PCRE2 (PHP>=7.3) Positive Lookbehind Regex to search for strings separated by ","

Question

I have this string:

'newProductsInfo: [[1] - $2 dollars,[2] New Product,[3] Hello,[4]Same, [5]Value]'

The word 'newProductsInfo', and a space ' ' always precede the array of strings. I want to be able to return

[1] - $2 dollars

[2] New Product

[3] Hello

[4]Same, [5]Value //should be returned since the comma is followed by a space ' '

in the Regex101 site.

Currently, using (?<=newProductsInfo: \[)[^,\]]+ only returns [1.

Edit: added possible return types in bubble plugin creator:

_"I have this string"_ - The first thing to do is usually to figure out in what format the string is. I don't recognize it. Where does this string come from? Most properly specified formats already have parsers so you don't have to go regex on the string. It looks like a complicated format so you will _most probably_ parse it erroneously if you try it with regex - if not with the data you see Today, then with something that pops up Tomorrow. — Ted Lyngmo, Mar 10 '23 at 14:57
@TedLyngmo I'm using it inside a [website builder](https://bubble.io), and in there I'm only allowed to return texts or a list of texts. aside from the object I have in my question, I'm also returning various other objects. Then again, I can only return it as a text. Limitations of the app made me use regex for recognizing patterns. — Prosy Arceno, Mar 10 '23 at 15:01
May I suggest returning text formatted as JSON? You then have a very well defined format and support in most (all?) languages to parse it properly. — Ted Lyngmo, Mar 10 '23 at 15:04
using JSON only displays something like [object Object] so it was useless for me. — Prosy Arceno, Mar 10 '23 at 15:05
Then you used it wrong. You need to build the JSON object and convert it to a string before returning it. Do the opposite on the receiving side. — Ted Lyngmo, Mar 10 '23 at 15:06
@NicoHaase I need to make it work [here](https://regex101.com/r/dHZWQ8/1) — Prosy Arceno, Mar 10 '23 at 15:48
@ProsyArceno How much time did you spend on trying to get JSON working? Did you give up when you noticed that your JSON object appeared as `[object Object]`? If you had just spent a few minutes extra there, I'm pretty sure you'd figure out how to make it appear as a fully parseable string - and all this regexing of your own format would be a non-issue. — Ted Lyngmo, Mar 10 '23 at 15:55
@TedLyngmo, I'm new to the web builder and newer to its plugin creator. But I am limited to the return of the plugin creator, see edit above — Prosy Arceno, Mar 10 '23 at 16:03
It looks like they've tried to make it simple to juggle JSON objects back and forth. I suggest spending some time investigating that a bit more. Perhaps ask about how to do it properly in one of their own user forums (if they have any) — Ted Lyngmo, Mar 10 '23 at 16:27
@TedLyngmo It's not so much a problem in the app but a problem in the return of the API I'm using. Mapping made it problematic because what supposedly should be returned as an array is being returned as a dictionary. What I'm currently doing is a workaround, which because of my limited knowledge of the app, may not be the most optimised. See [here](https://forum.bubble.io/t/extracting-return-of-a-plugin-and-saving-to-database/243599/3). I've gone past extracting the ids months ago. Now, I wanna get the names instead, so here I am. — Prosy Arceno, Mar 10 '23 at 16:40
@ProsyArceno The JSON you showed in that question is far from valid JSON, so I'd take their answer with a grain of salt. What if you asked the same question, but with valid JSON? — Ted Lyngmo, Mar 10 '23 at 16:49
@TedLyngmo that is the actual represented return from the API I'm currently using. See last edit a result from a call I made. — Prosy Arceno, Mar 10 '23 at 16:57
Your latest addition to the question contains pictures though. It's pretty hard to validate those. Put the data _as-is_ in the question as text and I'll give it a go. — Ted Lyngmo, Mar 10 '23 at 16:58

The fourth bird · Accepted Answer · 2023-03-10T20:21:05.493

2

Your pattern (was tagged JavaScript) only matches [1 because the negated character class [^,\]]+ can not cross matching a comma or ]

If you want the matches only, you can assert newProductsInfo: to the left followed by any character

Start the match with digits between square brackets up until either the next occurrence preceded by a comma, or the end of the string.

Using a lookbehind assertion (see the support) for it:

(?<=newProductsInfo: .*?)\[\d+].*?(?=,\[\d+]|$)

Regex demo

Edit

If you want to use PHP:

(?:newProductsInfo: \[|\G(?!^)),?\K[^,\n]+(?>,\h+[^,\n]+)*

Explanation

(?: Non capture group for the alternatives
- newProductsInfo: \[ Match newProductsInfo: [
- | Or
- \G(?!^) Assert the current position at the end of the previous match, not at the start
) Close the non capture group
,?\K Match an optional comma and forget what is matched so far
[^,\n]+
(?>,\h+[^,\n]+)* Optionally repeat matching a comma, 1+ spaces and 1+ chars other than , or newline

Regex demo

edited Mar 10 '23 at 20:21

answered Mar 10 '23 at 14:45

The fourth bird

154,723
16
55
70

I don't know why we're having different [results](https://regex101.com/r/6pex53/1). – Prosy Arceno Mar 10 '23 at 14:48
@ProsyArceno you should select Javascript as the flavor https://regex101.com/r/ADKUI3/1 – The fourth bird Mar 10 '23 at 14:49
This answer will only work on versions of JavaScript which support lookbehinds. On some browsers, lookbehinds would not be supported. – Tim Biegeleisen Mar 10 '23 at 14:53
@TimBiegeleisen That is correct, although the OP is using a lookbehind assertion – The fourth bird Mar 10 '23 at 14:54
Is there a way to do this on the default PCRE2 (PHP>=7.3)? I'm using bubble and I wasn't sure if JS was used – Prosy Arceno Mar 10 '23 at 14:56
@ProsyArceno Why did you tag Javascript? – The fourth bird Mar 10 '23 at 14:56
1

@ProsyArceno I have added a pattern that you can use with PHP . – The fourth bird Mar 10 '23 at 14:59
The php version is not working on PCRE2 (PHP>=7.3). Like [here](https://regex101.com/r/dHZWQ8/1). – Prosy Arceno Mar 10 '23 at 15:48
I might have used a wrong sample case because there are cases with which a string does not begin with a '[number]' – Prosy Arceno Mar 10 '23 at 15:51
@ProsyArceno Try it like this `(?:newProductsInfo: |\G(?!^)).*?\K\[[^][]*].*?(?=,\[[^][]*]|$)` See https://regex101.com/r/cffvF7/1 – The fourth bird Mar 10 '23 at 16:25
this worked somehow, but there should be four returns as opposed to just two. – Prosy Arceno Mar 10 '23 at 16:43
@ProsyArceno which 4? there are 2 parts with square brackets right? – The fourth bird Mar 10 '23 at 16:44
no, i just want the values separated by a comma (,), and preceded by a ‘newProductsInfo: [’ and ended with ‘]’. basically all csv inside the array. – Prosy Arceno Mar 10 '23 at 16:47
@ProsyArceno In that case `(?:newProductsInfo: \[|\G(?!^)),?\K[^,\n]+(?>,\h+[^,\n]+)*` https://regex101.com/r/sQ6rbN/1 – The fourth bird Mar 10 '23 at 17:00
1

this is perfect. worked nicely – Prosy Arceno Mar 10 '23 at 17:25
2

My version with a style from outerspace: https://regex101.com/r/EfwuEC/1 – Casimir et Hippolyte Mar 10 '23 at 19:44
1

@CasimiretHippolyte Awesome, like a real pro ;-) Some food for thought on this Friday evening. I remember seeing you use a marker before, but I have to read into this again. – The fourth bird Mar 10 '23 at 20:16
1

Thanks! Unfortunately, you can't see the markers with regex101, but `preg_match*` functions show them in the result array. – Casimir et Hippolyte Mar 10 '23 at 20:32
1

@CasimiretHippolyte Cool! I did not know that either about the preg_match functions. – The fourth bird Mar 10 '23 at 20:33

score 1 · Answer 2 · answered Mar 10 '23 at 14:52

Here is one approach using match():

var input = 'newProductsInfo: [[1] - $2 dollars,[2] New Product,[3] Hello,[4]Same, [5]Value]';
var matches = input.replace(/^.*?\[(?=\[\d+\])|\]$/g, "")
                   .split(/,(?=\[\d+\])/);
console.log(matches);

The call to replace() strips off the leading/trailing [ and ]. The call to split() splits the input string at every comma which is followed by a numbered bracket term such as [2].

PCRE2 (PHP>=7.3) Positive Lookbehind Regex to search for strings separated by ","

2 Answers2