From a872890d874c25a2729aca4c1e3ffb301a33e259 Mon Sep 17 00:00:00 2001
From: Thomas Newton <thomas.w.newton@gmail.com>
Date: Sun, 15 Oct 2023 14:58:38 +0100
Subject: [PATCH] Tidy path validation

---
 cpp/src/arrow/filesystem/azurefs.cc | 32 +++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 8 deletions(-)

diff --git a/cpp/src/arrow/filesystem/azurefs.cc b/cpp/src/arrow/filesystem/azurefs.cc
index baad308af557d..64dfb321e4d53 100644
--- a/cpp/src/arrow/filesystem/azurefs.cc
+++ b/cpp/src/arrow/filesystem/azurefs.cc
@@ -82,7 +82,8 @@ struct AzurePath {
     // http://127.0.0.1/accountName/pathToBlob
     if (internal::IsLikelyUri(s)) {
       return Status::Invalid(
-          "Expected an Azure object path of the form 'container/key...', got a URI: '", s, "'");
+          "Expected an Azure object path of the form 'container/key...', got a URI: '", s,
+          "'");
     }
     auto src = internal::RemoveTrailingSlash(s);
     if (arrow::internal::StartsWith(src, "https://127.0.0.1") ||
@@ -163,6 +164,25 @@ struct AzurePath {
   }
 };
 
+Status PathNotFound(const AzurePath& path) {
+  return ::arrow::fs::internal::PathNotFound(path.full_path);
+}
+
+Status NotAFile(const AzurePath& path) {
+  return ::arrow::fs::internal::NotAFile(path.full_path);
+}
+
+Status ValidateFilePath(const AzurePath& path) {
+  if (path.container.empty()) {
+    return PathNotFound(path);
+  }
+
+  if (path.container.empty() || path.path_to_file.empty()) {
+    return NotAFile(path);
+  }
+  return Status::OK();
+}
+
 template <typename ObjectResult>
 std::shared_ptr<const KeyValueMetadata> GetObjectMetadata(const ObjectResult& result) {
   auto md = std::make_shared<KeyValueMetadata>();
@@ -204,7 +224,7 @@ class ObjectInputFile final : public io::RandomAccessFile {
       return Status::OK();
     } catch (const Azure::Storage::StorageException& exception) {
       if (exception.StatusCode == Azure::Core::Http::HttpStatusCode::NotFound) {
-        // Could be either container or blob not found. 
+        // Could be either container or blob not found.
         return ::arrow::fs::internal::PathNotFound(path_.full_path);
       }
       return Status::IOError(exception.RawResponse->GetReasonPhrase());
@@ -346,12 +366,7 @@ class AzureFileSystem::Impl {
   Result<std::shared_ptr<ObjectInputFile>> OpenInputFile(const std::string& s,
                                                          AzureFileSystem* fs) {
     ARROW_ASSIGN_OR_RAISE(auto path, AzurePath::FromString(s));
-
-    // TODO: Return NotAFile if path.path_to_file is empty. Return PathNotFound if account
-    // name or container name is empty.
-    if (path.empty()) {
-      return ::arrow::fs::internal::PathNotFound(path.full_path);
-    }
+    ValidateFilePath(path);
 
     auto blob_client = std::make_shared<Azure::Storage::Blobs::BlobClient>(
         std::move(service_client_->GetBlobContainerClient(path.container)
@@ -372,6 +387,7 @@ class AzureFileSystem::Impl {
     }
 
     ARROW_ASSIGN_OR_RAISE(auto path, AzurePath::FromString(info.path()));
+    ValidateFilePath(path);
 
     auto blob_client = std::make_shared<Azure::Storage::Blobs::BlobClient>(
         std::move(service_client_->GetBlobContainerClient(path.container)