Improve remove_dot_segments

danielaparker · Dec 3, 2024 · 77ba89f · 77ba89f
1 parent 7f52a66
commit 77ba89f
Show file tree

Hide file tree

Showing 2 changed files with 92 additions and 2 deletions.
diff --git a/include/jsoncons/utility/uri.hpp b/include/jsoncons/utility/uri.hpp
@@ -821,7 +821,7 @@ namespace jsoncons { namespace utility {
    3.  Finally, the output buffer is returned as the result of
        remove_dot_segments.
 */
-        static std::string remove_dot_segments(jsoncons::string_view path)
+        static std::string remove_dot_segments2(jsoncons::string_view path)
         {
             std::string input{path};
             std::string output;
@@ -898,6 +898,95 @@ namespace jsoncons { namespace utility {
             return output;
         }
 
+        static std::string remove_dot_segments(jsoncons::string_view path)
+        {
+            std::string input{path};
+            std::string output;
+
+            std::size_t rel = 0;
+            const std::size_t buflen = input.size();
+            while (rel < buflen)
+            {
+                std::size_t len = buflen - rel;
+                if (len >= 3 && input[rel] == '.' && input[rel+1] == '.' && input[rel+2] == '/')
+                { 
+                    //input.erase(0,3);
+                    rel += 3;
+                }
+                else if (len >= 2 && input[rel] == '.' && input[rel+1] == '/')
+                {
+                    //input.erase(0,2);
+                    rel += 2;
+                }
+                else if (len >= 3 && input[rel] == '/' && input[rel+1] == '.' && input[rel+2] == '/')
+                { 
+                    //input.erase(0,2);
+                    rel += 2;
+                    input[rel] = '/';
+                }
+                else if (len == 2 && input[rel] == '/' && input[rel+1] == '.')
+                {
+                    //input.erase(0,1);
+                    ++rel;
+                    input[rel] = '/';
+                }
+                else if (len >= 4 && input[rel] == '/' && input[rel+1] == '.' && input[rel+2] == '.' && input[rel+3] == '/')
+                { 
+                    //input.erase(0,3);
+                    rel += 3;
+                    input[rel] = '/';
+                    auto rslash = output.rfind('/');
+                    if (rslash != std::string::npos)
+                    {
+                        output.erase(rslash);
+                    }
+                }
+                else if (len >= 3 && input[rel] == '/' && input[rel+1] == '.' && input[rel+2] == '.')
+                { 
+                    //input.erase(0,2);
+                    rel += 2;
+                    input[rel] = '/';
+                    auto rslash = output.rfind('/');
+                    if (rslash != std::string::npos)
+                    {
+                        output.erase(rslash);
+                    }
+                }
+                else if (len == 1 && input[rel] == '.')
+                {
+                    //input.erase(0,1);
+                    ++rel;
+                }
+                else if (len == 2 && input[rel] == '.' && input[rel+1] == '.')
+                {
+                    //input.erase(0,2);
+                    rel += 2;
+                }
+                else
+                {
+                    const auto first = input.data() + rel;
+                    const auto last = first+(len);
+                    auto it = std::find(first+1, last, '/');
+                    if (it != last)
+                    {
+                        output.append(first, it - first);
+                        //input.erase(0, it - first);
+                        rel += (it - first);
+                    }
+                    else
+                    {
+                        output.append(first, len);
+                        //input.erase(0, len);
+                        rel += len;
+                    }
+                }
+            }
+
+            //std::cout << "path: " << path << ", output: " << output << "\n";
+
+            return output;
+        }
+
         static std::string merge_paths(const uri& base, const uri& relative)
         {
             std::string result;

diff --git a/test/corelib/src/utility/uri_tests.cpp b/test/corelib/src/utility/uri_tests.cpp
@@ -504,12 +504,14 @@ TEST_CASE("cpp-netib uri resolve tests")
         CHECK(uri.encoded_query() == jsoncons::string_view("q=1"));
         CHECK("http://a/g/x/y?q=1#s" == uri.string());
     }
+
     SECTION("remove_dot_segments1")
     {
         jsoncons::uri reference{"./g"};
         auto uri = reference.resolve(base_uri);
         CHECK("http://a/b/c/g" == uri.string());
     }
+
     SECTION("base_has_path__path_is_merged_1")
     {
         jsoncons::uri reference{"g/"};
@@ -697,4 +699,3 @@ TEST_CASE("cpp-netib uri resolve tests")
         CHECK("http://a/b/c/g#s/../x" == uri.string());
     }
 }
-