From 43dcc539848d97c60cd01772b061d920c828c089 Mon Sep 17 00:00:00 2001
From: Ben Kelly <wanderview@chromium.org>
Date: Thu, 2 Sep 2021 15:59:14 +0000
Subject: [PATCH] URLPattern: Support constructor strings with ipv6 addresses.

This adds support for constructor strings like:

  new URLPattern("http://[\\:\\:1]/");

As discussed in spec issue #113:

  https://github.com/WICG/urlpattern/issues/113

Fixed: 1245760
Change-Id: I712341e72a5c2af745dbbdbb8673a79809a98425
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3138336
Commit-Queue: Ben Kelly <wanderview@chromium.org>
Reviewed-by: Jeremy Roman <jbroman@chromium.org>
Cr-Commit-Position: refs/heads/main@{#917682}
---
 .../modules/url_pattern/url_pattern_parser.cc | 17 ++++++-
 .../modules/url_pattern/url_pattern_parser.h  |  8 +++
 .../resources/urlpatterntestdata.json         | 49 +++++++++++++++++++
 3 files changed, 73 insertions(+), 1 deletion(-)

diff --git a/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.cc b/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.cc
index 1358eaa71bb87c..f960d42b37e187 100644
--- a/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.cc
+++ b/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.cc
@@ -210,8 +210,15 @@ void Parser::Parse(ExceptionState& exception_state) {
         break;
 
       case StringParseState::kHostname:
+        // Track whether we are inside ipv6 address brackets.
+        if (IsIPv6Open())
+          hostname_ipv6_bracket_depth_ += 1;
+        else if (IsIPv6Close())
+          hostname_ipv6_bracket_depth_ -= 1;
+
         // If we find a `:` then we transition to the port component state.
-        if (IsPortPrefix())
+        // However, we ignore `:` when parsing an ipv6 address.
+        else if (IsPortPrefix() && !hostname_ipv6_bracket_depth_)
           ChangeState(StringParseState::kPort, Skip(1));
 
         // If we find a `/` then we transition to the pathname component state.
@@ -424,6 +431,14 @@ bool Parser::IsGroupClose() const {
   return token_list_[token_index_].type == liburlpattern::TokenType::kClose;
 }
 
+bool Parser::IsIPv6Open() const {
+  return IsNonSpecialPatternChar(token_index_, "[");
+}
+
+bool Parser::IsIPv6Close() const {
+  return IsNonSpecialPatternChar(token_index_, "]");
+}
+
 String Parser::MakeComponentString() const {
   DCHECK_LT(token_index_, token_list_.size());
   const auto& token = token_list_[token_index_];
diff --git a/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.h b/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.h
index c3392018b79f8b..f6a757ca0bf342 100644
--- a/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.h
+++ b/third_party/blink/renderer/modules/url_pattern/url_pattern_parser.h
@@ -135,6 +135,11 @@ class Parser final {
   bool IsGroupOpen() const;
   bool IsGroupClose() const;
 
+  // These methods indicate if the current token is an opening or closing
+  // bracket for an ipv6 hostname; e.g. '[' or ']'.
+  bool IsIPv6Open() const;
+  bool IsIPv6Close() const;
+
   // This method returns a String consisting of the tokens between
   // `component_start_` and the current `token_index_`.
   String MakeComponentString() const;
@@ -177,6 +182,9 @@ class Parser final {
   // The current nesting depth of `{ }` pattern groupings.
   int group_depth_ = 0;
 
+  // The current netsting depth of `[ ]` in hostname patterns.
+  int hostname_ipv6_bracket_depth_ = 0;
+
   // The current parse state.  This should only be changed via `ChangeState()`
   // or `RewindAndSetState()`.
   StringParseState state_ = StringParseState::kInit;
diff --git a/third_party/blink/web_tests/external/wpt/urlpattern/resources/urlpatterntestdata.json b/third_party/blink/web_tests/external/wpt/urlpattern/resources/urlpatterntestdata.json
index e9967adf219778..5ef045e560acb5 100644
--- a/third_party/blink/web_tests/external/wpt/urlpattern/resources/urlpatterntestdata.json
+++ b/third_party/blink/web_tests/external/wpt/urlpattern/resources/urlpatterntestdata.json
@@ -2150,6 +2150,55 @@
       "pathname": { "input": "/data:channel.html", "groups": {} }
     }
   },
+  {
+    "pattern": [ "http://[\\:\\:1]/" ],
+    "inputs": [ "http://[::1]/" ],
+    "exactly_empty_components": [ "username", "password", "port", "search",
+                                  "hash" ],
+    "expected_obj": {
+      "protocol": "http",
+      "hostname": "[\\:\\:1]",
+      "pathname": "/"
+    },
+    "expected_match": {
+      "protocol": { "input": "http", "groups": {} },
+      "hostname": { "input": "[::1]", "groups": {} },
+      "pathname": { "input": "/", "groups": {} }
+    }
+  },
+  {
+    "pattern": [ "http://[\\:\\:1]:8080/" ],
+    "inputs": [ "http://[::1]:8080/" ],
+    "exactly_empty_components": [ "username", "password", "search", "hash" ],
+    "expected_obj": {
+      "protocol": "http",
+      "hostname": "[\\:\\:1]",
+      "port": "8080",
+      "pathname": "/"
+    },
+    "expected_match": {
+      "protocol": { "input": "http", "groups": {} },
+      "hostname": { "input": "[::1]", "groups": {} },
+      "port": { "input": "8080", "groups": {} },
+      "pathname": { "input": "/", "groups": {} }
+    }
+  },
+  {
+    "pattern": [ "http://[\\:\\:a]/" ],
+    "inputs": [ "http://[::a]/" ],
+    "exactly_empty_components": [ "username", "password", "port", "search",
+                                  "hash" ],
+    "expected_obj": {
+      "protocol": "http",
+      "hostname": "[\\:\\:a]",
+      "pathname": "/"
+    },
+    "expected_match": {
+      "protocol": { "input": "http", "groups": {} },
+      "hostname": { "input": "[::a]", "groups": {} },
+      "pathname": { "input": "/", "groups": {} }
+    }
+  },
   {
     "pattern": [ "https://foo{{@}}example.com" ],
     "inputs": [ "https://foo@example.com" ],