正则表达式嵌套方括号（忽略方括号和空格）

Question

I am trying to create a regex pattern that reads through a bibTex citation file and match everything inside the brackets. For those who don't know, a bibtex citation looks like the following :

@INPROCEEDINGS{Fogel95,
  AUTHOR =       {L. J. Fogel and P. J. Angeline and D. B. Fogel},
  TITLE =        {An evolutionary programming approach to self-adaptation
                    on finite state machines},
  BOOKTITLE =    {Proceedings of the Fourth International Conference on
                    Evolutionary Programming},
  YEAR =         {1995},
  pages =        {355--365}
}

@ARTICLE{Goldberg91,
  AUTHOR =       {D. Goldberg},
  TITLE =        {Real-coded genetic algorithms, virtual alphabets, and blocking},
  JOURNAL =      {Complex Systems},
  YEAR =         {1991},
  pages =        {139--167}
}

@INPROCEEDINGS{Yao96,
  AUTHOR =       {X. Yao and Y. Liu},
  TITLE =        {Fast evolutionary programming},
  BOOKTITLE =    {Proceedings of the 6$^{th}$ Annual Conference on Evolutionary
                    Programming},
  YEAR =         {1996},
  pages =        {451--460}
}

The current pattern I have is as follows:

@(\\w+)\{(\\w+),\\s*((\\w+)\\s*=\\s*(\\"|\\{)?(.+)(\\"|\\})?,?\\s*)+\\}

This pattern matches the second citation but only parts of the first and third. I know the reason it doesn't match the third citation is because of the brackets within the left hand side of the citation ( 6$^ { th } $ ) and I have figured out that it won't match citations that have whitespaces/newlines within the left hand side of the citation elements

BOOKTITLE =    {Proceedings of the Fourth International Conference on
                Evolutionary Programming},
//This part of the citation has a newline in the middle of it.

Now I have been slaving away trying to fix my pattern, but the thing with regular expressions that I have found, is that the longer I try to fix the expression/add new conditions to it, the more confusing it gets. I am just wondering how I capture the whole citation regardless of inner brackets/parenthesis. Some citations contain no brackets/parenthesis after the "=" sign at all. Any help, along with an explanation would be greatly appreciated. I have looked at similar examples which have only confused me more due to the difficulty of deciphering a regular expression by simply glancing at it. Thank you.

Answer 1

The simplest way to capture everything between curly braces is:

\{([^}]+)}

The negation [^}] includes all character not a curly bracket, including newlines.

Answer 2

Regex is not a good parser for text with nested blocks.

If you insist on using regex, you should match the outer part first:

@INPROCEEDINGS{Fogel95,
  ???
}

Capture the ??? , so you can match on that in a nested loop.

The outer regex would be something like @(\\w+)\\{(\\w+),([^{}]*(?:\\{[^{}]*\\}[^{}]*)*)\\}

The inner regex would be something like (\\w+)\\s*=\\s*\\{([^}]*)\\}

Since a field value may be wrapped on multiple lines, you need to unwrap that.

Code

Pattern pTag = Pattern.compile("@(\\w+)" + // tag
                               "\\{" +
                                  "(\\w+)" + // name
                                  "," +
                                  "([^{}]*(?:\\{[^{}]*\\}[^{}]*)*)" + // content
                               "\\}");
Pattern pField = Pattern.compile("(\\w+)" + // field
                                 "\\s*=\\s*" +
                                 "\\{" +
                                    "([^}]*)" + // value
                                 "\\}");
Pattern pNewline = Pattern.compile("\\s*(?:\\R\\s*)+");
for (Matcher mTag = pTag.matcher(input); mTag.find(); ) {
    String tag = mTag.group(1);
    String name = mTag.group(2);
    String content = mTag.group(3);
    for (Matcher mField = pField.matcher(content); mField.find(); ) {
        String field = mField.group(1);
        String value = mField.group(2);
        value = pNewline.matcher(value).replaceAll(" ");
        System.out.printf("%-15s %-12s %-11s %s%n", tag, name, field, value);
    }
}

Test input

String input = "@INPROCEEDINGS{Fogel95,\n" +
               "  AUTHOR =       {L. J. Fogel and P. J. Angeline and D. B. Fogel},\n" +
               "  TITLE =        {An evolutionary programming approach to self-adaptation\n" +
               "                    on finite state machines},\n" +
               "  BOOKTITLE =    {Proceedings of the Fourth International Conference on\n" +
               "                    Evolutionary Programming},\n" +
               "  YEAR =         {1995},\n" +
               "  pages =        {355--365}\n" +
               "}\n" +
               "\n" +
               "@ARTICLE{Goldberg91,\n" +
               "  AUTHOR =       {D. Goldberg},\n" +
               "  TITLE =        {Real-coded genetic algorithms, virtual alphabets, and blocking},\n" +
               "  JOURNAL =      {Complex Systems},\n" +
               "  YEAR =         {1991},\n" +
               "  pages =        {139--167}\n" +
               "}\n" +
               "\n" +
               "@INPROCEEDINGS{Yao96,\n" +
               "  AUTHOR =       {X. Yao and Y. Liu},\n" +
               "  TITLE =        {Fast evolutionary programming},\n" +
               "  BOOKTITLE =    {Proceedings of the 6$^{th}$ Annual Conference on Evolutionary\n" +
               "                    Programming},\n" +
               "  YEAR =         {1996},\n" +
               "  pages =        {451--460}\n" +
               "}";

Output

INPROCEEDINGS   Fogel95      AUTHOR      L. J. Fogel and P. J. Angeline and D. B. Fogel
INPROCEEDINGS   Fogel95      TITLE       An evolutionary programming approach to self-adaptation on finite state machines
INPROCEEDINGS   Fogel95      BOOKTITLE   Proceedings of the Fourth International Conference on Evolutionary Programming
INPROCEEDINGS   Fogel95      YEAR        1995
INPROCEEDINGS   Fogel95      pages       355--365
ARTICLE         Goldberg91   AUTHOR      D. Goldberg
ARTICLE         Goldberg91   TITLE       Real-coded genetic algorithms, virtual alphabets, and blocking
ARTICLE         Goldberg91   JOURNAL     Complex Systems
ARTICLE         Goldberg91   YEAR        1991
ARTICLE         Goldberg91   pages       139--167

Answer 3

作为最好的，我可以告诉大家，安德烈亚斯的解决方案可能是更好的，但如果你想只是打破了整个字符串到一个数组正则表达式的字符串，您可以使用此： @(.*){(.*),\\s*(.*?)\\s*=\\s*{(.*?)},(?:\\s*(.*) =\\s*{([\\s\\S]*?)},)*?(?:\\s*?(.*?) =\\s*?{(.*?)})*?\\s*?}