C++ TreeNode::assign_columns方法代码示例

//.........这里部分代码省略.........
    //  iterate over observations
    for (size_t i=0; i < num_observations; i++)
    {
        mxArray * cell = mxGetCell(prhs[0], i);
        
        if (!mxIsCell(cell)) {
            if (tree) {
                delete tree;
            }
            mexErrMsgIdAndTxt("create_ngram_tree:invalidInput",
                              "Input must be a cell array.");
        }
        
        size_t num_unigrams = mxGetNumberOfElements(cell);
        
        //  iterate over unigrams for this observation
        std::vector<std::string> unigrams_cleaned;
        unigrams_cleaned.reserve(num_unigrams);
        
        for (size_t n=0; n < num_unigrams; n++)
        {
            mxArray * gram_cell = mxGetCell(cell, n);
            
            if (!mxIsChar(gram_cell)) {
                if (tree) {
                    delete tree;
                }
                mexErrMsgIdAndTxt(  "create_ngram_tree:invalidInput",
                                    "Input cell arrays must contain strings.");
            }
            
            char * cstr = mxArrayToString(gram_cell);

            //  convert ngram to cpp string
            std::string unigram_string = std::string(cstr);
            mxFree(cstr);
            
            //  make lowercase
            std::transform(unigram_string.begin(), unigram_string.end(), unigram_string.begin(), ::tolower);
            
            //  remove everything except alphanumerics + spaces and tabs
            unigram_string.erase(std::remove_if(unigram_string.begin(), unigram_string.end(), clean_predicate), unigram_string.end());
            
            //  trim starting and ending whitespace
            unigram_string = trim_whitespace(unigram_string);
            
            if (unigram_string.empty()) {
                continue;
            }
            
            //mexPrintf("extracted unigram: %s\n", unigram_string.c_str());
            unigrams_cleaned.push_back(unigram_string);
        }
        
        //  build bigrams
        std::vector <std::string> bigrams;
        build_ngrams(bigrams, unigrams_cleaned, gram_count);
        
        //  append to tree
        for (std::vector <std::string> :: iterator it = bigrams.begin(); it != bigrams.end(); it++) {
            if (!tree) {
                tree = new TreeNode(*it);
            }
            tree->append_increment(*it, static_cast<int>(i));
        }
    }
    
    int col=0;
    if (tree) {
        tree->assign_columns(col);      //  lazy - traverse tree to assign column values
    }
    
    //mexPrintf("Done building tree, %lu instances.\n", tree->count_observations());
    
    //  debug
    //std::string left = tree->leftmost_token();
    //std::string right = tree->rightmost_token();
    //mexPrintf("Leftmost term: %s, rightmost term: %s\n", left.c_str(), right.c_str());
    
    //  pass back tree
    if (tree)
    {
        plhs[0] = tree->create_mex_struct();
        
        if (nlhs == 2) {
            plhs[1] = mxCreateDoubleScalar((double)tree->count_nodes());
        }
        
        //  cleanup
        delete tree;
    }
    else
    {
        //  no tree was created, not enough unigrams - pass back logical false
        plhs[0] = mxCreateLogicalScalar(false);
        if (nlhs == 2) {
            plhs[1] = mxCreateDoubleScalar(0);
        }
    }
}