fix path_normalize corner case

fix #674
boostorg · Mar 8, 2023 · e549f40 · e549f40
1 parent 23e0109
commit e549f40
Show file tree

Hide file tree

Showing 2 changed files with 74 additions and 14 deletions.
diff --git a/include/boost/url/impl/url_base.ipp b/include/boost/url/impl/url_base.ipp
@@ -1542,6 +1542,12 @@ normalize_path()
     auto skip_dot = 0;
     bool encode_colons = false;
     string_view first_seg;
+
+//------------------------------------------------
+//
+//  Determine unnecessary initial dot segments to skip and
+//  if we need to encode colons in the first segment
+//
     if (
         !has_authority() &&
         p.starts_with("/./"))
@@ -1596,6 +1602,11 @@ normalize_path()
             encode_colons = first_seg.contains(':');
         }
     }
+
+//------------------------------------------------
+//
+//  Encode colons in the first segment
+//
     if (encode_colons)
     {
         // prepend with "./"
@@ -1649,20 +1660,34 @@ normalize_path()
         p_dest = s_ + impl_.offset(id_path);
         p_end = s_ + impl_.offset(id_path + 1);
     }
+
+//------------------------------------------------
+//
+//  Remove "." and ".." segments
+//
     p.remove_prefix(skip_dot);
     p_dest += skip_dot;
     auto n = detail::remove_dot_segments(
         p_dest, p_end, p);
+
+//------------------------------------------------
+//
+//  Update path parameters
+//
     if (n != pn)
     {
         BOOST_ASSERT(n < pn);
         shrink_impl(id_path, n + skip_dot, op);
         p = encoded_path();
-        if (!p.empty())
+        if (p == "/")
+            impl_.nseg_ = 0;
+        else if (!p.empty())
             impl_.nseg_ = std::count(
                 p.begin() + 1, p.end(), '/') + 1;
         else
             impl_.nseg_ = 0;
+        impl_.decoded_[id_path] =
+            detail::decode_bytes_unsafe(impl_.get(id_path));
     }
     return *this;
 }
@@ -2113,6 +2138,7 @@ edit_segments(
     std::size_t nchar = 0;
     std::size_t prefix = 0;
     bool encode_colons = false;
+    bool cp_src_prefix = false;
     if(it0.index > 0)
     {
         // first segment unchanged
@@ -2125,7 +2151,20 @@ edit_segments(
         {
             if( src.front == "." &&
                     src.fast_nseg > 1)
-                prefix = 2 + absolute;
+                if (src.s.empty())
+                {
+                    // if front is ".", we need the extra "." in the prefix
+                    // which will maintain the invariant that segments represent
+                    // {"."}
+                    prefix = 2 + absolute;
+                }
+                else
+                {
+                    // if the "." prefix is explicitly required from set_path
+                    // we do not include an extra "." segment
+                    prefix = absolute;
+                    cp_src_prefix = true;
+                }
             else if(absolute)
                 prefix = 1;
             else if(has_scheme() ||
@@ -2290,7 +2329,7 @@ edit_segments(
         BOOST_ASSERT(size() == new_size);
         end = dest + nchar;
         impl_.nseg_ = impl_.nseg_ + nseg - (
-            it1.index - it0.index);
+            it1.index - it0.index) - cp_src_prefix;
         if(s_)
             s_[size()] = '\0';
     }

diff --git a/test/unit/url.cpp b/test/unit/url.cpp
@@ -23,6 +23,12 @@
 #include <iomanip>
 #include <sstream>
 
+#ifdef BOOST_TEST_CSTR_EQ
+#undef BOOST_TEST_CSTR_EQ
+#define BOOST_TEST_CSTR_EQ(expr1,expr2) \
+    BOOST_TEST_EQ( boost::urls::detail::to_sv(expr1), boost::urls::detail::to_sv(expr2) )
+#endif
+
 /*  Legend
 
     '#' 0x23    ':' 0x3a
@@ -303,19 +309,34 @@ struct url_test
             BOOST_TEST_EQ( u.buffer(), "./kyle:xy" );
         }
         {
-            // issue 674: 1
-            auto ok = [](string_view u0, string_view p)
+            // issue 674
+            {
+                auto ok = [](string_view u0, string_view p)
+                {
+                  urls::url u(u0);
+                  u.set_path(p);
+                  BOOST_TEST_CSTR_EQ(u.buffer(), p);
+                  u.normalize();
+                  BOOST_TEST_CSTR_EQ(u.buffer(), p);
+                };
+                ok("/", "/");
+                ok("/", "");
+                ok("", "/");
+                ok("", "");
+            }
             {
-                urls::url u(u0);
-                u.set_path(p);
-                BOOST_TEST_CSTR_EQ(u.buffer(), p);
+                urls::url u;
+                BOOST_TEST_EQ(u.encoded_segments().size(), 0);
+                u.set_path("/");
+                BOOST_TEST_EQ(u.encoded_segments().size(), 0);
+                u.set_path("/./");
+                BOOST_TEST_EQ(u.encoded_segments().size(), 1);
+                BOOST_TEST_CSTR_EQ(u.buffer(), "/./");
                 u.normalize();
-                BOOST_TEST_CSTR_EQ(u.buffer(), p);
-            };
-            ok("/", "/");
-            ok("/", "");
-            ok("", "/");
-            ok("", "");
+                BOOST_TEST_CSTR_EQ(u.buffer(), "/");
+                BOOST_TEST_CSTR_EQ(u.encoded_target(), "/");
+                BOOST_TEST_EQ(u.encoded_segments().size(), 0);
+            }
         }
 
         // set_encoded_path