From 4b04d73bfcc26c51b3fff49f373236ceb0456de3 Mon Sep 17 00:00:00 2001
From: Juan Cruz Viotti <jv@jviotti.com>
Date: Thu, 9 Jan 2025 22:12:52 -0400
Subject: [PATCH] Fix canonicalization of relative URI paths with leading
 slashes (#1430)

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>
---
 src/uri/uri.cc                                |  8 ++---
 .../jsonschema_frame_2019_09_test.cc          | 31 +++++++++++++++++++
 test/uri/uri_canonicalize_test.cc             |  6 ++++
 test/uri/uri_resolve_from_test.cc             |  7 +++++
 4 files changed, 48 insertions(+), 4 deletions(-)

diff --git a/src/uri/uri.cc b/src/uri/uri.cc
index f67d9e625..229f9b5c8 100644
--- a/src/uri/uri.cc
+++ b/src/uri/uri.cc
@@ -65,7 +65,7 @@ static auto uri_parse(const std::string &data, UriUriA *uri) -> void {
   uri_normalize(uri);
 }
 
-static auto canonicalize_path(const std::string &path, const bool is_relative)
+static auto canonicalize_path(const std::string &path)
     -> std::optional<std::string> {
   // TODO: This is a hack, as this whole function works badly for
   // relative paths with ".."
@@ -98,7 +98,7 @@ static auto canonicalize_path(const std::string &path, const bool is_relative)
 
   // Reconstruct the canonical path
   std::string canonical_path;
-  std::string separator = (is_relative && !has_leading_with_word) ? "/" : "";
+  std::string separator = "";
 
   for (const auto &seg : segments) {
     canonical_path += separator + seg;
@@ -430,8 +430,7 @@ auto URI::canonicalize() -> URI & {
   // Clean Path form ".." and "."
   const auto result_path{this->path()};
   if (result_path.has_value()) {
-    const auto canonical_path{
-        canonicalize_path(result_path.value(), this->is_relative())};
+    const auto canonical_path{canonicalize_path(result_path.value())};
     if (canonical_path.has_value()) {
       this->path_ = canonical_path.value();
     }
@@ -559,6 +558,7 @@ auto URI::try_resolve_from(const URI &base) -> URI & {
     // TODO: This only handles a very specific case. We should generalize this
     // function to perform proper base resolution on relative bases
   } else if (this->is_fragment_only() && !base.fragment().has_value()) {
+    this->data = base.data;
     this->path_ = base.path_;
     this->userinfo_ = base.userinfo_;
     this->host_ = base.host_;
diff --git a/test/jsonschema/jsonschema_frame_2019_09_test.cc b/test/jsonschema/jsonschema_frame_2019_09_test.cc
index 6c877c17f..75b1b6485 100644
--- a/test/jsonschema/jsonschema_frame_2019_09_test.cc
+++ b/test/jsonschema/jsonschema_frame_2019_09_test.cc
@@ -2004,3 +2004,34 @@ TEST(JSONSchema_frame_2019_09, relative_base_uri_with_ref) {
   EXPECT_STATIC_REFERENCE(frame, "/allOf/0/$ref", "common#foo", "common",
                           "foo");
 }
+
+TEST(JSONSchema_frame_2019_09, relative_id_leading_slash) {
+  const sourcemeta::jsontoolkit::JSON document =
+      sourcemeta::jsontoolkit::parse(R"JSON({
+    "$id": "/base",
+    "$schema": "https://json-schema.org/draft/2019-09/schema"
+  })JSON");
+
+  sourcemeta::jsontoolkit::Frame frame;
+  frame.analyse(document, sourcemeta::jsontoolkit::default_schema_walker,
+                sourcemeta::jsontoolkit::official_resolver);
+
+  EXPECT_EQ(frame.locations().size(), 3);
+  EXPECT_FRAME_STATIC_2019_09_RESOURCE(frame, "/base", "/base", "", "/base", "",
+                                       0);
+
+  // JSON Pointers
+
+  EXPECT_FRAME_STATIC_2019_09_POINTER(frame, "/base#/$id", "/base", "/$id",
+                                      "/base", "/$id", 0);
+  EXPECT_FRAME_STATIC_2019_09_POINTER(frame, "/base#/$schema", "/base",
+                                      "/$schema", "/base", "/$schema", 0);
+
+  // References
+
+  EXPECT_EQ(frame.references().size(), 1);
+
+  EXPECT_STATIC_REFERENCE(
+      frame, "/$schema", "https://json-schema.org/draft/2019-09/schema",
+      "https://json-schema.org/draft/2019-09/schema", std::nullopt);
+}
diff --git a/test/uri/uri_canonicalize_test.cc b/test/uri/uri_canonicalize_test.cc
index e75219e69..26ccbae00 100644
--- a/test/uri/uri_canonicalize_test.cc
+++ b/test/uri/uri_canonicalize_test.cc
@@ -92,6 +92,12 @@ TEST(URI_canonicalize, example_relative_4) {
   EXPECT_EQ(uri.recompose(), "foo/bar");
 }
 
+TEST(URI_canonicalize, example_relative_6) {
+  sourcemeta::jsontoolkit::URI uri{"/foo"};
+  uri.canonicalize();
+  EXPECT_EQ(uri.recompose(), "/foo");
+}
+
 TEST(URI_canonicalize, example_12) {
   sourcemeta::jsontoolkit::URI uri{"#foo"};
   uri.canonicalize();
diff --git a/test/uri/uri_resolve_from_test.cc b/test/uri/uri_resolve_from_test.cc
index 0a2b06b41..2c09511ac 100644
--- a/test/uri/uri_resolve_from_test.cc
+++ b/test/uri/uri_resolve_from_test.cc
@@ -56,6 +56,13 @@ TEST(URI_try_resolve_from, pointer_fragment_on_relative_path) {
   EXPECT_EQ(relative.recompose(), "foo#/bar");
 }
 
+TEST(URI_try_resolve_from, base_relative_path_leading_slash) {
+  const sourcemeta::jsontoolkit::URI base{"/foo"};
+  sourcemeta::jsontoolkit::URI relative{"#/bar"};
+  relative.try_resolve_from(base);
+  EXPECT_EQ(relative.recompose(), "/foo#/bar");
+}
+
 // RFC 3986, inspired from
 // https://cr.openjdk.org/~dfuchs/writeups/updating-uri/A Section "Resolutuon"