first commit

.gitignore

@@ -0,0 +1,4 @@
+*.layout
+*.depend
+bin
+obj

Query.cpp

@@ -0,0 +1,92 @@
+#include <iostream>
+#include <cstdio>
+#include <cstdlib>
+#include <cstring>
+#include <fstream>
+#include <sstream>
+#include "lib/Query.h"
+using namespace std;
+
+Query::Query(string query)
+{
+    text = query;
+    count = 0;
+}
+
+node* Query::open_list(Lexicon &lexicon, LexiconCursor &lexicon_cursor)
+{
+    int offset = 0, doc_id = 0, frequency = 0;
+    string inverted_list;
+    stringstream parseable_inverted_list;
+    node* head = NULL;
+
+    offset = get_offset( lexicon, lexicon_cursor );
+    inverted_list = get_inverted_list( offset ); // remember to seek to beg after done
+
+    parseable_inverted_list << inverted_list;
+
+    while( parseable_inverted_list >> doc_id >> frequency ) {
+        node *temp, *temp2;
+
+        count++;
+
+        temp = new node;
+        temp->doc_id = doc_id;
+        temp->frequency = frequency;
+        temp->next = NULL;
+
+        if (head == NULL)
+            head = temp;
+        else {
+            temp2 = head;
+
+            while (temp2->next != NULL)
+                temp2 = temp2->next;
+
+            temp2->next = temp;
+        }
+    }
+
+    return head;
+}
+
+int Query::get_offset(Lexicon &lexicon, LexiconCursor &lexicon_cursor)
+{
+    lexicon_cursor = lexicon.find( text );
+    return ( *lexicon_cursor ).second;
+}
+
+string Query::get_inverted_list(int offset)
+{
+    string inverted_list;
+    ifstream inverted_index;
+
+    inverted_index.open( "structures/inverted_index" );
+    inverted_index.seekg( offset, ios::beg );
+    getline( inverted_index, inverted_list );
+    inverted_index.close();
+
+    return inverted_list;
+}
+
+/*
+void Query::close_list(node* head)
+{
+    free( head );
+}
+
+int Query::get_frequency(int document_id)
+{
+    int doc_id = 0, frequency = 0;
+    string inverted_list;
+    stringstream parseable_inverted_list;
+
+    while( parseable_inverted_list >> doc_id >> frequency ) {
+        if( doc_id == document_id )
+            return frequency;
+    }
+
+    return 0;
+}
+
+*/

QueryProcessor.cpp

@@ -0,0 +1,62 @@
+#include <iostream>
+#include <cstdio>
+#include <cstdlib>
+#include <cstring>
+#include <fstream>
+#include <sstream>
+#include <cmath>
+#include "lib/QueryProcessor.h"
+
+#define CONSTANT_K 1.2
+#define CONSTANT_B 0.75
+using namespace std;
+
+QueryProcessor::QueryProcessor(int query_count, vector<Query> user_queries)
+{
+    num_queries = query_count;
+    queries = user_queries;
+}
+
+/*int QueryProcessor::nextGEQ( node* list_head, int doc_id)
+{
+    node* head = NULL;
+
+    head = list_head;
+
+    if( head->doc_id >= doc_id )
+        return doc_id;
+    else
+        head = head->next;
+
+    return -1;
+}
+
+double QueryProcessor::calculate_rank(int doc_id)
+{
+    int total_pages, total_pages_with_queryword, freq_of_query_in_doc, doc_length;
+    double K, page_rank, log_result,freq_result,avg_doc_length;
+
+    K = CONSTANT_K* ( (1-CONSTANT_B) + ( CONSTANT_B * (doc_length/avg_doc_length) ) );
+
+    total_pages_with_queryword = Query.count();
+    freq_of_query_in_doc = Query.get_frequency();
+
+    log_result = log( (total_pages - freq_of_query_in_doc + 0.5) / ( freq_of_query_in_doc + 0.5) );
+    freq_result = ( ((CONSTANT_K+1) * total_pages_with_queryword) / (K+total_pages_with_queryword) );
+
+    page_rank = log_result * freq_result;
+
+    return page_rank;
+}
+
+string QueryProcessor::get_url(int doc_id)
+{
+    url_table_iterator = kyon_url_table.find( doc_id );
+    return ( *url_table_iterator ).second.url;
+}
+
+void QueryProcessor::add_to_heap(string url, int doc_id)
+{
+    //add url and doc_id
+}
+*/