Regex matching takes forever

Question

I have a simple text file of about 30 lines, as follows:

A 45.0 X 250.0  
Y -23.8
A 22.0 X -1016.0
Y 4.9
L 0
M 16.1 T 0 N 0

...etc...
...etc...

Every line contains 1 to 3 pairs of a capital letter and a float. All separators are spaces.
I then made a program which checks every single line of that file with this simple regex:

std::ifstream stream;
try
{
    stream.open("/ArchivioProgrammi/p1.prg", std::ios::in);
}
catch(std::exception& e)
{
    std::stringstream st;
    st << "Error opening file.\nError:   " << e.what();
    throw MyLoadException(st.str());
}
std::string line;
unsigned int lineCount = 0;

while(std::getline(stream, line))  //  Per ogni linea del file estraggo fino a 6 token
{
    if(line.length() == 0) continue;

    /***  This "if" hangs indefinitely!!  ***/
    if(!std::regex_match(line, std::regex("([A-Z] [\\-\\+]?[0-9]+(\\.[0-9]+)?)( [A-Z] [\\-\\+]?[0-9]+(\\.[0-9]+)?){0,2}")))
    {
        std::stringstream st;
        st << "Line #" << lineCount << " is invalid.";
        throw MyLoadException(st.str());
    }

    //  Tokenize the line and do something with the tokens...

    ++lineCount;

}

if(stream.eof())
{
    stream.close();
}

Whenever the control reaches the if, the programs hangs forever! Unresponsive interface, no errors, no false or true! Why?

I'm developing with KDevelop under OpenSuse latest version.

Matching a regular expression is not trivial. It taking a long time is something you should expect.. — Jesper Juhl, Feb 12 '18 at 19:22
@JesperJuhl That does not agree with my experience using regexes. — melpomene, Feb 12 '18 at 19:23
I see nothing fishy about the regex; no loops, no ambiguities, no potentially overlapping matches. — melpomene, Feb 12 '18 at 19:27
`std::regex` can be quite expensive to create as the create a FSM. What happens if you move the creation of the regex out of the if statement and just pass it to `std::regex_match`? — NathanOliver, Feb 12 '18 at 19:31
So you have a problem. You think, "I know, I can solve it with a regex!". Now you have two problems; the original problem, and you are using regex. — Yakk - Adam Nevraumont, Feb 12 '18 at 19:42
See [this thread](https://stackoverflow.com/questions/36304204/%D0%A1-regex-segfault-on-long-sequences), try using `-stdlib=libc++` — Wiktor Stribiżew, Feb 12 '18 at 19:50
@WiktorStribiżew I get "unrecognized command line option" when inserting your flag in the KDevelop project configuration. — Ernesto_Che, Feb 12 '18 at 20:19
@Jarod42: Neither the `-` nor the `+` require escaping at all, as far as I can tell. — Lightness Races in Orbit, Feb 12 '18 at 21:07
@LightnessRacesinOrbit: `-` might need escaping inside `[]`, but indeed not here. I have just replace first part of OP regex in fact... would be relevant only for `\.`... — Jarod42, Feb 12 '18 at 22:16
@Jarod42: Not if it's at the start or end of the character class. — Lightness Races in Orbit, Feb 13 '18 at 00:34
`std::regex re(R"((?:[A-Z] [-+]?[0-9]+(\.[0-9]+)? ?){1,3})"); std::smatch matches; std::regex_match(line, matches, re); if(matches.size() != 1) { .... }` Still, the creation of this regex takes forever. Why? — Ernesto_Che, Feb 14 '18 at 11:15

Ernesto_Che · Accepted Answer · 2018-02-13T20:43:21.813

0

Using raw string it works like a charm. Even if I have never understood the meaning of beginning such strings with both double quotes and a parenthesis....

How coud I have written the regex in a shorter form?

edited Feb 13 '18 at 20:43

answered Feb 12 '18 at 20:52

Ernesto_Che

67
2
8

How coud I have written the regex in a shorter form? – Ernesto_Che Feb 13 '18 at 20:58

score 0 · Answer 2 · answered Feb 12 '18 at 21:00

0

Try confirming that your regex pattern is correct by testing it online (e.g. online Regex tester). You may also want to consider iterating through the input file line by line using regex on one line at a time. Or just not use regex at all.

answered Feb 12 '18 at 21:00

Cogitator

154
1
1

I'm not sure of what yo're advising. Going line by line? That's exactly what I said I'm doing! – Ernesto_Che Feb 13 '18 at 20:32
Sorry - I see from your code that you are reading line by line. I also tested your regex pattern and it does find matches in your text file. Start commenting out your code (e.g. if(!std::regex_match) until you find what is making it hang. Hope you solve it. – Cogitator Feb 20 '18 at 22:29

Regex matching takes forever

2 Answers2