Question

Is it possible to have a .net regex pattern that references the numeric value of a previously captured group in another part of the pattern? I need to match strings that repeat the following format:

• 10-byte alpha key followed by 2-byte operator (subset of sql operators) followed by 5 integer digits followed by an n-byte value, where n = the integer value of the 5 integer digits.

EX String:
“key1 = 00004val1key2 <=00006value2key3 >=00011value_three”

Where val1 is 4-bytes, value2 is 6-bytes, value_three is 11-bytes, etc…. This string represents repeating set of criteria that I need to be able to parse and identify the following strings:

1st Criteria:
“key1”, “= “, “val1”

2nd Criteria:
“key2”, “<=”, “value2”

3rd Criteria:
“key3”, “>=”, “value_three”

I’ve used .net regex’s in the past and should be able to figure out parsing the key, operator and digits in a repeating pattern, but don’t know how, or even if, it is possible to reference the numeric value of the digits (i.e. an integer with value of 4 for the string “00004”) later in the pattern.

Update 1
Originally I was thinking that I needed a regex pattern such as:

^(?<criteria>(?<key>\w{10})(?<operator>(= |<=|>=))(?<value_length>\d{5})(?<value>\w{n}))+$

where the {n} quantifier in the “value” capturing group needs to have a value based on the value of the “value_length” capturing group, for each “criteria” captured. If this is not the right way to think about forming the regex pattern, I can change to another approach that works.

Answer 1

You can instead capture the required length string in the match itself till the next \w{10} and operator (= |<=|>=) using lookahead

string regex=@"(?<key>\w{10})(?<operator>(= |<=|>=))(?<value_length>\d{5})(?<value>.*?(?=\w{10}(= |<=|>=)|$))";
foreach(Match m in Regex.Matches(input,regex,RegexOptions.Singleline))
{
    int vLen=int.Parse(m.Groups["value_length"].Value);
    string requiredString=m.Groups["value"].Value.Substring(0,vLen-1);
}