Regex not match with logical OR - regex

Need to match words that doesn't start with tss- or equal to tss. Tried multiple combinations but no positive results.
^(((?!(tss)).*)|(?!tss-).+)

To apply two negative lookahead checks against the input string, you need to simply chain them after the ^ anchor:
^(?!tss$)(?!tss-).*
The logical relationship is AND in this case:
^ - start of string
(?!tss$) - the string must not be equal to tss
AND
(?!tss-) - the string must not start with tss-
.* - match the rest.

If the words can also occur in a sentence, you might also use lookarounds to assert not tss followed by either a whitespace bounadry or - using a negative lookahead.
(?<!\S)(?!tss(?:(?!\S)|-))\S+
(?<!\S) Assert a whitespace boundary to the left
(?! Negative lookahead, assert to the right is not
tss Match literally
(?:(?!\S)|-) Match either a whitespace boundary or -
) Close Lookahead
\S+ Match 1+ non whitespace chars
Regex demo

Related

How to match names separated by "and" excluding "and" itself using regex?

I am trying to solve http://play.inginf.units.it/#/level/10
I have some strings as follows:
title={AUTOMATIC ROCKING DEVICE},
author={Diaz, Navarro David and Gines, Rodriguez Noe},
year={2006},
title={The sitting position in neurosurgery: a retrospective analysis of 488 cases},
author={Standefer, Michael and Bay, Janet W and Trusso, Russell},
journal={Neurosurgery},
title={Fuel cells and their applications},
author={Kordesch, Karl and Simader, G{"u}nter and Wiley, John},
volume={117},
I need to match the names in bold. I tried the following regex:
(?<=author={).+(?=})
But it matches the entire string inside {}. I understand why is it so but how can I break the pattern with and?
It took me a little while to get the samples to show up in your link. What about:
(?:^\s*author={|\G(?!^) and )\K(?:(?! and |},).)+
See an online demo
(?:^\s*author={|\G(?!^) and ) - Either match start of a line followed by 0+ whitespace chars and literally match 'author={` or assert position at end of previous match but negate start-line;
\K - Reset starting point of reported match;
(?:(?! and |},).)+ - Match any if it's not followed by ' and ' or match a '}' followed by a comma.
Above will also match 'others' as per last sample in linked test. If you wish to exclude 'others' then maybe add the option to the negated list as per:
(?:^\s*author={|\G(?!^) and )\K(?:(?! and |},|\bothers\b).)+
See an online demo
In the comment section we established above would not work for given linked website. Apparently its JS based which would support zero-width lookbehind. Therefor try:
(?<=\bauthor={(?:(?!\},).*?))\b[A-Z]\S*\b(?:,? [A-Z]\S*\b)*
See the demo
(?<= - Open lookbehind;
\bauthor={ - Match word-boundary and literally 'author={';
(?:(?!\},).*?)) - Open non-capture group to match a negative lookahead for '},' and 0+ (lazy) characters. Close lookbehind;
\b[A-Z]\S*\b - Match anything between two word-boundaries starting with a capital letter A-Z followed by 0+ non-whitespace chars;
(?:,? [A-Z]\S*\b)* - A 2nd non-capture group to keep matching comma/space seperated parts of a name.
If using a lookbehind assertion is supported and matching word characters, you might use:
(?<=\bauthor={[^{}]*(?:{[^{}]*}[^{}]*)*)[A-Z][^\s,]*,(?:\s+[A-Z][^\s,]*)+\b
Explanation
(?<= Postive lookahead, assert that to the left of the current position is
\bauthor={ Match author={ preceded by a word boundary
[^{}]*(?:{[^{}]*}[^{}]*)* Match optional chars other than { } or match {...}
) Close the lookbehind
[A-Z] Match an uppercase char A-Z
[^\s,]*, Optionally match non whitespace chars except , and then match ,
(?: Non capture group to repeat as a whole part
\s+[A-Z][^\s,]* Match 1+ whitespace chars, uppercase char A-Z, optional non whitespace chars except ,
)+ Close the non capture group and repeat it 1 or more times
\b a word boundary
See a regex101 demo.

Regex: Match pattern unless preceded by pattern containing element from the matching character class

I am having a hard time coming up with a regex to match a specific case:
This can be matched:
any-dashed-strings
this-can-be-matched-even-though-its-big
This cannot be matched:
strings starting with elem- or asdf- or a single -
elem-this-cannot-be-matched
asdf-this-cannot-be-matched
-
So far what I came up with is:
/\b(?!elem-|asdf-)([\w\-]+)\b/
But I keep matching a single - and the whole -this-cannot-be-matched suffix. I cannot figure it out how to not only ignore a character present inside the matching character class conditionally, and not matching anything else if a suffix is found
I am currently working with the Oniguruma engine (Ruby 1.9+/PHP multi-byte string module).
If possible, please elaborate on the solution. Thanks a lot!
If a lookbehind is supported, you can assert a whitespace boundary to the left, and make the alternation for both words without the hyphen optional.
(?<!\S)(?!(?:elem|asdf)?-)[\w-]+\b
Explanation
(?<!\S) Assert a whitespace boundary to the left
(?! Negative lookahead, assert the directly to the right is not
(?:elem|asdf)?- Optionally match elem or asdf followed by -
) Close the lookahead
[\w-]+ Match 1+ word chars or -
\b A word boundary
See a regex demo.
Or a version with a capture group and without a lookbehind:
(?:\s|^)(?!(?:elem|asdf)?-)([\w-]+)\b
See another regex demo.

Negative Lookahead not match suffix

I have an expression that is matching something, but am trying to get this not to match if it's followed by the suffix: one or more spaces, three dashes, one or more spaces, one or more digits, a slash, and finally one or more digits. Here is the expression:
(?<=(^|\s+))[A-Z]+[ ]+([0-9]+(\.[0-9]{1,3})?)/([0-9]+(\.[0-9]{1,3})?)(?!(\s+\-\-\-\s+[0-9]+/[0-9]+))
And here is the text:
January 10.5/13.5 --- 22/26 ---
It's matching January 10.5/13, but I don't want it to match anything.
As lookarounds are supported, you can change the positive lookbehind at the start to a negative lookbehind asserting a whitespace boundary to the left (?<!\S)
You can use .* to it to scan the whole line, instead of starting with 1+ more whitespace chars \s+
The negative lookahead (?!.*\s-{3}\s+[0-9]+/[0-9] asserts that what is on the right is not the suffix.
You can omit the quantifier + after the last character class, as it does not matter if there are 1 or more digits following...as long as it is not a digit.
Note that in the current pattern, the decimal part is an optional capturing group 2. If you want that whole value in group 1, you can make it an optional group.
(?<!\S)[A-Z]+[ ]+([0-9]+(\.[0-9]{1,3})?)/([0-9]+(\.[0-9]{1,3})?)(?!.*\s-{3}\s+[0-9]+/[0-9])
Regex demo

Regex to match character if not between digits

I need to match a character to split a big string, let's say -, but not if it's between two digits
In a-b it should match -
In a-4 it should match -
In 3-a it should match -
In 3-4 it should not match
I've tried negative lookahead and lookbehind, but I've only been able to come up with this (?<=\D)-(?=\D)|(?<=\d)-(?=\D)|(?<=\D)-(?=\d)
Is there a simpler way to specify this pattern?
Edit: using regex conditionals I think I can use (?(?<=\D)-|-(?=\D))
The following will work for this scenario. Be sure that your Regex flavor of choice has conditionals, otherwise this will not work:
-(?(?=\d)(?<=\D-))
- // match a dash
(? // If
(?=\d) // the next character is a digit
(?<= // then start a lookbehind (assert preceding characters are)
\D- // a non-digit then the dash we matched
) // end lookbehind
) // end conditional
With nothing as the substitution, as the dash is the only character captured.
Another option is to use an alternation to match a - when on the left is not a digit or match a - when on the right is not a digit:
(?<!\d)-|-(?!\d)
(?<!\d)- Negative lookbehind, assert what is on the left is not a digit and match -
| or
-(?!\d) Match - and assert what is on the right is not a digit using a negative lookahead
Regex demo

Unmatch complete words if a negative lookahead is satisfied

I need to match only those words which doesn't have special characters like # and :.
For example:
git#github.com shouldn't match
list should return a valid match
show should also return a valid match
I tried it using a negative lookahead \w+(?![#:])
But it matches gi out of git#github.com but it shouldn't match that too.
You may add \w to the lookahead:
\w+(?![\w#:])
The equivalent is using a word boundary:
\w+\b(?![#:])
Besides, you may consider adding a left-hand boundary to avoid matching words inside non-word non-whitespace chunks of text:
^\w+(?![\w#:])
Or
(?<!\S)\w+(?![\w#:])
The ^ will match the word at the start of the string and (?<!S) will match only if the word is preceded with whitespace or start of string.
See the regex demo.
Why not (?<!\S)\w+(?!\S), the whitespace boundaries? Because since you are building a lexer, you most probably have to deal with natural language sentences where words are likely to be followed with punctuation, and the (?!\S) negative lookahead would make the \w+ match only when it is followed with whitespace or at the end of the string.
You can use negative lookbehind and negative lookahead patterns around a word pattern to make sure that the word is not preceded or followed by a non-space character, or in other words, to make sure that it is surrounded by either a space or a string boundary:
(?<!\S)\w+(?!\S)
Demo: https://regex101.com/r/cjhUUM/2