src: fix leading backslash bug in URL

RaisinTen · targos · commit dcba37460493 · 2021-05-01T09:44:50.000+02:00
The associated condition mentioned in the URL parsing algorithm of the WHATWG URL Standard is: url is special and c is U+005C (\) So, `special_back_slash` must be updated whenever `special` is updated. Fixes: #36559 PR-URL: #36613 Reviewed-By: Rich Trott <rtrott@gmail.com> Reviewed-By: Daijiro Wachi <daijiro.wachi@gmail.com> Reviewed-By: James M Snell <jasnell@gmail.com>
diff --git a/src/node_url.cc b/src/node_url.cc
@@ -1426,7 +1426,7 @@ void URL::Parse(const char* input,
     const char ch = p < end ? p[0] : kEOL;
     bool special = (url->flags & URL_FLAGS_SPECIAL);
     bool cannot_be_base;
-    const bool special_back_slash = (special && ch == '\\');
+    bool special_back_slash = (special && ch == '\\');
 
     switch (state) {
       case kSchemeStart:
@@ -1476,6 +1476,7 @@ void URL::Parse(const char* input,
             url->flags &= ~URL_FLAGS_SPECIAL;
             special = false;
           }
+          special_back_slash = (special && ch == '\\');
           buffer.clear();
           if (has_state_override)
             return;
@@ -1520,6 +1521,7 @@ void URL::Parse(const char* input,
             url->flags &= ~URL_FLAGS_SPECIAL;
             special = false;
           }
+          special_back_slash = (special && ch == '\\');
           if (base->flags & URL_FLAGS_HAS_PATH) {
             url->flags |= URL_FLAGS_HAS_PATH;
             url->path = base->path;
@@ -1543,6 +1545,7 @@ void URL::Parse(const char* input,
           url->flags |= URL_FLAGS_SPECIAL;
           special = true;
           state = kFile;
+          special_back_slash = (special && ch == '\\');
           continue;
         }
         break;
@@ -1572,6 +1575,7 @@ void URL::Parse(const char* input,
           url->flags &= ~URL_FLAGS_SPECIAL;
           special = false;
         }
+        special_back_slash = (special && ch == '\\');
         switch (ch) {
           case kEOL:
             if (base->flags & URL_FLAGS_HAS_USERNAME) {
diff --git a/test/cctest/test_url.cc b/test/cctest/test_url.cc
@@ -81,6 +81,52 @@ TEST_F(URLTest, Base3) {
   EXPECT_EQ(simple.path(), "/baz");
 }
 
+TEST_F(URLTest, Base4) {
+  const char* input = "\\x";
+  const char* base = "http://example.org/foo/bar";
+
+  URL simple(input, strlen(input), base, strlen(base));
+
+  EXPECT_FALSE(simple.flags() & URL_FLAGS_FAILED);
+  EXPECT_EQ(simple.protocol(), "http:");
+  EXPECT_EQ(simple.host(), "example.org");
+  EXPECT_EQ(simple.path(), "/x");
+}
+
+TEST_F(URLTest, Base5) {
+  const char* input = "/x";
+  const char* base = "http://example.org/foo/bar";
+
+  URL simple(input, strlen(input), base, strlen(base));
+
+  EXPECT_FALSE(simple.flags() & URL_FLAGS_FAILED);
+  EXPECT_EQ(simple.protocol(), "http:");
+  EXPECT_EQ(simple.host(), "example.org");
+  EXPECT_EQ(simple.path(), "/x");
+}
+
+TEST_F(URLTest, Base6) {
+  const char* input = "\\\\x";
+  const char* base = "http://example.org/foo/bar";
+
+  URL simple(input, strlen(input), base, strlen(base));
+
+  EXPECT_FALSE(simple.flags() & URL_FLAGS_FAILED);
+  EXPECT_EQ(simple.protocol(), "http:");
+  EXPECT_EQ(simple.host(), "x");
+}
+
+TEST_F(URLTest, Base7) {
+  const char* input = "//x";
+  const char* base = "http://example.org/foo/bar";
+
+  URL simple(input, strlen(input), base, strlen(base));
+
+  EXPECT_FALSE(simple.flags() & URL_FLAGS_FAILED);
+  EXPECT_EQ(simple.protocol(), "http:");
+  EXPECT_EQ(simple.host(), "x");
+}
+
 TEST_F(URLTest, TruncatedAfterProtocol) {
   char input[2] = { 'q', ':' };
   URL simple(input, sizeof(input));
