is_supported_remote_url

cpp/src/io/utilities/datasource.cpp

@@ -33,6 +33,7 @@
 #include <sys/mman.h>
 #include <unistd.h>
 
+#include <regex>
 #include <unordered_map>
 #include <vector>
 
@@ -452,6 +453,16 @@ class remote_file_source : public datasource {
     return datasource::buffer::create(std::move(h_data));
   }
 
+  /**
+   * @brief Is `url` referring to a remote file using a protocol KvikIO support?
+   */
+  static bool is_supported_remote_url(std::string const& url)
+  {
+    // Regular expression to match "<s3|http|https>://"
+    std::regex pattern{R"(^s3|http|https://)", std::regex_constants::icase};
+    return std::regex_search(url, pattern);
+  }
+
  private:
   kvikio::RemoteHandle _kvikio_file;
 };
@@ -470,14 +481,11 @@ std::unique_ptr<datasource> datasource::create(std::string const& filepath,
 
     CUDF_FAIL("Invalid LIBCUDF_MMAP_ENABLED value: " + policy);
   }();
-
-  if (use_memory_mapping) {
+  if (remote_file_source::is_supported_remote_url(filepath)) {
+    return std::make_unique<remote_file_source>(filepath.c_str());
+  } else if (use_memory_mapping) {
     return std::make_unique<memory_mapped_source>(filepath.c_str(), offset, max_size_estimate);
   } else {
-    // If this is a S3 filepath (i.e. "s3://<bucket>/<object>"), we create a remote file source
-    if (filepath.size() > 5 && filepath.substr(0, 5) == "s3://") {
-      return std::make_unique<remote_file_source>(filepath.c_str());
-    }
     // `file_source` reads the file directly, without memory mapping
     return std::make_unique<file_source>(filepath.c_str());
   }