javascript regex select text in [] from a string

Question

I'm not too familiar with regular expressions in javascript. I'm writing a validation plugin with jquery. I got all my regular expressions done except these two.

//get number in [] from string
foo='text[number]'
//get two numbers from [] the numbers are separated by -
bar='string[num1-num2]'

the numbers can be floats, the length can be anything and there are no spaces. I know I can get the position of portions of the string and get the values that way but I know there should be a way to get it with regular expressions.

Out of interest, can the numbers be negative? For example, is `string[123--456]` valid? And if that is the case, would `abc[-123]` be considered valid in the first example or invalid in the second? — Jonathon Bolster, Oct 12 '11 at 16:19

score 2 · Answer 1 · answered Oct 12 '11 at 16:28

var myregexp = /\[([\-+]?\b\d*\.?\d+\b)(?:\b-([\-+]?\d*\.?\d+\b))?\]/;
var match = myregexp.exec(subject);
if (match != null) {
    num1= match[1];
    num2 = match[2]; //maybe null if you have single number!
} else {
    num1 = num2 = "";
}

This will match any floating points inside your brackets unless it is specified in a scientific form.

Test :

//get number in [] from string
foo='text[20938423.20938]'
//get two numbers from [] the numbers are separated by -
bar='string[23.20-23]'
//another test
bar='string[23.20-.9873]'

Output :

Group 1 :

20938423.20938
23.20
23.20

Group 2 :

23
.9873

score 1 · Accepted Answer · edited May 23 '17 at 12:19

Regex to match a decimal integer or floating point "number":

Crafting an accurate regex to match a "number" is not as simple as it may first appear! (Just ask tchrist who explained it quite thoroughly in his excellent: "What’s a Number?" answer). For the purpose of the answer here, lets assume a "number" is defined as follows:

All numbers are decimal and no suffixes are allowed. (No octal, hex or binary numbers.)
The integer, fraction and exponent parts consist of a sequence of decimal digits.
The integer and exponent parts may be prefixed with an optional plus or minus sign and may have leading zeroes.
Either the integer part or the fraction part may be missing, but not both.
The exponent part is optional.

Given these requirements, here are examples of valid "number" representations, divided into two types; Type 1: numbers having an integer portion and Type 2: numbers without an integer portion.

Type 1: Required integer part. Optional decimal point with optional fractional digits and optional exponent:

1 +1 -1
1. +1. -1.
1.1 +1.1 -1.1
1E1 1E+1 1E-1
1.E1 1.E+1 1.E-1
1.1E1 1.1E+1 1.1E-1

Type 2: No integer part. Required decimal point and fractional digits. Optional exponent.

.1 +.1 -.1
.1E1 .1E+1 .1E-1

Here is a regex which matches all of the above number representations (written in PHP free-spacing format for readability):

$re_number = '/
    # Match a decimal integer or floating point "number".
    [+\-]?      # Optional leading sign.
    (?:         # Non-capture group for types 1 & 2.
      \b        # Either type 1: Number has integer part.
      [0-9]+    # Required whole integer part.
      (?:       # Optional fractional part.
        \.      # Required decimal point.
        [0-9]*  # Zero or more fractional digits.
      )?        # Fractional part is optional.
    |           # or Type 2: Number has no integer part.
      \B        # Leading "." is NOT on a word boundary.
      \.        # Required decimal point.
      [0-9]+    # Required fractional digits.
    )           # End group of type 1 & 2 alternatives.
    (?:         # Optional exponent.
      [Ee]      # Exponent begins with required literal E
      [+\-]?    # Optional leading sign.
      [0-9]+    # Required exponent number part.
    )?          # End optional exponent.
    /x';

Here is the same regex in Javascript syntax (with all comments removed):

var re_number =
    /[+\-]?(?:\b[0-9]+(?:\.[0-9]*)?|\B\.[0-9]+)(?:[Ee][+\-]?[0-9]+)?/;

Solution to OP question:

The OP wants to match either one or two "numbers" enclosed within square brackets. If there are two numbers, they are separated by a dash. Given the above defintion of a "number", here is a regex solution which meets these requirements:

$re_1_or_2_numbers_in_brackets = '/
    # Match one or two "numbers" inside square brackets.
    \[            # Literal left opening square bracket.
    [+\-]?        # Optional leading sign.
    (?:           # Non-capture group for types 1 & 2.
      \b          # Either type 1: Number has integer part.
      [0-9]+      # Required whole integer part.
      (?:         # Optional fractional part.
        \.        # Required decimal point.
        [0-9]*    # Zero or more fractional digits.
      )?          # Fractional part is optional.
    |             # or Type 2: Number has no integer part.
      \B          # Leading "." is NOT on a word boundary.
      \.          # Required decimal point.
      [0-9]+      # Required fractional digits.
    )             # End group of type 1 & 2 alternatives.
    (?:           # Optional exponent.
      [Ee]        # Exponent begins with required literal E
      [+\-]?      # Optional leading sign.
      [0-9]+      # Required exponent number part.
    )?            # End optional exponent.
    (?:           # Group for optional second "number".
      -           # Required - separator for second number.
      [+\-]?      # Optional leading sign.
      (?:         # Non-capture group for types 1 & 2.
        \b        # Either type 1: Number has integer part.
        [0-9]+    # Required whole integer part.
        (?:       # Optional fractional part.
          \.      # Required decimal point
          [0-9]*  # Zero or more fractional digits.
        )?        # Fractional part is optional.
      |           # or Type 2: Number has no integer part.
        \B        # Leading "." is NOT on a word boundary.
        \.        # Required decimal point.
        [0-9]+    # Required fractional digits.
      )           # End group of type 1 & 2 alternatives.
      (?:         # Optional exponent.
        [Ee]      # Exponent begins with required literal E
        [+\-]?    # Optional leading sign.
        [0-9]+    # Required exponent number part.
      )?          # End optional exponent.
    )?            # Second number is optional.
    \]            # Literal right closing square bracket.
    /x';

Here is the same regex in Javascript syntax (with all comments removed):

var re_1_or_2_numbers_in_brackets =
    /\[[+\-]?(?:\b[0-9]+(?:\.[0-9]*)?|\B\.[0-9]+)(?:[Ee][+\-]?[0-9]+)?(?:-[+\-]?(?:\b[0-9]+(?:\.[0-9]*)?|\B\.[0-9]+)(?:[Ee][+\-]?[0-9]+)?)?\]/;

This solution correctly matches all of the following variations:

[1-1] [+1-+1] [-1--1]
[1.-1.] [+1.-+1.] [-1.--1.]
[1.1-1.1] [+1.1-+1.1] [-1.1--1.1]
[1E1-1E1] [1E+1-1E+1] [1E-1-1E-1]
[1.E1-1.E1] [1.E+1-1.E+1] [1.E-1-1.E-1]
[1.1E1-1.1E1] [1.1E+1-1.1E+1] [1.1E-1-1.1E-1]
[.1-.1] [+.1-+.1] [-.1--.1]
[.1E1-.1E1] [.1E+1-.1E+1] [.1E-1-.1E-1]

thank you. can you please explain \b and \B in more detail. what I have read on it online is vague and I dont quite understand it. a word boundary would be the first or last letter in a string? — Yamiko, Oct 12 '11 at 23:38
the other detail I don't understand is ?: E.g: /(d).\1/ matches and captures 'dad' in "abcdadef" while /(?:.d){2}/ matches but doesn't capture 'cdad'. from http://www.javascriptkit.com/javatutors/redev2.shtml doesnt seem to be right or am I greatly misunderstanding what they're saying. — Yamiko, Oct 12 '11 at 23:45
The `\b` is a word boundary assertion which matches a _position_ between a word character and a non-word character. With Javascript a word character (which can be represented with a `\w` or `[A-Za-z0-9_]` ), is a character that is a digit or a letter or an underscore. The `\B` is an assertion of a position that is NOT on a word boundary. As you have already surmised, `(...)` is a capturing parentheses and `(?:...)` is a non-capturing parentheses. Check out [regular-expressions.info](http://www.regular-expressions.info/) for more regex fun! — ridgerunner, Oct 13 '11 at 00:45

javascript regex select text in [] from a string

2 Answers2

Regex to match a decimal integer or floating point "number":

Solution to OP question: