Solved

load XML file iteration

Posted on 2013-05-28
5
514 Views
Last Modified: 2013-05-30
Hi,
I have an xml File (see attached file)
This file is composed of Blocks, Lines, words, characters:

Every Block  is composed of 1,...,n Lines
Every line is composed of 1,...,k  words
Every word is composed of 1,...,l  characters

I am trying to create objects as follows:
Block(Int top, Int left, Int bottom, int right, vector<Lines>)
Line(Int top, Int left, Int bottom, int right, vector<words>)
Word(Int top, Int left, Int bottom, int right, vector<characters>)



I am using TinyXML on C++, but i can't link them together, My code can take one object( block,line,word,character) at a time.

void Keywords::checkChild(TiXmlElement *child)
{
       if(child)
        {

            if((string)child->Value() == "block")
            {
                cout << child->Value()<<endl;

                double x1 = atoi(child->Attribute("left"));
                double y1 = atoi(child->Attribute("top"));
                double x2 = atoi(child->Attribute("right"));
                double y2=  atoi(child->Attribute("bottom"));
              //Vector<Line>lineList
              //  blockList.push_back(newBlock(y1,x1,y2,x2,lineList));
            }


          checkChild(child->FirstChildElement());
          
          checkChild(child->NextSiblingElement());

        }///end if child
}

Open in new window


Thank you.
00000012-1-R.xml
0
Comment
Question by:HaniDaher
[X]
Welcome to Experts Exchange

Add your voice to the tech community where 5M+ people just like you are talking about what matters.

  • Help others & share knowledge
  • Earn cash & points
  • Learn & ask questions
  • 2
5 Comments
 
LVL 37

Expert Comment

by:TommySzalapski
ID: 39204521
You need to have a different function for each type (or if they all have the same attributes you could use templates).

Something like
void Keywords::checkBlock(TiXmlElement *child)
{
       if(child)
        {

            if((string)child->Value() == "block")
            {
                cout << child->Value()<<endl;

                double x1 = atoi(child->Attribute("left"));
                double y1 = atoi(child->Attribute("top"));
                double x2 = atoi(child->Attribute("right"));
                double y2=  atoi(child->Attribute("bottom"));
                blockList.push_back(newBlock(y1,x1,y2,x2);
            }

          child = child->FirstChildElement();
          while(child)
          {
              getLine(child, blockList.Back());
              child = child->NextSiblingElement();
           }
        }///end if child
}
[code]
void Keywords::checkLine(TiXmlElement *child, Block* block)
{
       if(child)
        {

            if((string)child->Value() == "line")
            {
                cout << child->Value()<<endl;

                double m = atoi(child->Attribute("slope")); //or whatever
                double x0 = atoi(child->Attribute("intercept"));
                block->m_line_list.push_back(newLine(m,x0));
            }

          child = child->FirstChildElement();
          while(child)
          {
              getWord(child, block->m_line_list.Back());
              child = child->NextSiblingElement();
           }
        }///end if child
}
      

Open in new window

0
 

Author Comment

by:HaniDaher
ID: 39204588
Yes Tommy that's what i thought. I actually managed to find the following solution:
void parseFile(TiXmlElement* document, vector<Block*>& blocks)
{
  for (TiXmlElement* sub = document->GetFirstChildElement("block"); sub; sub = sub->GetNextSiblingElement("block"))
    blocks.push_back(parseBlock(sub));
}
Block* parseBlock(TiXmlElement* element)
{
  double x1 = atof(element->Attribute("left"));
  // ...
  vector<Line*> lines;
  for (TiXmlElement* sub = element->GetFirstChildElement("line"); sub; sub = sub->GetNextSiblingElement("line"))
    lines.push_back(parseLine(sub));
  return new Block(x1, ..., lines);
}
Line* parseLine(TiXmlElement* element)
{
  double x1 = atof(element->Attribute("left"));
  // ...
  vector<Word*> words;
  for (TiXmlElement* sub = element->GetFirstChildElement("word"); sub; sub = sub->GetNextSiblingElement("word"))
    words.push_back(parseWord(sub));
  return new Line(x1, ..., words);
}
Word* parseWord(TiXmlElement* element)
{
  double x1 = atof(element->Attribute("left"));
  // ...
  vector<Char*> chars;
  for (TiXmlElement* sub = element->GetFirstChildElement("char"); sub; sub = sub->GetNextSiblingElement("char"))
    chars.push_back(parseChar(sub));
  return new Word(x1, ..., chars);
}
Char* parseChar(TiXmlElement* element)
{
  double x1 = atof(element->Attribute("left"));
  // ...
  return new Char(x1, ...);
}

Open in new window


I think it is basically the same idea as yours.
What do you think about the above code?
0
 
LVL 37

Assisted Solution

by:TommySzalapski
TommySzalapski earned 250 total points
ID: 39204642
Yes, that is the same basic idea. Looks like it would work. Personally, I would try to avoid all those calls to new so you don't have to worry about cleaning up all the memory later.
Something like this
void parseFile(TiXmlElement* document, vector<Block>& blocks)
{
  for (TiXmlElement* sub = document->GetFirstChildElement("block"); sub; sub = sub->GetNextSiblingElement("block"))
  {
    
    blocks.push_back(Block());
    parseBlock(sub, blocks.back())
  }
}
void parseBlock(TiXmlElement* element, Block* block)
{
  double x1 = atof(element->Attribute("left"));
  // ...
  for (TiXmlElement* sub = element->GetFirstChildElement("line"); sub; sub = sub->GetNextSiblingElement("line"))
  {
    block->m_lines.push_back(Line);
    parseLine(sub, block->m_lines.back());
   } 

//etc
}

Open in new window


Either way works. I've just found that using dynamic memory like that can lead to segfaults and memory leaking down the road (unless this is just a small one-off thing).
If you are building this as part of a larger program that other people may modify later, I would recommend only using new in a constructor or in a function that also has the delete.
0
 
LVL 34

Accepted Solution

by:
sarabande earned 250 total points
ID: 39205123
if using vector<Block> instead of vector<Block*> the parseBlock function needs to get the second argument by reference and not by pointer:

...
   parseBlock(sub, blocks.back()); // the blocks.back returns a reference to the new Block
  }
}
void parseBlock(TiXmlElement* element, Block& block)

Open in new window


nevertheless as you already have a class 'Keywords' there is no need to turn to c function style. all the objects Block, Line, Word, Character share the same attributes of a rectangle. hence the following class tree seems to map:

struct Rectangle
{
    int left;
    int top;
    int right;
    int bottom;
    Rectangle() : left(0), top(0), right(0), bottom(0) { }
    Rectangle(int l, int t, int r, int b) : left(l), top(t), right(r), bottom(b) { }
};

class Base
{
    int id;
    Rectangle rect;
    std::vector<Base*> subs;
public:
    virtual ~Base()
    { while subs.empty() == false) { delete subs[0]; subs.erase(subs.begin(); } }
    void setRectangle(TiXmlElement* obj);
    virtual Base * createSub();
    virtual std::string getSubName();
    bool parseSubs(const std::string & keyword, TiXmlElement* obj);
};

class Block : public Base
{
...
    Base * createSub() { return new Line; }
    std::string getSubName() { return "line"; }
};

...

class Word : public Base
{
    std::string value;
    int confidence;
    std::string font;
    int type;
public:
    ...
    Base * createSub() { return new Character; }
    std::string getSubName() { return "character"; }
    
};

Open in new window


if doing so you could use the Base container std::vector<Base*> subs as container for lines, words, characters and implement the function parseSubs such that it works for all 4 class objects. you would create new pointers of the 'sub' class by calling the virtual function createSub.

note, the pointers in the containers were deleted when the Base object was destructed. so no need to worry for leaks.

Sara
0

Featured Post

How Do You Stack Up Against Your Peers?

With today’s modern enterprise so dependent on digital infrastructures, the impact of major incidents has increased dramatically. Grab the report now to gain insight into how your organization ranks against your peers and learn best-in-class strategies to resolve incidents.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

This article will show, step by step, how to integrate R code into a R Sweave document
Basic understanding on "OO- Object Orientation" is needed for designing a logical solution to solve a problem. Basic OOAD is a prerequisite for a coder to ensure that they follow the basic design of OO. This would help developers to understand the b…
This tutorial covers a step-by-step guide to install VisualVM launcher in eclipse.
The viewer will learn how to pass data into a function in C++. This is one step further in using functions. Instead of only printing text onto the console, the function will be able to perform calculations with argumentents given by the user.

705 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question