Recipe 6.14 Matching from Where the Last Pattern Left Off

6.14.1 Problem

You want to match again in the same string, starting from where the last match left off. This is a useful approach to take when repeatedly extracting data in chunks from a string.

6.14.2 Solution

Use a combination of the /g and /c match modifiers, the \G pattern anchor, and the pos function.

6.14.3 Discussion

The /g modifier on a pattern match makes the matching engine keep track of the position in the string where it finished matching. If the next match also uses /g on that string, the engine starts looking for a match from this remembered position. This lets you, for example, use a while loop to progressively extract repeated occurrences of a match. Here we find all non-negative integers:

while (/(\d+)/g) {     print "Found number $1\n"; }

Within a pattern, \G means the end of the previous match. For example, if you had a number stored in a string with leading blanks, you could change each leading blank into the digit zero this way:

$n = "   49 here"; $n =~ s/\G /0/g; print $n; 00049 here

You can also make good use of \G in a while loop. Here we use \G to parse a comma-separated list of numbers (e.g., "3,4,5,9,120"):

while (/\G,?(\d+)/g) {     print "Found number $1\n"; }

By default, when your match fails (when we run out of numbers in the examples, for instance) the remembered position is reset to the start. If you don't want this to happen, perhaps because you want to continue matching from that position but with a different pattern, use the modifier /c with /g:

$_ = "The year 1752 lost 10 days on the 3rd of September"; while (/(\d+)/gc) {     print "Found number $1\n"; } # the /c above left pos at end of final match if (/\G(\S+)/g) {     print "Found $1 right after the last number.\n"; } Found number 1752 Found number 10 Found number 3 Found rd after the last number.

Successive patterns can use /g on a string, which remembers the ending position of the last successful match. That position is associated with the scalar matched against, not with the pattern. It's reset if the string is modified.

The position of the last successful match can be directly inspected or altered with the pos function, whose argument is the string whose position you want to get or set. Assign to the function to set the position.

$a = "Didst thou think that the eyes of the White Tower were blind?"; $a =~ /(\w{5,})/g; print "Got $1, position in \$a is ", pos($a), "\n"; Got Didst, position in $a is 5 pos($a) = 30; $a =~ /(\w{5,})/g; print "Got $1, position in \$a now ", pos($a), "\n"; Got White, position in $a now 43

Without an argument, pos operates on $_:

$_ = "Nay, I have seen more than thou knowest, Grey Fool."; /(\w{5,})/g; print "Got $1, position in \$_ is ", pos, "\n"; pos = 42; /\b(\w+)/g; print "Next full word after position 42 is $1\n"; Got knowest, position in $_ is 39 Next full word after position 42 is Fool

6.14.4 See Also

The /g and /c modifiers are discussed in perlre(1) and the "The m// Operator (Matching)" section of Chapter 5 of Programming Perl