From 419bebcadeb077415aa17c98fc70276a2ca4eef6 Mon Sep 17 00:00:00 2001
From: Wang Xiaojian <xwa88@sfu.ca>
Date: Tue, 17 Jun 2014 21:33:55 +0800
Subject: [PATCH] Implemented binary search

---
 simple_ir/Dic.cpp     | 31 +++++++++++++++++++++++--------
 simple_ir/List.cpp    |  2 +-
 simple_ir/Posting.cpp |  1 +
 simple_ir/main.cpp    |  2 +-
 simple_ir/test.cpp    |  6 ++++++
 5 files changed, 32 insertions(+), 10 deletions(-)

diff --git a/simple_ir/Dic.cpp b/simple_ir/Dic.cpp
index d530a01..791279f 100644
--- a/simple_ir/Dic.cpp
+++ b/simple_ir/Dic.cpp
@@ -3,26 +3,33 @@
 #include<cstring>
 #include<iostream>
 #include <string>
+#include <algorithm>
 
 using namespace std;
 
 Dic::Dic()
 {
     this->lists = vector<List>();
+    sorted = false;
+}
+
+bool compareLists(List list1, List list2)
+{
+    return (list1.getTerm() < list2.getTerm());
 }
 
 List* Dic::getListByTerm(string term)
 {
-    // Should only allow this operation when we have the list sorted by term.
-    // In that case we can do a binary search.
-    // Currently we do a linear search.
-    for(vector<List>::iterator list = lists.begin(); list != lists.end(); list++)
+    if(!sorted)
     {
-        if(list->getTerm() == term)
-        {
-            return &(*list);
-        }
+        cout << "Lists not sorted yet!" << endl;
+        return NULL;
     }
+    List searchFor(term);
+    // Binary search
+    auto results = equal_range(lists.begin(), lists.end(), searchFor, compareLists);
+    if(results.first != lists.end())
+        return &(*results.first);
     return NULL;
 }
 
@@ -30,4 +37,12 @@ void Dic::addList(string term)
 {
     List list(term);
     this->lists.push_back(list);
+    sorted = false;
+}
+
+
+void Dic::sortLists()
+{
+    sort(lists.begin(), lists.end(), compareLists);
+    sorted = true;
 }
\ No newline at end of file
diff --git a/simple_ir/List.cpp b/simple_ir/List.cpp
index 7e05640..73e3fb3 100644
--- a/simple_ir/List.cpp
+++ b/simple_ir/List.cpp
@@ -13,7 +13,7 @@ List::List(string term)
     this->term = term;
 }
 
-string List::getTerm()
+string List::getTerm() const
 {
     return term;
 }
diff --git a/simple_ir/Posting.cpp b/simple_ir/Posting.cpp
index 2b568eb..e149568 100644
--- a/simple_ir/Posting.cpp
+++ b/simple_ir/Posting.cpp
@@ -9,6 +9,7 @@ Posting::Posting(string docId, int fq)
     this->fq = fq;
     next=NULL;
 }
+
 int Posting::freq()
 {
     return fq;
diff --git a/simple_ir/main.cpp b/simple_ir/main.cpp
index 1b14ebc..79966ab 100644
--- a/simple_ir/main.cpp
+++ b/simple_ir/main.cpp
@@ -9,7 +9,7 @@ using namespace std;
 
 int main(int argc, char** argv)
 {
-    testList();
+    testDic();
 
     //TokenReader::readAndLowerTokensFromFile("../Reuters/10.html");
 
diff --git a/simple_ir/test.cpp b/simple_ir/test.cpp
index 3f51b0f..8be85a1 100644
--- a/simple_ir/test.cpp
+++ b/simple_ir/test.cpp
@@ -7,15 +7,18 @@ using namespace std;
 void testDic()
 {
     Dic* dic = new Dic();
+    dic->sortLists();
     List* list = dic->getListByTerm("a");
     if(list != NULL)
         cout << "Error!" << endl;
     dic->addList("a");
+    dic->sortLists();
     list = dic->getListByTerm("a");
     if(list->getTerm() != "a")
         cout << "Error!" << endl;
     dic->addList("b");
     dic->addList("c");
+    dic->sortLists();
     list = dic->getListByTerm("a");
     if(list->getTerm() != "a")
         cout << "Error!" << endl;
@@ -35,6 +38,7 @@ void testDic()
 void testList()
 {
     List *list = new List("a");
+    
     if(list->getTerm() != "a")
         cout << "Error!" << endl;
     if(list->getLength() != 0)
@@ -54,5 +58,7 @@ void testList()
     posting = posting->next;
     if(posting != NULL)
         cout << "Error!" << endl;
+    if(list->getLength() != 3)
+        cout << "Error!" << endl;
     cout << "If no error, then success!" << endl;
 }
\ No newline at end of file