Archive

Posts Tagged ‘find’

Regular expressions in C++ with Glib::Regex

November 30, 2013 No comments

In our programming life, there are some “before and after”s, and one of them is when we learn to use regular expressions… and they are like Twitter, you start with it, follow some famous people and a couple of friends, write a test tweet and a “how wonderful life is” tweet, and forget it. But when it’s your time, you can’t stop using it. So regular expressions or “regex” are the same, when you discover them you say: “Oh! It’s nice!”, or “I could do a lot with it”, but after some time (maybe weeks, months or years), when you have a strings problem, the first solution you try is a regex.

One of the common commands using regex is grep, of course this system is too good to be used only in one place. This is the reason why lots of programming languages have functions or classes to use them easily, for example, PHP had ereg_* in the past, now we use preg_*, in Javascript we use RegExp class, in Java we can even use the String class to parse regex, and so on.

But when working in C++ we don’t have native solutions for that, at least in std, ok C++11 has, but we don’t always have a C++11 compiler ready. We have to use libraries as Boost or Glib to support them, if we don’t want to do it by hand.

We are going to do it with Glib. Imagine we are making a template. Some keywords will be replaced with calculated values. Keywords will begin and end with a %, so we want to get the position of these keywords, and which keyword has been discovered:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
#include <glibmm/regex.h>
#include <glibmm/ustring.h>
#include <iostream>

using namespace std;
using namespace Glib;

int main()
{
  Glib::ustring str1 = "Hi %name%, your friend %friend% told me you are a %job%.";
  cout << "Original string: "<< str1 << endl;

  Glib::RefPtr<Regex> myr = Regex::create("%[a-z]*%");
  MatchInfo minfo;

  myr->match(str1, minfo);
  int start, end;
  int i=0;

  while (minfo.matches())
    {
      cout << "Word: " << minfo.fetch(0)<<endl;
      if (minfo.fetch_pos(0, start, end))
        {
          cout << "   Start:  "<<start<<endl<<"   End: "<<end<<endl;
        }
      minfo.next();
      ++i;
    }
    cout << "Occurrences: "<<i<<endl;
}

To compile it, we must have glibmm installed, then:

$ g++ -o regex1 regex1.cpp `pkg-config –libs –cflags glibmm-2.4`

In this piece of code, we can see, the regex “%[a-z]*%” has been applied, so we can get lowercase letters from a to z enclosed between % symbols. In the sample string we’ve found 3 occurrences, printing on screen start position, end position and the matched string for each one.

It can be enough for many cases, but this example will return strings like %name% or %friend%, which in certain cases it is not useful, we want name or friend, ok, we can handle that, but we can get those values with regex too applying a parenthesis in the regex, enclosing what we want, this way: “%([a-z]*)%”, in other words, we are interested in this part of the string. But we will obtain several values. One of them will be the old string, the entire match and not only the part we are interested in. But if we change the code a little bit, we’ll be able to get it:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
#include <glibmm/regex.h>
#include <glibmm/ustring.h>
#include <iostream>

using namespace std;
using namespace Glib;

int main()
{
  Glib::ustring str1 = "Hi %name%, your friend %friend% told me you are a %job%.";
  cout << "Original string: "<< str1 << endl;

  Glib::RefPtr<Regex> myr = Regex::create("%([a-z]*)%");
  MatchInfo minfo;

  myr->match(str1, minfo);
  int start, end;
  int i=0;

  while (minfo.matches())
    {
      cout << "Match "<< i+1 << ": "<<endl;

      for (unsigned j = 0; j< minfo.get_match_count(); ++j)
    {
      cout << "Word ("<<j<<"): " << minfo.fetch(j)<<endl;
      if (minfo.fetch_pos(j, start, end))
        {
          cout << "   Start:  "<<start<<endl<<"   End: "<<end<<endl;
        }
    }
      minfo.next();
      ++i;
    }
    cout << "Occurrences: "<<i<<endl;
}

In this case, we are iterating get_match_count() times, so we will get the number of strings returned by each match of the expression (expressions can be so complex, and we can add more parenthesis). Calling minfo.fetch(1) we will get the strings: “name”, “friend” and “job”.

But, to write a better example, let’s parse a simple XML tag. As regex we are taking: “<([\\w:]*)( [^<>]*)?>([^<>]*)</\\1>“, that means:

  • Symbol <
  • a word, letters and numbers
  • maybe a space and several characters, different than < and >
  • Symbol >
  • Several characters. Neither < nor >
  • Symbols < and /
  • The same word found in the beginning
  • Symbol >

Then our text string will be: “Sample text

And the result will be:

Original String: <MyTag id=”123″>Sample text</MyTag>
Match 1:
Word (0): <MyTag id=”123″>Sample Text</MyTag>
Start: 0
End: 51
Word (1): MyTag
Start: 1
End: 6
Word (2): id=”123″
Start: 6
End: 15
Word (3): Sample text
Start: 16
End: 43
Occurrences: 1

(Note: Start and End position, won’t match reality, they were taken by another example)

So, with this little regex we have parsed this XML tag, this would be useful in little projects.

Foto: li xiang (Flickr) CC-by

Replace substrings inside strings in C++

October 12, 2013 No comments

One of the most useful tools when programming is searching and replacing text from a string. In other words, searching for substrings inside a bigger string and replacing them with other substrings. For example, we can use templates to generate a message or our desired output, we just have to replace some keywords located inside our template (our big string) with our generated data.

We will do it calling standard string methods to find this substring and then replacing it with another one:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
#include <iostream>
#include <string>

using namespace std;

int main()
{
  string original = "Going to sleep. Everyone knows past 12 we must go to sleep.";

  string::size_type pos = original.find("sleep", 0);

  cout << "Original string: "<<original<<endl;

  if (pos < string::npos)
    original.replace(pos, string("sleep").length(), "code");

  cout << "Resulting string: "<<original<<endl;
  return 0;
}

To compile:

$g++ -o replace replace.cpp

In this case, we use method find to search for the position where the substring “sleep” starts. If the string is not found, it returns string::npos (a constant), but if it is, we will replace it with the word “code”, so we will have to say the start position inside original, and the size of the substring we want to replace.

Now, what we want is to replace all occurrences of “sleep” inside the string, so we just have to loop find() and replace() while we are still finding the substring, this way:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
#include <iostream>
#include <string>

using namespace std;

int main()
{
  string original = "Going to sleep. Everyone knows past 12 we must go to sleep.";

  string::size_type pos = 0;

  // No we are using these variables with the substring to be replaced and the replacement.
  string fromStr = "sleep";
  string toStr = "code";

  cout << "Original: "<<original<<endl;

  // Loop while finding substrings
  while ((pos = original.find(fromStr, pos)) < string::npos)
    {
      original.replace(pos, fromStr.length(), toStr);
      pos+=toStr.size();    // Very important, add the replacement substring size
                                // to pos to avoid infinite loops
    }

  cout << "Resulting string: "<<original<<endl;
  return 0;
}

If we run this code, the two occurrences of “sleep” will be replaced by “code”. But we find a interesting comment in the code: “to avoid infinite loops”, what we do is adding the size of the string “code” to the pos variable. This would be a special case when the substring to be replaced is contained inside the replacement string (or they both are the same), in other words, if we search “to” and we want to replace it by “ton” and we comment this line. It will do the replacement with no ending, and if we cout the string inside the loop we will se “tonnnnnn … with a growing number of n”.

We have now a good way to do search and replace, so let’s create a function to simplify the usage, but with an added value: we can give an offset value (so we can choose where to start searching in the original string, just giving pos a non-zero value), and a counter, so we can replace fromStr with toStr a number of times, so if our keyword appears ten times, we can choose to replace it five times.

Let’s do this:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
#include <iostream>
#include <string>

using namespace std;

string replace(string source, string fromStr, string toStr, int offset=0, int times=0)
{
  int total = 0;
  string::size_type pos=offset;
  while ( ( (pos = source.find(fromStr, pos)) < string::npos) && ( (times==0) || (total++<times) ) )
    {
      source.replace(pos, fromStr.length(), toStr);
      pos+=toStr.size();
    }
  return source;
}

int main()
{
  string original = "Going to sleep. Everyone knows past 12 we must go to sleep. So if I sleep tonight, I don't have to sleep tomorrow morning";

  cout << "Original string: "<<original<<endl;

  cout << "Resulting string: "<<replace(original, "sleep", "code 20, 2)<<endl;

  return 0;
}

The output will be like this:

Original string: Going to sleep. Everyone knows past 12 we must go to sleep. So if I sleep tonight, I don’t have to sleep tomorrow morning
Resulting string: Going to sleep. Everyone knows past 12 we must go to code. So if I code tonight, I don’t have to sleep tomorrow morning.

So, we start searching substring “sleep” from character number 20, and then replace this substring twice. The replace function has initial values, so if we omit offset we will start from the beginning of the string, and if we omit the number of times, this will be zero, and it means all occurrences will be replaced.

To make it better, we can make use of Glib::ustring instead of string. Glib is a cross-platform utility library that implements its own version of string with UTF8 support. So if we use special characters like tildes or symbols we will have encoding problems. We can test it this way:

1
2
3
4
5
6
7
#include <string>

int main()
{
  cout << string("piñata").length() << endl;
  return 0;
}

Let’s see the output. We know “piñata” has six letters, but sometimes we will get seven because we are using a multibyte encoding (like UTF8 wich uses two bytes to encode “ñ”). There it goes Glib::ustring, this library (free, of course), behaves exactly like string (but with UTF-8 support. Let’s see:

1
2
3
4
5
6
7
8
9
10
11
12
13
#include <string>
#include <iostream>
#include <glibmm/ustring.h>

using namespace std;

int main()
{
  cout << Glib::ustring("piñata").length()<<endl;
  cout << string("piñata").length()<<endl;

  return 0;
}

To compile:

$ g++ -o wordlength wordlength.cpp `pkg-config –libs –cflags glibmm-2.4`

We will have to install the library glibmm (version 2.4 as October 2013)

In this example (if the source file is saved with utf-8 encoding), it will give us the right value

And of course, the replace function can be compatible with Glib::ustring, just replacing string with Glib::ustring:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
#include <iostream>
#include <glibmm/ustring.h>

using namespace std;

Glib::ustring replace(Glib::ustring source, Glib::ustring fromStr, Glib::ustring toStr, int offset=0, int times=0)
{
  int total = 0;
  Glib::ustring::size_type pos=offset;
  while ( ( (pos = source.find(fromStr, pos)) < Glib::ustring::npos) && ( (times==0) || (total++<times) ) )
    {
      source.replace(pos, fromStr.length(), toStr);
      pos+=toStr.size();
    }
  return source;
}

int main()
{
  Glib::ustring original = "Going to sleep. Everyone knows past 12 we must go to sleep. So if I sleep tonight, I don't have to sleep tomorrow morning.";

  cout << "Original string: "<<original.raw()<<endl;

  cout << "Resulting string: "<<replace(original, "sleep", "code").raw()<<endl;

  return 0;
}

Notice that I’ve used the method raw() to display the string with cout, this is because we have to pass just bytes to cout and not characters (in this case, multibyte characters).

Photo: Jon-Eric Melsæter (Flickr) CC-by

Top