Question

I would like some help with this email parser, if possible.

This code has all of its proper declaration and initialization, but I’m not sure where I am messing up this loop:

  while ( getline( fin, lines ) )
  {
     for ( int i = 0; i < lines.length( ); i++ )
     {
        if ( lines[ i ] == '@' )
        {
           for ( s = i; s < lines.length( ); s-- )
           {
              if ( s < 0 )
              {
                 break;
              }
              if ( validChar( lines[ s ] ) == false )
              {
                 break;
              }
           } //for

           for ( e = i; e > lines.length( ); e++ )
           {
              if ( e == lines.length( ) )
              {
                 break;
              }
              if ( validChar( lines[ e ] ) == false )
              {
                 break;
              }
              if ( lines[ e ] == '.' )
              {
                 hasDot = true;
              }
           } // for

           anEmail = lines.substr( s, e );

           cout << anEmail << endl;
        }
     } // if
  }    // while

And this is for the function:

  bool validChar( char a )
  {
     bool result = false;
     if ( a >= 'A' && a <= 'Z' || a >= 'a' && a <= 'z' || a >= '0' && a <= '9' || a == '.' || a == '-' || a == '+' )
     {
        result = true;
     }
     return result;
  }

EDIT: a test case being this string in a text file “this is an email file with [e-mail] and some other text for the test”, I want this “[e-mail]”, and I’m only getting this “@my.test and some other text for the test”

Where am I going wrong?

Answer 1

Your issue is here:

       for ( s = i; s < lines.length( ); s-- )
       {
          //...
          if ( validChar( lines[ s ] ) == false )
          {
             break;
          }
       } //for

You’re checking if lines[s] is valid to decide whether to terminate; but lines[s] == '@' initially, since you just found the @! If you initialize s to i - 1, you’ll be closer… but then you’ll discover that your substr has a bunch of off-by-ones in it. You’ll end up needing to do anEmail = lines.substr( s + 1, e + 1 );

But that will just cause the code to past your test cases. This is not a valid approach to parsing e-mail addresses. This approach will not work on all valid e-mail addresses, including "an@sign"@foo and "spaces are legal only in quotes"@foo. You’ll also want to extend validChar to deal with the actual set of valid characters, which varies for the name and domain; !#$%&'*+-/=?^_{}|~@[IPv6:2001:db8:1ff::a0b:dbd0] is perfectly legal. Finally, if it’s important to you to actually rule out illegal addresses, you’ll again be restricted by this approach: double..dot@foo is not legal, nor is double@at@foo.

The source for this is RFC822 (or it’s much newer siblings RFC5322 and RFC6531), where you’ll discover that a regular expression cannot parse e-mails, as name(comment(comment))@foo is legal, while name(comment))@foo is not.