0

I have a class Word that holds a string and a vector of structs, each struct containing a pointer to a container class, like so:

struct doc { Document* d; int timesMentioned = 0; };

I have an output function that outputs each Word's string and every doc's key in the Word's doc vector. (d1, d2, d3, etc..)

Whenever I output all of them to file, it looks like this:

foo = d3, bar = d3, foobar = d3, etc.

With every Word outputting only d3 (The last document that was looped through previous from a given input file. I can post the code for that if needed)

The weird thing is, whenever I output it to console, it works fine. The doc that's associated with each Word is outputted correctly. Any idea why this could happen? I thought it may be a dangling pointer but it outputting to console correctly is curious. I can post the adding docs function or more code if needed.

Here is the code for the Word class:

void Word::createWord(string str) { removePunctuation(str); toLower(str); word = str; } void Word::addDoc(Document& d) { doc newDoc; newDoc.d = &d; newDoc.timesMentioned++; docs.push_back(newDoc); } vector<doc> Word::getDocs() { return docs; } string Word::getWord() { return word; } //Formatting the string, should be irrelevant to this question void Word::removePunctuation(string& str) { string temp = ""; for(int i = 0; i < str.length(); i++) { bool punctual = false; for(int k = 0; k < 42; k++) if(str[i] == punctuals[k]) punctual = true; if(!punctual) temp += str[i]; } str = temp; } void Word::toLower(string& str) { string temp = ""; for(int i = 0; i < str.size(); i++) { char c = str[i]; temp += tolower(c); } str = temp; }

Here's where the Words/Documents are instantiated and added to the vector:

void InvertedFileIndex::parseFile(string fileName) { fstream fin, fout; fin.open(fileName.c_str(), fstream::in); if(fin.is_open()) { //Parse input into a single string to the XMLParsing xml_document<> doc; string str, parse = ""; while(fin >> str) parse += str + " "; doc.parse<0>(&parse[0]); xml_node<>* root; root = doc.first_node("mediawiki"); //Iterate through each page int i = 1; for (xml_node<>* page = root->first_node("page"); page; page = page->next_sibling()) { Document d; Word w; string docName = "d" + intToStr(i); xml_node<>* title = page->first_node("title"); d.setTitle(page->first_node("title")->value()); d.setKey(docName); //Find text from document xml_node<>* revision = page->first_node("revision"); xml_node<>* text = revision->first_node("text"); d.setText(text->value()); string docText = text->value(); ////Begin parsing text into words stringstream ss(docText); //create a string stream so we can break it into tokens string item; while (getline(ss, item, ' ')) { Word w; w.createWord(item); w.addDoc(d); toks.push_back(w); } documents.push_back(d); writeToIndex(); i++; } fin.close(); } else cout << "The sample file " << fileName << "could not be opened." << endl; }

And here's the output code:

void InvertedFileIndex::writeToIndex() { fstream fout; fout.open("index.txt", fstream::out); if(fout.is_open()) { for(int i = 0; i < toks.size(); i++) { fout << toks[i].getWord() << " = "; cout << toks[i].getWord() << " = "; for(int j = 0; j < toks[i].getDocs().size(); j++) { fout << toks[i].getDocs()[j].d->getKey(); cout << toks[i].getDocs()[j].d->getKey(); fout << ", "; cout << ", "; } fout << endl; cout << endl; } fout.close(); } else cout << "Index file could not be opened." << endl; }
Improve this question
3
  • The rest of the code (for Word and for the output) would be helpful. Commented Apr 20, 2013 at 3:01
  • 1
    Obviously we need to see the code which reads from the file... because it is wrong. Commented Apr 20, 2013 at 3:03
  • I've added the input/output code and the Word class function implementation. It is a rather weird problem. Commented Apr 20, 2013 at 3:19

1 Answer 1

Reset to default
2

Look's to me that you're saving pointers to objects which go out of scope. Essentially your code is

while (...) { Document d; ... w.add(d); // add takes a reference and stores pointer to d } // sometime later writeToIndex();

Obviously by the time you get to writeToIndex all your document objects, which were created in the while loop, have been destroyed. So you have pointers to objects which no longer exist. If you want to store pointers, then you should dynamically allocate the objects somewhere.

Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

Yeah, that's what I've gathered...I tried adding each document to a vector after adding a pointer to it, but I'm guessing addressing is different inside of a vector. This is play code for an inverted file index, so I'll stick to just add a key to the doc instead of a pointer for locating the documents. The structure will come later. Thanks for your help!
@TaylorBishop. Addressing is the same everywhere. The problem is that you have a pointer to an object which has been destroyed. That would happen whether the pointer was in a vector or not. This is one of the many dangers associated with pointers. Having a pointer in no way guarantees that the object being pointed to is valid. To understand this better you need to look into the lifetime of objects, in particular the difference between the lifetime of an object created on the stack vs. the lifetime of an object created on the heap using new.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.