Regex for IPv4 addresses with real octet range checks

The pattern everyone writes first is \d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}, and it matches 999.999.999.999 without hesitation. Digit counts cannot express a numeric range, so the only way to constrain an octet inside the pattern is to enumerate the shapes a number from 0 to 255 can take.

That enumeration is the three branch alternation the preset uses: 25 followed by 0 to 5, or a 2 followed by 0 to 4 and any digit, or an optional leading 0 or 1 followed by one or two digits. Between them the branches cover every value up to 255 and nothing above it, and the branches are ordered longest first so the greedy case is tried before the general one.

The word boundaries at each end do more work than they look like they do. Without them the pattern would find a valid address hiding inside a longer run of digits, and with them the whole of 999.999.999.999 fails: every starting position either begins a three digit run that cannot be followed by a dot, or sits between two digits where there is no boundary at all.

One thing it does allow is a leading zero, so 010.0.0.1 matches. That is a real ambiguity rather than a pedantic one, because some resolvers read a leading zero as an octal prefix and 010 then means eight. If your input can come from a user, either reject leading zeros explicitly or normalise the address before comparing it to anything.