Feedback

.gitignore

@@ -1,4 +1,5 @@
 data/
 cmake-build-debug/
 cmake-build-release/
+persistence_index/
 .idea/

CMakeLists.txt

@@ -1,6 +1,8 @@
-cmake_minimum_required(VERSION 3.20)
+cmake_minimum_required(VERSION 3.16)
 project(22s_final_proj)
 
-set(CMAKE_CXX_STANDARD 14)
+set(CMAKE_CXX_STANDARD 17)
 
-add_executable(22s_final_proj main.cpp catch_setup.cpp)
+configure_file(file_cleanup/stop_words.txt NEWLINE_STYLE UNIX)
+
+add_executable(22s_final_proj main.cpp data_structures/AVLTree.h Term.cpp Term.h engine_architecture/Parser.cpp engine_architecture/Parser.h file_cleanup/stop_words.txt file_cleanup/porter2_stemmer-master/porter2_stemmer.h file_cleanup/porter2_stemmer-master/porter2_stemmer.cpp file_cleanup/porter2_stemmer-master/util/hash.h file_cleanup/porter2_stemmer-master/util/string_view.h engine_architecture/IndexHandler.cpp engine_architecture/IndexHandler.h engine_architecture/QueryProcessor.cpp engine_architecture/QueryProcessor.h engine_architecture/SearchEngine.cpp engine_architecture/SearchEngine.h DSDoc.cpp DSDoc.h DSDoc.cpp DSDoc.h)

Catch_Tests/AVL_tests.cpp

@@ -0,0 +1,23 @@
+#include "../catch.hpp"
+#include "../data_structures/AVLTree.h"
+
+TEST_CASE("AVL Tree Implementation") {
+
+    SECTION("AVL Node") {
+        AVLNode<int> node (5, nullptr, nullptr, 0);
+        CHECK(node.left == nullptr);
+        CHECK(node.right == nullptr);
+        CHECK(node.element == 5);
+    }
+
+    SECTION("Case 1 functionality") {
+        AVLTree<int> tree;
+        tree.insert(3);
+        tree.insert(2);
+        tree.insert(1);
+        tree.insert(4);
+        tree.insert(6);
+
+    }
+
+}

DSDoc.cpp

@@ -0,0 +1,112 @@
+//
+// Created by wnahl on 5/2/2022.
+//
+
+#include "DSDoc.h"
+
+using namespace std;
+
+DSDoc::DSDoc(std::string filepath, std::vector<std::string> terms) {
+    this->filepath = filepath;
+    this->terms = terms;
+    string cur_line;
+    string whole_file;
+    ifstream in(filepath);
+    while (getline(in, cur_line)) {
+        whole_file += cur_line + '\n';
+    }
+    in.close();
+    doc.Parse(whole_file.c_str());
+    relevancy = get_relevancy();
+}
+
+DSDoc::DSDoc(const DSDoc& other) {
+    this->filepath = other.filepath;
+    string cur_line;
+    string whole_file;
+    ifstream in(other.filepath);
+    while (getline(in, cur_line)) {
+        whole_file += cur_line + '\n';
+    }
+    in.close();
+    this->doc.Parse(whole_file.c_str());
+    this->relevancy = other.relevancy;
+    this->terms = other.terms;
+}
+
+DSDoc& DSDoc::operator= (const DSDoc& other) {
+    this->filepath = other.filepath;
+    string cur_line;
+    string whole_file;
+    ifstream in(other.filepath);
+    while (getline(in, cur_line)) {
+        whole_file += cur_line + '\n';
+    }
+    in.close();
+    this->doc.Parse(whole_file.c_str());
+    this->relevancy = other.relevancy;
+    this->terms = other.terms;
+    return *this;
+}
+
+std::string DSDoc::get_title() {
+    string title = doc["title"].GetString();
+    return title;
+}
+
+std::string DSDoc::get_text() {
+    string text = doc["text"].GetString();
+    return text;
+}
+
+std::string DSDoc::get_publication() {
+    string publication = doc["thread"]["site"].GetString();
+    return publication;
+
+}
+
+std::string DSDoc::get_url() {
+    string url = doc["url"].GetString();
+    return url;
+}
+
+std::string DSDoc::get_date_published() {
+    string date = doc["published"].GetString();
+    return date;
+}
+
+double DSDoc::get_relevancy() {
+    double relevancy = 0.0;
+    double correct_words = 0;
+    int total_words = 0;
+    string text = doc["text"].GetString();
+
+    string word = "";
+    for (int i = 0; i < text.size(); i++) {
+        if (text[i] == ' ') {
+            //CHECK IF WORD IS STOP WORD?
+
+            for (int j = 0; j < terms.size(); j++) {
+                if (word == terms[j])
+                    correct_words+= 1;
+            }
+            total_words++;
+            word = "";
+        } else {
+            word += text[i];
+        }
+    }
+
+    relevancy = correct_words/total_words;
+
+    return relevancy;
+}
+
+bool DSDoc::operator<(const DSDoc &other) const {
+    if (this->relevancy < other.relevancy ) {
+        return true;
+    }
+    return false;
+}
+
+

DSDoc.h

@@ -0,0 +1,34 @@
+#ifndef INC_22S_FINAL_PROJ_DSDOC_H
+#define INC_22S_FINAL_PROJ_DSDOC_H
+
+#include <string>
+#include <vector>
+#include <fstream>
+#include "rapidjson/rapidjson.h"
+#include "rapidjson/document.h"
+
+class DSDoc {
+public:
+    DSDoc(std::string filepath, std::vector<std::string> terms);
+
+    DSDoc(const DSDoc& other);
+    DSDoc& operator= (const DSDoc& other);
+
+    std::string get_title();
+    std::string get_text();
+    std::string get_publication();
+    std::string get_url();
+    std::string get_date_published();
+    double get_relevancy();
+
+    bool operator<(const DSDoc& other) const;
+
+private:
+    std::string filepath;
+    std::vector<std::string> terms;
+    rapidjson::Document doc;
+    double relevancy;
+};
+
+
+#endif //INC_22S_FINAL_PROJ_DSDOC_H

Term.cpp

@@ -0,0 +1,53 @@
+
+
+using namespace std;
+
+#include "Term.h"
+
+Term::Term(std::string term) {
+    this->term = term;
+}
+
+void Term::add_document(string current_doc_number) {
+    documents_vector.push_back(current_doc_number);
+}
+
+bool Term::operator==(const Term &T) {
+    return (term == T.term);
+}
+
+// < operator is necessary for the AVL trees to balance correctly
+bool Term::operator<(const Term &T) {
+    return (term < T.term);
+}
+
+// Using the contains function so we don't duplicate documents for a term
+bool Term::contains(string& docID) {
+    for (int i = 0; i < documents_vector.size(); i++) {
+        if (docID == documents_vector[i]) {
+            return true;
+        }
+    }
+    return false;
+}
+
+vector<string>& Term::get_IDs() {
+    return documents_vector;
+}
+
+std::string& Term::get_term() {
+    return term;
+}
+
+ostream& operator<<(ostream& out, Term& t) {
+    out << t.get_term() << " ";
+    for (int i = 0; i < t.get_IDs().size(); i++) {
+        out << t.get_IDs()[i] << " ";
+    }
+    out << endl;
+    return out;
+}
+
+
+
+

Term.h

@@ -0,0 +1,32 @@
+#ifndef INC_22S_FINAL_PROJ_TERM_H
+#define INC_22S_FINAL_PROJ_TERM_H
+
+#include <vector>
+#include <string>
+#include <iostream>
+
+class Term {
+public:
+    Term(std::string term); //The AVL trees consist of "Terms"
+    void add_document(string current_doc_number);
+
+    bool operator==(const Term& T);
+    bool operator<(const Term& T);
+
+
+    bool contains(string& docID);
+    vector<string>& get_IDs();
+    std::string& get_term();
+
+    friend ostream& operator<<(ostream& out, Term& t);
+
+private:
+    //Each term has a string which represents a word
+    //and the vector represents all of the documents
+    //where this word is found
+    std::vector<string> documents_vector;
+    std::string term;
+};
+
+
+#endif //INC_22S_FINAL_PROJ_TERM_H