Fix for node issue 47889 (#368)

Author: lemire

include/ada/url-inl.h

@@ -96,9 +96,7 @@ size_t url::get_pathname_length() const noexcept { return path.size(); }
 
   out.pathname_start = uint32_t(running_index);
 
-  if (!path.empty()) {
-    running_index += path.size();
-  }
+  running_index += path.size();
 
   if (query.has_value()) {
     out.search_start = uint32_t(running_index);

src/helpers.cpp

@@ -114,8 +114,9 @@ ada_really_inline bool shorten_path(std::string& path,
   }
 
   // Remove path’s last item, if any.
-  if (!path.empty()) {
-    path.erase(path.rfind('/'));
+  size_t last_delimiter = path.rfind('/');
+  if (last_delimiter != std::string::npos) {
+    path.erase(last_delimiter);
     return true;
   }
 
@@ -142,8 +143,8 @@ ada_really_inline bool shorten_path(std::string_view& path,
     size_t slash_loc = path.rfind('/');
     if (slash_loc != std::string_view::npos) {
       path.remove_suffix(path.size() - slash_loc);
+      return true;
     }
-    return true;
   }
 
   return false;
@@ -507,8 +508,9 @@ ada_really_inline void parse_prepared_path(std::string_view input,
             input.substr(previous_location, new_location - previous_location);
         previous_location = new_location + 1;
         if (path_view == "..") {
-          if (!path.empty()) {
-            path.erase(path.rfind('/'));
+          size_t last_delimiter = path.rfind('/');
+          if (last_delimiter != std::string::npos) {
+            path.erase(last_delimiter);
           }
         } else if (path_view != ".") {
           path += '/';
@@ -539,8 +541,8 @@ ada_really_inline void parse_prepared_path(std::string_view input,
               ? path_buffer_tmp
               : path_view;
       if (unicode::is_double_dot_path_segment(path_buffer)) {
-        helpers::shorten_path(path, type);
-        if (location == std::string_view::npos) {
+        if ((helpers::shorten_path(path, type) || special) &&
+            location == std::string_view::npos) {
           path += '/';
         }
       } else if (unicode::is_single_dot_path_segment(path_buffer) &&

src/parser.cpp

@@ -421,6 +421,8 @@ result_type parse_url(std::string_view user_input,
             url.password = base_url->password;
             url.host = base_url->host;
             url.port = base_url->port;
+            // cloning the base path includes cloning the has_opaque_path flag
+            url.has_opaque_path = base_url->has_opaque_path;
             url.path = base_url->path;
             url.query = base_url->query;
           } else {
@@ -430,6 +432,8 @@ result_type parse_url(std::string_view user_input,
             // update_base_hostname
             url.set_hostname(base_url->get_hostname());
             url.update_base_port(base_url->retrieve_base_port());
+            // cloning the base path includes cloning the has_opaque_path flag
+            url.has_opaque_path = base_url->has_opaque_path;
             url.update_base_pathname(base_url->get_pathname());
             url.update_base_search(base_url->get_search());
           }
@@ -861,7 +865,6 @@ result_type parse_url(std::string_view user_input,
           else if (input_position != input_size) {
             // Set url’s query to null.
             url.clear_search();
-
             // If the code point substring from pointer to the end of input does
             // not start with a Windows drive letter, then shorten url’s path.
             if (!checkers::is_windows_drive_letter(file_view)) {

src/url_aggregator.cpp

@@ -1653,8 +1653,9 @@ inline void url_aggregator::consume_prepared_path(std::string_view input) {
             input.substr(previous_location, new_location - previous_location);
         previous_location = new_location + 1;
         if (path_view == "..") {
-          if (!path.empty()) {
-            path.erase(path.rfind('/'));
+          size_t last_delimiter = path.rfind('/');
+          if (last_delimiter != std::string::npos) {
+            path.erase(last_delimiter);
           }
         } else if (path_view != ".") {
           path += '/';
@@ -1685,8 +1686,8 @@ inline void url_aggregator::consume_prepared_path(std::string_view input) {
               ? path_buffer_tmp
               : path_view;
       if (unicode::is_double_dot_path_segment(path_buffer)) {
-        helpers::shorten_path(path, type);
-        if (location == std::string_view::npos) {
+        if ((helpers::shorten_path(path, type) || special) &&
+            location == std::string_view::npos) {
           path += '/';
         }
       } else if (unicode::is_single_dot_path_segment(path_buffer) &&

tests/basic_tests.cpp

@@ -272,3 +272,24 @@ TYPED_TEST(basic_tests, should_update_password_correctly) {
             "https://username:test@host:8000/path?query#fragment");
   SUCCEED();
 }
+
+// https://github.com/nodejs/node/issues/47889
+TYPED_TEST(basic_tests, node_issue_47889) {
+  auto urlbase = ada::parse<TypeParam>("a:b");
+  ASSERT_EQ(urlbase->get_href(), "a:b");
+  ASSERT_EQ(urlbase->get_protocol(), "a:");
+  ASSERT_EQ(urlbase->get_pathname(), "b");
+  ASSERT_TRUE(urlbase->has_opaque_path);
+  ASSERT_TRUE(urlbase);
+  auto expected_url = ada::parse<TypeParam>("a:b#");
+  ASSERT_TRUE(expected_url);
+  ASSERT_TRUE(expected_url->has_opaque_path);
+  ASSERT_EQ(expected_url->get_href(), "a:b#");
+  ASSERT_EQ(expected_url->get_pathname(), "b");
+  auto url = ada::parse<TypeParam>("..#", &*urlbase);
+  ASSERT_TRUE(url);
+  ASSERT_TRUE(url->has_opaque_path);
+  ASSERT_EQ(url->get_href(), "a:b#");
+  ASSERT_EQ(url->get_pathname(), "b");
+  SUCCEED();
+}

tests/wpt/ada_extra_urltestdata.json

@@ -255,5 +255,35 @@
     "pathname": "/%22quoted%22",
     "search": "",
     "hash": ""
+  },
+  {
+    "input": "a:b#",
+    "base": null,
+    "href": "a:b#",
+    "origin": null,
+    "protocol": "a:",
+    "username": "",
+    "password": "",
+    "host": "",
+    "hostname": "",
+    "port": "",
+    "pathname": "b",
+    "search": "",
+    "hash": ""
+  },
+  {
+    "input": "..#",
+    "base": "a:b",
+    "href": "a:b#",
+    "origin": null,
+    "protocol": "a:",
+    "username": "",
+    "password": "",
+    "host": "",
+    "hostname": "",
+    "port": "",
+    "pathname": "b",
+    "search": "",
+    "hash": ""
   }
 ]