[PATCH] Pull gitweb If-Modified-Since handling out into its own function and use for snapshots.

The current gitweb only generates Last-Modified and handles
If-Modified-Since headers for the git_feed action.  This patch breaks
the Last-Modified and If-Modified-Since handling code out from
git_feed into a new function modified_since.  This makes the code easy
to reuse for other actions where it is appropriate, and I've added the
code to do that in git_snapshot.

Signed-off-by: W Trevor King <redacted>

---
 gitweb/gitweb.perl |   54 ++++++++++++++++++++++++++++++++++++---------------
 1 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/gitweb/gitweb.perl b/gitweb/gitweb.perl
index a8b5fad..96a13e7 100755
--- a/gitweb/gitweb.perl
+++ b/gitweb/gitweb.perl

@@ -7003,6 +7003,31 @@ sub snapshot_name {
 	return wantarray ? ($name, $name) : $name;
 }
 
+sub modified_since {
+	my ($content_type, $latest_epoch, $timezone) = @_;
+	our $cgi;
+
+	my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
+	if (defined $if_modified) {
+		my $since;
+		if (eval { require HTTP::Date; 1; }) {
+			$since = HTTP::Date::str2time($if_modified);
+		} elsif (eval { require Time::ParseDate; 1; }) {
+			$since = Time::ParseDate::parsedate($if_modified, GMT => 1);
+		}
+		if (defined $since && $latest_epoch <= $since) {
+			my %latest_date = parse_date($latest_epoch, $timezone);
+			print $cgi->header(
+				-type => $content_type,
+				-charset => 'utf-8',
+				-last_modified => $latest_date{'rfc2822'},
+				-status => '304 Not Modified');
+			return 0;
+		}
+	}
+	return 1;
+}
+
 sub git_snapshot {
 	my $format = $input_params{'snapshot_format'};
 	if (!@snapshot_fmts) {

@@ -7029,6 +7054,14 @@ sub git_snapshot {
 
 	my ($name, $prefix) = snapshot_name($project, $hash);
 	my $filename = "$name$known_snapshot_formats{$format}{'suffix'}";
+
+	my %co = parse_commit($hash) or die_error(404, "Unknown commit object");
+	if (! modified_since($known_snapshot_formats{$format}{'type'},
+	                     $co{'committer_epoch'},
+	                     $co{'committer_tz'})) {
+		return;
+	}
+
 	my $cmd = quote_command(
 		git_cmd(), 'archive',
 		"--format=$known_snapshot_formats{$format}{'format'}",

@@ -7038,9 +7071,11 @@ sub git_snapshot {
 	}
 
 	$filename =~ s/(["\\])/\\$1/g;
+	my %latest_date = parse_date($co{'committer_epoch'}, $co{'committer_tz'});
 	print $cgi->header(
 		-type => $known_snapshot_formats{$format}{'type'},
 		-content_disposition => 'inline; filename="' . $filename . '"',
+		-last_modified => $latest_date{'rfc2822'},
 		-status => '200 OK');
 
 	open my $fd, "-|", $cmd

@@ -7821,22 +7856,9 @@ sub git_feed {
 		%latest_commit = %{$commitlist[0]};
 		my $latest_epoch = $latest_commit{'committer_epoch'};
 		%latest_date   = parse_date($latest_epoch, $latest_commit{'comitter_tz'});
-		my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
-		if (defined $if_modified) {
-			my $since;
-			if (eval { require HTTP::Date; 1; }) {
-				$since = HTTP::Date::str2time($if_modified);
-			} elsif (eval { require Time::ParseDate; 1; }) {
-				$since = Time::ParseDate::parsedate($if_modified, GMT => 1);
-			}
-			if (defined $since && $latest_epoch <= $since) {
-				print $cgi->header(
-					-type => $content_type,
-					-charset => 'utf-8',
-					-last_modified => $latest_date{'rfc2822'},
-					-status => '304 Not Modified');
-				return;
-			}
+		if (! modified_since($content_type, $latest_epoch,
+		                     $latest_commit{'comitter_tz'})) {
+			return;
 		}
 		print $cgi->header(
 			-type => $content_type,

-- 
1.7.3.4

I'm sorry I am only now answering, but unfortunately this patch has
fallen thought crack.

"W. Trevor King" [off-list ref] writes:

The current gitweb only generates Last-Modified and handles
If-Modified-Since headers for the git_feed action.  This patch breaks
the Last-Modified and If-Modified-Since handling code out from
git_feed into a new function modified_since.  This makes the code easy
to reuse for other actions where it is appropriate,

That is a very good idea.

Of course adding support for If-Modified-Since has sense only if we
can calculate modification time without doing all the work (and
calculating modification time separately doesn;t add to work to be
done), or at least we are to send large amount of data so even if we
cannot save CPU time and I/O hit we can save network bandwidth.

and I've added the code to do that in git_snapshot.

Nitpick about grammar: here you change from impersonal to personal
form in commit message, unnecessarily I think.

Passing $content_type to function named modified_since() looks quite
strange to me.

+	our $cgi;
+
+	my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
+	if (defined $if_modified) {
+		my $since;
+		if (eval { require HTTP::Date; 1; }) {
+			$since = HTTP::Date::str2time($if_modified);
+		} elsif (eval { require Time::ParseDate; 1; }) {
+			$since = Time::ParseDate::parsedate($if_modified, GMT => 1);
+		}
+		if (defined $since && $latest_epoch <= $since) {
+			my %latest_date = parse_date($latest_epoch, $timezone);

Shouldn't we pass \%latest_date (or rather \%modification_date to be
more generic)?  Ah, I see that parse_date() subroutine does not store
original epoch not original timezone.

By the way, for $latest_date{'rfc2822'} we don't need $timezone
argument, and I think we should not pass it to generic
modified_since() function (or whatever it will be called).

+			print $cgi->header(
+				-type => $content_type,
+				-charset => 'utf-8',

Do we need -charset here?  There is no HTTP body, and -charset => 'utf-8'
is incorrect for snapshots (with e.g. 'application/x-zip' $content_type)

For a function named modified_since(), it does too many things.

I think a better idea would be to make this function/subroutine (named
e.g. if_modified_since() or something like that) to work more like
die_error() -- it should simply end request instead of having caller
to use complcated calling convention, and care about eraly termination
by itself.

This means turning modified_since() into subroutine (and renaming it),
and ending it with

  goto DONE_GITWEB;

or

  goto DONE_REQUEST;

depending on the code base you are applying to.

Have you run gitweb tests?  This simply has no way of working, as %co
variable you use here is not defined anywhere in git_snapshot()
subroutine.

What if there is no commit, for example if we are rewuesting snapshot
of a tree by its SHA-1?  We need to be able to deal with sich
situation, if only by not handling Last-Modified / If-Modified-Since.

Do we use %latest_date beside handling If-Modified-Since?  Is there a
need for separate $latest_epoch and %;atest_date variables?

-		my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
-		if (defined $if_modified) {
-			my $since;
-			if (eval { require HTTP::Date; 1; }) {
-				$since = HTTP::Date::str2time($if_modified);
-			} elsif (eval { require Time::ParseDate; 1; }) {
-				$since = Time::ParseDate::parsedate($if_modified, GMT => 1);
-			}
-			if (defined $since && $latest_epoch <= $since) {
-				print $cgi->header(
-					-type => $content_type,
-					-charset => 'utf-8',
-					-last_modified => $latest_date{'rfc2822'},
-					-status => '304 Not Modified');
-				return;
-			}
+		if (! modified_since($content_type, $latest_epoch,
+		                     $latest_commit{'comitter_tz'})) {
+			return;
 		}
 		print $cgi->header(
 			-type => $content_type,
-- 
1.7.3.4

-- 
Jakub Narebski

The current gitweb only generates Last-Modified and handles                     
If-Modified-Since headers for the git_feed action.  This patch breaks           
the Last-Modified and If-Modified-Since handling code out from                  
git_feed into a new function modified_since.  This makes the code easy          
to reuse for other actions where it is appropriate and adds the code            
to do that to git_snapshot.                                                     

Signed-off-by: W Trevor King <redacted>                                 
---
Thanks for the feedback, Jakub.  I believe I've addressed your points
in this revised patch.

Changed since v1:
- Fixed impersonal/personal commit message
- Simplified modified_since() -> die_if_unmodified().  Now it only
  takes a $latest_epoch argument and we set fewer headers.  It also
  jump straight to DONE_GITWEB if the page is unmodified.
- Ran tests (which passed this time).  Somehow I missed t/README
  earlier.

I'd add tests for my patch, but I'd have to have commit timestamps
from the test repository, and I don't understand the test framework
well enough to know how to do that cleanly.

Things that haven't changed:

quoted

@@ -7038,9 +7071,11 @@ sub git_snapshot {
 	}
 
 	$filename =~ s/(["\\])/\\$1/g;
+	my %latest_date = parse_date($co{'committer_epoch'}, $co{'committer_tz'});

...
This simply has no way of working, as %co variable you use here is
not defined anywhere in git_snapshot() subroutine.

I defined it earlier:

quoted

@@ -7029,6 +7054,14 @@ sub git_snapshot {

  my ($name, $prefix) = snapshot_name($project, $hash);
  my $filename = "$name$known_snapshot_formats{$format}{'suffix'}";
+
+ my %co = parse_commit($hash) or die_error(404, "Unknown commit object");
+ if (! modified_since($known_snapshot_formats{$format}{'type'},
+                      $co{'committer_epoch'},

While modified_since() (now die_if_unmodified()) handles the 304 case,
we also want to set Last-Modified for 200s.  That's what the latter
code is trying to do.

What if there is no commit, for example if we are rewuesting snapshot
of a tree by its SHA-1?  We need to be able to deal with sich
situation, if only by not handling Last-Modified / If-Modified-Since.

The parse_commit() call should handle these cases.

diff --git a/gitweb/gitweb.perl b/gitweb/gitweb.perl
index a8b5fad..b944351 100755
--- a/gitweb/gitweb.perl
+++ b/gitweb/gitweb.perl

@@ -7003,6 +7003,28 @@ sub snapshot_name {
 	return wantarray ? ($name, $name) : $name;
 }
 
+sub die_if_unmodified {
+	my ($latest_epoch) = @_;
+	our $cgi;
+
+	my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
+	if (defined $if_modified) {
+		my $since;
+		if (eval { require HTTP::Date; 1; }) {
+			$since = HTTP::Date::str2time($if_modified);
+		} elsif (eval { require Time::ParseDate; 1; }) {
+			$since = Time::ParseDate::parsedate($if_modified, GMT => 1);
+		}
+		if (defined $since && $latest_epoch <= $since) {
+			my %latest_date = parse_date($latest_epoch);
+			print $cgi->header(
+				-last_modified => $latest_date{'rfc2822'},
+				-status => '304 Not Modified');
+			goto DONE_GITWEB;
+		}
+	}
+}
+
 sub git_snapshot {
 	my $format = $input_params{'snapshot_format'};
 	if (!@snapshot_fmts) {

@@ -7029,6 +7051,10 @@ sub git_snapshot {
 
 	my ($name, $prefix) = snapshot_name($project, $hash);
 	my $filename = "$name$known_snapshot_formats{$format}{'suffix'}";
+
+	my %co = parse_commit($hash) or die_error(404, "Unknown commit object");
+	die_if_unmodified($co{'committer_epoch'});
+
 	my $cmd = quote_command(
 		git_cmd(), 'archive',
 		"--format=$known_snapshot_formats{$format}{'format'}",

@@ -7038,9 +7064,11 @@ sub git_snapshot {
 	}
 
 	$filename =~ s/(["\\])/\\$1/g;
+	my %latest_date = parse_date($co{'committer_epoch'}, $co{'committer_tz'});
 	print $cgi->header(
 		-type => $known_snapshot_formats{$format}{'type'},
 		-content_disposition => 'inline; filename="' . $filename . '"',
+		-last_modified => $latest_date{'rfc2822'},
 		-status => '200 OK');
 
 	open my $fd, "-|", $cmd

@@ -7820,24 +7848,8 @@ sub git_feed {
 	if (defined($commitlist[0])) {
 		%latest_commit = %{$commitlist[0]};
 		my $latest_epoch = $latest_commit{'committer_epoch'};
+		die_if_unmodified($latest_epoch);
 		%latest_date   = parse_date($latest_epoch, $latest_commit{'comitter_tz'});
-		my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
-		if (defined $if_modified) {
-			my $since;
-			if (eval { require HTTP::Date; 1; }) {
-				$since = HTTP::Date::str2time($if_modified);
-			} elsif (eval { require Time::ParseDate; 1; }) {
-				$since = Time::ParseDate::parsedate($if_modified, GMT => 1);
-			}
-			if (defined $since && $latest_epoch <= $since) {
-				print $cgi->header(
-					-type => $content_type,
-					-charset => 'utf-8',
-					-last_modified => $latest_date{'rfc2822'},
-					-status => '304 Not Modified');
-				return;
-			}
-		}
 		print $cgi->header(
 			-type => $content_type,
 			-charset => 'utf-8',
--                                                                              

1.7.3.4