Regular Expressions Board

[paulmw]

<-->[^<]*<xx>

[JackDunaway]

---

@DavidU wrote:

I've been trying to get my head around regex but I'm stuck.... The string consists of blocks of data separated by markers, unfortunately not all data blocks were complete, so I have to search through the string to find complete blocks

...

As a bonus, if I could have every complete data block returned, that would be fantastic.

---

Hey David, have a go with Lookarounds and this regex: (?s)(?<=<-->)(?:.(?!<-->))*?(?=<xx>)

Note that I solved this with a pair of positive/negative lookbehinds on your starting match of <-->; you may be able to rewrite this with a pair of positive/negative lookaheads on the ending match of <xx>.

The leading (?s) is just a mode modifier which directs the regex engine to match newlines with the dot.

One thing I like about lookarounds is they are zero-width assertions, meaning you can return content between two anchors while excluding the anchors themselves from the match. You may want to include the whitespace \n as part of your <--> and <xx> anchors to avoid an additional Trim Whitespace on the content of the matches.

[Hornless.Rhino]

---

@JackDunaway wrote:

Hey David, have a go with Lookarounds and this regex: (?s)(?<=<-->)(?:.(?!<-->))*?(?=<xx>)

---

Wow. That looks like something I'd expect to see in the channel tunnel.

[DavidU]

Thanks to both paulmw and Jack for your replys, I think I understand Jack's solution...

Extra Kudos to Jack for his mind reading abilities. (Unintentionally) I hadn't actually specified the whole of the problem. The Data Blocks themselves contain sub blocks marked with <-1> ... <x1>  etc.  Jack's solution passes these through successfully.

David

[DavidU]

Hi,

I'm back.

I've been testing the regex from Jack. Under some circumstances LabVIEW crashes at the Match Regular Expression subVI. If the string contains at least one <--> then for a given string length then LabVIEW crashes. Attached is a test vi. The search string contains 1888 characters including the initial "<-->". If you add another character after <--> and run the vi it crashes. Add it before the <--> and the vi runs fine. Is this down to some limit on the lookaround search length?

LabVIEW 8.6.1f1

Any help appreciated

David

[JackDunaway]

---

@DavidU wrote:

Under some circumstances LabVIEW crashes at the Match Regular Expression subVI. If the string contains at least one <--> then for a given string length then LabVIEW crashes. Attached is a test vi. The search string contains 1888 characters including the initial "<-->". If you add another character after <--> and run the vi it crashes. Add it before the <--> and the vi runs fine. Is this down to some limit on the lookaround search length?

---

David, i'm unable to crash LabVIEW even with a string length of >1MB (with LV 12.0.1f4 32-bit). I notice your signature says LV8.6.1 -- it's possible that LabVIEW or the underlying PCRE library has been enhanced since then to address the issue you're seeing. (If this is the case, you may be relegated to slice-n-dice with more primitive string splitting methods 😕 )

[GregFreeman]

Need help with this regex/search and replace. I have a string that is [ABC][XYZ]. I need to search for ABC between the first set of brackets, and if there is a match, I need to replace XYZ. So, my output would be [ABC][NEW]. Essentially if there is a match, keep the ABC in brackets but replace whats in the second set of brackets with a user specified string. I have no problem searching for [ABC] and if a match creating [NEW]  then concatating the string back up. But I'm curious if there is a way to do it in a single primative call to search and replace.

[JackDunaway]

---

@for(imstuck) wrote:

Need help with this regex/search and replace. I have a string that is [ABC][XYZ]. I need to search for ABC between the first set of brackets, and if there is a match, I need to replace XYZ. So, my output would be [ABC][NEW]. Essentially if there is a match, keep the ABC in brackets but replace whats in the second set of brackets with a user specified string. I have no problem searching for [ABC] and if a match creating [NEW]  then concatating the string back up. But I'm curious if there is a way to do it in a single primative call to search and replace.

---

Looks like nothing a couple lookarounds as zero-width assertions can't handle.

Try (?<=\[ABC\]\[)(.*?)(?=\]) with the Search and Replace String Node, ensuring to turn on Regular Expression matching via it's right-click menu option.

[GregFreeman]

---

@JackDunaway wrote:

---

@for(imstuck) wrote:

Need help with this regex/search and replace. I have a string that is [ABC][XYZ]. I need to search for ABC between the first set of brackets, and if there is a match, I need to replace XYZ. So, my output would be [ABC][NEW]. Essentially if there is a match, keep the ABC in brackets but replace whats in the second set of brackets with a user specified string. I have no problem searching for [ABC] and if a match creating [NEW]  then concatating the string back up. But I'm curious if there is a way to do it in a single primative call to search and replace.

---

Looks like nothing a couple lookarounds as zero-width assertions can't handle.

Try (?<=\[ABC\]\[)(.*?)(?=\]) with the Search and Replace String Node, ensuring to turn on Regular Expression matching via it's right-click menu option.

---

Nice, Jack. I know the regex option, but as per usual, the real issue is always the regex!

[JackDunaway]

---

for(imstuck) wrote: Nice, Jack. I know the regex option...

---

Oh, totally, didn't intend to speak down to an old pro like yourself. 😄

But I speak for myself, as a "newer" developer when I first found out about this hidden feature of the "Search and Replace" node, it was like Christmas morning convolved with an orthogonal plane of programming capability....

My hope is that a well-documented post gives an early Christmas to someone else, just as I have enjoyed so many early Christmases from others on this helpful forum 🙂

Spoiler

(**EDIT** : Ah, and in the spirit of keeping high a SNR on this regex board, here's an actual recent regex with some lookarounds from production code: (?:/s*)?(?<!\\)"(.*?)(?<!\\)"(?:/s*)?:(?:/s*)?(?<!\\)"(.*?)(?<!\\)"  One might say, "It's not fun til your regular expressions start to look like Brain**bleep**." Likewise, a saner mind might say, "It's no fun once your regular expressions start to look like Brain**bleep**")

**EDIT 2** : LOL, Lithium forum parser, you're so silly when it comes to curse words and special character sequences