Re: split and join, what's the clever way to do it

Replies are listed 'Best First'.
Re^2: split and join, what's the clever way to do it by ikegami (Patriarch) on Jan 27, 2008 at 03:55 UTC
You've introduced a bug. The first arg of `split` is a regex even if you obfuscate that fact by using quotes, so the `\|` needs to be escaped: `split '\|', $line, 5; # WRONG split '\\\|', $line, 5; # ok split /\\|/', $line, 5; # Better split qr/\\|/', $line, 5; # Can also use qr// split $re, $line, 5; # or compiled regexes` [download] Other useful syntaxes: `my @rec = = split /\\|/, $line, 5;` [download] `my %rec; @rec{qw( var1 var2 var3 var4 var5 )} = split /\\|/, $line, 5;` [download]	[reply] [d/l] [select]
Re^3: split and join, what's the clever way to do it by dsheroh (Monsignor) on Jan 27, 2008 at 18:25 UTC
Having been bitten by that myself, are there any documents explaining why `split` behaves this way rather than the (IMO more sensible) alternative of splitting on a regex if it's /specified as a regex/ and splitting on a plain string if it's 'specified as a plain string'?	[reply] [d/l]
Re^4: split and join, what's the clever way to do it by ikegami (Patriarch) on Jan 28, 2008 at 01:10 UTC
That would have prevented the dynamic construction of split regex patterns in the pre-qr// world, and backwards compatibility prevents us from doing that now (whether it's wise or not).	[reply]
Re^4: split and join, what's the clever way to do it by ysth (Canon) on Jan 28, 2008 at 02:52 UTC
Not just split; `=~ EXPR` is implicitly a match operation using the result of EXPR as the regex. Though the tendency to use quotes instead of // for split, e.g. `split "\t"` seems to have introduced more confusion about split's operand. (Well, that and the special `split " "` mode.) -- Online Fortune Cookie Search	[reply] [d/l] [select]
Re^2: split and join, what's the clever way to do it by Fletch (Bishop) on Jan 27, 2008 at 03:51 UTC
The first argument to split is a regex, not a string. It makes a difference. `$ perl -le '$line = "a\|b\|c\|d"; print "-- $_:\n", join( "\n", split( $_ +, $line ) ) for ( qr/\\|/, "\|" )' -- (?-xism:\\|): a b c d -- \|: a \| b \| c \| d` [download] The cake is a lie. The cake is a lie. The cake is a lie.	[reply] [d/l]