Why does my Perl regular expression only find the last occurrence?

Question

I have the following input to a Perl script and I wish to get the first occurrence of NAME="..." strings in each of the `...

` structures. The entire file is read into a single string and the regex acts on that input. However, the regex always returns the last occurrence of `NAME="..."` strings. Can anyone explain what is going on and how this can be fixed? ``` Input file: ADSDF NAME="ORDERSAA" line1 line2 NAME="ORDERSA" line3 NAME="ORDERSAB"

line1 line2 NAME="ORDERSB" line3

line1 line2 NAME="ORDERSC" line3

line1 line2 NAME="ORDERSD" line3 line3 line3

line1 line2 NAME="QUOTES2" line3 NAME="QUOTES3" NAME="QUOTES4" line3 NAME="QUOTES5" line3

line1 line2 NAME="QUOTES6" NAME="QUOTES7" NAME="QUOTES8" NAME="QUOTES9" line3 line3

NAME="MyName IsKhan"

``` Perl Code starts here: ``` use warnings; use strict; my $nameRegExp = '(((NAME="(.+)")|(.*| ))*

Accepted Answer

Try this: ``` '(?> +(?:(?!

|NAME=).* +)*)NAME="([^"]+)"' ``` The `(?:.* +)*` consumes any unwanted lines, while the embedded lookahead -- `(?!|NAME=)` -- keeps it from overrunning the first NAME field or the end of the TABLE record. Just in case there's a record with no NAME field, I wrapped most of the expression in an atomic group -- `(?>...)` -- to prevent pointless backtracking. Notice that there's only one capturing group now. It's good practice to use them only when you really need to capture something; otherwise, use the non-capturing variety: `(?:...)`. EDIT: As to why your regex didn't work, the short answer is greediness. After matching the opening tag, this part takes over: ``` ((NAME="(.+)")|(.*| ))* ``` The part in the outermost parens can match anything: tags, `NAME=` lines, linefeeds--even empty lines. Wrap that in a group controlled by a greedy `*`, and now it matches everything. There's nothing in there to make it stop matching at the first NAME field, or even at the end of a record. So it's actually "finding" every occurrence of `NAME="..."` strings, but it's doing it in a single match attempt that consumes the entire input at once. With each iteration of the enclosing `*`, the capture groups are overwritten; when it's done, the final NAME value -- `MyName IsKhan` -- is what happens to be left in group 4. I used a negative lookahead to check the greediness, but you can also do that more directly, by using a non-greedy quantifier. Here's how my regex would look with a reluctant `*` in place of the negative lookahead: ``` ' +(?:.* +)*?NAME="([^"]+)"' ``` Simply switching to a non-greedy quantifier wouldn't help with your regex though; you'd have to make some structural changes as well.

Why does my Perl regular expression only find the last occurrence?

Solution

Problem