I want a regex pattern to grab text within a pattern that excludes matches followed by a particular string

Question

I'm using preg_match_all() like this:

preg_match_all('/("user":"(.*?)".*?-->)/', $input_lines, $output_array);

On a string, the idea is that I want it to get whatever comes after "user" in a commented out block (it's a long story). So let's say $input_lines is something like:

<!-- block {"user":"josh-hart", 1234566 ,"hide":false} /-->
<!-- block {"user":"jalen-brunson", 7744633 ,"hide":true} /-->
<!-- block {"user":"julius-randle", 333333,"hide":false} /-->
<!-- block {"user":"obi-toppin", 4hh3n33n33n, /-->
<!-- block {"user":"rj-barrett", nmremxxx!! ,"hide":true} /-->
<!-- block {"user":"mitch-robinson",yahaoao /-->

I want it to match the user. But here's the the thing, I only want the user if "hide":true does not appear before the /-->. So for this string I would want the matches to be:

josh-hart, julius-randle, obi-toppin, mitch-robinson

What is this called in regex terms and how do i do it?

Answer 1

Assuming that opening and closing a comment is from <!-- to --> and there is no other use of these in between, you can first get the matches out of the way that contain <!-- ... "hide":true ... --> without crossing the opening or closing of a comment.

Then you can get a single match of the username, still in between the comment markers and independent of the order of appearance.

<!--(?:(?!-->|"hide":).)*+"hide":true\b(?:(?!-->).)*/-->(*SKIP)(*F)|<!--(?:(?!-->|"user":).)*"user":"\K[^"]+(?="(?:(?!-->).)*-->)

The pattern matches:

• <!-- Match literaly
• (?:(?!-->|"hide":).)*+ Optionally repeat matching any character not directly followed by either --> or "hide": using a Tempered greedy token
• "hide":true\b Match "hide":true followed by a word boundary to prevent a partial word match
• (?:(?!-->).)*/--> Match until the closing -->
• (*SKIP)(*F) Skip the current match
• | Or
• <!-- Match literally
• (?:(?!-->|"user").)* Optionally repeat matching any character not directly followed by either --> or "user:
• "user":"\K Match "user":" and forget what is matched so far
• [^"]+ Match 1+ chars other than " (the username that you want to match)
• (?="(?:(?!-->).)*-->) Assert --> to the right

Note that you can make the matching of the username more specific, as for now it matches 1 or more characters other than a double quote with [^"]+ which can also be a space or a newline. If you want to match only non whitespace characters except for a double quote, than you can change it to [^\s"]+

See a regex demo.

Answer 2

You may try that:

user":"([^"]+)"(?!.*"hide":true(?=.*\/-->))

Explanation:

• user":"([^"]+)" - after matching user":" it looks for characters inside " until it reaches a ".
• () ensures 1st capture group.
• (?!.*"hide":true negative lookahead to figure there is no "hide":true after that.
• (?=.*\/-->) positive lookahead to ensure that /--> is not preceded by "hide":true

Demo

Answer 3

As suggested by @bobblebubble, you can use the (*SKIP)(*FAIL) combo to skip whatever you don't want to match:

^                     # Match at the start of a line
.*                    # anything
"hide"\s*:\s*true     # followed by '"hide":true'
.*                    # then anything until the end of the line
                      # (i.e. a line that contains '"hide":true')
(*SKIP)(*FAIL)        # which we tell the engine to skip entirely
|                     # or
"user"\s*:\s*"\K      # '"user":"', which we forfeit,
[^"]+                 # followed by 1 or more non-quote characters.

Try it on regex101.com.