You're determined to make me do some real work, aren't you? :P

As a test, I hacked my crawler to download a single 20MB JPEG image from a remote http server and used tcpdump to view the network traffic. I tested this with and without the ExifTool FastScan option, and repeated the test a few times to make sure I wasn't being fooled by other network traffic. I consistently saw about 50 packets transferred with FastScan set, and about 37000 packets without FastScan.

So I conclude that this definitely works for me. I can't say what the difference is for you. I ran my tests on Mac OS X 10.4.11 with Perl 5.8.6.

#! /usr/bin/perl -w
#
# File:         url_info
#
# Author:       Phil Harvey
#
# Syntax:       url_info URL
#
# Description:  test to get image info from image on the net
#
# Example:      url_info http://owl.phy.queensu.ca/~phil/big.jpg
#
# References:   Based on web crawler script:
#               http://www.linuxjournal.com/files/linuxjournal.com/lin
+uxjournal/articles/022/2200/2200l1.html
#
use strict;
use Image::ExifTool;

sub url_info($);

my $DEBUG = 0;  # set to 1 for debugging network stuff
my $url = shift or die "Syntax: url_info URL\n";

my $exifTool = new Image::ExifTool;
$exifTool->Options(FastScan => 1);

my $info = url_info($url);

die "No image info for $url\n" unless $info;

foreach (sort keys %$info) {
    print "$_ => $$info{$_}\n";
}

exit 1;

#---------------------------------------------------------------------
# Get the page at specified URL
# Inputs: 0) URL
# Returns: 0) 
#---------------------------------------------------------------------
sub url_info($)
{
    my $url = shift;
    my ($protocol, $host, $port, $document) =
        $url =~ m|^([^:/]*)://([^:/]*):*([0-9]*)/*([^:]*)$|;
    
    # Some constants used to access the TCP network.
    my $AF_INET=2;
    my $SOCK_STREAM=1;
    
    # Use default http port if none specified.
    $port = 80 unless $port;
    
    # Get the protocol number for TCP.
    my ($name,$aliases,$proto) = getprotobyname("tcp");
    
    # Get the IP addresses for the two hosts.
    my ($type,$len,$thataddr);
    ($name,$aliases,$type,$len,$thataddr) = gethostbyname($host);
    
    # Check we could resolve the server host name.
    return undef unless defined $thataddr;
    my ($a,$b,$c,$d) = unpack('C4', $thataddr);
    if (not defined $d or
        ($a eq '' && $b eq '' && $c eq '' && $d eq ''))
    {
        warn "Unknown host $host.\n";
        return undef;
    }
    print "Server: $host ($a.$b.$c.$d)\n" if $DEBUG;
    
    # Pack the AF_INET magic number, the port, and the (already
    # packed) IP addresses into the same format as the C structure
    # would use. Note this is architecture dependent: this pack format
    # works for 32 bit architectures.
    my $that = pack("S n a4 x8", $AF_INET, $port, $thataddr);

    # Create the socket and connect.
    unless (socket(S, $AF_INET, $SOCK_STREAM, $proto)) {
        warn "Cannot create socket.\n";
        alarm 0;
        return undef;
    }
    print "Socket OK\n" if $DEBUG;
    local $SIG{ALRM} = sub { die "ALARM\n" };
    alarm 3600; # set timeout of 1 hour
    my $result = eval 'connect(S, $that)';
    if ($@ or not $result) {
        warn "Cannot connect to server $host, port $port.\n";
        alarm 0;
        return undef;
    }
    print "Connect OK\n" if $DEBUG;
    
    # Turn buffering in the socket off, and send request to the server
    select(S); $| = 1; select(STDOUT);
    
    $result=eval 'print S "GET /$document HTTP/1.1\nHost: $host\n\n"';
    if ($@ or not $result) {
        warn "Timeout when sending to $host, port $port.\n";
        alarm 0;
        return undef;
    }
    
    # Receive the response. Check to ensure the response is of MIME
    # type text/html or text/plain.
    my $header = 1;
    my $header_text = "";

    for (;;) {
        $_ = eval '<S>';
        if ($@) {
            warn "Timeout when reading from $host, port $port\n";
            last;
        }
        last unless defined $_;
        # Check if we've hit the end of the HTTP header (empty line)
        # If we have, check for a content-type header line, and ensure
        # it is valid.
        if( m|^[\n\r]*$| ){
            $header = 0;
            my ($content) = $header_text=~/Content-type: ([^\s;]+)/i;
            if ($content and $content =~ m{^image/(.*)}) {
                # extract image metadata
                my $raf = new File::RandomAccess(\*S);
                $info = $exifTool->ImageInfo($raf);
            } else {
                warn "Not an image\n";
            }
            last; # all done
        } elsif($header == 1){
            # Save to a header string if we're still working on the
            # HTTP header.
            $header_text .= "   " . $_;
        }
    }
    eval 'close S';
    alarm 0;

    print "HTTP header: \n $header_text" if $DEBUG;
    return $info;
}
[download]