Introduce a utility function to transform a path into a URI (#1706)

Fixes: #1790
Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>

Author: jviotti

src/core/uri/include/sourcemeta/core/uri.h

@@ -11,6 +11,8 @@
 // NOLINTEND(misc-include-cleaner)
 
 #include <cstdint>     // std::uint32_t
+#include <filesystem>  // std::filesystem
+#include <istream>     // std::istream
 #include <memory>      // std::unique_ptr
 #include <optional>    // std::optional
 #include <span>        // std::span
@@ -439,6 +441,19 @@ class SOURCEMETA_CORE_URI_EXPORT URI {
   /// ```
   static auto from_fragment(std::string_view fragment) -> URI;
 
+  /// Create a URI from a file system path. For example:
+  ///
+  /// ```cpp
+  /// #include <sourcemeta/core/uri.h>
+  /// #include <cassert>
+  /// #include <filesystem>
+  ///
+  /// const std::filesystem::path path{"/foo/bar"};
+  /// const sourcemeta::core::URI uri{sourcemeta::core::URI::from_path(path)};
+  /// assert(uri.recompose() == "file:///foo/bar");
+  /// ```
+  static auto from_path(const std::filesystem::path &path) -> URI;
+
   /// A convenient method to canonicalize and recompose a URI from a string. For
   /// example:
   ///

src/core/uri/uri.cc

@@ -2,12 +2,12 @@
 
 #include <sourcemeta/core/uri.h>
 
+#include <algorithm>  // std::replace
 #include <cassert>    // assert
 #include <cstdint>    // std::uint32_t
 #include <filesystem> // std::filesystem
-#include <istream>    // std::istream
 #include <optional>   // std::optional
-#include <sstream>    // std::ostringstream
+#include <sstream>    // std::ostringstream, std::istringstream
 #include <stdexcept>  // std::length_error, std::runtime_error
 #include <string>     // std::stoul, std::string, std::tolower
 #include <tuple>      // std::tie
@@ -114,6 +114,20 @@ auto canonicalize_path(const std::string &path) -> std::optional<std::string> {
   return canonical_path;
 }
 
+auto uri_escape_for_path(const std::string &value) -> std::string {
+  std::istringstream input{value};
+  std::ostringstream output;
+  uri_escape(input, output, sourcemeta::core::URIEscapeMode::SkipSubDelims);
+  auto result{output.str()};
+  // We don't want to escape ":" for Windows paths
+  std::string::size_type position = 0;
+  while ((position = result.find("%3A", position)) != std::string::npos) {
+    result.replace(position, 3, ":");
+  }
+
+  return result;
+}
+
 } // namespace
 
 namespace sourcemeta::core {
@@ -708,4 +722,44 @@ auto URI::canonicalize(const std::string &input) -> std::string {
   return URI{input}.canonicalize().recompose();
 }
 
+auto URI::from_path(const std::filesystem::path &path) -> URI {
+  auto normalized{path.lexically_normal().string()};
+  const auto is_unc{normalized.starts_with("\\\\")};
+  const auto is_windows_absolute{normalized.size() >= 2 &&
+                                 normalized[1] == ':'};
+  std::replace(normalized.begin(), normalized.end(), '\\', '/');
+  const auto is_unix_absolute{normalized.starts_with("/")};
+  if (!is_unix_absolute && !is_windows_absolute && !is_unc) {
+    throw URIError(
+        "It is not valid to construct a file:// URI out of a relative path");
+  }
+
+  normalized.erase(0, normalized.find_first_not_of('/'));
+  const std::filesystem::path final_path{normalized};
+
+  URI result{"file://"};
+
+  auto iterator{final_path.begin()};
+  if (is_unc) {
+    result.host_ = uri_escape_for_path(iterator->string());
+    std::advance(iterator, 1);
+  }
+
+  for (; iterator != final_path.end(); ++iterator) {
+    if (iterator->empty()) {
+      result.append_path("/");
+    } else if (*iterator == "/") {
+      if (std::next(iterator) == final_path.end()) {
+        result.append_path("/");
+      }
+    } else if (result.path_.has_value()) {
+      result.append_path(uri_escape_for_path(iterator->string()));
+    } else {
+      result.path_ = uri_escape_for_path(iterator->string());
+    }
+  }
+
+  return result;
+}
+
 } // namespace sourcemeta::core

test/uri/CMakeLists.txt

@@ -5,6 +5,7 @@ sourcemeta_googletest(NAMESPACE sourcemeta PROJECT core NAME uri
     uri_empty_test.cc
     uri_host_test.cc
     uri_path_test.cc
+    uri_from_path_test.cc
     uri_parse_test.cc
     uri_port_test.cc
     uri_scheme_test.cc

test/uri/uri_from_path_test.cc

@@ -0,0 +1,122 @@
+#include <gtest/gtest.h>
+
+#include <sourcemeta/core/uri.h>
+
+TEST(URI_from_path, unix_absolute) {
+  const std::filesystem::path example{"/foo/bar/baz"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///foo/bar/baz");
+}
+
+TEST(URI_from_path, unix_with_space_and_reserved) {
+  const std::filesystem::path example{"/foo/My Folder/has#hash?value%"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///foo/My%20Folder/has%23hash%3Fvalue%25");
+}
+
+TEST(URI_from_path, unix_trailing_slash) {
+  const std::filesystem::path example{"/foo/bar/"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///foo/bar/");
+}
+
+TEST(URI_from_path, windows_drive_absolute) {
+  const std::filesystem::path example{R"(C:\Program Files\Test)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///C:/Program%20Files/Test");
+}
+
+TEST(URI_from_path, windows_drive_lowercase) {
+  const std::filesystem::path example{R"(c:\temp\logs)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///c:/temp/logs");
+}
+
+TEST(URI_from_path, windows_drive_root) {
+  const std::filesystem::path example{R"(D:\)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///D:/");
+}
+
+TEST(URI_from_path, windows_trailing_slash) {
+  const std::filesystem::path example{R"(C:\foo\bar\)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///C:/foo/bar/");
+}
+
+TEST(URI_from_path, windows_percent_and_plus) {
+  // '%' → %25, '+' is allowed unencoded
+  const std::filesystem::path example{R"(C:\path\50%+plus.txt)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///C:/path/50%25+plus.txt");
+}
+
+TEST(URI_from_path, windows_unc_simple) {
+  const std::filesystem::path example{R"(\\server\share\file.txt)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(),
+            // For UNC, host=server, path=/share/file.txt
+            "file://server/share/file.txt");
+}
+
+TEST(URI_from_path, windows_unc_with_space) {
+  const std::filesystem::path example{R"(\\srv\My Docs\a b.txt)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file://srv/My%20Docs/a%20b.txt");
+}
+
+TEST(URI_from_path, unicode_unix) {
+  // U+00E9 (é) should be UTF-8 percent-encoded as %C3%A9
+  const std::filesystem::path example{u8"/data/éclair.txt"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///data/%C3%A9clair.txt");
+}
+
+TEST(URI_from_path, unicode_windows) {
+  // U+00E9 (é) should be UTF-8 percent-encoded as %C3%A9
+  const std::filesystem::path example{u8R"(C:\data\résumé.doc)"};
+  const auto uri{sourcemeta::core::URI::from_path(example)};
+  EXPECT_EQ(uri.recompose(), "file:///C:/data/r%C3%A9sum%C3%A9.doc");
+}
+
+TEST(URI_from_path, unix_relative_simple) {
+  const std::filesystem::path example{"foo/bar/baz"};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}
+
+TEST(URI_from_path, unix_relative_with_dot) {
+  const std::filesystem::path example{"./foo/bar"};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}
+
+TEST(URI_from_path, unix_relative_with_dotdot) {
+  const std::filesystem::path example{"../parent/dir"};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}
+
+TEST(URI_from_path, unix_empty_path) {
+  const std::filesystem::path example{""};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}
+
+TEST(URI_from_path, windows_relative_simple) {
+  const std::filesystem::path example{"folder\\file.txt"};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}
+
+TEST(URI_from_path, windows_relative_with_dot) {
+  const std::filesystem::path example{".\\foo\\bar"};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}
+
+TEST(URI_from_path, windows_relative_with_dotdot) {
+  const std::filesystem::path example{"..\\up\\one\\level"};
+  EXPECT_THROW(sourcemeta::core::URI::from_path(example),
+               sourcemeta::core::URIError);
+}