Regular expression using negative lookbehind not w

问题:

I have a source file with literally hundreds of occurrences of strings flecha.jpg and flecha1.jpg, but I need to find occurrences of any other .jpg image (i.e. casa.jpg, moto.jpg, whatever)

I have tried using a regular expression with negative lookbehind, like this:

(?<!flecha|flecha1).jpg

but it doesn't work! Notepad++ simply says that it is an invalid regular expression.

I have tried the regex elsewhere and it works, here is an example so I guess it is a problem with NPP's handling of regexes or with the syntax of lookbehinds/lookaheads.

So how could I achieve the same regex result in NPP?

If useful, I am using Notepad++ version 6.3 Unicode

As an extra, if you are so kind, what would be the syntax to achieve the same thing but with optional numbers (in this case only '1') as a suffix of my string? (even if it doesn't work in NPP, just to know)...

I tried (?<!flecha[1]?).jpg but it doesn't work. It should work the same as the other regex, see here (RegExr)

回答1:

Notepad++ seems to not have implemented variable-length look-behinds (this happens with some tools). A workaround is to use more than one fixed-length look-behind:

(?<!flecha)(?<!flecha1)\.jpg

As you can check, the matches are the same. But this works with npp.

Notice I escaped the ., since you are trying to match extensions, what you want is the literal .. The way you had, it was a wildcard - could be any character.

About the extra question, unfortunately, as we can't have variable-length look-behinds, it is not possible to have optional suffixes (numbers) without having multiple look-behinds.

回答2:

Solving the problem of the variable-length-negative-lookbehind limitation in Notepad++

Given here are several strategies for working around this limitation in Notepad++ (or any regex engine with the same limitation)

Defining the problem

Notepad++ does not support the use of variable-length negative lookbehind assertions, and it would be nice to have some workarounds. Let's consider the example in the original question, but assume we want to avoid occurrences of files named flecha with any number of digits after flecha, and with any characters before flecha. In that case, a regex utilizing a variable-length negative lookbehind would look like (?<!flecha[0-9]*)\.jpg.

Strings we don't want to match in this example

• flecha.jpg
• flecha1.jpg
• flecha00501275696.jpg
• aflecha.jpg
• img_flecha9.jpg
• abcflecha556677.jpg

---

The Strategies

1. Inserting Temporary Markers

   Begin by performing a find-and-replace on the instances that you want to avoid working with - in our case, instances of flecha[0-9]*\.jpg. Insert a special marker to form a pattern that doesn't appear anywhere else. For this example, we will insert an extra . before .jpg, assuming that ..jpg doesn't appear elsewhere. So we do:

   Find: (flecha[0-9]*)(\.jpg)

   Replace with: $1.$2

   Now you can search your document for all the other .jpg filenames with a simple regex like \w+\.jpg or (?<!\.)\.jpg and do what you want with them. When you're done, do a final find-and-replace operation where you replace all instances of ..jpg with .jpg, to remove the temporary marker.

2. Using a negative lookahead assertion

   A negative lookahead assertion can be used to make sure that you're not matching the undesired file names:

   (?<!\S)(?!\S*flecha\d*\.jpg)\S+\.jpg

   Breaking it down:

   • (?<!\S) ensures that your match begins at the start of a file name, and not in the middle, by asserting that your match is not preceded by a non-whitespace character.
   • (?!\S*flecha\d*\.jpg) ensures that whatever is matched does not contain the pattern we want to avoid
   • \S+\.jpg is what actually gets matched -- a string of non-whitespace characters followed by .jpg.

3. Using multiple fixed-length negative lookbehinds

   This is a quick (but not-so-elegant) solution for situations where the pattern you don't want to match has a small number of possible lengths.

   For example, if we know that flecha is only followed by up to three digits, our regex could be:

   (?<!flecha)(?<!flecha[0-9])(?<!flecha[0-9][0-9])(?<!flecha[0-9][0-9][0-9])\.jpg

回答3:

Are you aware that you're only matching (in the sense of consuming) the extension (.jpg)? I would think you wanted to match the whole filename, no? And that's much easier to do with a lookahead:

\b(?!flecha1?\b)\w+\.jpg

The first \b anchors the match to the beginning of the name (assuming it's really a filename we're looking at). Then (?!flecha1?\b) asserts that the name is not flecha or flecha1. Once that's done, the \w+ goes ahead and consumes the name. Then \.jpg grabs the extension to finish off the match.