如何使用正则表达式在Java中查找不包含特定字符串的字符串

Question

I am trying to filter out those strings in a list of strings which do not contain a specific substring (which is nothing but today's date in the YYYY-MM-DD format in my case), but unable to do so.

This is what I have tried so far.

My String is in this format - ABC-TEST.20181206.20181208- 20180215 .log

The string may also be in this format - ABC-TEST.20181206.20181208- 20180215 -1.log (the 1 before .log may go on up to infinite)

If I see that the string is ending with today's date at the end (before .log or -.log), I need to find it.

So, since it is not ending with today's date, I need to filter this out.

I have tried this pattern to identify the file which contains today's date, but I am unable to find that string which does not contain it

(.*?)-20180221-?(\\d+)?.log

This is one of the patterns I have tried without luck

(.*?)-^((?!20180221))-?(\\d+)?.log

Answer 1

If you are willing to accept some reasonable limit on the number of digits, then you could use a basic negative lookbehind, like this:

String pattern = ".*(?<!20180215(-[0-9]{1,7})?\\.log)$";
String false1 = "ABC-TEST.20181206.20181208-20180215.log";
String false2 = "ABC-TEST.20181206.20181208-20180215-1.log";
String true1 = "ABC-TEST.20181206.20181208-20180216.log";
String true2 = "ABC-TEST.20181206.20181208-20180216-1.log";

System.out.println(false1.matches(pattern)); // false
System.out.println(false2.matches(pattern)); // false
System.out.println(true1.matches(pattern)); // true
System.out.println(true2.matches(pattern)); // true

I would like to point out that if the trailing counter is allowed to have more than 7 digits then it creates an ambiguous situation based on the algorithm you have outlined, because at that point there is no way to distinguish between the counter 20180215 and the date 20180215.

---

Question Evolution #1

The question writer has used the comment section on this answer to change his requirements as follows:

the string should contain "TEST"

Answer

You would just add .*TEST to the front of the pattern in this answer. Like so:

String pattern = ".*TEST.*(?<!20180215(-[0-9]{1,7})?\\.log)$";

---

Question Evolution #2

The question writer has used the comment section on this answer to change his requirements as follows:

to pick up the String which does NOT contain TEST and which does NOT contain today's date

Answer

You could use a negative lookahead of "TEST" for every repetition of the initial wildcard, like so:

String pattern = "((?!TEST).)*(?<!20180215(-[0-9]{1,7})?\\.log)";

Answer 2

Well, in my timezone, it is still the 14th of february, so I used:

egrep -- "-$(date "+%Y%m%d" -d now+1day )-?.log" sample
My String is in this format - ABC-TEST.20181206.20181208-20180215.log

Well - you're using Java? Why not?

-> import java.util.Date
-> Date d = new Date()
-> String today = String.format ("%tY%tm%td", d, d, d)
-> String s1 = "My String is in this format - ABC-TEST.20181206.20181208-20180214.log"
-> String s2 = "The string may also be in this format - ABC-TEST.20181206.20181208-20180214-1.log (the 1 before .log may go on up to infinite)"
-> String pattern = ".*" + today + "\\.log"
-> s1.matches (pattern) 
|  Expression value is: true
|    assigned to temporary variable $39 of type boolean
-> s2.matches (pattern) 
|  Expression value is: false
|    assigned to temporary variable $40 of type boolean

That's copied from the jshell, a fine tool, for fast ad hoc testing.

 (.*?)-20180221-?(\\d+)?.log

Well compared with today - be it the 14th or the 15th - we don't want to be to strict about it, but 02/21 is pretty off, isn't it?

What is the first question mark supposed to do?

Can the fine line be

   -20180221.log
   -20180221-.log
   -20180221888.log
   -20180221-888.log

? If lines don't match - do you still need to find those, which contain

or isn't it either

   -20180221.log  

or -20180221-888.log

Then:

   String pattern = ".*" + today + "(-[0-9]+)?\\.log";

If there might be something behind log:

   String pattern = ".*" + today + "(-[0-9]+)?\\.log.*";