Develop Reference

.NET Regex - get parts of string that do not match pattern

I have this string
TEST_TEXT_ONE_20112017
I want to eliminate _20112017, which is a underscore with numbers, those numbers can vary; my goal is to have only
TEST_TEXT_ONE
So far I have this but I get the entire string, is there something I'm missing?
Regex r = new Regex(#"\b\w+[0-9]+\b");
MatchCollection words = r.Matches("TEST_TEXT_ONE_20112017");
foreach(Match word in words)
{
string w = word.Groups[0].Value;
//I still get the entire string
}

Notes for your consideration:
You should use parenthesis to mark groups for capture -or- use named group. The first group (index=0) is the entire match. you probably want index=1 instead.
\w stands for word character and it already includes both underscore and digits. If you want to match anything before the numbers then you should consider using . instead of \w.
by default +is greedy and your \w+ will consume your last undescore and all but the very last number as well. You probably want to explicitly require an underscore before last block of numbers.
I would suggest considering if you want to find a matching substring or the entire string to match. if the latter, then consider using the start and end markers: ^ and $.
if you know you want to eliminate 8 digits, then you could giving explicit count like \d{8}
For example this should work:
Regex r = new Regex(#"^(.+)_\d+$");
MatchCollection words = r.Matches("TEST_TEXT_ONE_20112017");
foreach (Match word in words)
{
string w = word.Groups[1].Value;
}
Alternative
Use a Zero-Width Positive Lookahead Assertions construct to check what comes next without capturing it. This uses the syntax on (?=stuff). So you could use a shorter code and avoid surfing in Groups altogether:
Regex r = new Regex(#"^.+(?=_\d+$)");
String result = r.Match("TEST_TEXT_ONE_20112017").Value;
Note that we require the end marker $ within the positive lookahead group.

Regex r = new Regex(#"(\b.+)_([0-9]+)\b");
String w = r.Match("TEST_TEXT_ONE_20112017").Groups[1].Value; //TEST_TEXT_ONE
or:
String w = r.Match("TEST_TEXT_ONE_20112017").Groups[2].Value; //20112017

This seems a bit overkill for Regex in my opinion. As an alternative you could just split on the _ character and rebuild the string:
private static string RemoveDate(string input)
{
string[] parts = input.Split('_');
return string.Join("_", parts.Take(parts.Length - 1));
}
Or if the date suffix is always the same length, you could also just substring:
private static string RemoveDateFixedLength(string input)
{
//Removes last 9 characters (8 for date, 1 for underscore)
return input.Substring(0, input.Length - 9);
}
However I feel like the first approach is better, this is just another option.
Fiddle here

How to split string by another string

I have this string (it's from EDI data):
ISA*ESA?ISA*ESA?
The * indicates it could be any character and can be of any length.
? indicates any single character.
Only the ISA and ESA are guaranteed not to change.
I need this split into two strings which could look like this: "ISA~this is date~ESA|" and
"ISA~this is more data~ESA|"
How do I do this in c#?
I can't use string.split, because it doesn't really have a delimeter.

You can use Regex.Split for accomplishing this
string splitStr = "|", inputStr = "ISA~this is date~ESA|ISA~this is more data~ESA|";
var regex = new Regex($#"(?<=ESA){Regex.Escape(splitStr)}(?=ISA)", RegexOptions.Compiled);
var items = regex.Split(inputStr);
foreach (var item in items) {
Console.WriteLine(item);
}
Output:
ISA~this is date~ESA
ISA~this is more data~ESA|
Note that if your string between the ISA and ESA have the same pattern that we are looking for, then you will have to find some smart way around it.
To explain the Regex a bit:
(?<=ESA) Look-behind assertion. This portion is not captured but still matched
(?=ISA) Look-ahead assertion. This portion is not captured but still matched
Using these look-around assertions you can find the correct | character for splitting

Simply use the
int x = whateverString.indexOf("?ISA"); // replace ? with the actual character here
and then just use the substring from 0 to that indexOf, indexOf to length.
Edit:
If ? is not known,
can we just use the regex Pattern and Matcher.
Matcher matcher = Patter.compile("ISA.*ESA").match(whateverString);
if(matcher.find()) {
matcher.find();
int x = matcher.start();
}
Here x would give that start index of that match.
Edit: I mistakenly saw it as java one, for C#
string pattern = #"ISA.*ESA";
Regex myRegex = new Regex(pattern, RegexOptions.IgnoreCase);
Match m = myRegex.Match(whateverString); // m is the first match
while (m.Success)
{
Console.writeLine(m.value);
m = m.NextMatch(); // more matches
}

RegEx will probably be the best for this. See this link
Mask would be
ISA(?<data1>.*?)ESA.ISA(?<data2>.*?)ESA.
This will give you 2 groups with data you need
Match match = Regex.Match(input, #"ISA(?<data1>.*?)ESA.ISA(?<data2>.*?)ESA.",RegexOptions.IgnoreCase);
if (match.Success)
{
var data1 = match.Groups["data1"].Value;
var data2 = match.Groups["data2"].Value;
}
Use Regex.Matches If you need multiple matches found, and specify different RegexOptions if needed.

It's kinda hacky but you could do...
string x = "ISA*ESA?ISA*ESA?";
x = x.Replace("*","~"); // OR SOME OTHER DELIMITER
string[] y = x.Split('~');
Not perfect in all situations, but it could solve your problem simply.

You could split by "ISA" and "ESA" and then put the parts back together.
string input = "ISA~this is date~ESA|ISA~this is more data~ESA|";
string start = "ISA",
end = "ESA";
var splitedInput = input.Split(new[] { start, end }, StringSplitOptions.None);
var firstPart = $"{start}{splitedInput[1]}{end}{splitedInput[2]}";
var secondPart = $"{start}{splitedInput[3]}{end}{splitedInput[4]}";
firstPart = "ISA~this is date~ESA|"
secondPart = "ISA~this is more data~ESA|";

Use a Regex like ISA(.+?)ESA and select the first group
string input = "ISA~mycontent+ESA";
Match match = Regex.Match(input, #"ISA(.+?)ESA",RegexOptions.IgnoreCase);
if (match.Success)
{
string key = match.Groups[1].Value;
}

Instead of "splitting" by a string, I would instead describe your question as "grouping" by a string. This can easily be done using a regular expression:
Regular expression: ^(ISA.*?(?=ESA)ESA.)(ISA.*?(?=ESA)ESA.)$
Explanation:
^ - asserts position at start of the string
( - start capturing group
ISA - match string ISA exactly
.*?(?=ESA) - match any character 0 or more times, positive lookahead on the
string ESA (basically match any character until the string ESA is found)
ESA - match string ESA exactly
. - match any character
) - end capturing group
repeat one more time...
$ - asserts position at end of the string
Try it on Regex101
Example:
string input = "ISA~this is date~ESA|ISA~this is more data~ESA|";
Regex regex = new Regex(#"^(ISA.*?(?=ESA)ESA.)(ISA.*?(?=ESA)ESA.)$",
RegexOptions.Compiled);
Match match = regex.Match(input);
if (match.Success)
{
string firstValue = match.Groups[1].Value; // "ISA~this is date~ESA|"
string secondValue = match.Groups[2].Value; // "ISA~this is more data~ESA|"
}

There are two answers to the question "How to split a string by another string".
var matches = input.Split(new [] { "ISA" }, StringSplitOptions.RemoveEmptyEntries);
and
var matches = Regex.Split(input, "ISA").ToList();
However, the first removes empty entries, while the second does not.

Improve RegEx search

Using DirectoryServices.AccountManagement I'm getting users DistinguishedName which looks like so:
CN=Adam West,OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu
I need to get first OU value from this.
I found similar solution: C# Extracting a name from a string
And using some tweaks I created this code:
string input = #"CN=Adam West,OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu";
Match m = Regex.Match(input, #"OU=([a-zA-Z\\]+)\,.*$");
Console.WriteLine(m.Groups[1].Value);
This code returns STORE as expected, but if I change Groups[1] to Groups[0] I get almost same result as input string:
OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu
How can I change this regex so it will return only values of OU? SO that in this example I get array of 2 matches. If I would have more OU in my string then array would be longer.
EDIT:
I've converted my code (using #dasblinkenlight suggestions) into function:
private static List<string> GetOUs()
{
var input = #"CN=Adam West,OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu";
var mm = Regex.Matches(input, #"OU=([a-zA-Z\\]+)");
return (from Match m in mm select m.Groups[1].Value).ToList();
}
Is that correct?

Your regular expression is fine (almost), you are just using a wrong API.
Remove the parts of the regexp that match up to the ending anchor $, and change the call of Match for a call of Matches, and get the matches in a loop, like this:
var input = #"CN=Adam West,OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu";
var mm = Regex.Matches(input, #"OU=([a-zA-Z\\]+)");
foreach (Match m in mm)
Console.WriteLine(m.Groups[1].Value);
}

Your existing regex:
#"OU=([a-zA-Z\\]+)\,.*$"
Matches OU=, then some letters and backslashes ([a-zA-Z\\]+), then a comma, then any characters (.*) to the end of the line ($).
Thus a single match will always match the entire line after the first OU section.
Modify your regex by removing the ,.*$ at the end, at it will match each OU group:
#"OU=([a-zA-Z\\]+)"
Also note that the parentheses are a capturing group. They are useful if you also want to capture just the value part by itself, but if you are not using that, they are not necessary, and you can just have this:
#"OU=[a-zA-Z\\]+"

It's beacuse you are mixing up matches and groups
string input = #"CN=Adam West,OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu";
MatchCollection mc = Regex.Matches(input, #"OU=([a-zA-Z\\]+),");
foreach(Match m in mc)
{
Console.WriteLine(m.Result("$1"));
}

Group[0] returns the full match:
Group[1] returns the first Pattern in the match [i.e. everything in the first parenthesis '(' ')' ]
So if you wanted to get exactly those 2 occurances of OU.. you could do this:
Match m = Regex.Match(input, #"OU=([a-zA-Z\\]+)\,OU=([a-zA-Z\\]+)\,.*$");
Console.WriteLine(m.Groups[1].Value);
Console.WriteLine(m.Groups[2].Value);
Group[0] returns the full match: (which you don't want)
Group[1] returns the first Pattern in the match [i.e everything in the first parenthesis '(' ')' ]
Group[2] returns the second Pattern in the match [i.e. everything in the second parenthesis '(' ')' ]
Giving:
STORE
COMPANY
But I'm assuming you don't want to be so explicit with your Regex for each Pattern you are interested in.
If you want to get multiple matches, then you need to do Regex's Matches call that returns a Matchcollection.
MatchCollection ms = Regex.Matches(...);
This still won't work with your current Regex though, because everything from STORE so the end of the line will be in the first match. If you only want to get the pattern "1-or-more-letters" after a "OU="
You only need:
#"OU=([a-zA-Z\\]+)"
So your code would be:
string input = #"CN=Adam West,OU=STORE,OU=COMPANY,DC=mycompany,DC=group,DC=eu";
MatchCollection ms = Regex.Matches(input, #"OU=([a-zA-Z\\]+)");
foreach (Match m in ms)
{
Console.WriteLine(m.Groups[1].Value);// get the string in the first "(" ")"
}

Match the pattern TEXT[0,0]

I have a lot of source codes with SQL queries like below:
c_query := "SELECT * FROM TABLE WHERE FIELD_NAME_ONE[2] = 'AB' AND FIELD_NAME_TWO[1,8] = 'ABCDEFGH'"
I would like to match these: FIELD_NAME_ONE[2] and FIELD_NAME_TWO[1,8] and these patterns must be found between double quotes (").
Edit
c_query := "SELECT * FROM TABLE WHERE FIELD_NAME_ONE[2] = 'AB' AND FIELD_NAME_TWO[1,8] = 'ABCDEFGH' AND TESTE[9] = 'XXXXXXXXX' AND FOO = '" + is_an_array[2] + "'"
It shouldn´t match is_an_array[2] because is not inside the double quotes.

I'm assuming you want to be able to match more than just those two specific fields, otherwise you wouldn't have gone to the trouble of applying regular expressions:
var tokens = Regex.Matches(sql, "\"([^\"]+)\"");
foreach (Match token in tokens) {
string str = token.Groups[1].Value;
var fields = Regex.Matches(str, #"(\w+\[\d+(,\d+)*\])");
foreach (Match field in fields)
Console.WriteLine(field.Value);
}
This will find any sequence of letters, numbers and underscores followed by square brackets, with 1 or more comma-separated numbers.
If you only want to match a sequence of letters and underscores before the square brackets, ammend the pattern to:
#"([a-zA-Z_]+\[\d+(,\d+)*\])"

Here's a regex pattern for what you seek (edited based on comment):
".*?[A-Z_]+\[\d(,\d)?\].*?"
I made the match non-greedy (.*?) in case there are multiple matches on the one line (unlikely, but for completeness...)
Edited
It's not clear if you want to match just the target field names or the whole SQL. If you want to match just the field names alone, use non-capturing groups for the rest:
(?:".*?)[A-Z_]+\[\d(,\d)?\](?:.*?")

c# Regex question

I have a problem dealing with the # symbol in Regex, I am trying to remove #sometext
from a text string can't seem to find anywhere where it uses the # as a literal. I have tried myself but doesn't remove the word from the string. Any ideas?
public string removeAtSymbol(string input)
{
Regex findWords = new Regex(______);//Find the words like "#text"
Regex[] removeWords;
string test = input;
MatchCollection all = findWords.Matches(test);
removeWords = new Regex[all.Count];
int index = 0;
string[] values = new string[all.Count];
YesOutputBox.Text = " you got here";
foreach (Match m in all) //List all the words
{
values[index] = m.Value.Trim();
index++;
YesOutputBox.Text = YesOutputBox.Text + " " + m.Value;
}
for (int i = 0; i < removeWords.Length; i++)
{
removeWords[i] = new Regex(" " + values[i]);
// If the words appears more than one time
if (removeWords[i].Matches(test).Count > 1)
{
removeWords[i] = new Regex(" " + values[i] + " ");
test = removeWords[i].Replace(test, " "); //Remove the first word.
}
}
return test;
}

You can remove all occurences of "#sometext" from string test via the method
Regex.Replace(test, "#sometext", "")
or for any word starting with "#" you can use
Regex.Replace(test, "#\\w+", "")
If you need specifically a separate word (i.e. nothing like #comp within tom#comp.com) you may preceed the regex with a special word boundary (\b does not work here):
Regex.Replace(test, "(^|\\W)#\\w+", "")

You can use:
^\s#([A-Za-z0-9_]+)
as the regex to recognize Twitter usernames.

Regex to remove #something from this string: I want to remove #something from this string.
var regex = new Regex("#\\w*");
string result = regex.Replace(stringWithAt, "");
Is that what you are looking for?

I've had good luck applying this pattern:
\B#\w+
This will match any string starting with an # character that contains alphanumeric characters, plus some linking punctuation like the underscore character, if it does not occur on a boundary between alphanumeric and non-alphanumeric characters.
The result of executing this code:
string result = Regex.Replace(
#"#This1 #That2_thing this2#3that #the5Others #alpha#beta#gamma",
#"\B#\w+",
#"redacted");
is the following string:
redacted redacted this2#3that redacted redacted#beta#gamma
If this question is Twitter-specific, then Twitter provides an open source library that helps capture Twitter-specific entities like links, mentions and hashtags. This java file contains the code defining the regular expressions that Twitter uses, and this yml file contains test strings and expected outcomes of many unit tests that exercise the regular expressions in the Twitter library.
Twitter's mention-matching pattern (extracted from their library, modified to remove unnecessary capture groups, and edited to make sense in the context of a replacement) is shown below. The match should be performed in a case-insensitive manner.
(^|[^a-z0-9_])[#\uFF20][a-z0-9_]{1,20}
Here is an example which reproduces the results of the first replacement in my answer:
string result = Regex.Replace(
#"#This1 #That2_thing this2#3that #the5Others #alpha#beta#gamma",
#"(^|[^a-z0-9_])[#\uFF20][a-z0-9_]{1,20}",
#"$1redacted",
RegexOptions.IgnoreCase);
Note the need to include the substitution $1 since the first capture group can't be directly converted into an atomic zero-width assertion.