c ++将字符串向量转换为给定.dat文件中的双精度向量

Question

So I have many questions, but I'll start with what I believe should be an easy one. I've been given an assignment to compare template files to query files, calculate the dot product, and return the 10 nearest neighbors. I think I can do the calculations fairly easily, but I'm having a hard time with the file i/o. I'm able to read in the data to a vector of strings, but I'm not sure how to convert it to a vector of doubles while maintaining the integrity of the vertex. If I try using a string stream, or an iterator, I end up getting each number assigned it's own index number, instead of each line getting it's own index number. Here's what I have...can you please help me?

Edited for clarification purposes:

I am comparing query files to template files that contain collections of images to get the 10 nearest neighbors. A query file contains one "set" (for lack of a better description). A template file contains 138 lines of data. Right now, all I would like to do is print each line of data from the template file with it's corresponding index number from the .dat file, but in a format that allows me to do the necessary calculations. Once all is said and done, I will need to compute the cosine between two vectors (a query and the ith row in the template), so I will actually need to breakout the ith row of the template in order to compute the cosine between it and the query file. Is that more clear?

Here's a link to the query file: https://www.dropbox.com/s/6xytafmojrct3lh/001_AU01_query.dat?dl=0 Here's a link to the template file: https://www.dropbox.com/s/vnqi7h1btxdsf9u/001_template.dat?dl=0

Sample output would be something like: "001_AU01_query: 15 20 135 19 36 22 105 95 55 68" where the numbers represent the line numbers of the corresponding template file that most closely match the query data.

Again, I really appreciate your help.

void NearestNeighbor::readQuery(){
        vector<string> queryVector;
        string line;
            ifstream queryData;
        queryData.open("001_AU01_query.dat");
        if (queryData.fail()) {
            cout << "Unable to read query.dat file";
            exit(1);
        }
        //populate the vector with the template info
        while(getline(queryData, line, '\n')){
            queryVector.push_back(line);
        }
        //this prints the contents of the queryVector to the console
        for ( unsigned int i = 0; i < (queryVector.size()); i++){
            cout << "Index[" << i << "] " << queryVector[i] << endl;
        }
        queryData.close();
    }//end readQuery()

I'm happy to post a sample of the input and expected output, if you think it will help. Thanks in advance!

Answer 1

You just need to choose the correct data structure and everything else should flow from that.

A single 1-dimensional vector<double> will not work because you can't keep track of the values in each line of the data. However, a vector<double> is appropriate to store the values in a single line of the data. Then you just need one of these for each line.

So a more appropriate data structure would be vector<vector<double>> . ie a 2D vector:

void readQuery(std::istream& queryData){
  std::vector<std::vector<double>> queryVector;
  std::string line;

  while(getline(queryData, line, '\n'))
    queryVector.push_back(splitData(line));

  for (unsigned i = 0u; i != queryVector.size(); ++i) {
    std::cout << "Index[" << i << "] ";
    for(double value : queryVector[i])
        std::cout << value << " ";
    std::cout << "\n";
  }
}

This requires a function splitData to split a string into a vector<double> . You can find plenty of examples (and debate) on the best way of splitting a string here on SO but an example implementation might be:

std::vector<double> splitData(const std::string& line) {
  std::istringstream iss(line);
  std::istream_iterator<double> begin(iss);
  std::istream_iterator<double> end;
  return {begin, end};
}

Live demo Live demo C++03