made stop word class so I can access stop words in all classes

CMakeLists.txt

@@ -4,4 +4,4 @@ project(22s_final_proj)
 set(CMAKE_CXX_STANDARD 17)
 
 #add_executable(22s_final_proj main.cpp catch_setup.cpp DSAVLTree.h DocParser.cpp DocParser.h Word.cpp Word.h Document.cpp Document.h catchTests.cpp)
-add_executable(22s_final_proj main.cpp DSAVLTree.h DocParser.cpp DocParser.h Word.cpp Word.h Document.cpp Document.h porter2_stemmer.cpp porter2_stemmer.h QueryProcessor.cpp QueryProcessor.h UserInterface.cpp UserInterface.h)
+add_executable(22s_final_proj main.cpp DSAVLTree.h DocParser.cpp DocParser.h Word.cpp Word.h Document.cpp Document.h porter2_stemmer.cpp porter2_stemmer.h QueryProcessor.cpp QueryProcessor.h UserInterface.cpp UserInterface.h StopWord.cpp StopWord.h)

DocParser.cpp

@@ -3,24 +3,9 @@
 //
 
 #include "DocParser.h"
-DocParser::DocParser()
-{
-    ifstream stop;//make the stop words AVL tree
-    stop.open("stopWords.txt");
-    string curr;
-    while (getline(stop, curr))//make an avl tree of stop words
-    {
-        string s = curr.substr(0, curr.length()-1);//cut off end char
-        stops.insert(s);
-    }
-}
-
-DocParser::DocParser(DSAVLTree<string>& stopWords)
-{
-    //TODO
-}
+DocParser::DocParser() {}
 
-void DocParser::parse(const string& filename) {
+void DocParser::parse(const string& filename, StopWord& stop) {
     cout << "NEW DOC: " << filename << endl;
     //TODO parse for org and person, put unique ones into avl tree
     //parse main text
@@ -58,7 +43,7 @@ void DocParser::parse(const string& filename) {
         Word curr(text.substr(0, space));
 //        cout << "current: " << curr.getStr() << endl;
         curr.toLower();//remove caps
-        if (isStopWord(curr.getStr())){
+        if (stop.isStopWord(curr.getStr())){
             text = text.substr(space + 1);//cut off curr word
             space = text.find(" ");
             continue;//don't add to tree
@@ -89,13 +74,13 @@ void DocParser::parse(const string& filename) {
     }
 }
 
-void DocParser::getFiles(const string& directory)
+void DocParser::getFiles(const string& directory, StopWord& stop)
 {
     for (const auto & entry : fs::recursive_directory_iterator(directory)){
         if (entry.is_regular_file()) {
             if (entry.path().extension().string() == ".json") {
                 string filename = entry.path().c_str();
-                parse(filename);
+                parse(filename, stop);
             }
         }
     }
@@ -107,11 +92,6 @@ vector<Document>& DocParser::findIndex(Word& obj)
 //    return words.find(words.getRoot(), obj)->getData().getDocs();
 }
 
-bool DocParser::isStopWord(string& str)
-{
-    return stops.contains(str);//if str is in the avl tree, its a stop word
-}
-
 vector<Document>& DocParser::findWordIndex(Word& w) { return w.getDocs(); }
 vector<Document>& DocParser::findOrgIndex(Word& org) { return org.getDocs(); }
 vector<Document>& DocParser::findPersonIndex(Word& p) {return p.getDocs(); }

DocParser.h

@@ -14,6 +14,7 @@
 #include "DSAVLTree.h"
 #include "Word.h"
 #include "include/rapidjson/document.h"
+#include "StopWord.h"
 
 using namespace std;
 namespace fs = std::filesystem;
@@ -23,16 +24,12 @@ class DocParser {
     DSAVLTree<Word> words;
     DSAVLTree<Word> orgs;
     DSAVLTree<Word> people;
-    DSAVLTree<string> stops;
 
 public:
     DocParser();
-    DocParser(DSAVLTree<string>& stopWords);
-    void parse(const string& filename);//parse the documents and create 3 AVLTrees
-//    void parse(DSAVLTree<Word>& words, DSAVLTree<Word> orgs, DSAVLTree<Word> people);
-    void getFiles(const string& directory);//returns filenames for traversal through directory
+    void parse(const string& filename, StopWord& stop);//parse the documents and create 3 AVLTrees
+    void getFiles(const string& directory, StopWord& stop);//returns filenames for traversal through directory
     vector<Document>& findIndex(Word& obj);//return the document index of a given Word object
-    bool isStopWord(string& str);
 
     vector<Document>& findWordIndex(Word& w);
     vector<Document>& findOrgIndex(Word& org);

QueryProcessor.cpp

@@ -12,15 +12,16 @@ QueryProcessor::QueryProcessor(DSAVLTree<Word>& w, DSAVLTree<Word>& o, DSAVLTree
 //    people = p;
 }
 
-void QueryProcessor::parseQuery(string& query)//parse query
+void QueryProcessor::parseQuery(string& query, StopWord& stop)//parse query
 {
     std::cout << "NEW QUERY: " << query << std::endl;
     int space;
     while (space != -1)
     {
         space = query.find(" ");
         string curr = query.substr(0, space);
-        //TODO remove stop words and stem
+        if (stop.isStopWord(curr))
+            continue;//skip stop words
 
         if (curr == "AND" || curr == "OR"){//2 arg operators
             query = query.substr(space + 1);//cut off operator

QueryProcessor.h

@@ -11,6 +11,7 @@
 #include "Document.h"
 #include "DSAVLTree.h"
 #include "Word.h"
+#include "StopWord.h"
 
 class QueryProcessor {
 private:
@@ -23,7 +24,7 @@ class QueryProcessor {
 public:
     QueryProcessor();
     QueryProcessor(DSAVLTree<Word>& w, DSAVLTree<Word>& o, DSAVLTree<Word>& p);
-    void parseQuery(string& query);//parse query
+    void parseQuery(string& query, StopWord& stop);//parse query
     void setUnion(vector<Document>& a, vector<Document>& b);//OR keyword
     void intersection(vector<Document>& a, vector<Document>&b);//AND keyword
     void complement(vector<Document>& a);//set subtraction

StopWord.cpp

@@ -0,0 +1,21 @@
+//
+// Created by Karina Shin on 4/23/2022.
+//
+
+#include "StopWord.h"
+
+StopWord::StopWord()
+{
+    ifstream stop;//make the stop words AVL tree
+    stop.open("stopWords.txt");
+    string curr;
+    while (getline(stop, curr))//make an avl tree of stop words
+    {
+        string s = curr.substr(0, curr.length()-1);//cut off end char
+        stops.insert(s);
+    }
+}
+bool StopWord::isStopWord(string& str)
+{
+    return stops.contains(str);//if str is in the avl tree, its a stop word
+}

StopWord.h

@@ -0,0 +1,24 @@
+//
+// Created by Karina Shin on 4/23/2022.
+//
+
+#ifndef INC_22S_FINAL_PROJ_STOPWORD_H
+#define INC_22S_FINAL_PROJ_STOPWORD_H
+
+#include <iostream>
+#include <fstream>
+#include <string>
+#include "DSAVLTree.h"
+
+using namespace std;
+
+class StopWord {
+private:
+    DSAVLTree<string> stops;
+public:
+    StopWord();
+    bool isStopWord(string& str);
+};
+
+
+#endif //INC_22S_FINAL_PROJ_STOPWORD_H

UserInterface.cpp

@@ -4,17 +4,7 @@
 
 #include "UserInterface.h"
 
-UserInterface::UserInterface()
-{
-    ifstream stop;//make the stop words AVL tree
-    stop.open("stopWords.txt");
-    string curr;
-    while (getline(stop, curr))//make an avl tree of stop words
-    {
-        string s = curr.substr(0, curr.length()-1);//cut off end char
-        stops.insert(s);
-    }
-}
+UserInterface::UserInterface() {}
 
 void UserInterface::clearIndex()
 {
@@ -24,10 +14,10 @@ void UserInterface::clearIndex()
     //TODO also erase contents of persistence file?
 }
 
-void UserInterface::parseDocs(string& direct)
+void UserInterface::parseDocs(const string& direct)
 {
     std::cout << "parsing documents..." << std::endl;
-    docReader.parse(direct);
+    docReader.getFiles(direct, stops);
     std::cout << "done parsing!" << std::endl;
 }
 
@@ -38,10 +28,12 @@ void UserInterface::persistentIndex()
 
 void UserInterface::enterQuery(string& query)
 {
-    process.parseQuery(query);
+    process.parseQuery(query, stops);
 }
 
 void UserInterface::stats()
 {
 
-}
+}
+
+DocParser& UserInterface::getDocParser() { return docReader; }

UserInterface.h

@@ -10,24 +10,26 @@
 #include "DSAVLTree.h"
 #include "Word.h"
 #include "Document.h"
+#include "StopWord.h"
 
 class UserInterface {
 private:
     DocParser docReader;
     QueryProcessor process;
+    StopWord stops;
 
     DSAVLTree<Word> words;
     DSAVLTree<Word> orgs;
     DSAVLTree<Word> persons;
-    DSAVLTree<string> stops;
 
 public:
     UserInterface();
     void clearIndex();//delete every tree
-    void parseDocs(string& direct);//parse all documents
+    void parseDocs(const string& direct);//parse all documents
     void persistentIndex();//read in persistence file to index words
     void enterQuery(string& query);
     void stats();
+    DocParser& getDocParser();//used to access word/org/person avl trees
 };
 
 

main.cpp

@@ -7,20 +7,21 @@
 #include "DSAVLTree.h"
 #include "Word.h"
 #include "QueryProcessor.h"
+#include "UserInterface.h"
 
 using namespace std;
 int main(int argc, char** argv) {
     //TODO find out where to put stop words list so that it can be used in all classes
     //just have the avl tree indexes in the user interface class and then pass it everywhere else?
     //can't have multiple avl trees in each class (doc parser and query), should just have one set
 
-    DocParser parse;
+    UserInterface parse;
     cout << "parsing..." << endl;
-    parse.getFiles(argv[2]);//absolute path
+    parse.parseDocs(argv[2]);//absolute path
     cout << "done!" << endl;
     Word w(argv[1]);//search term
     w.stemming();
-    parse.getWordTree().find(parse.getWordTree().getRoot(), w).printDocs();
+    parse.getDocParser().getWordTree().find(parse.getDocParser().getWordTree().getRoot(), w).printDocs();
 //    parse.getWordTree().find(parse.getWordTree().getRoot(), w)->getData().printDocs();
 
 //    vector<int> finalIndex;