Module 12: Pattern Hash

a.c

@@ -22,30 +22,29 @@ string ToUpper(string s) {
 struct Word {
   Word() {}  // default empty constructor
   Word(string s) : word(s) {}  // standard way of initialization
+  int len() const { return word.length(); }
   string word;
 };
-typedef vector<Word> Words;
-typedef unordered_map<string, Word> WordMap;  // hash-table from stl
+typedef vector<Word*> Words;  // will store pointers to Words (will not delete by itself)
+typedef unordered_map<string, Words> WordMap;  // hash-table from stl
 
 // --------------------------------------------------------------------------------
 //-Library
 class Library {
 public:
   Library() {}  // hash-tables are automatically initialized
+  ~Library() {  // destructor of the pointers (instead of the unique pointer)
+    for (Word* w : words_) {
+      delete w;
+    }
+  }
   void FindWord(const string& s) const { // references are prefered instead of copies 
-    int len = s.length();  // cache the length to avoid multiple calls
-    for (const Word& w : words_) {  // references are prefered instead of copies
-      string temp = w.word;
-      if (len == temp.length()) {
-        for (int i=0; i<len; i++) {
-          if (s[i] == '-') {
-            temp[i] = '-';
-          }
-        }
-        if (s == temp) {
-          cout << w.word << "\n";  // print the original word
-        }
+    auto it = word_map_.find(s);
+    if (it != word_map_.end()) {
+      for (const Word* w : it->second) {
+        cout << " " << w->word;
       }
+      cout << "\n";
     }
   }
   bool IsWord(string s) const {
@@ -60,8 +59,8 @@ public:
   void ComputeStats() {
     assert(counts_.empty());
     counts_.resize(18);
-    for (Word w : words_) {
-      int len = w.word.length();
+    for (const Word* w : words_) {  // Word is a pointer!
+      int len = w->word.length();  // w is not an actual object (it's a pointer)
       if (len < 18) {
         counts_[len]++;
       }
@@ -75,7 +74,23 @@ public:
   }
   string GetWord(int i) const {
     assert(i >= 0 && i < words_.size());
-    return words_[i].word;
+    return words_[i]->word;
+  }
+  void CreatePatternHash(Word* w) {
+    int len = w->len();
+    int num_patterns = 1 << len;  // create 2^len patterns
+    // cout << "PATTERN HASH on " << w->word << "\n";
+    for (int i=0; i<num_patterns; i++) {
+      // cout << "  " << i << "\n";
+      string temp = w->word;
+      for (int j=0; j<len; j++) {
+        if ((i >> j) & 1) {  // get every bit and check if it's 1
+          temp[j] = '-';
+        }
+      }
+      // cout << "  " << temp << "\n";
+      word_map_[temp].push_back(w);
+    }
   }
   void ReadFromFile(string filename) {
     ifstream f;
@@ -90,8 +105,9 @@ public:
         if (line[len - 1] == '\r') {
           line = line.substr(0, len - 1);
         }
-        words_.push_back(Word(line));
-        word_map_[line] = Word(line);  // create entry for the hash-table
+        Word* w = new Word(line);
+        words_.push_back(w);  // Word would be allocated on the heap
+        CreatePatternHash(w);
       }
     }
     cout << "Read " << words_.size() << " words from file '"
@@ -104,8 +120,8 @@ public:
   }
 
 private:  // _ is used to indicate privacy
-  Words words_;
-  WordMap word_map_;
+  Words words_;  // master vector of words
+  WordMap word_map_;  // pattern hash
   vector<int> counts_;
 };