Update files to better handle network shares.

In some of the path processing, adopt the root-name concept from `std::filesystem::path`; this should better detect absolute vs. relative urls in windows as well as network shares.

Translate network shares to the file://host/path syntax for uris.

Fixes: #250
PiperOrigin-RevId: 814286407
Change-Id: I0d84be09fa85f82ed170a60166b05333ddda82a0

Author: laramiel

tensorstore/BUILD

@@ -755,6 +755,7 @@ tensorstore_cc_test(
         "//tensorstore/internal/json_binding:bindable",
         "//tensorstore/internal/json_binding:gtest",
         "//tensorstore/internal/testing:json_gtest",
+        "//tensorstore/internal/testing:on_windows",
         "//tensorstore/kvstore",
         "//tensorstore/kvstore/file",
         "//tensorstore/kvstore/memory",

tensorstore/driver/auto/driver_test.cc

@@ -55,7 +55,7 @@ auto MakeInitial(const Context &context) {
 
 template <typename... Args>
 std::string AsFileUri(std::string_view path, Args... args) {
-  return absl::StrCat("file://", tensorstore::internal::OsPathToUriPath(path),
+  return absl::StrCat(tensorstore::internal::OsPathToFileUri(path).value(),
                       std::forward<Args>(args)...);
 }
 

tensorstore/driver/driver_testutil.cc

@@ -173,9 +173,13 @@ void TestTensorStoreDriverSpecRoundtrip(
     ReplaceStringInJson(options.full_base_spec, tempdir_key, tempdir->path());
 
     // For the URL, the tempdir path must begin with a leading slash.
+    // Generate the replacement url and strip off the "file://" prefix.
+    static constexpr size_t kFileUriPrefixLen =
+        std::string_view("file://").size();
+    auto file_uri = internal::OsPathToFileUri(tempdir->path());
+    TENSORSTORE_ASSERT_OK(file_uri.status());
     options.url = absl::StrReplaceAll(
-        options.url,
-        {{tempdir_key, internal::OsPathToUriPath(tempdir->path())}});
+        options.url, {{tempdir_key, file_uri->substr(kFileUriPrefixLen)}});
   }
   Transaction transaction(mode);
   auto context = Context::Default();

tensorstore/driver/zarr/BUILD

@@ -110,8 +110,8 @@ tensorstore_cc_test(
         ":compressor",
         ":zlib_compressor",
         "//tensorstore/internal/testing:json_gtest",
-        "//tensorstore/util:status",
         "//tensorstore/util:status_testutil",
+        "@abseil-cpp//absl/status",
         "@googletest//:gtest_main",
         "@nlohmann_json//:json",
     ],

tensorstore/driver/zarr/compressor_test.cc

@@ -19,7 +19,6 @@
 #include "absl/status/status.h"
 #include <nlohmann/json.hpp>
 #include "tensorstore/internal/testing/json_gtest.h"
-#include "tensorstore/util/status.h"
 #include "tensorstore/util/status_testutil.h"
 
 namespace {

tensorstore/internal/BUILD

@@ -1367,6 +1367,7 @@ tensorstore_cc_library(
     srcs = ["path.cc"],
     hdrs = ["path.h"],
     deps = [
+        ":ascii_set",
         "@abseil-cpp//absl/strings",
     ],
 )
@@ -1377,6 +1378,7 @@ tensorstore_cc_test(
     srcs = ["path_test.cc"],
     deps = [
         ":path",
+        "//tensorstore/internal/testing:on_windows",
         "@googletest//:gtest_main",
     ],
 )
@@ -1613,6 +1615,10 @@ tensorstore_cc_library(
     hdrs = ["uri_utils.h"],
     deps = [
         ":ascii_set",
+        ":path",
+        "//tensorstore/util:quote_string",
+        "//tensorstore/util:result",
+        "//tensorstore/util:status",
         "@abseil-cpp//absl/status",
         "@abseil-cpp//absl/strings",
     ],
@@ -1624,6 +1630,9 @@ tensorstore_cc_test(
     deps = [
         ":ascii_set",
         ":uri_utils",
+        "//tensorstore/internal/testing:on_windows",
+        "//tensorstore/util:status_testutil",
+        "@abseil-cpp//absl/status",
         "@googletest//:gtest_main",
     ],
 )

tensorstore/internal/path.cc

@@ -21,19 +21,29 @@
 #include <string_view>
 #include <utility>
 
+#include "absl/strings/ascii.h"  // IWYU pragma: keep
+#include "absl/strings/match.h"  // IWYU pragma: keep
 #include "absl/strings/str_cat.h"
 #include "absl/strings/string_view.h"
+#include "tensorstore/internal/ascii_set.h"  // IWYU pragma: keep
 
+namespace tensorstore {
 namespace {
 
 #ifdef _WIN32
 constexpr inline bool IsDirSeparator(char c) { return c == '\\' || c == '/'; }
+
+bool IsWindowsDriveLetter(std::string_view path) {
+  return path.length() >= 2 && path[1] == ':' && absl::ascii_isalpha(path[0]);
+}
+
+static constexpr internal::AsciiSet kIllegalUNCCharacters{"<>:/\\\"|'?*"};
+
 #else
 constexpr inline bool IsDirSeparator(char c) { return c == '/'; }
 #endif
 
 }  // namespace
-namespace tensorstore {
 namespace internal_path {
 
 std::string JoinPathImpl(std::initializer_list<std::string_view> paths) {
@@ -56,20 +66,22 @@ namespace internal {
 // Splits a path into the pair {dirname, basename}
 std::pair<std::string_view, std::string_view> PathDirnameBasename(
     std::string_view path) {
+  // Find the root directory.
+  auto root_dir = PathRootName(path);
+  if (root_dir.size() < path.size() && IsDirSeparator(path[root_dir.size()])) {
+    root_dir = path.substr(0, root_dir.size() + 1);
+  }
+
   size_t pos = path.size();
-  while (pos != 0 && !IsDirSeparator(path[pos - 1])) {
+  while (pos > root_dir.size() && !IsDirSeparator(path[pos - 1])) {
     --pos;
   }
   size_t basename = pos;
-  --pos;
-  if (pos == std::string_view::npos) {
-    return {"", path};
-  }
-  while (pos != 0 && IsDirSeparator(path[pos - 1])) {
+  if (pos > root_dir.size()) {
     --pos;
-  }
-  if (pos == 0) {
-    return {"/", path.substr(basename)};
+    while (pos > root_dir.size() && IsDirSeparator(path[pos - 1])) {
+      --pos;
+    }
   }
   return {path.substr(0, pos), path.substr(basename)};
 }
@@ -99,20 +111,50 @@ void AppendPathComponent(std::string& path, std::string_view component) {
   }
 }
 
+std::string_view PathRootName(std::string_view path) {
+#ifdef _WIN32
+  if (path.empty()) return {};
+  if (IsWindowsDriveLetter(path)) {
+    return path.substr(0, 2);
+  }
+
+  // Handle windows network shares, only on Windows
+  if (absl::StartsWith(path, "\\\\") || absl::StartsWith(path, "//")) {
+    size_t prefix = 2;
+    while (prefix < path.size() && path[prefix] >= 31 &&
+           !kIllegalUNCCharacters.Test(path[prefix])) {
+      ++prefix;
+    }
+    if (prefix > 2 && prefix < path.length() &&
+        (path[prefix] == '/' || path[prefix] == '\\')) {
+      return path.substr(0, prefix);
+    }
+  }
+#endif
+  return {};
+}
+
 std::string LexicalNormalizePath(std::string path) {
   if (path.empty()) return path;
 
   const char* src = path.c_str();
   auto dst = path.begin();
 
-  // Remove initial '/' characters.
-  // Assume that Windows paths do not begin with '/'.
-  const bool is_absolute_path = (*src == '/');
+  // Skip the root name if present.
+  if (auto root_name = PathRootName(path); !root_name.empty()) {
+    dst += root_name.size();
+    src += root_name.size();
+  }
+
+  // A root directory begins after the root name; if it is present,
+  // copy it and skip any duplicate separators.
+  const bool is_absolute_path = IsDirSeparator(*src);
   if (is_absolute_path) {
-    dst++;
+    *dst++ = '/';
     src++;
-    while (*src == '/') ++src;
+    while (IsDirSeparator(*src)) ++src;
   }
+
   auto limit = dst;
 
   // Process all parts
@@ -175,5 +217,12 @@ std::string LexicalNormalizePath(std::string path) {
   return path;
 }
 
+bool IsAbsolutePath(std::string_view path) {
+  if (path.empty()) return false;
+  auto root_name = PathRootName(path);
+  return path.size() > root_name.size() &&
+         IsDirSeparator(path[root_name.size()]);
+}
+
 }  // namespace internal
 }  // namespace tensorstore

tensorstore/internal/path.h

@@ -73,6 +73,17 @@ void EnsureNonDirectoryPath(std::string& path);
 /// as many relative path segments ('.' and '..') as possible.
 std::string LexicalNormalizePath(std::string path);
 
+// Returns the length of the root name. Generally this is the empty
+// string, however in Windows this may be drive-letter prefix or the
+// network share name. For example:
+//  * "/tmp/foo" -> ""
+//  * "C:\foo" -> "C:"
+//  * "\\server\share\foo" -> "\\server"
+std::string_view PathRootName(std::string_view path);
+
+/// Returns true if the path is an absolute path.
+bool IsAbsolutePath(std::string_view path);
+
 }  // namespace internal
 }  // namespace tensorstore