From e39354cb15cc5875b0eaff953d9b9de3aed107a2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?David=20Doblas=20Jim=C3=A9nez?=
 <david.doblas-jimenez@xfel.eu>
Date: Mon, 19 Jul 2021 22:15:11 +0200
Subject: [PATCH] Module 9: Search the Library (C++ code only)

---
 a.c | 174 +++++++++++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 125 insertions(+), 49 deletions(-)

diff --git a/a.c b/a.c
index 8a20779..486509f 100644
--- a/a.c
+++ b/a.c
@@ -8,53 +8,124 @@ using namespace std;
 // For compiling C++ code
 // g++ a.c -o a
 
-class Library {
-  public:
-    void ComputeStats() {
-      assert(counts.empty());
-      counts.resize(18);
-      for (string s : words) {
-        int len = s.length();
-        if (len < 18) {
-          counts[len]++;
-        }
-      }
-    }
-    void PrintStats() const {
-      cout << "Here are the counts of each word length:\n";
-      for (int i=1; i<counts.size(); i++) {
-        cout << "[" << i << "] " << counts[i] << "\n";
-      }
-    }
-    string GetWord(int i) const {
-      assert(i >= 0 && i < words.size());
-      return words[i];
-    }
-    void ReadFromFile(string filename) {
-      ifstream f;
-      f.open(filename);
-      while (!f.eof())
-      {
-        string line;
-        getline(f, line);
-        // cout << line << "\n";
-        if (!line.empty())
-        {
-          int len = line.length();
-          if (line[len-1] == '\r') {
-            line = line.substr(0, len-1);
-          }
-          words.push_back(line);
-        }
-      }
-      cout << "Read " << words.size() << " words from file '"
-          << filename << "'\n";
-    }
-  private:
-    vector<string> words;
-    vector<int> counts;
-};
+string ToUpper(string s)
+{
+  string s2;
+  for (char c : s)
+  {
+    s2.push_back(toupper(c));
+  }
+  return s2;
+}
 
+struct Word
+{
+  Word(string s)
+  {
+    word = s;
+  }
+  string word;
+};
+typedef vector<Word> Words;
+
+const int num_buckets = 1001;
+
+// hash function
+int bucket(string s)
+{
+  assert(!s.empty());
+  int i = 0;
+  for (char c : s)
+  {
+    i = (i * 217) + c;
+  }
+  i = abs(i);
+  int b = i % num_buckets;
+  assert(b >= 0 && b < num_buckets);
+  return b;
+}
+
+class Library
+{
+public:
+  Library()
+  {
+    shelves.resize(num_buckets);
+  }
+  bool IsWord(string s) const
+  {
+    for (Word w : shelves[bucket(s)])
+    {
+      if (s == w.word)
+      {
+        return true;
+      }
+    }
+    return false;
+  }
+  void ComputeStats()
+  {
+    assert(counts.empty());
+    counts.resize(18);
+    for (Word w : words)
+    {
+      int len = w.word.length();
+      if (len < 18)
+      {
+        counts[len]++;
+      }
+    }
+  }
+  void PrintStats() const
+  {
+    cout << "Here are the counts of each word length:\n";
+    for (int i = 1; i < counts.size(); i++)
+    {
+      cout << "[" << i << "] " << counts[i] << "\n";
+    }
+  }
+  string GetWord(int i) const
+  {
+    assert(i >= 0 && i < words.size());
+    return words[i].word;
+  }
+  void ReadFromFile(string filename)
+  {
+    ifstream f;
+    f.open(filename);
+    while (!f.eof())
+    {
+      string line;
+      getline(f, line);
+      // cout << line << "\n";
+      if (!line.empty())
+      {
+        line = ToUpper(line);
+        int len = line.length();
+        if (line[len - 1] == '\r')
+        {
+          line = line.substr(0, len - 1);
+        }
+        words.push_back(Word(line));
+        shelves[bucket(line)].push_back(Word(line));
+      }
+    }
+    cout << "Read " << words.size() << " words from file '"
+         << filename << "'\n";
+  }
+  void DebugBuckets() const
+  {
+    for (int i = 0; i < shelves.size(); i++)
+    {
+      cout << "[" << i << "] " << shelves[i].size() << "\n";
+    }
+  }
+
+private:
+  Words words;
+  vector<Words> shelves;
+  vector<int> counts;
+};
 
 struct Grid
 {
@@ -106,7 +177,7 @@ struct Grid
       cout << "  " << s << "\n";
     }
   }
-  string name; // string are initialized empty
+  string name; // strings are initialized empty
   vector<string> lines;
 };
 
@@ -114,8 +185,13 @@ int main()
 {
   Library lib;
   lib.ReadFromFile("top_12000.txt");
-  lib.ComputeStats();
-  lib.PrintStats();
+  //lib.ComputeStats();
+  //lib.PrintStats();
+
+  cout << lib.IsWord("DOG") << "\n";
+  cout << lib.IsWord("CAT") << "\n";
+  cout << lib.IsWord("THANKS") << "\n";
+  lib.DebugBuckets();
 
   //Grid grid("MY GRID");
   //grid.LoadFromFile("test");