Read files in bulk and report output stream failures

benchmarks/README.md

@@ -38,6 +38,16 @@ to only run some of the benchmarks:
 ./build/benchmarks/all/reflect-cpp-all-format-benchmarks --benchmark_filter=canada_read
 ```
 
+The `BM_load_file` cases measure the shared binary and text file loaders with
+1 KiB, 1 MiB, and 32 MiB inputs. Fixture creation and content checks happen before
+timing; timed work includes opening, reading, closing, and destroying the result.
+The files are freshly written and read once for validation, so these measure
+loading with a warm filesystem cache, not cold-storage throughput.
+
+```
+./build/benchmarks/all/reflect-cpp-all-format-benchmarks --benchmark_filter=BM_load_file
+```
+
 ## Results summary
 
 ### How reflect-cpp compares to other JSON libraries

benchmarks/all/file_io.cpp

@@ -0,0 +1,85 @@
+#include <benchmark/benchmark.h>
+
+#include <chrono>
+#include <filesystem>
+#include <fstream>
+#include <rfl/io/load_bytes.hpp>
+#include <rfl/io/load_string.hpp>
+#include <stdexcept>
+#include <string>
+
+namespace file_io {
+
+class InputFile {
+ public:
+  explicit InputFile(std::size_t size) {
+    const auto stamp =
+        std::chrono::steady_clock::now().time_since_epoch().count();
+    for (unsigned i = 0; i < 100; ++i) {
+      dir_ = std::filesystem::temp_directory_path() /
+             ("reflectcpp-io-bench-" + std::to_string(stamp) + "-" +
+              std::to_string(i));
+      if (std::filesystem::create_directory(dir_)) {
+        name_ = (dir_ / "input").string();
+        std::ofstream out;
+        out.exceptions(std::ios::badbit | std::ios::failbit);
+        out.open(name_, std::ios::binary);
+        const std::string data(size, 'x');
+        out.write(data.data(), static_cast<std::streamsize>(data.size()));
+        out.close();
+        return;
+      }
+    }
+    throw std::runtime_error("Cannot create benchmark directory");
+  }
+  ~InputFile() {
+    std::error_code ec;
+    std::filesystem::remove_all(dir_, ec);
+  }
+  const std::string& name() const { return name_; }
+
+ private:
+  std::filesystem::path dir_;
+  std::string name_;
+};
+
+template <bool Binary>
+void BM_load_file(benchmark::State& state) {
+  const auto size = static_cast<std::size_t>(state.range(0));
+  const InputFile file(size);
+  const auto load = [&] {
+    if constexpr (Binary)
+      return rfl::io::load_bytes(file.name());
+    else
+      return rfl::io::load_string(file.name());
+  };
+  {
+    const auto result = load();
+    if (!result || result->size() != size) {
+      state.SkipWithError("File loading failed");
+      return;
+    }
+    for (const char c : *result) {
+      if (c != 'x') {
+        state.SkipWithError("File contents changed");
+        return;
+      }
+    }
+  }
+  for (auto _ : state) {
+    auto result = load();
+    benchmark::DoNotOptimize(result);
+  }
+  state.SetBytesProcessed(state.iterations() * state.range(0));
+}
+
+BENCHMARK_TEMPLATE(BM_load_file, true)
+    ->Arg(1024)
+    ->Arg(1024 * 1024)
+    ->Arg(32 * 1024 * 1024);
+BENCHMARK_TEMPLATE(BM_load_file, false)
+    ->Arg(1024)
+    ->Arg(1024 * 1024)
+    ->Arg(32 * 1024 * 1024);
+
+}  // namespace file_io

include/rfl/io/load_bytes.hpp

@@ -1,26 +1,17 @@
 #ifndef RFL_IO_LOAD_BYTES_HPP_
 #define RFL_IO_LOAD_BYTES_HPP_
 
-#include <fstream>
-#include <iostream>
 #include <string>
 #include <vector>
 
 #include "../Result.hpp"
+#include "load_file.hpp"
 
 namespace rfl {
 namespace io {
 
 inline Result<std::vector<char>> load_bytes(const std::string& _fname) {
-  std::ifstream input(_fname, std::ios::binary);
-  if (input.is_open()) {
-    std::istreambuf_iterator<char> begin(input), end;
-    const auto bytes = std::vector<char>(begin, end);
-    input.close();
-    return bytes;
-  } else {
-    return error("File '" + _fname + "' not found!");
-  }
+  return internal::load_file<std::vector<char>>(_fname, std::ios::binary);
 }
 
 }  // namespace io

include/rfl/io/load_file.hpp

@@ -0,0 +1,66 @@
+#ifndef RFL_IO_LOAD_FILE_HPP_
+#define RFL_IO_LOAD_FILE_HPP_
+
+#include <array>
+#include <cstdint>
+#include <exception>
+#include <filesystem>
+#include <fstream>
+#include <limits>
+#include <string>
+#include <system_error>
+
+#include "../Result.hpp"
+
+namespace rfl::io::internal {
+
+template <class Container>
+Result<Container> load_file(const std::string& _fname,
+                            const std::ios::openmode _mode) {
+  try {
+    std::ifstream input(_fname, _mode);
+    if (!input.is_open()) {
+      return error("Unable to open file '" + _fname + "' for reading.");
+    }
+    Container bytes;
+    std::error_code ec;
+    const auto size = std::filesystem::file_size(_fname, ec);
+    if (!ec) {
+      if (size > bytes.max_size() ||
+          size > static_cast<std::uintmax_t>(
+                     (std::numeric_limits<std::streamsize>::max)())) {
+        return error("File '" + _fname + "' is too large to read.");
+      }
+      bytes.resize(static_cast<std::size_t>(size));
+      if (!bytes.empty()) {
+        input.read(bytes.data(), static_cast<std::streamsize>(bytes.size()));
+        bytes.resize(static_cast<std::size_t>(input.gcount()));
+        if (input.bad() || (input.fail() && !input.eof())) {
+          return error("Unable to read file '" + _fname + "'.");
+        }
+        if (input.eof()) {
+          return bytes;
+        }
+      }
+    } else if (std::filesystem::is_directory(_fname, ec)) {
+      return error("Cannot read directory '" + _fname + "' as a file.");
+    }
+    // The size is only a hint: text conversion, growing files, and special
+    // files can produce a different number of bytes. Never require seeking.
+    std::array<char, 64 * 1024> buffer;
+    do {
+      input.read(buffer.data(), static_cast<std::streamsize>(buffer.size()));
+      if (input.bad() || (input.fail() && !input.eof())) {
+        return error("Unable to read file '" + _fname + "'.");
+      }
+      bytes.insert(bytes.end(), buffer.data(), buffer.data() + input.gcount());
+    } while (!input.eof());
+    return bytes;
+  } catch (const std::exception& e) {
+    return error("Unable to read file '" + _fname + "': " + e.what());
+  }
+}
+
+}  // namespace rfl::io::internal
+
+#endif

include/rfl/io/load_string.hpp

@@ -1,25 +1,16 @@
 #ifndef RFL_IO_LOAD_STRING_HPP_
 #define RFL_IO_LOAD_STRING_HPP_
 
-#include <fstream>
 #include <string>
 
 #include "../Result.hpp"
+#include "load_file.hpp"
 
 namespace rfl {
 namespace io {
 
 inline Result<std::string> load_string(const std::string& _fname) {
-  std::ifstream infile(_fname);
-  if (infile.is_open()) {
-    auto r = std::string(std::istreambuf_iterator<char>(infile),
-                         std::istreambuf_iterator<char>());
-    infile.close();
-    return r;
-  } else {
-    return error("Unable to open file '" + _fname +
-                 "' or file could not be found.");
-  }
+  return internal::load_file<std::string>(_fname, std::ios::in);
 }
 
 }  // namespace io

include/rfl/io/save_bytes.hpp

@@ -2,7 +2,6 @@
 #define RFL_IO_SAVE_BYTES_HPP_
 
 #include <fstream>
-#include <iostream>
 #include <string>
 
 #include "../Result.hpp"
@@ -20,7 +19,10 @@ Result<Nothing> save_bytes(const std::string& _fname, const T& _obj,
     }
     _write(_obj, output);
     output.close();
-  } catch (std::exception& e) {
+    if (!output) {
+      return error("Unable to write or close file '" + _fname + "'.");
+    }
+  } catch (const std::exception& e) {
     return error(e.what());
   }
   return Nothing{};

include/rfl/io/save_string.hpp

@@ -20,7 +20,10 @@ Result<Nothing> save_string(const std::string& _fname, const T& _obj,
     }
     _write(_obj, outfile);
     outfile.close();
-  } catch (std::exception& e) {
+    if (!outfile) {
+      return error("Unable to write or close file '" + _fname + "'.");
+    }
+  } catch (const std::exception& e) {
     return error(e.what());
   }
   return Nothing{};