[Boost-users] boost::regex: Unexpected behaviour

Subject: [Boost-users] boost::regex: Unexpected behaviour
From: Pascal Kesseli (pascal_kesseli_at_[hidden])
Date: 2009-07-31 06:39:35

Hi everyone

Please have a look at the following code:

1:    const regex REPORT_HEADER("^[^()]+((?:\\([^()]*\\)){4,})$");
2:    const regex REPORT_DATA_ITEM("\\(([^()]*?)\\)(.*)");
3:    string headerLine =
"P.01(1090618220000)(0008)(15)(3)(1.5.0)(kW)(5.5.0)(kvar)(8.5.0)(kvar)";
4:    smatch match;
5:    if(regex_match(headerLine, match, REPORT_HEADER)) {
6:        regex_match(lexical_cast<string>(match[1]), match,
REPORT_DATA_ITEM);
7:        clog << match[1] << endl;
8:        regex_match(lexical_cast<string>(match[2]), match,
REPORT_DATA_ITEM);
9:        clog << match[1] << endl;
10:       regex_match(lexical_cast<string>(match[2]), match,
REPORT_DATA_ITEM);
11:       clog << match[1] << endl;
12:       regex_match(lexical_cast<string>(match[2]), match,
REPORT_DATA_ITEM);
13:       unsigned int numberOfEntries = lexical_cast<unsigned
int>(match[1]);
14:       string tmp;
15:       for(unsigned int i = 0; i < numberOfEntries; ++i) {
16:           tmp = match[2];
17:           clog << "1: " << tmp << endl;
18:           if(regex_match(tmp, match, REPORT_DATA_ITEM)) {
19:               clog << "2: " << match[1] << endl;
20:               tmp = match[2];
21:               clog << "3: " << tmp << endl;
22:               if(regex_match(tmp, match, REPORT_DATA_ITEM)) {
23:                   clog << "4: " << match[1] << endl;
24:               }
25:           }
26:       }
27:   }

Important is actually only the trailing for loop. Running the code as it
reads above, the output is as expected:

1090618220000
0008
15
1: (1.5.0)(kW)(5.5.0)(kvar)(8.5.0)(kvar)
2: 1.5.0
3: (kW)(5.5.0)(kvar)(8.5.0)(kvar)
4: kW
1: (5.5.0)(kvar)(8.5.0)(kvar)
2: 5.5.0
3: (kvar)(8.5.0)(kvar)
4: kvar
1: (8.5.0)(kvar)
2: 8.5.0
3: (kvar)
4: kvar

However, changing line 18 to:
if(regex_match(lexical_cast<string>(match[2]), match, REPORT_DATA_ITEM)) {

and thus eliminating the temporary variable tmp, results in the following
output:

kW
kvar
1090618220000
0008
15
1: (1.5.0)(kW)(5.5.0)(kvar)(8.5.0)(kvar)
1: 
1:

What might be the problem here? Am I misusing the match_results class in
some way?

Thanks for any help and best regards
Pascal