Find string that does not contain some substring

Question

I have a one liner string that looks like this:

My db objects are db.main_flow_tbl, 'main_flow_audit_tbl', main_request_seq and MAIN_SUBFLOW_TBL.

I want to use regular expressions to return database tables that start with main but do not contain words audit or seq, and irrespective of the case. So in the above example strings main_flow_tbl and MAIN_SUBFLOW_TBL shall return. Can someone help me with this please?

Maybe this links could help http://stackoverflow.com/questions/4813497/java-regex-case-insensitivity-not-working and http://stackoverflow.com/questions/406230/regular-expression-to-match-string-not-containing-a-word — holap, Dec 11 '13 at 08:27
What did you try so far? Does it need to use a regular expression? — jwueller, Dec 11 '13 at 08:29

Boris the Spider · Accepted Answer · 2013-12-11T08:42:08.447

2

Here is a fully regex based solution:

public static void main(String[] args) throws Exception {
    final String in = "My db objects are db.main_flow_tbl, 'main_flow_audit_tbl', main_request_seq and MAIN_SUBFLOW_TBL.";
    final Pattern pat = Pattern.compile("main_(?!\\w*?(?:audit|seq))\\w++", Pattern.CASE_INSENSITIVE);
    final Matcher m = pat.matcher(in);
    while(m.find()) {
        System.out.println(m.group());
    }
}

Output:

main_flow_tbl
MAIN_SUBFLOW_TBL

This assumes that table names can only contain A-Za-Z_ which \w is the shorthand for.

Pattern breakdown:

main_ is the liternal "main" that you want tables to start with
(?!\\w*?(?:audit|seq)) is a negative lookahead (not followed by) which takes any number of \w characters (lazily) followed by either "audit" or "seq". This excludes tables names that contain those sequences.
\\w++ consume any table characters possesively.

EDIT

OP's comment they may contain numbers as well

In this case use this pattern:

main_(?![\\d\\w]*?(?:audit|seq))[\\d\\w]++

i.e. use [\\d\\w] rather than \\w

edited Dec 11 '13 at 08:42

answered Dec 11 '13 at 08:35

Boris the Spider

59,842
6
106
166

they may contain numbers as well – Yasser Al Masri Dec 11 '13 at 08:40
Nice solution! I was still tossing around on the negative lookahead – Michel Feldheim Dec 11 '13 at 08:43
Can you please explain it to me to get the maximum benefit? – Yasser Al Masri Dec 11 '13 at 08:48
@YasserAlMasri There is a breakdown of the pattern, what parts do you need me to expand? – Boris the Spider Dec 11 '13 at 08:56
my particular question is that I tried this: (?i)main_(?!.*audit|seq)[a-zA-Z0-9_]* but it kept eliminating all the line once audit or seq is found. What's the difference that made yours do the trick? – Yasser Al Masri Dec 11 '13 at 09:50
@YasserAlMasri two things. 1) I used `[\\d\\w]*?` rather than `.*`, this matches only inside the table name - otherwise it will wander along until it finds an `audit` or `seq` _anywhere_ further in the `String`. It is also lazy rather than greedy. 2) I used `(?:audit|seq)` - i.e. I encapsulated the or in a non-capturing group; your pattern `.*audit|seq` is actually `.*audit` or `seq`. – Boris the Spider Dec 11 '13 at 10:26
Appreciated. This is a real regex guru! – Yasser Al Masri Dec 11 '13 at 11:33

score 0 · Answer 2 · answered Dec 11 '13 at 08:25

0

String str
while ((str.startsWith("main"))&&!str.contains("audit")||!str.contains("seq")){
   //your code here
}

answered Dec 11 '13 at 08:25

crzbt

183
4
14

I asked for regular expressions – Yasser Al Masri Dec 11 '13 at 08:30
@YasserAlMasri: Is this homework? Why does it have to be a regular expression if there is a working alternative? – jwueller Dec 11 '13 at 08:31
because it will be very tedious otherwise. What is str in your case? I don't have anything to capture an array of strings that really contain the tables. It just can be anything with in the string – Yasser Al Masri Dec 11 '13 at 08:34

score 0 · Answer 3 · answered Dec 11 '13 at 08:41

0

If the string matches

    ^main_(\w_)*(?!(?:audit|seq))

it should be what you want...

answered Dec 11 '13 at 08:41

nozzleman

9,529
4
37
58

Almost. You don't have the case insensitive flag. You have a start anchor so it won't find patterns _in_ `String`s; which is what the OP wants. And your match group would capture _up to_ the sequences the OP doesn't want. So rather than excluding names containing `seq` like `table_seq` this would capture `table_`. In short **test your code before posting**. – Boris the Spider Dec 11 '13 at 09:06

Find string that does not contain some substring

3 Answers3