updated ranking, tested

DocParser.cpp

@@ -75,6 +75,12 @@ void DocParser::parse(const string& filename, StopWord& stop) {
         curr.removePunc();//remove punctuation
         curr.stemming();
 
+        if (curr.getStr().empty()){//don't insert an empty string
+            text = text.substr(space + 1);
+            space = text.find(" ");
+            continue;
+        }
+
         //put unique words into the avl tree
         if (!words.contains(curr)){//if the word is not already in the tree/new unique word
             curr.incrFreq(currDoc);

QueryProcessor.cpp

@@ -88,7 +88,7 @@ void QueryProcessor::parseQuery(string& q, DSAVLTree<Word>& words, DSAVLTree<Wor
         space = query.find(" ");//to check if youve reached the end of the query
     }
 
-//    rankIndex();//TODO
+    rankIndex();//TODO
 }
 
 vector<Word> QueryProcessor::parseAndOr()
@@ -268,6 +268,7 @@ void QueryProcessor::addPersonOrg(vector<Document>& a)//remove any docs from fin
 
 void QueryProcessor::rankIndex()
 {
+    //TODO something with NOT operator doesn't work with ranking system
     cout << "Rank index" << endl;
 //    cout << "finalIndex size " << finalIndex.size() << endl;
 //    cout << "query words size: " << queryWords.size() << endl;
@@ -279,20 +280,22 @@ void QueryProcessor::rankIndex()
         {
             //get the each words frequency in the current doc and add them all together
             sum += queryWords.at(i).getDocFreq(finalIndex.at(queryIndex));//add total freq of each word for this doc
-            cout << queryWords.at(i).getDocFreq(finalIndex.at(queryIndex)) << " " << finalIndex.at(queryIndex).getPath() << endl;
+//            cout << queryWords.at(i).getDocFreq(finalIndex.at(queryIndex)) << " " << finalIndex.at(queryIndex).getPath() << endl;
         }
         freqs.push_back(sum);
 //        cout << "sum: " << sum << endl;
     }
     //result: total frequency for each doc
 
-    cout << "Frequency" << endl;
-    for (int i = 0; i < freqs.size(); i++)
-        cout << freqs.at(i) << " " << finalIndex.at(i).getPath() << endl;
+//    cout << "Frequency" << endl;
+//    for (int i = 0; i < freqs.size(); i++)
+//        cout << freqs.at(i) << " " << finalIndex.at(i).getPath() << endl;
 
 
     //get the top 15 docs with the highest freq
     for (int n = 0; n < 15; n++){
+//        if (n > freqs.size() || freqs.size() == 0)//less that 15 docs in the finalIndex
+//            break;
         int highest = freqs.at(0);
         int index = 0;
         if (n > freqs.size())//less that 15 docs in the finalIndex
@@ -305,7 +308,7 @@ void QueryProcessor::rankIndex()
             }
         }
         best.push_back(finalIndex.at(index));//get the corresponding doc for that freq
-        cout << "next higheset frequency: " << freqs.at(index) << endl;
+//        cout << "next higheset frequency: " << freqs.at(index) << endl;
         freqs.erase(freqs.begin() + index);
         finalIndex.erase(finalIndex.begin() + index);
     }

UserInterface.cpp

@@ -1,5 +1,5 @@
 //
-// Created by 18476 on 4/23/2022.
+// Created by Karina Shin on 4/23/2022.
 //
 
 #include "UserInterface.h"
@@ -19,7 +19,11 @@ void UserInterface::run(const string& file)
         cin >> choice;
         if (choice == "1"){
             cout << "parsing..." << endl;
+            start = std::chrono::high_resolution_clock::now();
             docReader.getFiles(file, stops);
+            end = std::chrono::high_resolution_clock::now();
+            std::chrono::duration<double> time_in_seconds = end - start;
+            cout << std::fixed << "Parsing Time: " << time_in_seconds.count() << endl;
             cout << "done!" << endl;
         }
         else if (choice == "2"){
@@ -97,36 +101,36 @@ void UserInterface::parseDocs(const string& direct)
     std::cout << "done parsing!" << std::endl;
 }
 
-//void UserInterface::displayResults()//with ranking
-//{
-//    if (process.getBest().size() == 0)
-//        cout << "No results found" << endl;
-//
-//    for (int i = 0; i < process.getBest().size(); i++)
-//    {
-//        if (i == 15)
-//            break;
-//        cout << i + 1 << ") ";
-//        cout << "Title: " << process.getBest().at(i).getTitle() << ", " << process.getBest().at(i).getPub() << ", Date: " << process.getBest().at(i).getDate() << endl;
-//        cout << "Path: " << process.getBest().at(i).getPath() << endl;
-//    }
-//}
-
-void UserInterface::displayResults()//without ranking
+void UserInterface::displayResults()//with ranking
 {
-    if (process.getFinal().size() == 0)
+    if (process.getBest().size() == 0)
         cout << "No results found" << endl;
 
-    for (int i = 0; i < process.getFinal().size(); i++)
+    for (int i = 0; i < process.getBest().size(); i++)
     {
         if (i == 15)
             break;
         cout << i + 1 << ") ";
-        cout << "Title: " << process.getFinal().at(i).getTitle() << ", " << process.getFinal().at(i).getPub() << ", Date: " << process.getFinal().at(i).getDate() << endl;
-        cout << "Path: " << process.getFinal().at(i).getPath() << endl;
+        cout << "Title: " << process.getBest().at(i).getTitle() << ", " << process.getBest().at(i).getPub() << ", Date: " << process.getBest().at(i).getDate() << endl;
+        cout << "Path: " << process.getBest().at(i).getPath() << endl;
     }
 }
 
+//void UserInterface::displayResults()//without ranking
+//{
+//    if (process.getFinal().size() == 0)
+//        cout << "No results found" << endl;
+//
+//    for (int i = 0; i < process.getFinal().size(); i++)
+//    {
+//        if (i == 15)
+//            break;
+//        cout << i + 1 << ") ";
+//        cout << "Title: " << process.getFinal().at(i).getTitle() << ", " << process.getFinal().at(i).getPub() << ", Date: " << process.getFinal().at(i).getDate() << endl;
+//        cout << "Path: " << process.getFinal().at(i).getPath() << endl;
+//    }
+//}
+
 void UserInterface::showText(Document& d)
 {
     rapidjson::Document doc;