Commit b398dc78 authored by jan.koester's avatar jan.koester
Browse files

test

parents 3b24881b 88364f09

CMakeLists.txt

+26 −1
Original line number Diff line number Diff line 
project(libhtmlpp CXX)

cmake_minimum_required(VERSION 3.12)

cmake_minimum_required(VERSION 3.23)

enable_testing ()



set(LIBV "1.0.0")

set(Upstream_VERSION 1.0.0)



find_package(Doxygen QUIET)

if (DOXYGEN_FOUND)

    set(DOXYFILE_IN  ${CMAKE_CURRENT_SOURCE_DIR}/Doxyfile.in)

    set(DOXYFILE_OUT ${CMAKE_CURRENT_BINARY_DIR}/Doxyfile)

    set(DOC_INSTALL_DIR "${CMAKE_INSTALL_PREFIX}/share/libhtmlpp/html")



    # Allow @LIBV@ substitution from your existing variable

    configure_file(${DOXYFILE_IN} ${DOXYFILE_OUT} @ONLY)



    add_custom_target(docs ALL

        COMMAND ${DOXYGEN_EXECUTABLE} ${DOXYFILE_OUT}

        WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}

        COMMENT "Generating API documentation with Doxygen (sources: ${CMAKE_SOURCE_DIR}/src)"

        VERBATIM)



   install(

        DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}/docs/html/"

        DESTINATION "${DOC_INSTALL_DIR}"

        COMPONENT docs

   )



else()

    message(STATUS "Doxygen not found; 'docs' target will not be available.")

endif()



if(NOT ${CMAKE_CXX_COMPILER} MATCHES "MSVC")

set(CMAKE_CXX_FLAGS "-fPIC -Wall")

endif()

Doxyfile.in

0 → 100644
+59 −0
Original line number Diff line number Diff line 
# Doxygen configuration for libhtmlpp

# This file is configured by CMake (configure_file) to produce a Doxyfile.



PROJECT_NAME           = "libhtmlpp"

PROJECT_NUMBER         = "@LIBV@"



OUTPUT_DIRECTORY       = "@CMAKE_BINARY_DIR@/docs"

CREATE_SUBDIRS         = NO



# What to document

INPUT                  = "@CMAKE_SOURCE_DIR@/src" "@CMAKE_SOURCE_DIR@"

RECURSIVE              = YES

FILE_PATTERNS          = *.h *.hpp *.hh *.hxx *.c *.cc *.cpp *.cxx



# Extraction options

EXTRACT_ALL            = YES

EXTRACT_PRIVATE        = NO

EXTRACT_STATIC         = NO

EXTRACT_LOCAL_CLASSES  = YES

EXTRACT_ANON_NSPACES   = YES



# Warnings

WARNINGS               = YES

WARN_IF_UNDOCUMENTED   = NO

WARN_IF_DOC_ERROR      = YES

WARN_NO_PARAMDOC       = YES

WARN_LOGFILE           = "@CMAKE_BINARY_DIR@/docs/doxygen-warnings.log"



# Source browser

SOURCE_BROWSER         = YES

INLINE_SOURCES         = NO

STRIP_CODE_COMMENTS    = YES



# HTML output

GENERATE_HTML          = YES

HTML_OUTPUT            = html

SEARCHENGINE           = YES



# Disable LaTeX/PDF by default

GENERATE_LATEX         = NO



# Diagrams (optional, requires Graphviz "dot")

HAVE_DOT               = YES

DOT_IMAGE_FORMAT       = svg

DOT_TRANSPARENT        = YES

CALL_GRAPH             = NO

CALLER_GRAPH           = NO

CLASS_GRAPH            = YES

COLLABORATION_GRAPH    = NO

INCLUDE_GRAPH          = YES

INCLUDED_BY_GRAPH      = YES



# Misc

TAB_SIZE               = 4

JAVADOC_AUTOBRIEF      = YES

MULTILINE_CPP_IS_BRIEF = NO

OPTIMIZE_OUTPUT_FOR_C  = NO

OPTIMIZE_OUTPUT_JAVA   = NO

OPTIMIZE_FOR_FORTRAN   = NO

src/encode.h

+1 −0
Original line number Diff line number Diff line
@@ -34,6 +34,7 @@ namespace libhtmlpp { 
        {"<","&lt;"},

        {">","&gt;"},

        {"'","&apos;"},

        {"+","&plus;"},

        {nullptr,nullptr}

    };

};

src/html.cpp

+829 −282

File changed.

Preview size limit exceeded, changes collapsed.

src/html.h

+120 −4
Original line number Diff line number Diff line 
/**

 * @file html.h

 * @brief Public declarations for libhtmlpp HTML element types and utilities.

 * @date 2025-10-30

 *

 * This file is part of libhtmlpp and provides HTML parsing, DOM-like element

 * types, serialization helpers, and encoding utilities.

 */

/*******************************************************************************

Copyright (c) 2014, Jan Koester jan.koester@gmx.net

All rights reserved.
@@ -34,10 +42,23 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
#include <memory>



#pragma once

/**

 * @namespace libhtmlpp

 * @brief Core namespace for the libhtmlpp HTML parsing and printing library.

 */



namespace libhtmlpp {

/**

 * @brief Internal helper frame used while building the tree during parsing.

 */

    class DocElements;

/**

 * @brief Concrete element representing a generic HTML tag with attributes and child/next links.

 */

    class HtmlElement;

/**

 * @brief Mutable string buffer for HTML input/output with parsing facilities.

 */

    class HtmlString;



    enum ElementType{
@@ -45,8 +66,12 @@ namespace libhtmlpp { 
        HtmlEl=1,

        CommentEl=2,

        ScriptEL=3,

        SvgEL=4

        SvgEL=4,

        TextAreaEL=5

    };

/**

 * @brief Abstract base class for all nodes in the HTML tree. Provides linkage and common operations.

 */



    class Element {

    public:
@@ -115,6 +140,8 @@ namespace libhtmlpp { 


        int    getType() const;

        void   remove(Element* el);



        static size_t parseElement(const std::vector<char> &in,std::unique_ptr<libhtmlpp::Element> &el,size_t start,bool &termination);

    protected:



        std::unique_ptr<Element>  _childElement=nullptr;
@@ -127,6 +154,7 @@ namespace libhtmlpp { 
            std::unique_ptr<Attributes> _nextAttr;

        };



        void _serialelize(const std::vector<char> &in);

    private:

        //if text tagname must be zero

        std::vector<char> _TagName;
@@ -140,6 +168,9 @@ namespace libhtmlpp { 
        friend void  print(const Element& element, HtmlString &output,bool formated);

        friend void _copy(libhtmlpp::Element *dest,const libhtmlpp::Element *src);

    };

/**

 * @brief Leaf node representing plain text content of an HTML document.

 */



    class TextElement : public Element {

    public:
@@ -155,12 +186,18 @@ namespace libhtmlpp { 
        void                       setText(const std::string &txt);



        int         getType() const;



        static size_t parseElement(const std::vector<char> &in,std::unique_ptr<libhtmlpp::Element> &el,size_t start,bool &termination);



    protected:

        std::vector<char> _Text;

        friend class HtmlString;

        friend void  print(const Element& element, HtmlString &output,bool formated);

        friend void _copy(libhtmlpp::Element *dest,const libhtmlpp::Element *src);

    };

/**

 * @brief Leaf node representing an HTML comment (<!-- -->).

 */



    class CommentElement : public Element{

    public:
@@ -175,12 +212,17 @@ namespace libhtmlpp { 
        void        setComment(const std::string &txt);



        int         getType() const;



        static size_t parseElement(const std::vector<char> &in,std::unique_ptr<libhtmlpp::Element> &el,size_t start,bool &termination);

    protected:

        std::vector<char> _Comment;

        friend class HtmlString;

        friend void  print(const Element& element, HtmlString &output,bool formated);

        friend void _copy(libhtmlpp::Element *dest,const libhtmlpp::Element *src);

    };

/**

 * @brief Element representing a <script> tag and its text content.

 */



    class ScriptElement : public HtmlElement{

    public:
@@ -201,6 +243,7 @@ namespace libhtmlpp { 
        void         appendChild(const Element* el)=delete;

        void         appendChild(const Element& el)=delete;



        static size_t parseElement(const std::vector<char> &in,std::unique_ptr<libhtmlpp::Element> &el,size_t start,bool &termination);

    protected:



        std::unique_ptr<Element> _childElement=nullptr;
@@ -210,6 +253,9 @@ namespace libhtmlpp { 
        friend void  print(const Element& element, HtmlString &output,bool formated);

        friend void _copy(libhtmlpp::Element *dest,const libhtmlpp::Element *src);

    };

/**

 * @brief Element representing an embedded <svg> tag and its attributes/content.

 */



    class SvgElement : public HtmlElement{

    public:
@@ -230,6 +276,7 @@ namespace libhtmlpp { 
        void         appendChild(const Element* el)=delete;

        void         appendChild(const Element& el)=delete;



        static size_t parseElement(const std::vector<char> &in,std::unique_ptr<libhtmlpp::Element> &el,size_t start,bool &termination);

    protected:



        std::unique_ptr<Element> _childElement=nullptr;
@@ -240,6 +287,48 @@ namespace libhtmlpp { 
        friend void _copy(libhtmlpp::Element *dest,const libhtmlpp::Element *src);

    };



/**

 * @brief Element representing an embedded <textarea> tag and its attributes/content.

 */



    class TextArea : public HtmlElement{

    public:

        TextArea();

        TextArea(const TextArea &textsrc);

        ~TextArea();



        TextArea& operator=(const Element &hel);

        TextArea& operator=(const Element *hel);



        const std::vector<char> getText();

        void                    setText(const std::string  &text);



        int            getType() const;



        void         insertChild(const Element* el)=delete;

        void         insertChild(const Element& el)=delete;

        void         appendChild(const Element* el)=delete;

        void         appendChild(const Element& el)=delete;



        static size_t parseElement(const std::vector<char> &in,std::unique_ptr<libhtmlpp::Element> &el,size_t start,bool &termination);

    protected:



        std::unique_ptr<Element> _childElement=nullptr;



        std::vector<char> _Text;

        friend class HtmlString;

        friend void  print(const Element& element, HtmlString &output,bool formated);

        friend void _copy(libhtmlpp::Element *dest,const libhtmlpp::Element *src);

    };



/**

 * @brief Serializes an element (and its subtree) into an HtmlString.

 * @param element Root element to print.

 * @param output Output buffer to append serialized HTML to.

 * @param formated If true, pretty-prints the output with line breaks/indentation.

 */





    void print(const Element& element, HtmlString &output,bool formated=false);



    class HtmlString {
@@ -247,7 +336,7 @@ namespace libhtmlpp { 


        using value_type = char;



        HtmlString();

        HtmlString()=default;

        HtmlString(const HtmlString &str);

        HtmlString(char str);

        HtmlString(const std::string &str);
@@ -284,21 +373,46 @@ namespace libhtmlpp { 
        const std::vector<char>&  data() const;

        const std::string                  str() const;

        const char                         *c_str() const;

/**

 * @brief Parses the current buffer into a DOM-like tree and returns the root element.

 * @return Reference to the root Element stored internally.

 * @throws HTMLException on malformed input.

 */



        libhtmlpp::Element &parse();



    private:

        std::unique_ptr<Element> _rootEl;

        void                 _serialelize(std::vector<char> in, HtmlElement* out);

        void                 _buildTree();

        void _buildtreenode(DocElements *firstel,DocElements *lastel,std::unique_ptr<Element>&html);

        std::vector<char>    _Data;

        friend void HtmlEncode(const std::string &input,HtmlString *output);

        friend class HtmlPage;

    };

/**

 * @brief Decodes special HTML characters in a string and appends to an HtmlString.

 * @param input Plain input string.

 * @param output Destination HtmlString that receives encoded characters.

 */

    void HtmlDecode(const std::string &input,HtmlString &output);



/**

 * @brief Decodes special HTML characters in a string and appends to an std::string.

 * @param input Plain input string.

 * @param output Destination HtmlString that receives encoded characters.

 */

    void HtmlDecode(const std::string &input,std::string &output);



/**

 * @brief Encodes special HTML characters in a string and writes into std::string.

 * @param input Plain input string.

 * @param output Receives encoded HTML string.

 */



    void HtmlEncode(const std::string &input,HtmlString *output);

    void HtmlEncode(const std::string &input,std::string &output);

/**

 * @brief High level loader/saver for HTML documents (files and strings).

 */



    class HtmlPage {

    public:
@@ -374,3 +488,5 @@ namespace libhtmlpp { 
        size_t   _count;

    };

};



std::ostream& operator<<(std::ostream& os, const libhtmlpp::HtmlString& p);