Safer weakly_canonical and canonical improved implementations (#1933)

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>

Author: jviotti

src/lang/io/CMakeLists.txt

@@ -1,4 +1,4 @@
-sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME io)
+sourcemeta_library(NAMESPACE sourcemeta PROJECT core NAME io SOURCES io.cc)
 
 if(SOURCEMETA_CORE_INSTALL)
   sourcemeta_library_install(NAMESPACE sourcemeta PROJECT core NAME io)

src/lang/io/include/sourcemeta/core/io.h

@@ -1,6 +1,10 @@
 #ifndef SOURCEMETA_CORE_IO_H_
 #define SOURCEMETA_CORE_IO_H_
 
+#ifndef SOURCEMETA_CORE_IO_EXPORT
+#include <sourcemeta/core/io_export.h>
+#endif
+
 #include <cassert>    // assert
 #include <filesystem> // std::filesystem
 #include <fstream>    // std::basic_ifstream
@@ -17,6 +21,37 @@
 
 namespace sourcemeta::core {
 
+/// @ingroup io
+///
+/// A safe variant of `std::filesystem::canonical` that takes into account
+/// platform-specific oddities like FIFO on GNU/Linux. For example:
+///
+/// ```cpp
+/// #include <sourcemeta/core/io.h>
+/// #include <cassert>
+///
+/// const auto output{sourcemeta::core::canonical("/tmp/../foo.json")};
+/// assert(output == "/foo.json");
+/// ```
+SOURCEMETA_CORE_IO_EXPORT
+auto canonical(const std::filesystem::path &path) -> std::filesystem::path;
+
+/// @ingroup io
+///
+/// A safe variant of `std::filesystem::weakly_canonical` that takes into
+/// account platform-specific oddities like FIFO on GNU/Linux. For example:
+///
+/// ```cpp
+/// #include <sourcemeta/core/io.h>
+/// #include <cassert>
+///
+/// const auto output{sourcemeta::core::weakly_canonical("/tmp/../foo.json")};
+/// assert(output == "/foo.json");
+/// ```
+SOURCEMETA_CORE_IO_EXPORT
+auto weakly_canonical(const std::filesystem::path &path)
+    -> std::filesystem::path;
+
 /// @ingroup io
 ///
 /// A convenience function to open a stream from a file. For example:
@@ -37,12 +72,7 @@ auto read_file(const std::filesystem::path &path)
         std::make_error_code(std::errc::is_a_directory));
   }
 
-  std::ifstream stream{
-      // On Linux, FIFO files (like /dev/fd/XX due to process substitution)
-      // cannot be
-      // made canonical
-      // See https://github.com/sourcemeta/jsonschema/issues/252
-      std::filesystem::is_fifo(path) ? path : std::filesystem::canonical(path)};
+  std::ifstream stream{sourcemeta::core::canonical(path)};
   stream.exceptions(std::ifstream::badbit);
   assert(!stream.fail());
   assert(stream.is_open());

src/lang/io/io.cc

@@ -0,0 +1,23 @@
+#include <sourcemeta/core/io.h>
+
+namespace sourcemeta::core {
+
+auto canonical(const std::filesystem::path &path) -> std::filesystem::path {
+  // On Linux, FIFO files (like /dev/fd/XX due to process substitution)
+  // cannot be made canonical
+  // See https://github.com/sourcemeta/jsonschema/issues/252
+  return std::filesystem::is_fifo(path) ? path
+                                        : std::filesystem::canonical(path);
+}
+
+auto weakly_canonical(const std::filesystem::path &path)
+    -> std::filesystem::path {
+  // On Linux, FIFO files (like /dev/fd/XX due to process substitution)
+  // cannot be made canonical
+  // See https://github.com/sourcemeta/jsonschema/issues/252
+  return std::filesystem::is_fifo(path)
+             ? path
+             : std::filesystem::weakly_canonical(path);
+}
+
+} // namespace sourcemeta::core

test/io/CMakeLists.txt

@@ -1,5 +1,5 @@
 sourcemeta_googletest(NAMESPACE sourcemeta PROJECT core NAME io
-  SOURCES io_read_file_test.cc)
+  SOURCES io_canonical_test.cc io_weakly_canonical_test.cc io_read_file_test.cc)
 
 target_link_libraries(sourcemeta_core_io_unit
   PRIVATE sourcemeta::core::io)

test/io/io_canonical_test.cc

@@ -0,0 +1,15 @@
+#include <gtest/gtest.h>
+
+#include <sourcemeta/core/io.h>
+
+TEST(IO_canonical, test_txt) {
+  const auto path{sourcemeta::core::canonical(
+      std::filesystem::path{TEST_DIRECTORY} / ".." / "io" / "test.txt")};
+  EXPECT_EQ(path, std::filesystem::path{TEST_DIRECTORY} / "test.txt");
+}
+
+TEST(IO_canonical, not_exists) {
+  EXPECT_THROW(sourcemeta::core::canonical(
+                   std::filesystem::path{TEST_DIRECTORY} / "foo.txt"),
+               std::filesystem::filesystem_error);
+}

test/io/io_weakly_canonical_test.cc

@@ -0,0 +1,15 @@
+#include <gtest/gtest.h>
+
+#include <sourcemeta/core/io.h>
+
+TEST(IO_weakly_canonical, test_txt) {
+  const auto path{sourcemeta::core::weakly_canonical(
+      std::filesystem::path{TEST_DIRECTORY} / ".." / "io" / "test.txt")};
+  EXPECT_EQ(path, std::filesystem::path{TEST_DIRECTORY} / "test.txt");
+}
+
+TEST(IO_weakly_canonical, not_exists) {
+  const auto path{sourcemeta::core::weakly_canonical(
+      std::filesystem::path{TEST_DIRECTORY} / "foo.txt")};
+  EXPECT_EQ(path, std::filesystem::path{TEST_DIRECTORY} / "foo.txt");
+}