libxml-cxx/src/xml.cxx

/*! @file

    @id $Id$
*/
//       1         2         3         4         5         6         7         8
// 45678901234567890123456789012345678901234567890123456789012345678901234567890

#include <xml-cxx/xml.hxx>
#include <iostream>
#include <sstream>
#include <cstdlib>

#include <iomanip>
class MethodTrace {
  public:
    MethodTrace(const void* addr, const std::string& name) throw():
        _addr(addr), _name(name) {
      std::clog<<std::hex<<std::setw(15)<<_addr<<": "
               <<std::dec<<std::setw(2+_level)
               <<std::setfill(' ')<<"\\ "<<_name<<std::endl;
                ++_level;
    }
    ~MethodTrace() throw() {
        --_level;
        std::clog<<std::hex<<std::setw(15)<<_addr<<": "<<std::dec
                 <<std::setw(2+_level)<<std::setfill(' ')
                 <<"/ "<<_name<<std::endl;
    }
  private:
    const void* _addr;
    const std::string _name;
    static unsigned int _level;  
};
unsigned int MethodTrace::_level(0);
#define TRACE MethodTrace XXX_METHOD(this, __PRETTY_FUNCTION__)
#define LOG(X) std::clog<<__PRETTY_FUNCTION__<<"\t**** "<<X<<std::endl

namespace xml {
  
  //================================================================= EXCEPTIONS

  //----------------------------------------------------------------------------
  exception::exception(std::string reason) throw():
      _what(reason), _node(0) {
  }
  exception::exception(std::string reason, const Node& t) throw():
      _what(reason), _node(t.clone().release()) {
  }
  exception::~exception() throw() {
    delete _node;
  }
  void exception::line(unsigned long line) throw() {
    std::stringstream ss;
    ss<<line;
    _what += "\nat line: "+ss.str();
  }
  const char* exception::what() const throw() {
    static std::string w;
    if (!w.size()) {
      if (_node)
        if (_node->isChild())
          w = _what+"\ntag: "+_node->name()+"\nparent: "+_node->parent().name();
        else
          w = _what+"\ntag: "+_node->name()+" (root element)";
      else
        w = _what;
    }
    return w.c_str();
  }
  //----------------------------------------------------------------------------
  access_error::access_error(const Node& t, const std::string& name) throw():
      exception("child not available", t), _name(name) {
  }
  const char* access_error::what() const throw() {
    static std::string w;
    if (!w.size()) w = std::string(exception::what())+"\nchild name: "+_name;
    return w.c_str();
  }
  //----------------------------------------------------------------------------
  out_of_range::out_of_range(const Node& t, size_t pos) throw():
      exception("child not available", t), _pos(pos) {
  }
  const char* out_of_range::what() const throw() {
    static std::stringstream w;
    if (!w.str().size()) w<<exception::what()<<"\nchild number: "<<_pos;
    return w.str().c_str();
  }
  //----------------------------------------------------------------------------
  cannot_have_children::cannot_have_children(const Node& parent,
                                             const Node& child) throw():
      exception("node does not accept child nodes", parent),
      _child(child.clone().release()) {
                                             }
  cannot_have_children::~cannot_have_children() throw() {
    delete _child;
  }
  const char* cannot_have_children::what() const throw() {
    static std::string w;
    if (!w.size())
      w = std::string(exception::what())+"\nchild name: "+_child->name();
    return w.c_str();
  }
  //----------------------------------------------------------------------------
  stream_error::stream_error(const std::string& reason, const Node& t,
                             std::istream& is, const Tag& tag, char c) throw():
      exception(reason, t), _pos(is.tellg()), _tag(new Tag(tag)), _char(c) {
  }
  stream_error::~stream_error() throw() {
    delete _tag;
  }
  const char* stream_error::what() const throw() {
    static std::string w;
    if (!w.size()) {
      std::stringstream ss;
      ss<<exception::what()
        <<"\nposition in stream: "<<_pos;
      if (_char)
        ss<<"\nactual character: "<<(_char>31?_char:'?')
          <<" (ascii="<<(int)_char<<")";
      ss<<"\ntag type: "<<(_tag->type==START?"START"
                           :(_tag->type==END?"END"
                             :(_tag->type==EMPTY?"EMPTY"
                               :"???? <UNKNOWN>")));
      if (_tag->name.size()) ss<<"\nnode name read: "<<_tag->name;
      if (_tag->text.size()) ss<<"\ncontained text: "<<_tag->text;
      w = ss.str();
    }
    return w.c_str();
  }

  //============================================================================

  //----------------------------------------------------------------------------
  //! Copy an attribute.
  Attributes::Value::Value(const value_type& o) throw():
      Attributes::value_type(o) {
  }
  //! Construct an empty attribute.
  Attributes::Value::Value(const std::string& name) throw():
      Attributes::value_type(name, std::string()) {
  }
  //! Construct an attribute with name an value.
  Attributes::Value::Value(const std::string& name,
                           const std::string& value) throw():
      Attributes::value_type(name, value) {
  }
  //! Assign a value.
  Attributes::Value& Attributes::Value::operator=(const std::string& value)
      throw() {
    second = value;
    return *this;
  }
  //! Get the attribute name.
  const std::string& Attributes::Value::name() const throw() {
    return first;
  }
  //! Get the attribute value.
  const std::string& Attributes::Value::value() const throw() {
    return second;
  }
  //! Get the attribute value.
  std::string& Attributes::Value::value() throw() {
    return second;
  }
  //! Convert the attribute to a boolean.
  Attributes::Value::operator bool() const throw() {
    /*! @return @c true if the value is set and not equal to one of:
        @c false @c no @c 0. */
    return !(second.size()||second=="false"||second=="no"||second=="0");
  }
  //! Convert the attribute to a number.
  Attributes::Value::operator unsigned long() const throw() {
    std::stringstream ss(second);
    int i(0);
    ss>>i;
    return i;
  }
  //! Convert the attribute to a space separated list.
  Attributes::Value::operator List() const throw() {
    return toList();
  }
  //! Convert the attribute to a space separated list.
  Attributes::List Attributes::Value::toList(const std::string& separators)
      const throw() {
    List l;
    for (std::string::size_type it(0), pos(0);
         it!=std::string::npos &&
           ((pos=second.find_first_of(separators, it)), true);
         it=pos!=std::string::npos?++pos:std::string::npos)
      if (pos==std::string::npos)
        l.push_back(second.substr(it));
      else
        l.push_back(second.substr(it, pos-it));
    return l;
  }
  //----------------------------------------------------------------------------
  //! Empty attribute list
  Attributes::Attributes() throw() {}
  //! Attribute list with first one empty attribute given.
  Attributes::Attributes(const std::string& empty) throw() {
    insert(Value(empty));
  }
  //! Attribute list with first attribute given.
  Attributes::Attributes(const std::string& key,
                         const std::string& value) throw() {
    insert(Value(key, value));
  }
  //! Add a new key-value pair to the attribute list.
  Attributes& Attributes::operator<<(const Attributes::Value& v) throw() {
    insert(v);
    return *this;
  }
  //! Add a new empty key to the attribute list.
  Attributes& Attributes::operator<<(const std::string& key) throw() {
    insert(Value(key));
    return *this;
  }
  //----------------------------------------------------------------------------
  //! Nodes must be given a name.
  /*! Unnamed nodes are not allowed, therefore there's no default
      constructor. */
  Node::Node(std::string name) throw():
      _name(name), _parent(0) {
  }
  Node::Node(const Node& o) throw():
      _attributes(o._attributes), _name(o.name()), _parent(0) {
    for (Contents::const_iterator it(o._contents.begin());
         it!=o._contents.end(); ++it)
      _contents.push_back((*it)->clone(this).release());
  }
  Node& Node::operator=(const Node& o) throw() {
    _attributes=o._attributes;
    _name = o.name();
    _parent = 0;
    for (Contents::const_iterator it(o._contents.begin());
         it!=o._contents.end(); ++it)
      _contents.push_back((*it)->clone(this).release());
  }
  Node::~Node() throw() {
    clear();
  }
  //! Clones But clears the parent.
  std::auto_ptr<Node> Node::clone() const throw() {
    std::auto_ptr<Node> res(new Node(*this));
    res->_parent = 0;
    return res;
  }
  std::ostream& Node::out(std::ostream& o, unsigned int level) const throw() {
    if (_contents.size()) {
      o<<std::string(level, '\t')<<'<'<<name();
      for (Attributes::const_iterator it(_attributes.begin());
           it!=_attributes.end(); ++it)
        o<<' '<<it->first<<"=\""<<it->second<<'"';
      o<<'>'<<std::endl;
      for (Contents::const_iterator it(_contents.begin());
           it!=_contents.end(); ++it)
        (*it)->out(o, level+1)<<std::endl;
      o<<std::string(level, '\t')<<"</"<<name()<<'>';
    } else {
      o<<std::string(level, '\t')<<'<'<<name();
      for (Attributes::const_iterator it(_attributes.begin());
           it!=_attributes.end(); ++it)
        o<<' '<<it->first<<"=\""<<it->second<<'"';
      o<<"/>";
    }
    return o;
  }
  std::string Node::text() const throw() {
    std::string s;
    for (Contents::const_iterator it(_contents.begin());
         it!=_contents.end(); ++it)
      s+=***it;
    return s;
  }
  Node& Node::text(const std::string& txt) throw(tag_expected, type_mismatch) {
    throw tag_expected(*this, txt);
  }
  Node& Node::append(const Node& o) throw(cannot_have_children) {
    _contents.push_back(o.clone(this).release());
    return *this;
  }
  Node& Node::set(const Attributes& o) throw() {
    _attributes.insert(o.begin(), o.end());
    return *this;
  }
  Node& Node::clear() throw () {
    for (Contents::const_iterator it(_contents.begin());
         it!=_contents.end(); ++it)
      delete *it;
    _contents.clear();
    _attributes.clear();
    return *this;
  }
  std::string Node::name() const throw() {
    return _name;
  }
  bool Node::isChild() const throw() {
    return _parent;
  }
  Node& Node::parent() const throw(no_parent) {
    if (!_parent) throw no_parent(*this);
    return *_parent;
  }
  bool Node::hasAttr(const std::string& name) const throw() {
    return _attributes.find(name)!=_attributes.end();
  }
  Node& Node::attr(const std::string& name, bool mandatory) throw() {
    _attributes[name] = mandatory?"mandatory":"optional";
    return *this;
  }
  std::string Node::attr(const std::string& name) const throw() {
    Attributes::const_iterator it(_attributes.find(name));
    if (it!=_attributes.end()) return it->second;
    return std::string();
  }
  std::string& Node::attr(const std::string& name) throw() {
    return _attributes[name];
  }
  const Attributes::Value Node::attribute(const std::string& name)
      const throw(attribute_not_available) {
    Attributes::const_iterator it(_attributes.find(name));
    if (it!=_attributes.end()) return *it;
    throw attribute_not_available(*this, name);
  }
  Node::List Node::list(const std::string& name) const throw() {
    List res;
    for (Contents::const_iterator it(_contents.begin());
         it!=_contents.end(); ++it)
      if ((*it)->_name==name) res.push_back(*it);
    return res;
  }
  bool Node::operator()(const std::string& child) const throw() {
    return find(child);
  }
  Node& Node::operator<<(const Node& o) throw(cannot_have_children) {
    return append(o);
  }
  Node& Node::operator<<(const Attributes& o) throw() {
    return set(o);
  }
  Node::size_type Node::children() const throw() {
    return _contents.size();
  }
  const Node& Node::operator[](Node::size_type child) const
      throw(access_error) try {
    return *_contents.at(child);
  } catch (...) {
    throw out_of_range(*this, child);
  }
  Node& Node::operator[](Node::size_type child) throw(access_error) try {
    return *_contents.at(child);
  } catch (...) {
    throw out_of_range(*this, child);
  }
  const Node& Node::operator[](const std::string& child) const
      throw(access_error) {
    const Node* const t(find(child));
    if (!t) throw access_error(*this, child);
    return *t;
  }
  Node& Node::operator[](const std::string& child) throw(access_error) {
    Node* const t(find(child));
    if (!t) throw access_error(*this, child);
    return *t;
  }
  std::string Node::operator*() const throw() {
    return text();
  }
  Node& Node::operator=(const std::string& contents) throw(tag_expected,
                                                           type_mismatch) {
    return text(contents);
  }
  std::ostream& operator<<(std::ostream& o, const Node& t) throw() {
    return t.out(o);
  }
  Node* Node::find(const std::string& child) const throw() {
    for (Contents::const_iterator it(_contents.begin());
         it!=_contents.end(); ++it) {
      if ((*it)->name()==child) return *it;
    }
    return 0;
  }
  std::auto_ptr<Node> Node::clone(Node* p) const throw() {
    std::auto_ptr<Node> c(clone());
    c->_parent = p;
    return c;
  }

  //----------------------------------------------------------------------------
  String::String(std::string name, const std::string& text) throw():
      Node(name), _text(text) {
  }
  std::auto_ptr<Node> String::clone() const throw() {
    return std::auto_ptr<Node>(new String(*this));
  }
  std::string String::text() const throw() {
    return _text;
  }
  String& String::text(const std::string& txt) throw(tag_expected,
                                                     type_mismatch) {
    _text = txt;
    return *this;
  }
  std::ostream& String::out(std::ostream& o, unsigned int level) const throw() {
    if (_text.size()) {
      o<<std::string(level, '\t')<<'<'<<name();
      for (Attributes::const_iterator it(_attributes.begin());
           it!=_attributes.end(); ++it)
        o<<' '<<it->first<<"=\""<<it->second<<'"';
      o<<'>'<<_text<<"</"<<name()<<'>';
    } else {
      o<<std::string(level, '\t')<<'<'<<name();
      for (Attributes::const_iterator it(_attributes.begin());
           it!=_attributes.end(); ++it)
        o<<' '<<it->first<<"=\""<<it->second<<'"';
      o<<"/>";
    }
    return o;
  }
  String& String::append(const Node& o) throw(cannot_have_children) {
    throw cannot_have_children(*this, o);
  }
  Node& String::operator=(const std::string& contents) throw() {
    return text(contents);
  }

  //----------------------------------------------------------------------------
  UnsignedInteger::UnsignedInteger(std::string name, unsigned long i) throw():
      String(name,
             dynamic_cast<std::stringstream&>(std::stringstream()<<i).str()) {
  }
  std::auto_ptr<Node> UnsignedInteger::clone() const throw() {
    return std::auto_ptr<Node>(new UnsignedInteger(*this));
  }
  UnsignedInteger& UnsignedInteger::text(const std::string& txt)
      throw(tag_expected, type_mismatch) {
    std::string::size_type
      start(txt.find_first_not_of(" \t\n\r")),
      last(txt.find_last_not_of(" \t\n\r"));
    if (start==std::string::npos) { // empty - set to 0
      _text = "0";
      return *this;
    }
    std::string::size_type pos(txt.find_first_not_of("0123456789", start));
    if (pos<last || pos==last && last==start)
      throw type_mismatch(*this, txt,
                          "unsigned integer may only contain numbers");
    unsigned long i(0);
    std::stringstream(txt)>>i;
    std::stringstream ss;
    ss<<i;
    _text = ss.str();
    return *this;
  }
  unsigned long UnsignedInteger::number() const throw() {
    return number(*this);
  }
  unsigned long UnsignedInteger::number(const Node& node) throw() {
    unsigned long i(0);
    std::stringstream ss(node.text());
    ss>>i;
    return i;
  }

  //----------------------------------------------------------------------------
  Factory::Factory(const Node& t) throw(): _template(t.clone()), _line(0) {}
  const Node& Factory::operator*() const throw() {
    return *_template;
  }
  std::auto_ptr<Node> Factory::read(std::istream& is)
      throw(wrong_end_tag, wrong_start_tag, tag_expected, type_mismatch,
            second_slash_in_tag,
            character_after_slash, missing_end_tag, attribute_value_not_quoted,
            access_error, duplicate_attribute,
            attributes_in_end_tag) try {
    _line=1;
    std::auto_ptr<Node> node(_template->clone());
    node->clear();
    Tag res;
    while (true) {
      res = tag(is, *node);
      *node<<res.attributes;
      switch (res.type) {
        case END: throw wrong_end_tag(*node, is, res);
        case EMPTY: return node; // empty
        case START:
          if (!res.name.size()) throw tag_expected(*node, res.text);
          if (node->name()!=res.name) throw wrong_start_tag(*node, is, res);
          return read(is, *_template);
          //! @todo store instead of ignore
        case SPECIAL: break;
      }
    }
  } catch (exception& e) {
    e.line(_line);
    throw;
  }
  bool Factory::ws(char c) throw() {
    static char last(0);
    if ((c=='\n'||c=='\r')&&last!='\n'&&last!='\r') ++_line;
    last = c;
    return c==' '||c=='\t'||c=='\n'||c=='\r';
  }
  std::auto_ptr<Node> Factory::read(std::istream& is, const Node& node)
      throw(wrong_end_tag, wrong_start_tag, tag_expected, type_mismatch,
            second_slash_in_tag,
            character_after_slash, missing_end_tag, attribute_value_not_quoted,
            access_error, duplicate_attribute, attributes_in_end_tag) {
    std::auto_ptr<Node> result(node.clone());
    result->clear();
    for (bool finished(false); is && !finished;) {
      Tag res(tag(is, node));
      if (res.text.size()) result->text(res.text);
      switch (res.type) {
        case END: {
          if (res.attributes.size()) throw attributes_in_end_tag(node, is, res);
          if (res.name!=node.name()) throw wrong_end_tag(node, is, res);
          finished = true;
        } break;
        case EMPTY: {
          *result<<(node[res.name].clone()->clear()<<res.attributes);
        } break;
        case START: {
          *result<<(*read(is, node[res.name])<<res.attributes);
        } break;
        case SPECIAL: break; //! @todo ignored could be stored
      }
    }
    return result;
  }
  Tag Factory::tag(std::istream& is, const Node& position)
      throw(second_slash_in_tag, character_after_slash,
            missing_end_tag, attribute_value_not_quoted,
            access_error, duplicate_attribute) {
    char c(0);
    Tag tag((Tag){"", START, "", Attributes()});
    while (is && is.get(c) && ws(c)); // skip ws
    if (!is) throw missing_end_tag(position, is, tag);
    if (c!='<') do tag.text+=c; while (is && is.get(c) && c!='<');
    bool nameRead(false);
    for (char last(c); is && is.get(c) && c!='>'; last=c) switch (c) {
      case '\n': case '\r':
        if (last!='\n'&&last!='\r') ++_line;
      case ' ': case '\t':
        if (!nameRead && tag.name.size()) nameRead=true;
        break;
      case '/':
        if (!nameRead && tag.name.size()) nameRead=true;
        if (tag.type!=START) throw second_slash_in_tag(position, is, tag, c);
        tag.type = nameRead?EMPTY:END;
        break;
      case '!': case '?':
        if (last=='<') { // <! tag or <?xml.. starts
          tag.special+=last;
          do tag.special+=c; while (c!='>' && is && is.get(c));
          tag.type=SPECIAL;
          return tag;
        }
      default:
        if (tag.type==EMPTY) throw character_after_slash(position, is, tag, c);
        if (nameRead) { // read attribute
          std::string attrname(1, c), attrvalue;
          while (is && is.get(c) && c!='=' && c!='>' && !ws(c)) attrname+=c;
          while (ws(c) && is && is.get(c)); // skip ws, search '='
          if (c=='=') { // get the value
            while (is && is.get(c) && ws(c)); // skip ws
            if (c!='"')
              throw attribute_value_not_quoted(position, is, tag, c, attrname);
            while (is && is.get(c) && c!='"') attrvalue+=c;
            if (c!='"')
              throw attribute_value_not_quoted(position, is, tag, c, attrname);
          } else {
            is.unget();
          }// read too far
          if (tag.attributes.find(attrname)!=tag.attributes.end())
            throw duplicate_attribute(position, is, tag, attrname);
          tag.attributes[attrname] = attrvalue;
        } else { // part of a tag name
          tag.name+=c;
        }
    }
    return tag;
  }

}