Expand support of uri parser to some missing cases and a subset of re…

…lative references
DataDog · Jun 17, 2024 · 7fdef07 · 7fdef07
1 parent 0678e68
commit 7fdef07
Show file tree

Hide file tree

Showing 2 changed files with 78 additions and 24 deletions.
diff --git a/src/uri_utils.cpp b/src/uri_utils.cpp
@@ -14,6 +14,11 @@
                  / path-absolute
                  / path-rootless
                  / path-empty
+   relative-ref  = relative-part [ "?" query ] [ "#" fragment ]
+   relative-part = "//" authority path-abempty
+                 / path-absolute
+                 / path-noscheme -> Not supported
+                 / path-empty -> Not supported
    scheme        = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
    authority     = [ userinfo "@" ] host [ ":" port ]
    userinfo      = *( unreserved / pct-encoded / sub-delims / ":" )
@@ -68,7 +73,6 @@ enum class token_type {
     ipv6address,
     regname_or_ipv4address,
     path,
-    path_no_authority,
     query,
     fragment,
 };
@@ -172,26 +176,10 @@ std::optional<uri_decomposed> uri_parse(std::string_view uri)
                 i += 2;
             } else {
                 // Otherwise we expect a path (path-absolute, path-rootless, path-empty)
-                expected_token = token_type::path_no_authority;
+                expected_token = token_type::path;
             }
             break;
         }
-        case token_type::path_no_authority: {
-            auto token_begin = i;
-            // The path can be empty but we wouldn't be here...
-            while (i < uri.size()) {
-                const auto c = uri[i++];
-                if (!is_path_char(c) && c != '/') {
-                    return std::nullopt;
-                }
-            }
-
-            decomposed.path_index = token_begin;
-            decomposed.path = uri.substr(token_begin, i - token_begin);
-
-            // We're done, nothing else to parse
-            return decomposed;
-        }
         case token_type::authority: {
             auto token_begin = i;
             authority_end = uri.find_first_of("/?#", i);

diff --git a/tests/uri_utils_test.cpp b/tests/uri_utils_test.cpp
@@ -140,12 +140,7 @@ TEST(TestURI, SchemeAndPath)
     }
 }
 
-TEST(TestURI, SchemeMalformedPath)
-{
-    EXPECT_FALSE(ddwaf::uri_parse("file:[][][]"));
-    EXPECT_FALSE(ddwaf::uri_parse("file:?query"));
-    EXPECT_FALSE(ddwaf::uri_parse("file:#fragment"));
-}
+TEST(TestURI, SchemeMalformedPath) { EXPECT_FALSE(ddwaf::uri_parse("file:[][][]")); }
 
 TEST(TestURI, SchemeHost)
 {
@@ -178,6 +173,40 @@ TEST(TestURI, SchemeHost)
     }
 }
 
+TEST(TestURI, SchemeQuery)
+{
+    auto uri = ddwaf::uri_parse("http:?hello");
+    ASSERT_TRUE(uri);
+    EXPECT_STRV(uri->scheme, "http");
+    EXPECT_STRV(uri->authority.host, "");
+    EXPECT_TRUE(uri->authority.userinfo.empty());
+    EXPECT_TRUE(uri->authority.port.empty());
+    EXPECT_STRV(uri->query, "hello");
+}
+
+TEST(TestURI, SchemeFragment)
+{
+    auto uri = ddwaf::uri_parse("http:#hello");
+    ASSERT_TRUE(uri);
+    EXPECT_STRV(uri->scheme, "http");
+    EXPECT_STRV(uri->authority.host, "");
+    EXPECT_TRUE(uri->authority.userinfo.empty());
+    EXPECT_TRUE(uri->authority.port.empty());
+    EXPECT_STRV(uri->fragment, "hello");
+}
+
+TEST(TestURI, SchemeQueryFragment)
+{
+    auto uri = ddwaf::uri_parse("http:?hello#bye");
+    ASSERT_TRUE(uri);
+    EXPECT_STRV(uri->scheme, "http");
+    EXPECT_STRV(uri->authority.host, "");
+    EXPECT_TRUE(uri->authority.userinfo.empty());
+    EXPECT_TRUE(uri->authority.port.empty());
+    EXPECT_STRV(uri->query, "hello");
+    EXPECT_STRV(uri->fragment, "bye");
+}
+
 TEST(TestURI, SchemeIPv4Host)
 {
     auto uri = ddwaf::uri_parse("http://1.2.3.4");
@@ -602,4 +631,41 @@ TEST(TestURI, RelativeRefAbsolutePathQueryFragment)
     EXPECT_STRV(uri->fragment, "f");
 }
 
+TEST(TestURI, RelativeRefQuery)
+{
+    auto uri = ddwaf::uri_parse("/?hello");
+    ASSERT_TRUE(uri);
+    EXPECT_STRV(uri->scheme, "");
+    EXPECT_STRV(uri->authority.host, "");
+    EXPECT_TRUE(uri->authority.userinfo.empty());
+    EXPECT_TRUE(uri->authority.port.empty());
+    EXPECT_STRV(uri->path, "/");
+    EXPECT_STRV(uri->query, "hello");
+}
+
+TEST(TestURI, RelativeRefFragment)
+{
+    auto uri = ddwaf::uri_parse("/#hello");
+    ASSERT_TRUE(uri);
+    EXPECT_STRV(uri->scheme, "");
+    EXPECT_STRV(uri->authority.host, "");
+    EXPECT_TRUE(uri->authority.userinfo.empty());
+    EXPECT_TRUE(uri->authority.port.empty());
+    EXPECT_STRV(uri->path, "/");
+    EXPECT_STRV(uri->fragment, "hello");
+}
+
+TEST(TestURI, RelativeRefQueryFragment)
+{
+    auto uri = ddwaf::uri_parse("/?hello#bye");
+    ASSERT_TRUE(uri);
+    EXPECT_STRV(uri->scheme, "");
+    EXPECT_STRV(uri->authority.host, "");
+    EXPECT_TRUE(uri->authority.userinfo.empty());
+    EXPECT_TRUE(uri->authority.port.empty());
+    EXPECT_STRV(uri->path, "/");
+    EXPECT_STRV(uri->query, "hello");
+    EXPECT_STRV(uri->fragment, "bye");
+}
+
 } // namespace