C++ - Line parser stalls when searching for string with underscore

Question

I have written a function that parses a line from a text file and replaces cerain keywords with the appropriate counterpart in a list.

the code looks like this:

std::string __iec_parse_cl(const std::string &line){
    std::string ret = line;

    static const size_t num_tokens = 6;
    static const char *tokens_search[6]{
        filled ...
    };

    static const char *tokens_replace[6]{
        filled ...
    };

    for(size_t x = 0; x < num_tokens; x++){
        size_t yolo;
        do{
            yolo = ret.find(tokens_search[x];
            if(yolo != std::string::npos){
                ret.erase(yolo, strlen(tokens_search[x]));
                ret.insert(yolo, tokens_replace[x]);
            }
        } while(yolo != std::string::npos);
    }

    return ret;
}

When I parse a token that looks like this: globalid and replace it with this: get_global_id everything is fine...

BUT when I try to parse a token that looks like THIS: global_id and try to replace it with this: get_global_id the program stalls somewhere in the function :/

What could be causing this?

You should change the name of your function, otherwise the behavior of your code is undefined. http://stackoverflow.com/questions/228783/what-are-the-rules-about-using-an-underscore-in-a-c-identifier — Benjamin Lindley, Dec 08 '13 at 02:55
I understand the naming conventions, but I'm fairly certain the standard library or the gcc developers aren't going to use that function name :P haha — RamblingMad, Dec 08 '13 at 03:08

Joe Z · Accepted Answer · 2013-12-08T02:57:53.343

That's because your search needs to start after the replacement. If you were to print out your string after each replacement, you would see that it's got get_global_id then get_get_global_id then get_get_get_global_id, etc.

What you need to do is to tell find to start after your last replacement.

    size_t yolo = 0;
    do{
        yolo = ret.find(tokens_search[x], yolo);
        if (yolo != std::string::npos){
            ret.erase(yolo, strlen(tokens_search[x]));
            ret.insert(yolo, tokens_replace[x]);
            yolo += strlen(tokens_replace[x]);
        }
    } while(yolo != std::string::npos);

Doing this will prevent other issues as well, if there are other overlaps between your tokens and their replacements.

It's an easy error to make. I think my first attempt at a macro preprocessor umpteen years ago died a similar ignominious death. Only, because it was in C, with ugly pointers and mallocs and such, it had many more reasons for crashing, making it that much harder to find. — Joe Z, Dec 08 '13 at 03:20

C++ - Line parser stalls when searching for string with underscore

1 Answers1