How do I write a simple regular expression pattern matching function in C or C++?

Question

This is a question in my paper test today, the function signature is

int is_match(char* pattern,char* string)

The pattern is limited to only A

Answer 1

Here is recursive extendable implementation. Tested for first order of pattern complexity.

#include 
#include 
#include 
#include 

struct Match {
  Match():_next(0) {}
  virtual bool match(const char * pattern, const char * input) const {
    return !std::strcmp(pattern, input);
  }
  bool next(const char * pattern, const char * input) const {
    if (!_next) return false;
    return _next->match(pattern, input);
  }
  const Match * _next;
};

class MatchSet: public Match {
  typedef std::vector Set;
  Set toTry;
public:
  virtual bool match(const char * pattern, const char * input) const {
    for (Set::const_iterator i = toTry.begin(); i !=toTry.end(); ++i) {
      if ((*i)->match(pattern, input)) return true;
    }
    return false;
  }
  void add(Match * m) {
    toTry.push_back(m);
    m->_next = this;
  }
  ~MatchSet() {
    for (Set::const_iterator i = toTry.begin(); i !=toTry.end(); ++i)
      if ((*i)->_next==this) (*i)->_next = 0;
  }
};

struct MatchQuestion: public Match  {
  virtual bool match(const char * pattern, const char * input) const {
    if (pattern[0] != '?')
      return false;
    if (next(pattern+1, input))
      return true;
    if (next(pattern+1, input+1))
      return true;
    return false;
  }
};

struct MatchEmpty: public Match {
  virtual bool match(const char * pattern, const char * input) const {
    if (pattern[0]==0 && input[0]==0)
      return true;
    return false;
  }
};

struct MatchAsterisk: public Match {
  virtual bool match(const char * pattern, const char * input) const {
    if (pattern[0] != '*')
      return false;
    if (pattern[1] == 0) {
      return true;
    }
    for (int i = 0; input[i] != 0; ++i) {
      if (next(pattern+1, input+i))
        return true;
    }
    return false;
  }
};

struct MatchSymbol: public Match {
  virtual bool match(const char * pattern, const char * input) const {
    // TODO: consider cycle here to prevent unnecessary recursion
    // Cycle should detect special characters and call next on them
    // Current implementation abstracts from that
    if (pattern[0] != input[0])
      return false;
    return next(pattern+1, input+1);
  }
};

class DefaultMatch: public MatchSet {
  MatchEmpty empty;
  MatchQuestion question;
  MatchAsterisk asterisk;
  MatchSymbol symbol;
public:
  DefaultMatch() {
    add(&empty);
    add(&question);
    add(&asterisk);
    add(&symbol);
  }
  void test(const char * p, const char * input) const {
    testOneWay(p, input);
    if (!std::strcmp(p, input)) return;
    testOneWay(input, p);
  }
  bool testOneWay(const char * p, const char * input) const {
    const char * eqStr = " == ";
    bool rv = match(p, input);
    if (!rv) eqStr = " != ";
    std::cout << p << eqStr << input << std::endl;
    return rv;
  }

};


int _tmain(int argc, _TCHAR* argv[])
{
  using namespace std;

  typedef vector Strings;
  Strings patterns;

  patterns.push_back("*");
  patterns.push_back("*hw");
  patterns.push_back("h*w");
  patterns.push_back("hw*");

  patterns.push_back("?");
  patterns.push_back("?ab");
  patterns.push_back("a?b");
  patterns.push_back("ab?");

  patterns.push_back("c");
  patterns.push_back("cab");
  patterns.push_back("acb");
  patterns.push_back("abc");

  patterns.push_back("*this homework?");
  patterns.push_back("Is this homework?");
  patterns.push_back("This is homework!");
  patterns.push_back("How is this homework?");

  patterns.push_back("hw");
  patterns.push_back("homework");
  patterns.push_back("howork");

  DefaultMatch d;
  for (unsigned i = 0; i < patterns.size(); ++i)
    for (unsigned j =i; j < patterns.size(); ++j)
      d.test(patterns[i].c_str(), patterns[j].c_str());

    return 0;
}

If something is unclear, ask.