Libravatar: Refactor cache handling to avoid unncessary stat

[debbugs.git] / Debbugs / Libravatar.pm

diff --git a/Debbugs/Libravatar.pm b/Debbugs/Libravatar.pm
index 08915fd4f1dab8612f41fae02c738bbb5afeeb18..0dafaf27adbdee63fc7634dd243223d98be49151 100644 (file)
--- a/Debbugs/Libravatar.pm
+++ b/Debbugs/Libravatar.pm
@@ -31,7 +31,7 @@ None known.
 use warnings;
 use strict;
 use vars qw($VERSION $DEBUG %EXPORT_TAGS @EXPORT_OK @EXPORT);
-use base qw(Exporter);
+use Exporter qw(import);
 
 use Debbugs::Config qw(:config);
 use Debbugs::Common qw(:lock);
@@ -39,8 +39,9 @@ use Libravatar::URL;
 use CGI::Simple;
 use Debbugs::CGI qw(cgi_parameters);
 use Digest::MD5 qw(md5_hex);
-use LWP::UserAgent;
 use File::Temp qw(tempfile);
+use File::LibMagic;
+use Cwd qw(abs_path);
 
 use Carp;
 
@@ -49,26 +50,19 @@ BEGIN{
      $DEBUG = 0 unless defined $DEBUG;
 
      @EXPORT = ();
-     %EXPORT_TAGS = (libravatar => [qw(cache_valid serve_cache retrieve_libravatar cache_location)]
+     %EXPORT_TAGS = (libravatar => [qw(retrieve_libravatar cache_location)]
                    );
      @EXPORT_OK = ();
      Exporter::export_ok_tags(keys %EXPORT_TAGS);
      $EXPORT_TAGS{all} = [@EXPORT_OK];
 }
 
-sub cache_valid{
-    my ($cache_location) = @_;
-    if (-e $cache_location) {
-        if (time - (stat($cache_location))[9] < 60*60) {
-            return 1;
-        }
-    }
-    return 0;
-}
 
-=item retreive_libravatar
+=over
+
+=item retrieve_libravatar
 
-     $cache_location = retreive_libravatar(location => $cache_location,
+     $cache_location = retrieve_libravatar(location => $cache_location,
                                            email => lc($param{email}),
                                           );
 
@@ -78,7 +72,7 @@ there isn't a matching avatar, or there is an error, returns undef.
 
 =cut
 
-sub retreive_libravatar{
+sub retrieve_libravatar{
     my %type_mapping =
         (jpeg => 'jpg',
          png => 'png',
@@ -90,7 +84,7 @@ sub retreive_libravatar{
         );
     my %param = @_;
     my $cache_location = $param{location};
-    $cache_location =~ s/\.[^\.]+$//;
+    $cache_location =~ s/\.[^\.\/]+$//;
     # take out a lock on the cache location so that if another request
     # is made while we are serving this one, we don't do double work
     my ($fh,$lockfile,$errors) =
@@ -100,11 +94,13 @@ sub retreive_libravatar{
     } else {
         # figure out if the cache is now valid; if it is, return the
         # cache location
-        my $temp_location = cache_location(email => $param{email});
-        if (cache_valid($temp_location)) {
+        my ($temp_location, $is_valid) = cache_location(email => $param{email});
+        if ($is_valid) {
             return $temp_location;
         }
     }
+    require LWP::UserAgent;
+
     my $dest_type;
     eval {
         my $uri = libravatar_url(email => $param{email},
@@ -155,7 +151,7 @@ sub retreive_libravatar{
                    $temp_fn,
                    $cache_location.'.'.$dest_type) == 0 or
                        die "convert file failed";
-            unlink($temp_fh);
+            unlink($temp_fn);
         };
         if ($@) {
             unlink($cache_location.'.'.$dest_type) if -e $cache_location.'.'.$dest_type;
@@ -173,9 +169,29 @@ sub retreive_libravatar{
     return $cache_location.'.'.$dest_type;
 }
 
+sub blocked_libravatar {
+    my ($email,$md5sum) = @_;
+    my $blocked = 0;
+    for my $blocker (@{$config{libravatar_blacklist}||[]}) {
+        for my $element ($email,$md5sum) {
+            next unless defined $element;
+            eval {
+                if ($element =~ /$blocker/) {
+                    $blocked=1;
+                }
+            };
+        }
+    }
+    return $blocked;
+}
+
+# Returns ($path, $is_valid)
+# - For blocked images, $path will be undef
+# - If $is_valid is false (and $path is not undef), the image should
+#   be re-fetched.
 sub cache_location {
     my %param = @_;
-    my $md5sum;
+    my ($md5sum, $stem);
     if (exists $param{md5sum}) {
         $md5sum = $param{md5sum};
     }elsif (exists $param{email}) {
@@ -183,12 +199,16 @@ sub cache_location {
     } else {
         croak("cache_location must be called with one of md5sum or email");
     }
-    for my $ext (qw(.png .jpg)) {
-        if (-e $config{libravatar_cache_dir}.'/'.$md5sum.$ext) {
-            return $config{libravatar_cache_dir}.'/'.$md5sum.$ext;
+    return (undef, 0) if blocked_libravatar($param{email},$md5sum);
+    $stem = $config{libravatar_cache_dir}.'/'.$md5sum;
+    for my $ext ('.png', '.jpg', '') {
+        my $path = $stem.$ext;
+        if (-e $path) {
+            my $is_valid = (time - (stat(_))[9] < 60*60) ? 1 : 0;
+            return ($path, $is_valid);
         }
     }
-    return $config{libravatar_cache_dir}.'/'.$md5sum;
+    return ($stem, 0);
 }
 
 ## the following is mod_perl specific
@@ -225,7 +245,7 @@ sub handler {
     my $location = $r->location();
     my ($email) = $uri =~ m/\Q$location\E\/?(.*)$/;
     if (not length $email) {
-        return Apache2::Const::NOT_FOUND;
+        return Apache2::Const::NOT_FOUND();
     }
     my $q = CGI::Simple->new();
     my %param = cgi_parameters(query => $q,
@@ -238,13 +258,13 @@ sub handler {
         return Apache2::Const::DECLINED();
     }
     # figure out what the md5sum of the e-mail is.
-    my $cache_location = cache_location(email => $email);
+    my ($cache_location, $is_valid) = cache_location(email => $email);
     # if we've got it, and it's less than one hour old, return it.
-    if (cache_valid($cache_location)) {
+    if ($is_valid) {
         serve_cache_mod_perl($cache_location,$r);
         return Apache2::Const::DECLINED();
     }
-    $cache_location = retreive_libravatar(location => $cache_location,
+    $cache_location = retrieve_libravatar(location => $cache_location,
                                           email => $email,
                                          );
     if (not defined $cache_location) {
@@ -258,17 +278,29 @@ sub handler {
 }
 
 
+
+our $magic;
+
 sub serve_cache_mod_perl {
     my ($cache_location,$r) = @_;
     if (not defined $cache_location or not length $cache_location) {
         # serve the default image
         $cache_location = $config{libravatar_default_image};
     }
+    $magic = File::LibMagic->new() if not defined $magic;
+
+    return Apache2::Const::DECLINED() if not defined $magic;
+
+    $r->content_type($magic->checktype_filename(abs_path($cache_location)));
+
     $r->filename($cache_location);
     $r->path_info('');
-    $r->finfo(APR::Finfo::stat($cache_location, APR::Const::FINFO_NORM, $r->pool));
+    $r->finfo(APR::Finfo::stat($cache_location, APR::Const::FINFO_NORM(), $r->pool));
 }
 
+=back
+
+=cut
 
 1;