From 36e834ac13f26e99b83ed0dc756687d56c3dac79 Mon Sep 17 00:00:00 2001
From: karinashin <karinashin0531@gmail.com>
Date: Mon, 18 Apr 2022 11:44:00 -0500
Subject: [PATCH] using avl tree for stop words, moved to parse class

---
 DocParser.cpp | 20 +++++++++++++++++++-
 DocParser.h   |  3 +++
 Word.cpp      | 29 ++++++++++++++---------------
 Word.h        |  6 ++++--
 main.cpp      | 11 ++++++-----
 5 files changed, 46 insertions(+), 23 deletions(-)

diff --git a/DocParser.cpp b/DocParser.cpp
index dce8bcc..033c81b 100644
--- a/DocParser.cpp
+++ b/DocParser.cpp
@@ -3,6 +3,17 @@
 //
 
 #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);
+    }
+}
 
 void DocParser::parse(const string& filename) {
 //    cout << "NEW DOC: " << filename << endl;
@@ -39,7 +50,7 @@ void DocParser::parse(const string& filename) {
 //        cout << "text: " << text << endl;
         Word curr(text.substr(0, space));
         curr.toLower();//remove caps
-        if (curr.isStopWord()){
+        if (isStopWord(curr.getStr())){
             text = text.substr(space + 1);//cut off curr word
             space = text.find(" ");
             continue;//don't add to tree
@@ -48,6 +59,8 @@ void DocParser::parse(const string& filename) {
         curr.stemming();
 //        cout << "current: " << curr.getStr() << endl;
         //put unique words into the avl tree
+//        Node<Word>& found = words.find(words.getRoot(), curr);//ref to word in tree
+
         if (!words.contains(curr)){//if the word is not already in the tree/new unique word
             curr.incrFreq(currDoc);//TODO combine contains and find
             words.insert(curr);
@@ -80,6 +93,11 @@ vector<Document>& DocParser::findIndex(Word& obj)
     return words.find(words.getRoot(), obj).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(); }
diff --git a/DocParser.h b/DocParser.h
index 4c50ecd..4188dbd 100644
--- a/DocParser.h
+++ b/DocParser.h
@@ -23,11 +23,14 @@ class DocParser {
     DSAVLTree<Word> words;
     DSAVLTree<Word> orgs;
     DSAVLTree<Word> people;
+    DSAVLTree<string> stops;
 
 public:
+    DocParser();
     void parse(const string& filename);//parse the documents and create 3 AVLTrees
     void getFiles(const string& directory);//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);
diff --git a/Word.cpp b/Word.cpp
index d227050..5470429 100644
--- a/Word.cpp
+++ b/Word.cpp
@@ -47,20 +47,19 @@ void Word::removePunc()
     str = buffer;
 }
 
-bool Word::isStopWord()
-{
-    ifstream stop;
-    stop.open("stopWords.txt");
-    string curr;
-    while (getline(stop, curr))//go through entire list of stop words
-    {
-        string s = curr.substr(0, curr.length()-1);//cut off end char
-        if (s == str){//the current string is a stop word
-            return true;
-        }
-    }
-    return false;
-}
+//bool Word::isStopWord()
+//{
+//    DSAVLTree<string> stops;
+//    ifstream stop;
+//    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);
+//    }
+//    return stops.contains(str);//if str is in the avl tree, its a stop word
+//}
 
 void Word::stemming()
 {
@@ -68,7 +67,7 @@ void Word::stemming()
     Porter2Stemmer::stem(str);
 }
 
-string Word::getStr() { return str; }
+string& Word::getStr() { return str; }
 vector<Document>& Word::getDocs() { return docs; }
 void Word::printDocs()
 {
diff --git a/Word.h b/Word.h
index 3f43311..05c187f 100644
--- a/Word.h
+++ b/Word.h
@@ -11,6 +11,7 @@
 #include <iostream>
 #include "Document.h"
 #include "porter2_stemmer.h"
+#include "DSAVLTree.h"
 
 using namespace std;
 class Word {
@@ -18,6 +19,7 @@ class Word {
     string str;
     vector<Document> docs;//documents word appears in
     vector<int> frequency;//frequency for each doc word appears in
+//    DSAVLTree<string> stops;//stop words
     int total = 0;
 
 public:
@@ -28,10 +30,10 @@ class Word {
     bool operator==(const Word& w);
     void toLower();
     void removePunc();
-    bool isStopWord();
+//    bool isStopWord();
     void stemming();//using porter2 stemming library
 
-    string getStr();
+    string& getStr();
     vector<Document>& getDocs();
     void printDocs();
     void incrFreq(Document& doc);//given a doc, incrememnt its corresponding freq.
diff --git a/main.cpp b/main.cpp
index 2c0d3c6..3a49e08 100644
--- a/main.cpp
+++ b/main.cpp
@@ -16,12 +16,13 @@ int main(int argc, char** argv) {
 //    cout << parse.getWordTree().getCount();
     Word w("investors");
     w.stemming();
-    if (parse.getWordTree().contains(w)){
-        cout << "true" << endl;
-        parse.getWordTree().find(parse.getWordTree().getRoot(), w).printDocs();
-    }
-    //data folder took 1:45 4/17
+//    if (parse.getWordTree().contains(w)){
+//        cout << "true" << endl;
+//        parse.getWordTree().find(parse.getWordTree().getRoot(), w).printDocs();
+//    }
+    parse.getWordTree().find(parse.getWordTree().getRoot(), w).printDocs();
 
+    //data folder took 1:45 4/17
 //    DSAVLTree<Word> tree;
 //    Word a("a");
 //    Word z("z");