feat(text.hpp): newline normalization

Adamska1008 · Aug 30, 2024 · 4f3852b · 4f3852b
1 parent 52cf642
commit 4f3852b
Show file tree

Hide file tree

Showing 11 changed files with 143 additions and 48 deletions.
diff --git a/include/myxml/document.hpp b/include/myxml/document.hpp
@@ -19,7 +19,6 @@ namespace myxml
         /* Exportable */
         virtual std::string ExportRaw() const;
         virtual std::string ExportFormatted(int indentLevel = 0, int indentSize = 4) const;
-        virtual void SetEntityEncoding(bool flag) override;
     };
 
     class Document : public Exportable

diff --git a/include/myxml/element.hpp b/include/myxml/element.hpp
@@ -47,6 +47,5 @@ namespace myxml
         /* Implement Exportable */
         virtual std::string ExportRaw() const override;
         virtual std::string ExportFormatted(int indentLevel = 0, int indentSize = 4) const override;
-        virtual void SetEntityEncoding(bool) override;
     };
 }
diff --git a/include/myxml/exportable.hpp b/include/myxml/exportable.hpp
@@ -3,12 +3,25 @@
 
 namespace myxml
 {
+    class ExportConfig
+    {
+    public:
+        bool EntityEncoding;
+        bool PlatformSpecificNewline;
+
+        ExportConfig();
+    };
+
     class Exportable
     {
+    protected:
+        ExportConfig config;
+
     public:
         virtual ~Exportable() = default;
         virtual std::string ExportRaw() const = 0;
         virtual std::string ExportFormatted(int indentLevel = 0, int indentSize = 4) const = 0;
-        virtual void SetEntityEncoding(bool) = 0;
+        virtual void SetEntityEncoding(bool) {};
+        virtual void SetPlatformSpecificNewline(bool) {};
     };
 }
diff --git a/include/myxml/node.hpp b/include/myxml/node.hpp
@@ -28,6 +28,10 @@ namespace myxml
         std::enable_if_t<std::is_base_of_v<Node, T>,
                          std::optional<std::shared_ptr<T>>>
         As();
+
+        /** Implement Export */
+        virtual void SetEntityEncoding(bool) override;
+        virtual void SetPlatformSpecificNewline(bool) override;
     };
 
     // Element are Composite Node.
@@ -52,6 +56,10 @@ namespace myxml
         std::shared_ptr<Node> InsertAtFront(const std::shared_ptr<Node> &);
         std::shared_ptr<Node> InsertAtEnd(const std::shared_ptr<Node> &);
         void Unlink(const std::shared_ptr<Node> &);
+
+        /** Implement Export */
+        virtual void SetEntityEncoding(bool) override;
+        virtual void SetPlatformSpecificNewline(bool) override;
     };
 
     template <typename T>

diff --git a/include/myxml/text.hpp b/include/myxml/text.hpp
@@ -8,7 +8,6 @@ namespace myxml
     {
     private:
         std::string inner;
-        bool encodeOnExport;
 
     public:
         explicit Text(std::string_view str);
@@ -21,6 +20,10 @@ namespace myxml
         /* Implment Exportable*/
         virtual std::string ExportRaw() const override;
         virtual std::string ExportFormatted(int indentLevel = 0, int indentSize = 4) const override;
-        virtual void SetEntityEncoding(bool) override;
+    };
+
+    namespace util
+    {
+        const char *const platformSpecificNewline();
     };
 }
diff --git a/src/document.cpp b/src/document.cpp
@@ -103,10 +103,6 @@ namespace myxml
         return std::string(' ', indentLevel * indentSize) + this->ExportRaw();
     }
 
-    void Declaration::SetEntityEncoding(bool flag)
-    {
-    }
-
     namespace util
     {
         bool isValidXmlVersion(std::string_view version)

diff --git a/src/element.cpp b/src/element.cpp
@@ -96,12 +96,4 @@ namespace myxml
         builder += indent + "</" + std::string(this->GetName()) + ">\n";
         return builder;
     }
-
-    void Element::SetEntityEncoding(bool flag)
-    {
-        for (auto it = this->FirstChild(); it != nullptr; it = it->next)
-        {
-            it->SetEntityEncoding(flag);
-        }
-    }
 }
diff --git a/src/exportable.cpp b/src/exportable.cpp
@@ -0,0 +1,10 @@
+#include "myxml/exportable.hpp"
+
+namespace myxml
+{
+  ExportConfig::ExportConfig()
+      : EntityEncoding(true),
+        PlatformSpecificNewline(false)
+  {
+  }
+}
diff --git a/src/node.cpp b/src/node.cpp
@@ -4,7 +4,15 @@
 
 namespace myxml
 {
+    void Node::SetEntityEncoding(bool flag)
+    {
+        this->config.EntityEncoding = flag;
+    }
 
+    void Node::SetPlatformSpecificNewline(bool flag)
+    {
+        this->config.PlatformSpecificNewline = flag;
+    }
 
     std::shared_ptr<Node> CompositeNode::LastChild()
     {
@@ -120,4 +128,21 @@ namespace myxml
         elem->parent = nullptr;
     }
 
+    void CompositeNode::SetEntityEncoding(bool flag)
+    {
+        this->config.EntityEncoding = flag;
+        for (auto it = this->FirstChild(); it != nullptr; it = it->next)
+        {
+            it->SetEntityEncoding(flag);
+        }
+    }
+
+    void CompositeNode::SetPlatformSpecificNewline(bool flag)
+    {
+        this->config.PlatformSpecificNewline = flag;
+        for (auto it = this->FirstChild(); it != nullptr; it = it->next)
+        {
+            it->SetPlatformSpecificNewline(flag);
+        }
+    }
 }
diff --git a/src/text.cpp b/src/text.cpp
@@ -6,51 +6,65 @@ namespace myxml
 {
 
     Text::Text(std::string_view input)
-        : encodeOnExport(true)
     {
-        // entity encoding
-        static std::map<std::string, char, std::less<>> entityMap = {
-            {"&lt;", '<'},
-            {"&gt;", '>'},
-            {"&amp;", '&'},
-            {"&quot;", '"'},
-            {"&apos;", '\''},
-        };
-        std::size_t len = input.length();
-        std::size_t start = 0; // start of current segment
-        for (std::size_t i = 0; i < len; i++)
+        if (config.EntityEncoding)
         {
-            if (input[i] == '&')
+            // entity encoding
+            static std::map<std::string, char, std::less<>> entityMap = {
+                {"&lt;", '<'},
+                {"&gt;", '>'},
+                {"&amp;", '&'},
+                {"&quot;", '"'},
+                {"&apos;", '\''},
+            };
+            std::size_t len = input.length();
+            std::size_t start = 0; // start of current segment
+
+            for (std::size_t i = 0; i < len; i++)
             {
-                if (auto semicolonPos = input.find(';', i); semicolonPos != std::string::npos)
+                // Newline Normalization
+                if (input[i] == '\r')
                 {
-                    std::string_view entity = input.substr(i, semicolonPos - i + 1);
-                    if (auto it = entityMap.find(entity); it != entityMap.end())
+                    this->inner += input.substr(start, i - start);
+                    if (i + 1 < len && input[i + 1] == '\n')
                     {
-                        this->inner += input.substr(start, i - start); // append unmodified segment
-                        this->inner += it->second;                     // append decoded character
-                        i = semicolonPos;                              // skip past the entity
-                        start = semicolonPos + 1;                      // update last unappend position
+                        i += 1;
+                    }
+                    this->inner += '\n';
+                    start = i + 1;
+                }
+                // Entity Decoding
+                if (input[i] == '&')
+                {
+                    if (auto semicolonPos = input.find(';', i); semicolonPos != std::string::npos)
+                    {
+                        std::string_view entity = input.substr(i, semicolonPos - i + 1);
+                        if (auto it = entityMap.find(entity); it != entityMap.end())
+                        {
+                            this->inner += input.substr(start, i - start); // append unmodified segment
+                            this->inner += it->second;                     // append decoded character
+                            i = semicolonPos;                              // skip past the entity
+                            start = semicolonPos + 1;                      // update last unappend position
+                        }
                     }
                 }
             }
+            this->inner += input.substr(start, len - start); // append the remaining
+        }
+        else
+        {
+            this->inner = input;
         }
-        this->inner += input.substr(start, len - start); // append the remaining
     }
 
     bool Text::IsAllSpace() const
     {
         return std::all_of(this->inner.begin(), this->inner.end(), isspace);
     }
 
-    void Text::SetEntityEncoding(bool flag)
-    {
-        this->encodeOnExport = flag;
-    }
-
     std::string Text::ExportRaw() const
     {
-        if (!this->encodeOnExport)
+        if (!this->config.EntityEncoding && !this->config.PlatformSpecificNewline)
         {
             return this->inner;
         }
@@ -68,11 +82,23 @@ namespace myxml
             std::string builder;
             for (std::size_t i = 0; i < len; i++)
             {
-                if (auto it = entityMap.find(this->inner[i]); it != entityMap.end())
+                if (this->config.EntityEncoding)
                 {
-                    builder += this->inner.substr(start, i - start);
-                    builder += it->second;
-                    start = i + 1;
+                    if (auto it = entityMap.find(this->inner[i]); it != entityMap.end())
+                    {
+                        builder += this->inner.substr(start, i - start);
+                        builder += it->second;
+                        start = i + 1;
+                    }
+                }
+                if (this->config.PlatformSpecificNewline)
+                {
+                    if (this->inner[i] == '\n')
+                    {
+                        builder += this->inner.substr(start, i - start);
+                        builder += util::platformSpecificNewline();
+                        start = i + 1;
+                    }
                 }
             }
             builder += this->inner.substr(start, len - start);
@@ -85,4 +111,16 @@ namespace myxml
         // TODO: better implementation
         return std::string(indentLevel * indentSize, ' ') + this->inner + '\n';
     }
+
+    namespace util
+    {
+        const char *const platformSpecificNewline()
+        {
+#ifdef _WIN32
+            return "\r\n";
+#else
+            return "\n";
+#endif
+        }
+    }
 }
diff --git a/tests/parser_test.cpp b/tests/parser_test.cpp
@@ -208,4 +208,16 @@ TEST_CASE("Parsing simple xml elements", "[parser]")
 
         REQUIRE(elem->FirstChild()->As<myxml::CData>().value()->ExportRaw() == "<![CDATA[Hello!]]>\n");
     }
+
+    SECTION("Newline Normalization")
+    {
+        std::string nl = "<root>hello\r\n</root>";
+        auto elem = myxml::Element::Parse(nl);
+
+        REQUIRE(elem->FirstChild()->As<myxml::Text>().value()->ExportRaw() == "hello\n");
+        nl = "<root>hello\r</root>";
+
+        elem = myxml::Element::Parse(nl);
+        REQUIRE(elem->FirstChild()->As<myxml::Text>().value()->ExportRaw() == "hello\n");
+    }
 }