use strict;
-use Cwd;
-use Date::Parse;
-use Date::Format;
-use Getopt::Long;
-use LWP::UserAgent;
-use Time::Local;
+use Date::Calc::Object; # For date/time math
+use File::stat; # For getting mtimes
+use Getopt::Long; # For parsing command-line options
+use HTTP::Status; # For HTTP status codes
+use LWP::UserAgent; # For HTTP client functionality
+use HTTP::Date; # For HTTP-compatible str2time and time2str
+
##################
# Global variables
##################
-# by default, this is not a test
+# Verbose? (Debugging messages)
my $verbose = 0;
# by default, this is the working directory
# LWP user agent for fetching files
# keep_alive is important, to avoid the overhead of
# establishing a new connection for each file we fetch
-my $ua = LWP::UserAgent->new(agent => "MirMirror/0.1",
- keep_alive => 1);
+my $ua = new LWP::UserAgent( agent => "MirMirror/0.1",
+ keep_alive => 1 );
###############
# if the file with the update time has disappeared, alert the admin
# and use the datestamp on the startpage file ( /en/index.html )
sub findLastUpdateTime() {
- open (UPDATETIME, "<", $lastupdatefile) or return $timenow;
+ open (UPDATETIME, "<", $lastupdatefile) or return Date::Calc->today()->date2time();
my $lastupdatetimestr = <UPDATETIME>;
close (UPDATETIME);
chomp ($lastupdatetimestr);
+ my $lastupdatetime = str2time ($lastupdatetimestr);
+ die "Can't parse last update time" if not $lastupdatetime;
return str2time ($lastupdatetimestr);
}
-# convert the date into a correctly formatted string
-sub date2ISOstr($) {
- return time2str ("%Y:%m:%dT%T", $_[0]);
-}
-
-# convert the date into RFC2616 format
-sub date2HTTPstr($) {
- return time2str ("%Y:%m:%dT%T", $_[0]);
-}
-
# write the time now into the last update file
sub saveLastUpdateTime() {
open (UPDATETIME, ">", $lastupdatefile) or die "Can't open $lastupdatefile for writing ($!)";
- print UPDATETIME date2ISOstr($timenow);
+ print UPDATETIME time2str($timenow);
close (UPDATETIME);
}
{
my ($fromtime, $totime) = @_;
my @files;
- for (my $time = str2time(time2str("%Y:%m:%dT00:00:00", $fromtime));
- $time < $totime;
- $time += 86400) {
- push @files, time2str("changes%Y%m%d.txt", $time);
+ my $time = Date::Calc->time2date($fromtime);
+ my $maxtime = Date::Calc->time2date($totime);
+ for (; $time <= $maxtime; ++$time) {
+ my ($y,$m,$d) = $time->date;
+ push @files, sprintf("changes%04d%02d%02d.txt",$y,$m,$d);
}
return @files;
}
print STDERR " as $localfile" if $localfile;
}
- my $req = HTTP::Request->new(GET => "$remotefile");
- # TODO: If-Modified-Since
- my $resp = $ua->request(HTTP::Request->new(GET => "$remotefile"));
- if ($resp->is_success) {
+ my $req = new HTTP::Request(GET => "$remotefile");
+
+ if ($localfile and -e $localfile) {
+ # Don't fetch unless more recent than local copy
+ my $stat = stat($localfile);
+ $req->header("If-Modified-Since" => time2str($stat->mtime));
+ }
+ my $resp = $ua->request($req);
+ if ($resp->is_success) { # 2xx codes
+ my $mtime = str2time($resp->header("Last-Modified"));
if ($localfile) {
- print STDERR " -> success\n" if $verbose;
+ if ($verbose) {
+ print STDERR " -> success";
+ print STDERR "; mtime ".time2str($mtime) if $mtime;
+ print STDERR "\n";
+ }
ensureDir(dirPart($localfile));
open (LOCAL, ">", "$localfile") or die "Can't open $localfile for writing ($!)";
print LOCAL $resp->content or die "Error writing $localfile ($!)";
close LOCAL or die "Error writing $localfile ($!)";
- # TODO: set mtime from Last-Modified
+
+ if ($mtime) {
+ utime $mtime, $mtime, $localfile;
+ }
}
return $resp->content;
}
+ elsif ($resp->is_redirect) { # 3xx codes
+ if ($resp->code == RC_NOT_MODIFIED) { # 304
+ print STDERR " -> not modified\n" if $verbose;
+ open (LOCAL, "<", "$localfile") or die "Can't open $localfile ($!)";
+ local $/; # slurp whole file
+ my $content = <LOCAL>;
+ close LOCAL;
+ return $content;
+ }
+ print STDERR " -> redirect (".$resp->code.")\n" if $verbose;
+ die "Can't fetch $remotefile (got redirect, not yet handled)";
+ }
else {
- print STDERR " -> failed\n" if $verbose;
+ if ($resp->code == RC_NOT_FOUND) { # 404
+ print STDERR " -> not found\n" if $verbose;
+ return undef;
+ }
+
+ print STDERR " -> failed (".$resp->code.")\n" if $verbose;
die "Can't fetch $remotefile (".$resp->status_line.")";
}
}
my $timeoflastupdate = findLastUpdateTime();
if ($verbose) {
- print STDERR "timenow is ".time2str("%c",$timenow)." \n";
- print STDERR "timeoflastupdate is ".time2str("%c",$timeoflastupdate)." \n\n";
+ print STDERR "timenow is ".time2str($timenow)." \n";
+ print STDERR "timeoflastupdate is ".time2str($timeoflastupdate)." \n\n";
}
# Now we know which days' changes we need to get from the server
my @changesfiles = getChangesFileList($timeoflastupdate, $timenow);
-if ($verbose or $listchangefiles) {
+if ($verbose) {
foreach my $file (@changesfiles) { print STDERR "using changes file $file\n"; }
- exit 0 if $listchangefiles;
+}
+
+if ($listchangefiles) {
+ foreach my $file (@changesfiles) { print "$file\n"; }
+ exit 0;
}
# get the changes files
-my @changesfilecontent;
-foreach my $file (@changesfiles) { push @changesfilecontent, getChangesFile($file); }
+my %changesfilecontent;
+foreach my $file (@changesfiles) { $changesfilecontent{$file} = getChangesFile($file); }
# if the file has not changed (response code 304) then ignore it
# iterate over all the fetched files, building up a list of files
# to fetch/delete
my %files;
-foreach my $changes (@changesfilecontent) {
- my @changes = split /[\r\n]+/, $changes;
+foreach my $changesfile (@changesfiles) {
+ my $changesfilecontent = $changesfilecontent{$changesfile};
+ if (not defined $changesfilecontent) {
+ print STDERR "Skipping changes file $changesfile; not present at remote end\n" if $verbose;
+ next;
+ }
+
+ my $date = $changesfile;
+ $date =~ s{^(?:.*/)?changes([0-9]+)\.txt$}{$1} or die "Can't extract date from changes filename $changesfile";
+
+ print STDERR "Processing changes file $changesfile\n" if $verbose;
+
+ my @changes = split /[\r\n]+/, $changesfilecontent;
foreach my $change (@changes) {
my ($time, $op, $path) = split ' ', $change;
+
+ # Ignore malformed lines, especially wacky paths that could be malicious
+ if ($time =~ /[^0-9:]/) {
+ die "Invalid time $time in $changesfile";
+ }
+
+ if ($path =~ m{(?:^|/)\.\.(?:/|$)}) {
+ die "Invalid path $path (contains ..) in $changesfile";
+ }
+
# Strip scheme and host from absolute URLs
$path =~ s{^[a-z]+://[a-z0-9\-\.]+/}{/};
- # TODO: Ignore changes prior to $timeoflastupdate
- # TODO: Ignore malformed lines, especially wacky paths that could be malicious
+
+ # Combine time with date and parse
+ $time = str2time("$date $time");
+ if (not defined $time) {
+ die "Failed to parse datetime '$date $time'";
+ }
+
+ # Ignore changes prior to $timeoflastupdate
+ next if $time < $timeoflastupdate;
+
$files{$path} = $op;
print STDERR "Marked $path as '$op'\n" if $verbose;
}
}
elsif ($op eq "add" or $op eq "change" or $op eq "Modification") {
# add/change: re-fetch the file
+ # FIXME: don't insist on reading entire file into memory
my $content = fetchFile("$remoteroot/$file","$workingdir/$file");
+ die "File $remoteroot/$file not found" if not defined $content;
}
else {
die "Unknown operation '$op'";