The notion that regex doesn't support inverse matching is not entirely true. You can mimic this behavior by using negative look-arounds:

^((?!hede).)*$

The regex above will match any string, or line without a line break, not containing the (sub) string 'hede'. As mentioned, this is not something regex is "good" at (or should do), but still, it is possible.

And if you need to match line break chars as well, use the DOT-ALL modifier (the trailing s in the following pattern):

/^((?!hede).)*$/s

or use it inline:

/(?s)^((?!hede).)*$/

(where the /.../ are the regex delimiters, ie, not part of the pattern)

If the DOT-ALL modifier is not available, you can mimic the same behavior with the character class [\s\S]:

/^((?!hede)[\s\S])*$/

Explanation

A string is just a list of n characters. Before, and after each character, there's an empty string. So a list of n characters will have n+1 empty strings. Consider the string "ABhedeCD":

    +--+---+--+---+--+---+--+---+--+---+--+---+--+---+--+---+--+
S = |e1| A |e2| B |e3| h |e4| e |e5| d |e6| e |e7| C |e8| D |e9|
    +--+---+--+---+--+---+--+---+--+---+--+---+--+---+--+---+--+

index    0      1      2      3      4      5      6      7

where the e's are the empty strings. The regex (?!hede). looks ahead to see if there's no substring "hede" to be seen, and if that is the case (so something else is seen), then the . (dot) will match any character except a line break. Look-arounds are also called zero-width-assertions because they don't consume any characters. They only assert/validate something.

So, in my example, every empty string is first validated to see if there's no "hede" up ahead, before a character is consumed by the . (dot). The regex (?!hede). will do that only once, so it is wrapped in a group, and repeated zero or more times: ((?!hede).)*. Finally, the start- and end-of-input are anchored to make sure the entire input is consumed: ^((?!hede).)*$

As you can see, the input "ABhedeCD" will fail because on e3, the regex (?!hede) fails (there is "hede" up ahead!).

Note that the solution to does not start with “hede”:

^(?!hede).*$

is generally much more efficient than the solution to does not contain “hede”:

^((?!hede).)*$

The former checks for “hede” only at the input string’s first position, rather than at every position.

If you're just using it for grep, you can use grep -v hede to get all lines which do not contain hede.

ETA Oh, rereading the question, grep -v is probably what you meant by "tools options".

The given answers are perfectly fine, just an academic point:

Regular Expressions in the meaning of theoretical computer sciences ARE NOT ABLE do it like this. For them it had to look something like this:

^([^h].*$)|(h([^e].*$|$))|(he([^h].*$|$))|(heh([^e].*$|$))|(hehe.+$) 

This only does a FULL match. Doing it for sub-matches would even be more awkward.

^((?!hede).)*$

Explanation:

^the beginning of the string

( group and capture to \1 (0 or more times (matching the most amount possible))

(?! look ahead to see if there is not:

hede your string

) end of look-ahead

. any character except \n

)* end of \1 (NOTE: because you are using a quantifier on this capture, only the LAST repetition of the captured pattern will be stored in \1)

$ before an optional \n, and the end of the string

Here's a good explanation of why it's not easy to negate an arbitrary regex. I have to agree with the other answers, though: if this is anything other than a hypothetical question, then a regex is not the right choice here.

If you want the regex test to only fail if the entire string matches, the following will work:

^(?!hede$).*

e.g. -- If you want to allow all values except "foo" (i.e. "foofoo", "barfoo", and "foobar" will pass, but "foo" will fail), use: ^(?!foo$).*

Of course, if you're checking for exact equality, a better general solution in this case is to check for string equality, i.e.

myStr !== 'foo'

You could even put the negation outside the test if you need any regex features (here, case insensitivity and range matching):

!/^[a-f]oo$/i.test(myStr)

The regex solution at the top may be helpful, however, in situations where a positive regex test is required (perhaps by an API).

Not regex, but I've found it logical and useful to use serial greps with pipe to eliminate noise.

eg. search an apache config file without all the comments-

grep -v '\#' /opt/lampp/etc/httpd.conf      # this gives all the non-comment lines

and

grep -v '\#' /opt/lampp/etc/httpd.conf |  grep -i dir

The logic of serial grep's is (not a comment) and (matches dir)

Benchmarks

I decided to evaluate some of the presented Options and compare their performance, as well as use some new Features. Benchmarking on .NET Regex Engine: http://regexhero.net/tester/

Benchmark Text:

The first 7 lines should not match, since they contain the searched Expression, while the lower 7 lines should match!

Regex Hero is a real-time online Silverlight Regular Expression Tester.
XRegex Hero is a real-time online Silverlight Regular Expression Tester.
Regex HeroRegex HeroRegex HeroRegex HeroRegex Hero is a real-time online Silverlight Regular Expression Tester.
Regex Her Regex Her Regex Her Regex Her Regex Her Regex Her Regex Hero is a real-time online Silverlight Regular Expression Tester.
Regex Her is a real-time online Silverlight Regular Expression Tester.Regex Hero
egex Hero egex Hero egex Hero egex Hero egex Hero egex Hero Regex Hero is a real-time online Silverlight Regular Expression Tester.
RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRegex Hero is a real-time online Silverlight Regular Expression Tester.

Regex Her
egex Hero
egex Hero is a real-time online Silverlight Regular Expression Tester.
Regex Her is a real-time online Silverlight Regular Expression Tester.
Regex Her Regex Her Regex Her Regex Her Regex Her Regex Her is a real-time online Silverlight Regular Expression Tester.
Nobody is a real-time online Silverlight Regular Expression Tester.
Regex Her o egex Hero Regex  Hero Reg ex Hero is a real-time online Silverlight Regular Expression Tester.

Results:

Results are Iterations per second as the median of 3 runs - Bigger Number = Better

01: ^((?!Regex Hero).)*$                    3.914   // Accepted Answer
02: ^(?:(?!Regex Hero).)*$                  5.034   // With Non-Capturing group
03: ^(?>[^R]+|R(?!egex Hero))*$             6.137   // Lookahead only on the right first letter
04: ^(?>(?:.*?Regex Hero)?)^.*$             7.426   // Match the word and check if you're still at linestart
05: ^(?(?=.*?Regex Hero)(?#fail)|.*)$       7.371   // Logic Branch: Find Regex Hero? match nothing, else anything

P1: ^(?(?=.*?Regex Hero)(*FAIL)|(*ACCEPT))  ?????   // Logic Branch in Perl - Quick FAIL
P2: .*?Regex Hero(*COMMIT)(*FAIL)|(*ACCEPT) ?????   // Direct COMMIT & FAIL in Perl

Since .NET doesn't support action Verbs (*FAIL, etc.) I couldn't test the solutions P1 and P2.

Summary:

I tried to test most proposed solutions, some Optimizations are possible for certain words. For Example if the First two letters of the search string are not the Same, answer 03 can be expanded to ^(?>[^R]+|R+(?!egex Hero))*$ resulting in a small performance gain.

But the overall most readable and performance-wise fastest solution seems to be 05 using a conditional statement or 04 with the possesive quantifier. I think the Perl solutions should be even faster and more easily readable.

With negative lookahead, regular expression can match something not contains specific pattern. This is answered and explained by Bart Kiers. Great explanation!

However, with Bart Kiers' answer, the lookahead part will test 1 to 4 characters ahead while matching any single character. We can avoid this and let the lookahead part check out the whole text, ensure there is no 'hede', and then the normal part (.*) can eat the whole text all at one time.

Here is the improved regex:

/^(?!.*?hede).*$/

Note the (*?) lazy quantifier in the negative lookahead part is optional, you can use (*) greedy quantifier instead, depending on your data: if 'hede' does present and in the beginning half of the text, the lazy quantifier can be faster; otherwise, the greedy quantifier be faster. However if 'hede' does not present, both would be equal slow.

Here is the demo code.

For more information about lookahead, please check out the great article: Mastering Lookahead and Lookbehind.

Also, please check out RegexGen.js, a JavaScript Regular Expression Generator that helps to construct complex regular expressions. With RegexGen.js, you can construct the regex in a more readable way:

var _ = regexGen;

var regex = _(
    _.startOfLine(),             
    _.anything().notContains(       // match anything that not contains:
        _.anything().lazy(), 'hede' //   zero or more chars that followed by 'hede',
                                    //   i.e., anything contains 'hede'
    ), 
    _.endOfLine()
);

FWIW, since regular languages (aka rational languages) are closed under complementation, it's always possible to find an regular expression (aka rational expression) that negates another expression. But not many tools implement this.

Vcsn supports this operator (which it denotes {c}, postfix).

You first define the type of your expressions: labels are letter (lal_char) to pick from a to z for instance (defining the alphabet when working with complementation is, of course, very important), and the "value" computed for each word is just a Boolean: true the word is accepted, false, rejected.

In Python:

In [5]: import vcsn
        c = vcsn.context('lal_char(a-z), b')
        c
Out[5]: {a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z} → 𝔹

then you enter your expression:

In [6]: e = c.expression('(hede){c}'); e
Out[6]: (hede)^c

convert this expression to an automaton:

In [7]: a = e.automaton(); a

finally, convert this automaton back to a simple expression.

In [8]: print(a.expression())
        \e+h(\e+e(\e+d))+([^h]+h([^e]+e([^d]+d([^e]+e[^]))))[^]*

where + is usually denoted |, \e denotes the empty word, and [^] is usually written . (any character). So, with a bit of rewriting ()|h(ed?)?|([^h]|h([^e]|e([^d]|d([^e]|e.)))).*.

You can see this example here, and try Vcsn online there.

with this, you avoid to test a lookahead on each positions:

/^(?:[^h]+|h++(?!ede))*+$/

equivalent to (for .net):

/^(?>(?:[^h]+|h+(?!ede))*)$/

Old answer:

/^(?>[^R]+|R+(?!egex Hero))*$/

Here's how I'd do it:

^[^h]*(h(?!ede)[^h]*)*$

Accurate and more efficient than the other answers. It implements Friedl's "unrolling-the-loop" efficiency technique and requires much less backtracking.

If you want to match a character to negate a word similar to negate character class:

For example, a string:

<?
$str="aaa        bbb4      aaa     bbb7";
?>

Do not use:

<?
preg_match('/aaa[^bbb]+?bbb7/s', $str, $matches);
?>

Use:

<?
preg_match('/aaa(?:(?!bbb).)+?bbb7/s', $str, $matches);
?>

Notice "(?!bbb)." is neither lookbehind nor lookahead, it's lookcurrent, for example:

"(?=abc)abcde", "(?!abc)abcde"

The OP did not specify or Tag the post to indicate the context (programming language, editor, tool) the Regex will be used within.

For me, I sometimes need to do this while editing a file using Textpad.

Textpad supports some Regex, but does not support lookahead or lookbehind, so it takes a few steps.

If I am looking to retain all lines that Do NOT contain the string hede, I would do it like this:

1. Search/replace the entire file to add a unique "Tag" to the beginning of each line containing any text.

    Search string:^(.)  
    Replace string:<@#-unique-#@>\1  
    Replace-all  

2. Delete all lines that contain the string hede (replacement string is empty):

    Search string:<@#-unique-#@>.*hede.*\n  
    Replace string:<nothing>  
    Replace-all  

3. At this point, all remaining lines Do NOT contain the string hede. Remove the unique "Tag" from all lines (replacement string is empty):

    Search string:<@#-unique-#@>
    Replace string:<nothing>  
    Replace-all  

Now you have the original text with all lines containing the string hede removed.

If I am looking to Do Something Else to only lines that Do NOT contain the string hede, I would do it like this:

1. Search/replace the entire file to add a unique "Tag" to the beginning of each line containing any text.

    Search string:^(.)  
    Replace string:<@#-unique-#@>\1  
    Replace-all  

2. For all lines that contain the string hede, remove the unique "Tag":

    Search string:<@#-unique-#@>(.*hede)
    Replace string:\1  
    Replace-all  

3. At this point, all lines that begin with the unique "Tag", Do NOT contain the string hede. I can now do my Something Else to only those lines.

4. When I am done, I remove the unique "Tag" from all lines (replacement string is empty):

    Search string:<@#-unique-#@>
    Replace string:<nothing>  
    Replace-all  

Through PCRE verb (*SKIP)(*F)

^hede$(*SKIP)(*F)|^.*$

This would completely skips the line which contains the exact string hede and matches all the remaining lines.

DEMO

Execution of the parts:

Let us consider the above regex by splitting it into two parts.

1. Part before the | symbol. Part shouldn't be matched.

   ^hede$(*SKIP)(*F)

2. Part after the | symbol. Part should be matched.

   ^.*$

PART 1

Regex engine will start its execution from the first part.

^hede$(*SKIP)(*F)

Explanation:

• ^ Asserts that we are at the start.
• hede Matches the string hede
• $ Asserts that we are at the line end.

So the line which contains the string hede would be matched. Once the regex engine sees the following (*SKIP)(*F) (Note: You could write (*F) as (*FAIL)) verb, it skips and make the match to fail. | called alteration or logical OR operator added next to the PCRE verb which inturn matches all the boundaries exists between each and every character on all the lines except the line contains the exact string hede. See the demo here. That is, it tries to match the characters from the remaining string. Now the regex in the second part would be executed.

PART 2

^.*$

Explanation:

• ^ Asserts that we are at the start. ie, it matches all the line starts except the one in the hede line. See the demo here.

• .* In the Multiline mode, . would match any character except newline or carriage return characters. And * would repeat the previous character zero or more times. So .* would match the whole line. See the demo here.

  Hey why you added .* instead of .+ ?

  Because .* would match a blank line but .+ won't match a blank. We want to match all the lines except hede , there may be a possibility of blank lines also in the input . so you must use .* instead of .+ . .+ would repeat the previous character one or more times. See .* matches a blank line here.

• $ End of the line anchor is not necessary here.

It may be more maintainable to two regexes in your code, one to do the first match, and then if it matches run the second regex to check for outlier cases you wish to block for example ^.(hede). then have appropriate logic in your code.

Ok, I admit this is not really an answer to the posted question posted and it may also use slightly more processing than a single regex. But for developers who came here looking for a fast emergency fix for an outlier case then this solution should not be overlooked.

The TXR Language supports regex negation.

$ txr -c '@(repeat)
@{nothede /~hede/}
@(do (put-line nothede))
@(end)'  Input

A more complicated example: match all lines that start with a and end with z, but do not contain the substring hede:

$ txr -c '@(repeat)
@{nothede /a.*z&~.*hede.*/}
@(do (put-line nothede))
@(end)' -
az         <- echoed
az
abcz       <- echoed
abcz
abhederz   <- not echoed; contains hede
ahedez     <- not echoed; contains hede
ace        <- not echoed; does not end in z
ahedz      <- echoed
ahedz

Regex negation is not particularly useful on its own but when you also have intersection, things get interesting, since you have a full set of boolean set operations: you can express "the set which matches this, except for things which match that".