* name, mode_t mode);
 int open(const char* name, int flags, ...);
 int unlink(const char* name);
 #else
+using ::fopen;
+using ::mkdir;
 using ::open;
 using ::unlink;
 #endif

base/utf8.cpp

@@ -19,7 +19,9 @@
 #include "android-base/utf8.h"
 
 #include <fcntl.h>
+#include <stdio.h>
 
+#include <algorithm>
 #include <string>
 
 #include "android-base/logging.h"
@@ -153,12 +155,58 @@ bool UTF8ToWide(const std::string& utf8, std::wstring* utf16) {
   return UTF8ToWide(utf8.c_str(), utf8.length(), utf16);
 }
 
+static bool isDriveLetter(wchar_t c) {
+  return (c >= L'a' && c <= L'z') || (c >= L'A' && c <= L'Z');
+}
+
+bool UTF8PathToWindowsLongPath(const char* utf8, std::wstring* utf16) {
+  if (!UTF8ToWide(utf8, utf16)) {
+    return false;
+  }
+  // Note: Although most Win32 File I/O API are limited to MAX_PATH (260
+  //       characters), the CreateDirectory API is limited to 248 characters.
+  if (utf16->length() >= 248) {
+    // If path is of the form "x:\" or "x:/"
+    if (isDriveLetter((*utf16)[0]) && (*utf16)[1] == L':' &&
+        ((*utf16)[2] == L'\\' || (*utf16)[2] == L'/')) {
+      // Append long path prefix, and make sure there are no unix-style
+      // separators to ensure a fully compliant Win32 long path string.
+      utf16->insert(0, LR"(\\?\)");
+      std::replace(utf16->begin(), utf16->end(), L'/', L'\\');
+    }
+  }
+  return true;
+}
+
 // Versions of standard library APIs that support UTF-8 strings.
 namespace utf8 {
 
+FILE* fopen(const char* name, const char* mode) {
+  std::wstring name_utf16;
+  if (!UTF8PathToWindowsLongPath(name, &name_utf16)) {
+    return nullptr;
+  }
+
+  std::wstring mode_utf16;
+  if (!UTF8ToWide(mode, &mode_utf16)) {
+    return nullptr;
+  }
+
+  return _wfopen(name_utf16.c_str(), mode_utf16.c_str());
+}
+
+int mkdir(const char* name, mode_t mode) {
+  std::wstring name_utf16;
+  if (!UTF8PathToWindowsLongPath(name, &name_utf16)) {
+    return -1;
+  }
+
+  return _wmkdir(name_utf16.c_str());
+}
+
 int open(const char* name, int flags, ...) {
   std::wstring name_utf16;
-  if (!UTF8ToWide(name, &name_utf16)) {
+  if (!UTF8PathToWindowsLongPath(name, &name_utf16)) {
     return -1;
   }
 
@@ -175,7 +223,7 @@ int open(const char* name, int flags, ...) {
 
 int unlink(const char* name) {
   std::wstring name_utf16;
-  if (!UTF8ToWide(name, &name_utf16)) {
+  if (!UTF8PathToWindowsLongPath(name, &name_utf16)) {
     return -1;
   }
 

base/utf8_test.cpp

@@ -18,7 +18,12 @@
 
 #include <gtest/gtest.h>
 
+#include <fcntl.h>
+#include <stdlib.h>
+
 #include "android-base/macros.h"
+#include "android-base/test_utils.h"
+#include "android-base/unique_fd.h"
 
 namespace android {
 namespace base {
@@ -408,5 +413,76 @@ TEST(SysStrings, SysUTF8ToWide) {
   EXPECT_EQ(expected_null, SysUTF8ToWide(utf8_null));
 }
 
+TEST(UTF8PathToWindowsLongPathTest, DontAddPrefixIfShorterThanMaxPath) {
+  std::string utf8 = "c:\\mypath\\myfile.txt";
+
+  std::wstring wide;
+  EXPECT_TRUE(UTF8PathToWindowsLongPath(utf8.c_str(), &wide));
+
+  EXPECT_EQ(std::string::npos, wide.find(LR"(\\?\)"));
+}
+
+TEST(UTF8PathToWindowsLongPathTest, AddPrefixIfLongerThanMaxPath) {
+  std::string utf8 = "c:\\mypath";
+  while (utf8.length() < 300 /* MAX_PATH is 260 */) {
+    utf8 += "\\mypathsegment";
+  }
+
+  std::wstring wide;
+  EXPECT_TRUE(UTF8PathToWindowsLongPath(utf8.c_str(), &wide));
+
+  EXPECT_EQ(0U, wide.find(LR"(\\?\)"));
+  EXPECT_EQ(std::string::npos, wide.find(L"/"));
+}
+
+TEST(UTF8PathToWindowsLongPathTest, AddPrefixAndFixSeparatorsIfLongerThanMaxPath) {
+  std::string utf8 = "c:/mypath";
+  while (utf8.length() < 300 /* MAX_PATH is 260 */) {
+    utf8 += "/mypathsegment";
+  }
+
+  std::wstring wide;
+  EXPECT_TRUE(UTF8PathToWindowsLongPath(utf8.c_str(), &wide));
+
+  EXPECT_EQ(0U, wide.find(LR"(\\?\)"));
+  EXPECT_EQ(std::string::npos, wide.find(L"/"));
+}
+
+namespace utf8 {
+
+TEST(Utf8FilesTest, CanCreateOpenAndDeleteFileWithLongPath) {
+  TemporaryDir td;
+
+  // Create long directory path
+  std::string utf8 = td.path;
+  while (utf8.length() < 300 /* MAX_PATH is 260 */) {
+    utf8 += "\\mypathsegment";
+    EXPECT_EQ(0, mkdir(utf8.c_str(), 0));
+  }
+
+  // Create file
+  utf8 += "\\test-file.bin";
+  int flags = O_WRONLY | O_CREAT | O_TRUNC | O_BINARY;
+  int mode = 0666;
+  android::base::unique_fd fd(open(utf8.c_str(), flags, mode));
+  EXPECT_NE(-1, fd.get());
+
+  // Close file
+  fd.reset();
+  EXPECT_EQ(-1, fd.get());
+
+  // Open file with fopen
+  FILE* file = fopen(utf8.c_str(), "rb");
+  EXPECT_NE(nullptr, file);
+
+  if (file) {
+    fclose(file);
+  }
+
+  // Delete file
+  EXPECT_EQ(0, unlink(utf8.c_str()));
+}
+
+}  // namespace utf8
 }  // namespace base
 }  // namespace android