CommonPathPrefix utility: return the largest common subpath

PiperOrigin-RevId: 344193277
Change-Id: I262cf8f44c3d63bf8eaa7db6818d7c204c2b82ee

tensorflow/core/lib/io/path_test.cc

@@ -141,5 +141,14 @@ TEST(PathTest, CreateParseURI) {
 }
 #undef EXPECT_PARSE_URI
 
+TEST(PathTest, CommonPathPrefix) {
+  EXPECT_EQ(CommonPathPrefix({"/alpha/beta/c", "/alpha/beta/g"}),
+            "/alpha/beta/");
+  EXPECT_EQ(CommonPathPrefix({"/a/b/c", "/a/beta/gamma"}), "/a/");
+  EXPECT_EQ(CommonPathPrefix({}), "");
+  EXPECT_EQ(CommonPathPrefix({"/a/b/c", "", "/a/b/"}), "");
+  EXPECT_EQ(CommonPathPrefix({"alpha", "alphabeta"}), "");
+}
+
 }  // namespace io
 }  // namespace tensorflow

tensorflow/core/platform/BUILD

@@ -450,6 +450,7 @@ cc_library(
         ":strcat",
         ":stringpiece",
         ":types",
+        "@com_google_absl//absl/algorithm:container",
     ],
     alwayslink = True,
 )

tensorflow/core/platform/path.cc

@@ -28,6 +28,7 @@ limitations under the License.
 
 #include <vector>
 
+#include "absl/algorithm/container.h"
 #include "tensorflow/core/platform/logging.h"
 #include "tensorflow/core/platform/mutex.h"
 #include "tensorflow/core/platform/scanner.h"
@@ -273,6 +274,35 @@ int64 UniqueId() {
   return ++id;
 }
 
+string CommonPathPrefix(absl::Span<const string> paths) {
+  if (paths.empty()) return "";
+  size_t min_filename_size =
+      absl::c_min_element(paths, [](const string& a, const string& b) {
+        return a.size() < b.size();
+      })->size();
+  if (min_filename_size == 0) return "";
+
+  size_t common_prefix_size = [&] {
+    for (size_t prefix_size = 0; prefix_size < min_filename_size;
+         prefix_size++) {
+      char c = paths[0][prefix_size];
+      for (int f = 1; f < paths.size(); f++) {
+        if (paths[f][prefix_size] != c) {
+          return prefix_size;
+        }
+      }
+    }
+    return min_filename_size;
+  }();
+
+  size_t rpos = absl::string_view(paths[0])
+                    .substr(0, common_prefix_size)
+                    .rfind(internal::kPathSep);
+  return rpos == std::string::npos
+             ? ""
+             : std::string(absl::string_view(paths[0]).substr(0, rpos + 1));
+}
+
 string GetTempFilename(const string& extension) {
 #if defined(__ANDROID__)
   LOG(FATAL) << "GetTempFilename is not implemented in this platform.";

tensorflow/core/platform/path.h

@@ -64,6 +64,15 @@ tensorflow::StringPiece Basename(tensorflow::StringPiece path);
 // there is no "." in the basename, the result is empty.
 tensorflow::StringPiece Extension(tensorflow::StringPiece path);
 
+// Returns the largest common subpath of `paths`.
+//
+// For example, for "/alpha/beta/gamma" and "/alpha/beta/ga" returns
+// "/alpha/beta/". For "/alpha/beta/gamma" and "/alpha/beta/gamma" returns
+// "/alpha/beta/".
+//
+// Does not perform any path normalization.
+string CommonPathPrefix(absl::Span<string const> paths);
+
 // Collapse duplicate "/"s, resolve ".." and "." path elements, remove
 // trailing "/".
 //