Add parsing regular expressions

CMakeLists.txt

@@ -19,7 +19,9 @@ set(SOURCE_FILES
   src/regular/RegularTree.cpp
   )
 
-set(TEST_FILES)
+set(TEST_FILES
+  tests/regular/parse_regular.cpp
+  )
 
 add_executable(Formalang src/main.cpp ${SOURCE_FILES})
 

include/regular/RegularTree.hpp

@@ -10,20 +10,30 @@ class RegularTree {
  public:
   class Node {
    public:
+    enum class Type {
+      Addition, Concatenation, Word
+    };
+    enum class Modifier {
+      Plus, Star, None
+    };
+
     Node();
+    Node(Type);
 
     void Parse(const std::string&);
-    void Parse(const std::string_view);
+    void Print() const;
-
+    std::vector<std::unique_ptr<Node>> children;
-    enum class Type {
+    std::string word;
-      Addition, Concatenation
-    };
-    std::variant<std::vector<std::unique_ptr<Node>>, std::string> value;
     Type type;
+    Modifier modifier = Modifier::None;
    private:
+    void ParseCurrentType(const std::string_view);
+    void Print(int nesting_level) const;
   };
 
   RegularTree(const std::string&);
+  const Node& GetNode() const;
+  void Print() const;
 
  private:
   Node node_;

src/main.cpp

@@ -3,5 +3,8 @@
 
 int main() {
   using namespace regular;
-  RegularTree reg_tree("(ab)*");
+  std::string str;
+  std::cin >> str;
+  RegularTree reg_tree(str);
+  reg_tree.Print();
 }

src/regular/RegularTree.cpp

@@ -1,22 +1,188 @@
 #include "regular/RegularTree.hpp"
 #include <iostream>
+#include <exception>
 
 namespace regular {
 RegularTree::Node::Node() {}
+RegularTree::Node::Node(Type type) : type(type) {}
 
 void RegularTree::Node::Parse(const std::string& regular) {
-  Parse(std::string_view(regular.c_str(), regular.size()));
+  type = Type::Addition;
+  ParseCurrentType(std::string_view(regular.c_str(), regular.size()));
 }
 
-void RegularTree::Node::Parse(const std::string_view regular) {
+void RegularTree::Node::Print() const {
-  const int n = regular.size();
+  Print(0);
-  for (int i = 0; i < n; ++i) {
+}
-    std::cout << regular[i];
+
+void RegularTree::Node::ParseCurrentType(const std::string_view regular) {
+  const size_t n = regular.size();
+  children.clear();
+  word.clear();
+
+  auto AddChild = [this](const std::string_view regular) {
+    this->children.push_back(std::make_unique<Node>(Type::Addition));
+    this->children.back()->ParseCurrentType(regular);
+  };
+
+  if (n == 1) {
+    type = Type::Word;
+  }
+
+  if (type == Type::Addition) {
+    int balance = 0;
+    int begin_child = 0;
+    bool wrapped_brackets = (regular[0] == '(');
+
+    for (size_t i = 0; i < n; ++i) {
+      if (regular[i] == '(') {
+        ++balance;
+      } else if (regular[i] == ')') {
+        --balance;
+      }
+
+      if (regular[i] == '|') {
+        if (balance == 0) {
+          AddChild(regular.substr(begin_child + wrapped_brackets, i -
+                                     begin_child - 2 *
+                                     wrapped_brackets));
+          begin_child = i + 1;
+
+          if (i + 1 < n)
+            wrapped_brackets = (regular[i + 1] == '(');
+        }
+      } else {
+        if (i + 1 == n) {
+          if (children.size() == 0) {
+            type = Type::Concatenation;
+            break;
+          } else {
+            AddChild(regular.substr(begin_child + wrapped_brackets, i -
+                                    begin_child + 1 - 2 *
+                                    wrapped_brackets));
+            begin_child = i + 1;
+            if (i + 1 < n)
+              wrapped_brackets = (regular[i + 1] == '(');
+          } 
+        } else if (balance == 0) {
+          wrapped_brackets = false;
+        }
+      } 
+    }
+  }
+
+  if (type == Type::Concatenation) {
+    int balance = 0;
+    int begin_child = 0;
+
+    for (size_t i = 0; i < n; ++i) {
+      if (regular[i] == '(') {
+        ++balance;
+        if (balance == 1) {
+          if (begin_child < i) {
+            AddChild(regular.substr(begin_child, i - begin_child));
+            begin_child = i + 1;
+          }
+        } 
+      } else if (regular[i] == ')') {
+        --balance;
+        if (balance == 0) {
+          AddChild(regular.substr(begin_child + 1, i - begin_child - 1));
+          begin_child = i + 1;
+        }
+      } else if (i + 1 == n) {
+        if (balance != 0) {
+          throw std::logic_error("invalid regular");
+        }
+        if (children.size() == 0) {
+          type = Type::Word;
+          break;
+        } else {
+          AddChild(regular.substr(begin_child, i - begin_child + 1));
+          begin_child = i + 1;
+        } 
+      } else if (balance == 0) {
+        if (regular[i] == '+') {
+          if (begin_child < i) {
+            AddChild(regular.substr(begin_child, i - begin_child));
+          }
+          children.back()->modifier = Modifier::Plus;
+          begin_child = i + 1;
+        } else if (regular[i] == '*') {
+          if (begin_child < i) {
+            AddChild(regular.substr(begin_child, i - begin_child));
+          }
+          children.back()->modifier = Modifier::Star;
+          begin_child = i + 1;
+        }
+      }
+    }
+  }
+
+  if (type == Type::Word) {
+    bool exist_modifire = regular.back() == '+' || regular.back() == '*';
+    if (regular.back() == '+') {
+      modifier = Modifier::Plus;
+    } else if (regular.back() == '*') {
+      modifier = Modifier::Star;
+    }
+    for (size_t i = 0; i < n - exist_modifire; ++i) {
+      if (regular[i] == '|' || regular[i] == '(' || regular[i] == ')') {
+        throw std::logic_error("invalid regular");
+      }
+      word += regular[i];
+    }
+  }
+}
+
+void RegularTree::Node::Print(int nesting_level) const {
+  auto PrintNesingLevel = [](int nesting_level) {
+    for (int i = 0; i < nesting_level; ++i) {
+      std::cout << "  ";
+    }
+  };
+
+  PrintNesingLevel(nesting_level);
+  if (type == Type::Addition) {
+    std::cout << "Addition";
+    if (modifier == Modifier::Plus) {
+      std::cout << "+";
+    } else if (modifier == Modifier::Star) {
+      std::cout << "*";
+    }
+    std::cout << " " << std::to_string(children.size()) << ":" << std::endl;
+  } else if (type == Type::Concatenation) {
+    std::cout << "Concatenation";
+    if (modifier == Modifier::Plus) {
+      std::cout << "+";
+    } else if (modifier == Modifier::Star) {
+      std::cout << "*";
+    }
+    std::cout << " " << children.size() << ":" << std::endl;
+  } else if (type == Type::Word) {
+    std::cout << "Word";
+    if (modifier == Modifier::Plus) {
+      std::cout << "+";
+    } else if (modifier == Modifier::Star) {
+      std::cout << "*";
+    }
+    std::cout << ": " << word << std::endl;
+  }
+
+  for (const auto& i: children) {
+    i->Print(nesting_level + 1);
   }
-  std::cout << std::endl;
 }
 
 RegularTree::RegularTree(const std::string& regular) {
   node_.Parse(regular);
 }
+
+const RegularTree::Node& RegularTree::GetNode() const {
+  return node_;
+}
+
+void RegularTree::Print() const {
+  node_.Print();
+}
 }

tests/regular/parse_regular.cpp

@@ -0,0 +1,40 @@
+#include <gtest/gtest.h>
+#include "regular/RegularTree.hpp"
+
+using namespace regular;
+
+TEST(parse_regular, only_addition) {
+  RegularTree("a");
+  RegularTree("aa");
+  RegularTree("aaa");
+  RegularTree("aaaa");
+  RegularTree("aaaaa");
+  RegularTree("a|aaaa");
+  RegularTree("a|a|a|a|a");
+  RegularTree("hello|world");
+  RegularTree("qe|wr|lkj|alk");
+}
+
+TEST(parse_regular, only_folding) {
+  RegularTree("(kajfkasf(aksdjf)jka(((aksjdf)K)))jakd");
+  RegularTree("(kajsdfk(aksdjf)kajsdf)kjasdkfja(skdjf(((aksjdkadf)ksjf(kdja))))");
+  RegularTree("((((kdjf))))");
+  RegularTree("kasjf(akjsfkjasdg)kajsdg");
+  RegularTree("akjsdf(akjdf(kdjfak(jkasdf)))");
+  RegularTree("123k4j1k351kk21jkj21k6j2k36j1(((((ajkfajsdfkafdalkdjflk)))))");
+}
+
+TEST(parse_regular, only_modifiers) {
+  RegularTree("jlakjdf*aksdjflaf8*laksfj*lakjsf*alksjdf");
+  RegularTree("jlakjdf*aksdjflaf8+laksfj*lakjsf*alksjdf*");
+  RegularTree("jlakjdf*aksdjflaf8*laksfj*lakjsf+alksjdf");
+  RegularTree("jlakjd+aksdjflaf8*laksfj*lakjsf*alksjdf+");
+}
+
+TEST(parse_regular, all_operations) {
+  RegularTree("(alkjdfaksdf*|lkasdj*|(kasjdf|kajdf*|kjd)*|laksjf*)+");
+  RegularTree("(alkjdfaksdf|lkasdj*|(kas+jdf|kajdf*|kjd)*|laksjf*)+");
+  RegularTree("a|(a|(a|(a|)*))*");
+  RegularTree("kj*|kjadf*|(kajsdf|(kajsd|kjadf|(kasjdf)|kajs)*)*");
+}
+