匹配给定的正则表达式，除非存在给定的单词（向前或向后）

Question

I am using javascript regex to parse a series of URLs. I need to match a digit in a URL (it's actually more complicated, but I'm simplifying), but only want to match a number where a given word is not in the URL.

Namely, I want to exclude lines with the word 'changelogs' in them, and would therefore capture ' 1047 ', ' 1048 ', ' 1245 ' and ' 1049 ' from the following list;

http://www.opera.com/docs/changelogs/unified/1215/
http://www.whatever.com/docs/changelogs/anythingelse/anything/1215/
http://www.blabblah/security/advisory/1047
http://booger/security/advisory/1048/
ftp://msn.global.whatever/somethingelse/1245
whatever/it/doesnt/matter/could/be/anything/i/still/want/this/number/1049/

I know I need some kind of look-around look-ahead look-behind, but I'm striking out. Here is the last pattern I've tried;

(?!changelogs)(\d+)

Here is the regex101 sandbox I'm using .

Also, it's important that the only match is the actual number. I don't want anything else to match .

---

Here is what my .NET code looks like (note the "BulletinOrAdvisoryPattern" is the regex in question)...

Regex bulletinPattern = new Regex(@matchingDomain.Vendor.BulletinOrAdvisoryPattern, RegexOptions.IgnoreCase );
Match bulletinMatch = bulletinPattern.Match(referenceTitle);

                    if (bulletinMatch.Success)
                    {
                        //Found the bulletin ID in the NVD Reference Title 
                        return bulletinMatch.Value;
                    }

Answer 1

The "ugly" regex you need is

(?<=http://www\.opera\.com\b(?!.*/changelogs(?:/|$))\S*)\d+

See the .NET regex demo

However, all you need is

var result = input.Contains("/changelogs/") ? "" : input.Trim('/').Split('/').LastOrDefault();

See the IDEONE C# demo :

var lst = new List<string>() {"http://w...content-available-to-author-only...a.com/docs/changelogs/unified/1215/",
    "http://w...content-available-to-author-only...a.com/docs/changelogs/anythingelse/anything/1215/",
    "http://w...content-available-to-author-only...a.com/security/advisory/1047",
    "http://w...content-available-to-author-only...a.com/security/advisory/1048/",
    "http://w...content-available-to-author-only...a.com/doesnt/matter/could/be/anything/1049/"};
lst.ForEach(m => Console.WriteLine(
        m.Contains("/changelogs/") ? "" : m.Trim('/').Split('/').LastOrDefault()
    ));

UPDATE

You switched the language from C# to JavaScript that changes the situation drastically since JS regex engine does not support a lookbehind.

Thus, you have to work around it, and there are means to mimick the lookbehind, or just use capturing mechanism.

If you can use capturing , try

/^(?!.*\/changelogs(?:\/|$)).*\/(\d+)/

See the regex demo

 var re = /^(?!.*\\/changelogs(?:\\/|$)).*\\/(\\d+)/gmi; var str = 'http://www.opera.com/docs/changelogs/unified/1215/\\nhttp://www.whatever.com/docs/changelogs/anythingelse/anything/1215/\\nhttp://www.blabblah/security/advisory/1047\\nhttp://booger/security/advisory/1048/\\nftp://msn.global.whatever/somethingelse/1245\\nwhatever/it/doesnt/matter/could/be/anything/i/still/want/this/number/1049/'; var res = []; while ((m = re.exec(str)) !== null) { res.push(m[1]); } document.body.innerHTML = JSON.stringify(res, 0, 4); 

Or, use an optional group (if you are replacing):

 var re = /(\\/changelogs\\/.*)?\\/(\\d+)/gi; var str = 'http://www.opera.com/docs/changelogs/unified/1215/\\nhttp://www.whatever.com/docs/changelogs/anythingelse/anything/1215/\\nhttp://www.blabblah/security/advisory/1047\\nhttp://booger/security/advisory/1048/\\nftp://msn.global.whatever/somethingelse/1245\\nwhatever/it/doesnt/matter/could/be/anything/i/still/want/this/number/1049/'; var result = str.replace(re, function (m, g1, g2){ return g1 ? m : "NEW_VAL"; }); document.body.innerHTML = result; 

Answer 2

something like the below should do it. If you are not only interested in opera, you would be able to tweak this to be more general by replacing opera with .+ Additionally you could match things like com and net with something like (com|net|org|gov) in place of com:

http:\/\/www\.opera\.com(?!.*changelogs)(\/[^\/]+)*\/(\d+)\/{0,1}

Here is your regex 101 updated to reflect this

Answer 3

This pattern excludes lines with 'changelogs' in them and finds the last occurrence of a number encapsulated by slashes.

(?:\/)(?!.*changelogs)(?:\/[^\/]+)*\/(\d+)\/{0,1}

Here is the updated regex 101 .