From f30abe29f0072ce2fa9acbef1c8ab998f5dc7723 Mon Sep 17 00:00:00 2001
From: Yagiz Nizipli <yagiz@cloudflare.com>
Date: Fri, 24 Jan 2025 16:01:25 -0500
Subject: [PATCH] update url_pattern to pass more wpt on node.js

---
 include/ada/implementation.h   |  2 --
 include/ada/url_pattern-inl.h  |  5 -----
 src/url_pattern_regex.cpp      |  3 ++-
 tests/wpt_urlpattern_tests.cpp | 26 ++++++++++++++++++++++++++
 4 files changed, 28 insertions(+), 8 deletions(-)

diff --git a/include/ada/implementation.h b/include/ada/implementation.h
index 33e166715..e79670031 100644
--- a/include/ada/implementation.h
+++ b/include/ada/implementation.h
@@ -53,8 +53,6 @@ bool can_parse(std::string_view input,
  * @param input valid UTF-8 string or URLPatternInit struct
  * @param base_url an optional valid UTF-8 string
  * @param options an optional url_pattern_options struct
- * @param provider an optional regex provider. if not provided, it will
- * use ada::url_pattern_regex::std_regex_provider
  * @return url_pattern instance
  */
 template <url_pattern_regex::regex_concept regex_provider =
diff --git a/include/ada/url_pattern-inl.h b/include/ada/url_pattern-inl.h
index 1823d4ac8..d12ed55e4 100644
--- a/include/ada/url_pattern-inl.h
+++ b/include/ada/url_pattern-inl.h
@@ -47,11 +47,6 @@ url_pattern_component<regex_provider>::create_component_match_result(
   auto result =
       url_pattern_component_result{.input = std::string(input), .groups = {}};
 
-  // If input is empty, then groups will always be empty.
-  if (input.empty()) {
-    return result;
-  }
-
   // Optimization: Let's reserve the size.
   result.groups.reserve(exec_result.size());
 
diff --git a/src/url_pattern_regex.cpp b/src/url_pattern_regex.cpp
index ec62edf2c..59cac71b9 100644
--- a/src/url_pattern_regex.cpp
+++ b/src/url_pattern_regex.cpp
@@ -31,7 +31,8 @@ std_regex_provider::regex_search(std::string_view input,
     return std::nullopt;
   }
   std::vector<std::optional<std::string>> matches;
-  if (match_result.empty()) {
+  // If input is empty, let's assume the result will be empty as well.
+  if (input.empty() || match_result.empty()) {
     return matches;
   }
   matches.reserve(match_result.size());
diff --git a/tests/wpt_urlpattern_tests.cpp b/tests/wpt_urlpattern_tests.cpp
index 5f58434ce..93b694aee 100644
--- a/tests/wpt_urlpattern_tests.cpp
+++ b/tests/wpt_urlpattern_tests.cpp
@@ -14,6 +14,32 @@ using namespace simdjson;
 constexpr std::string_view URL_PATTERN_TEST_DATA =
     "wpt/urlpatterntestdata.json";
 
+TEST(wpt_urlpattern_tests, test_regex_difference) {
+  // {
+  //   "pattern": [{ "pathname": "/foo/bar" }],
+  //   "inputs": [{ "pathname": "/foo/bar" }],
+  //   "expected_match": {
+  //     "pathname": { "input": "/foo/bar", "groups": {} }
+  //   }
+  // }
+  auto init = ada::url_pattern_init{};
+  init.pathname = "/foo/bar";
+  auto u =
+      ada::parse_url_pattern<ada::url_pattern_regex::std_regex_provider>(init);
+  ASSERT_TRUE(u);
+  auto match = u->exec(init, nullptr);
+  ASSERT_TRUE(match);
+  ASSERT_TRUE(match->has_value());
+
+  std::unordered_map<std::string, std::optional<std::string>> empty_object{};
+
+  ASSERT_EQ(match->value().protocol.input, "");
+  ASSERT_EQ(match->value().protocol.groups, empty_object);
+  ASSERT_EQ(match->value().pathname.input, "/foo/bar");
+  ASSERT_EQ(match->value().pathname.groups, empty_object);
+  SUCCEED();
+}
+
 TEST(wpt_urlpattern_tests, parser_tokenize_basic_tests) {
   auto tokenize_result =
       tokenize("*", ada::url_pattern_helpers::token_policy::STRICT);