reproduce and fix urlpattern bug(s) (#992)

Author: anonrig

src/url_pattern.cpp

@@ -341,7 +341,11 @@ tl::expected<std::string, errors> url_pattern_init::process_search(
   if (value.starts_with("?")) {
     value.remove_prefix(1);
   }
-  ADA_ASSERT_TRUE(!value.starts_with("?"));
+  // We cannot assert that the value is no longer starting with a single
+  // question mark because technically it can start. The question is whether or
+  // not we should remove the first question mark. Ref:
+  // https://github.com/ada-url/ada/pull/992 The spec is not clear on this.
+
   // If type is "pattern" then return strippedValue.
   if (type == process_type::pattern) {
     return std::string(value);

src/url_pattern_helpers.cpp

@@ -410,7 +410,10 @@ tl::expected<std::string, errors> canonicalize_search(std::string_view input) {
   url->set_search(input);
   if (url->has_search()) {
     const auto search = url->get_search();
-    return std::string(search.substr(1));
+    if (!search.empty()) {
+      return std::string(search.substr(1));
+    }
+    return "";
   }
   return tl::unexpected(errors::type_error);
 }
@@ -430,7 +433,10 @@ tl::expected<std::string, errors> canonicalize_hash(std::string_view input) {
   // Return dummyURL's fragment.
   if (url->has_hash()) {
     const auto hash = url->get_hash();
-    return std::string(hash.substr(1));
+    if (!hash.empty()) {
+      return std::string(hash.substr(1));
+    }
+    return "";
   }
   return tl::unexpected(errors::type_error);
 }

tests/basic_tests.cpp

@@ -524,6 +524,7 @@ TYPED_TEST(basic_tests, test_issue_970) {
   SUCCEED();
 }
 
+// Ref: https://github.com/cloudflare/workerd/issues/5144
 TYPED_TEST(basic_tests, test_workerd_issue_5144_1) {
   auto url = ada::parse<TypeParam>("https://example.sub.com/??");
   ASSERT_TRUE(url);
@@ -533,6 +534,7 @@ TYPED_TEST(basic_tests, test_workerd_issue_5144_1) {
   SUCCEED();
 }
 
+// Ref: https://github.com/cloudflare/workerd/issues/5144
 TYPED_TEST(basic_tests, test_workerd_issue_5144_2) {
   auto url = ada::parse<TypeParam>("https://example.sub.com/???");
   ASSERT_TRUE(url);
@@ -541,6 +543,7 @@ TYPED_TEST(basic_tests, test_workerd_issue_5144_2) {
   SUCCEED();
 }
 
+// Ref: https://github.com/cloudflare/workerd/issues/5144
 TYPED_TEST(basic_tests, test_workerd_issue_5144_3) {
   auto url = ada::parse<TypeParam>("https://example.sub.com/????");
   ASSERT_TRUE(url);
@@ -549,6 +552,7 @@ TYPED_TEST(basic_tests, test_workerd_issue_5144_3) {
   SUCCEED();
 }
 
+// Ref: https://github.com/cloudflare/workerd/issues/5144
 TYPED_TEST(basic_tests, test_workerd_issue_5144_4) {
   using regex_provider = ada::url_pattern_regex::std_regex_provider;
   auto init = ada::url_pattern_init{};
@@ -559,5 +563,8 @@ TYPED_TEST(basic_tests, test_workerd_issue_5144_4) {
   ASSERT_TRUE(pattern->match("https://example.com/?"));
   ASSERT_TRUE(pattern->match("https://example.com/??"));
 
+  auto dummy_init = ada::url_pattern_init{};
+  dummy_init.search = "???";
+  ASSERT_TRUE(pattern->exec(std::move(dummy_init)));
   SUCCEED();
 }