Here's a simple example, iterating through the %food_color
hash from the introduction.
# %food_color per the introduction
while(($food, $color) = each(%food_color)) {
print "$food is $color.\n";
}
Banana is yellow.
Apple is red.
Carrot is orange.
Lemon is yellow.
foreach $food (keys %food_color) {
my $color = $food_color{$food};
print "$food is $color.\n";
}
Banana is yellow.
Apple is red.
Carrot is orange.
Lemon is yellow.
We didn't really need the $color
variable in the foreach
example because we only use it once. Instead, we could have just written:
print
"$food
is
$food_color{$food}.\n"
Every time each
is called on the same hash, it returns the "next" key-value pair. We say "next" because the pairs are returned in the order the underlying lookup structure imposes on them, and this order is almost never alphabetic or numeric. When each
runs out of hash elements, it returns the empty list ()
, which tests false and terminates the while
loop.
The foreach
example uses keys
, which constructs an entire list containing every key from hash, before the loop even begins executing. The advantage to using each
is that it gets the keys and values one pair at a time. If the hash contains many keys, not having to pre-construct a complete list of them can save substantial memory. The each
function, however, doesn't let you control the order in which pairs are processed.
Using foreach
and keys
to loop over the list lets you impose an order. For instance, if we wanted to print the food names in alphabetical order:
foreach $food (sort keys %food_color) {
print "$food is $food_color{$food}.\n";
}
Apple is red.
Banana is yellow.
Carrot is orange.
Lemon is yellow.
This is a common use of foreach
. We use keys
to obtain a list of keys in the hash, and then we use foreach
to iterate over them. The danger is that if the hash contains a large number of elements, the list returned by keys
will use a lot of memory. The trade-off lies between memory use and the ability to process the entries in a particular order. We cover sorting in more detail in Recipe 5.9.
Because keys
, values
, and each
all use the same internal data structures, be careful about mixing calls to these functions or prematurely exiting an each
loop. Each time you call keys
or values
, the current location for each
is reset. This code loops forever, printing the first key returned by each
:
while ( ($k,$v) = each %food_color ) {
print "Processing $k\n";
keys %food_color; # goes back to the start of %food_color
}
Modifying a hash while looping over it with each
or foreach
is, in general, fraught with danger. The each
function can behave differently with tie
d and untied hashes when you add or delete keys from a hash. A foreach
loops over a pre-generated list of keys, so once the loop starts, foreach
can't know whether you've added or deleted keys. Keys added in the body of the loop aren't automatically appended to the list of keys to loop over, nor are keys deleted by the body of the loop deleted from this list.
Example 5.1 reads a mailbox file and reports on the number of messages from each person. It uses the From:
line to determine the sender. (It isn't smart in this respect, but we're showing hash manipulation, not mail-file processing.) Supply the mailbox filename as a command-line argument, or use "-"
to indicate you're piping the mailbox to the program.
#!/usr/bin/perl
# countfrom - count number of messages from each sender
$filename = $ARGV[0] || "-";
open(FILE, "<$filename") or die "Can't open $filename : $!";
while(<FILE>) {
if (/^From: (.*)/) { $from{$1}++ }
}
foreach $person (sort keys %from) {
print "$person: $from{$person}\n";
}