www: use PerlIO::scalar (zfh) for buffering

Calling Compress::Raw::Zlib::deflate is fairly expensive.
Relying on the `.=' (concat) operator inside ->zadd operator is
faster, but the method dispatch overhead is noticeable compared
to the original code where we had bare `.=' littered throughout.

Fortunately, `print' and `say' with the PerlIO::scalar IO layer
appears to offer better performance without high method dispatch
overhead.  This doesn't allow us to save as much memory as I
originally hoped, but does allow us to rely less on concat
operators in other places and just pass a list of args to
`print' and `say' as a appropriate.

This does reduce scratchpad use, however, allowing for large
memory savings, and we still ->deflate every single $eml.

7 files changed, 94 insertions, 95 deletions

diff --git a/lib/PublicInbox/GzipFilter.pm b/lib/PublicInbox/GzipFilter.pm
index 1f11acb8..848370ce 100644
--- a/lib/PublicInbox/GzipFilter.pm
+++ b/lib/PublicInbox/GzipFilter.pm
@@ -127,38 +127,39 @@ sub write {
         http_out($_[0])->write(translate($_[0], $_[1]));
 }
 
-sub zadd {
-        my $self = shift;
-        $self->{pbuf} .= $_ for @_; # perl internal pad memory use here
+sub zfh {
+        $_[0]->{zfh} // do {
+                open($_[0]->{zfh}, '>>', \($_[0]->{pbuf} //= '')) or
+                        die "open: $!";
+                $_[0]->{zfh}
+        };
 }
 
 # similar to ->translate; use this when we're sure we know we have
 # more data to buffer after this
 sub zmore {
-        my $self = shift; # $_[1] => input
+        my $self = shift;
+        my $zfh = delete $self->{zfh};
+        if (@_ > 1 || $zfh) {
+                print { $zfh // zfh($self) } @_;
+                @_ = (delete $self->{pbuf});
+                delete $self->{zfh};
+        };
         http_out($self);
-        my $x;
-        defined($x = delete($self->{pbuf})) and unshift(@_, $x);
-        for (@_) {
-                ($x = $self->{gz}->deflate($_, $self->{zbuf})) == Z_OK or
-                        die "gzip->deflate: $x";
-        }
-        undef;
+        my $err;
+        ($err = $self->{gz}->deflate($_[0], $self->{zbuf})) == Z_OK or
+                die "gzip->deflate: $err";
 }
 
 # flushes and returns the final bit of gzipped data
 sub zflush ($;@) {
         my $self = shift; # $_[1..Inf] => final input (optional)
+        zmore($self, @_) if scalar(@_) || $self->{zfh};
+        # not a bug, recursing on DS->write failure
+        my $gz = delete $self->{gz} // return '';
+        my $err;
         my $zbuf = delete $self->{zbuf};
-        my $gz = delete $self->{gz};
-        my $x;
-        defined($x = delete($self->{pbuf})) and unshift(@_, $x);
-        for (@_) { # it's a bug iff $gz is undef if @_ isn't empty, here:
-                ($x = $gz->deflate($_, $zbuf)) == Z_OK or
-                        die "gzip->deflate: $x";
-        }
-        $gz // return ''; # not a bug, recursing on DS->write failure
-        ($x = $gz->flush($zbuf)) == Z_OK or die "gzip->flush: $x";
+        ($err = $gz->flush($zbuf)) == Z_OK or die "gzip->flush: $err";
         $zbuf;
 }
 
diff --git a/lib/PublicInbox/Mbox.pm b/lib/PublicInbox/Mbox.pm
index cfe34d9c..2ef8ff2b 100644
--- a/lib/PublicInbox/Mbox.pm
+++ b/lib/PublicInbox/Mbox.pm
@@ -20,7 +20,7 @@ sub getline {
         my $ibx = $ctx->{ibx};
         my $eml = delete($ctx->{eml}) // $ibx->smsg_eml($smsg) // return;
         my $n = $ctx->{smsg} = $ibx->over->next_by_mid(@{$ctx->{next_arg}});
-        $ctx->zadd(msg_hdr($ctx, $eml));
+        $ctx->zmore(msg_hdr($ctx, $eml));
         if ($n) {
                 $ctx->translate(msg_body($eml));
         } else { # last message
diff --git a/lib/PublicInbox/SearchView.pm b/lib/PublicInbox/SearchView.pm
index b18947ee..8932c73d 100644
--- a/lib/PublicInbox/SearchView.pm
+++ b/lib/PublicInbox/SearchView.pm
@@ -331,10 +331,10 @@ sub mset_thread {
 # callback for PublicInbox::WwwStream::getline
 sub mset_thread_i {
         my ($ctx, $eml) = @_;
-        $ctx->zadd($ctx->html_top) if exists $ctx->{-html_tip};
+        print { $ctx->zfh } $ctx->html_top if exists $ctx->{-html_tip};
         $eml and return PublicInbox::View::eml_entry($ctx, $eml);
         my $smsg = shift @{$ctx->{msgs}} or
-                $ctx->zmore(${delete($ctx->{skel})});
+                print { $ctx->zfh } ${delete($ctx->{skel})};
         $smsg;
 }
 
diff --git a/lib/PublicInbox/View.pm b/lib/PublicInbox/View.pm
index 85dc3bd8..0753c06e 100644
--- a/lib/PublicInbox/View.pm
+++ b/lib/PublicInbox/View.pm
@@ -40,7 +40,7 @@ sub msg_page_i {
                                 "../${\mid_href($smsg->{mid})}/" : '';
                 if (_msg_page_prepare($eml, $ctx)) {
                         $eml->each_part(\&add_text_body, $ctx, 1);
-                        $ctx->zadd('</pre><hr>');
+                        print { $ctx->{zfh} } '</pre><hr>';
                 }
                 html_footer($ctx, $ctx->{first_hdr}) if !$ctx->{smsg};
                 ''; # XXX TODO cleanup
@@ -58,7 +58,7 @@ sub no_over_html ($) {
         PublicInbox::WwwStream::init($ctx);
         if (_msg_page_prepare($eml, $ctx)) { # sets {-title_html}
                 $eml->each_part(\&add_text_body, $ctx, 1);
-                $ctx->zadd('</pre><hr>');
+                print { $ctx->{zfh} } '</pre><hr>';
         }
         html_footer($ctx, $eml);
         $ctx->html_done;
@@ -240,12 +240,11 @@ sub eml_entry {
                 my $html = ascii_html($irt);
                 $rv .= qq(In-Reply-To: &lt;<a\nhref="$href">$html</a>&gt;\n)
         }
-        $rv .= "\n";
+        say { $ctx->zfh } $rv;
 
         # scan through all parts, looking for displayable text
         $ctx->{mhref} = $mhref;
         $ctx->{changed_href} = "#e$id"; # for diffstat "files? changed,"
-        $ctx->zadd($rv); # XXX $rv is small, reuse below
         $eml->each_part(\&add_text_body, $ctx, 1); # expensive
 
         # add the footer
@@ -386,8 +385,8 @@ sub pre_thread  { # walk_thread callback
 sub thread_eml_entry {
         my ($ctx, $eml) = @_;
         my ($beg, $end) = thread_adj_level($ctx, $ctx->{level});
-        $ctx->zadd($beg.'<pre>');
-        eml_entry($ctx, $eml) . '</pre>' . $end;
+        print { $ctx->zfh } $beg, '<pre>';
+        print { $ctx->{zfh} } eml_entry($ctx, $eml), '</pre>', $end;
 }
 
 sub next_in_queue ($$) {
@@ -414,15 +413,15 @@ sub stream_thread_i { # PublicInbox::WwwStream::getline callback
                                 if (!$ghost_ok) { # first non-ghost
                                         $ctx->{-title_html} =
                                                 ascii_html($smsg->{subject});
-                                        $ctx->zadd($ctx->html_top);
+                                        print { $ctx->zfh } $ctx->html_top;
                                 }
                                 return $smsg;
                         }
                         # buffer the ghost entry and loop
-                        $ctx->zadd(ghost_index_entry($ctx, $lvl, $smsg));
+                        print { $ctx->zfh } ghost_index_entry($ctx, $lvl, $smsg)
                 } else { # all done
-                        $ctx->zadd(join('', thread_adj_level($ctx, 0)));
-                        $ctx->zadd(${delete($ctx->{skel})});
+                        print { $ctx->zfh } thread_adj_level($ctx, 0),
+                                                ${delete($ctx->{skel})};
                         return;
                 }
         }
@@ -491,7 +490,7 @@ sub thread_html_i { # PublicInbox::WwwStream::getline callback
                 my $smsg = $ctx->{smsg};
                 if (exists $ctx->{-html_tip}) {
                         $ctx->{-title_html} = ascii_html($smsg->{subject});
-                        $ctx->zadd($ctx->html_top);
+                        print { $ctx->zfh } $ctx->html_top;
                 }
                 return eml_entry($ctx, $eml);
         } else {
@@ -499,7 +498,7 @@ sub thread_html_i { # PublicInbox::WwwStream::getline callback
                         return $smsg if exists($smsg->{blob});
                 }
                 my $skel = delete($ctx->{skel}) or return; # all done
-                $ctx->zadd($$skel);
+                print { $ctx->zfh } $$skel;
                 undef;
         }
 }
@@ -560,8 +559,9 @@ sub add_text_body { # callback for each_part
         my $ct = $part->content_type || 'text/plain';
         my $fn = $part->filename;
         my ($s, $err) = msg_part_text($part, $ct);
-        $s // return $ctx->zadd(attach_link($ctx, $ct, $p, $fn) // '');
-        my $buf = $part->{is_submsg} ? submsg_hdr($ctx, $part)."\n" : '';
+        my $zfh = $ctx->zfh;
+        $s // return print $zfh (attach_link($ctx, $ct, $p, $fn) // '');
+        say $zfh submsg_hdr($ctx, $part) if $part->{is_submsg};
 
         # makes no difference to browsers, and don't screw up filename
         # link generation in diffs with the extra '%0D'
@@ -609,21 +609,19 @@ sub add_text_body { # callback for each_part
         undef $s; # free memory
         if (defined($fn) || ($depth > 0 && !$part->{is_submsg}) || $err) {
                 # badly-encoded message with $err? tell the world about it!
-                $buf .= attach_link($ctx, $ct, $p, $fn, $err) . "\n";
+                say $zfh attach_link($ctx, $ct, $p, $fn, $err);
         }
         delete $part->{bdy}; # save memory
-        $ctx->zadd($buf);
-        undef $buf;
         for my $cur (@sections) { # $cur may be huge
                 if ($cur =~ /\A>/) {
                         # we use a <span> here to allow users to specify
                         # their own color for quoted text
-                        $ctx->zadd(qq(<span\nclass="q">),
-                                        $l->to_html($cur), '</span>');
+                        print $zfh qq(<span\nclass="q">),
+                                        $l->to_html($cur), '</span>';
                 } elsif ($diff) {
                         flush_diff($ctx, \$cur);
                 } else { # regular lines, OK
-                        $ctx->zadd($l->to_html($cur));
+                        print $zfh $l->to_html($cur);
                 }
                 undef $cur; # free memory
         }
@@ -685,10 +683,10 @@ sub _msg_page_prepare {
                 $hbuf .= qq[Message-ID: &lt;$x&gt; (<a href="raw">raw</a>)\n];
         }
         if (!$nr) { # first (and only) message, common case
-                $ctx->zadd($ctx->html_top, $hbuf);
+                print { $ctx->zfh } $ctx->html_top, $hbuf;
         } else {
                 delete $ctx->{-title_html};
-                $ctx->zadd($ctx->{-html_tip}, $hbuf);
+                print { $ctx->zfh } $ctx->{-html_tip}, $hbuf;
         }
         $ctx->{-linkify} //= PublicInbox::Linkify->new;
         $hbuf = '';
@@ -699,7 +697,7 @@ sub _msg_page_prepare {
                         $hbuf .= "$h: $_\n" for ($eml->header_raw($h));
                 }
                 $ctx->{-linkify}->linkify_mids('..', \$hbuf, 1); # escapes HTML
-                $ctx->zadd($hbuf);
+                print { $ctx->{zfh} } $hbuf;
                 $hbuf = '';
         }
         my @irt = $eml->header_raw('In-Reply-To');
@@ -717,7 +715,7 @@ sub _msg_page_prepare {
                 $hbuf .= 'References: <'.join(">\n\t<", @$refs).">\n" if @$refs;
         }
         $ctx->{-linkify}->linkify_mids('..', \$hbuf); # escapes HTML
-        $ctx->zadd($hbuf .= "\n");
+        say { $ctx->{zfh} } $hbuf;
         1;
 }
 
@@ -771,7 +769,7 @@ sub thread_skel ($$$) {
 sub html_footer {
         my ($ctx, $hdr) = @_;
         my $upfx = '../';
-        my ($related, $skel);
+        my (@related, $skel);
         my $foot = '<pre>';
         my $qry = delete $ctx->{-qry};
         if ($qry && $ctx->{ibx}->isrch) {
@@ -786,7 +784,7 @@ sub html_footer {
                 $q = wrap('', '', $q);
                 my $rows = ($q =~ tr/\n/\n/) + 1;
                 $q = ascii_html($q);
-                $related = <<EOM;
+                $related[0] = <<EOM;
 <form id=related
 action=$upfx
 ><pre>find likely ancestor, descendant, or conflicting patches for <a
@@ -799,7 +797,7 @@ EOM
         if ($ctx->{ibx}->over) {
                 my $t = ts2str($ctx->{-t_max});
                 my $t_fmt = fmt_ts($ctx->{-t_max});
-                my $fallback = $related ? "\t" : "<a id=related>\t</a>";
+                my $fallback = @related ? "\t" : "<a id=related>\t</a>";
                 $skel = <<EOF;
 ${fallback}other threads:[<a
 href="$upfx?t=$t">~$t_fmt UTC</a>|<a
@@ -834,10 +832,10 @@ EOF
         } else { # unindexed inboxes w/o over
                 $skel = qq( <a\nhref="$upfx">latest</a>);
         }
-        $foot .= qq(<a\nhref="#R">reply</a>);
         # $skel may be big for big threads, don't append it to $foot
-        $skel .= '</pre>' . ($related // '');
-        $ctx->zadd($foot, $skel .= msg_reply($ctx, $hdr));
+        print { $ctx->zfh } $foot, qq(<a\nhref="#R">reply</a>),
+                                $skel, '</pre>', @related,
+                                msg_reply($ctx, $hdr);
 }
 
 sub ghost_parent {
diff --git a/lib/PublicInbox/ViewDiff.pm b/lib/PublicInbox/ViewDiff.pm
index 95b615dc..5d23881b 100644
--- a/lib/PublicInbox/ViewDiff.pm
+++ b/lib/PublicInbox/ViewDiff.pm
@@ -67,8 +67,8 @@ sub oid ($$$) {
 }
 
 # returns true if diffstat anchor written, false otherwise
-sub anchor0 ($$$$) {
-        my ($dst, $ctx, $fn, $rest) = @_;
+sub anchor0 ($$$) {
+        my ($ctx, $fn, $rest) = @_;
 
         my $orig = $fn;
 
@@ -84,15 +84,12 @@ sub anchor0 ($$$$) {
         # long filenames will require us to check in anchor1()
         push(@{$ctx->{-long_path}}, $fn) if $fn =~ s!\A\.\.\./?!!;
 
-        if (defined(my $attr = to_attr($ctx->{-apfx}.$fn))) {
-                $ctx->{-anchors}->{$attr} = 1;
-                my $spaces = ($orig =~ s/( +)\z//) ? $1 : '';
-                $$dst .= " <a\nid=i$attr\nhref=#$attr>" .
-                        ascii_html($orig) . '</a>' . $spaces .
+        my $attr = to_attr($ctx->{-apfx}.$fn) // return;
+        $ctx->{-anchors}->{$attr} = 1;
+        my $spaces = ($orig =~ s/( +)\z//) ? $1 : '';
+        print { $ctx->{zfh} } " <a\nid=i$attr\nhref=#$attr>",
+                        ascii_html($orig), '</a>', $spaces,
                         $ctx->{-linkify}->to_html($rest);
-                return 1;
-        }
-        undef;
 }
 
 # returns "diff --git" anchor destination, undef otherwise
@@ -156,33 +153,34 @@ sub diff_header ($$$) {
                 warn "BUG? <$$x> had no ^index line";
         }
         $$x =~ s!^diff --git!anchor1($ctx, $pb) // 'diff --git'!ems;
-        $ctx->zadd(qq(<span\nclass="head">$$x</span>));
+        print { $ctx->{zfh} } qq(<span\nclass="head">), $$x, '</span>';
         $dctx;
 }
 
 sub diff_before_or_after ($$) {
         my ($ctx, $x) = @_;
-        if (exists $ctx->{-anchors} && $$x =~ /\A(.*?) # likely "---\n"
+        if (exists $ctx->{-anchors} && $$x =~ /\A(.*?) # likely "---\n" # \$1
                         # diffstat lines:
                         ((?:^\x20(?:[^\n]+?)(?:\x20+\|\x20[^\n]*\n))+)
                         (\x20[0-9]+\x20files?\x20)changed,([^\n]+\n)
                         (.*?)\z/msx) { # notes, commit message, etc
                 my @x = ($5, $4, $3, $2, $1);
+                undef $$x;
                 my $lnk = $ctx->{-linkify};
-                $$x = $lnk->to_html(pop @x); # uninteresting prefix
-                for my $l (split(/^/m, pop(@x))) { # per-file diffstat lines
+                my $zfh = $ctx->{zfh};
+                print $zfh $lnk->to_html(pop @x); # $1 uninteresting prefix
+                for my $l (split(/^/m, pop(@x))) { # $2 per-file stat lines
                         $l =~ /^ (.+)( +\| .*\z)/s and
-                                anchor0($x, $ctx, $1, $2) and next;
-                        $$x .= $lnk->to_html($l);
+                                anchor0($ctx, $1, $2) and next;
+                         print $zfh $lnk->to_html($l);
                 }
-                $$x .= pop @x; # $3 /^ \d+ files? /
                 my $ch = $ctx->{changed_href} // '#related';
-                $$x .= qq(<a href="$ch">changed</a>,);
-                $$x .= ascii_html(pop @x); # $4: insertions/deletions
-                # notes, commit message, etc
-                $ctx->zadd($$x .= $lnk->to_html(pop @x));
+                print $zfh pop(@x), # $3 /^ \d+ files? /
+                        qq(<a href="$ch">changed</a>,),
+                        ascii_html(pop @x), # insertions/deletions
+                        $lnk->to_html(@x); # notes, commit message, etc
         } else {
-                $ctx->zadd($ctx->{-linkify}->to_html($$x));
+                print { $ctx->{zfh} } $ctx->{-linkify}->to_html($$x);
         }
 }
 
@@ -195,6 +193,7 @@ sub flush_diff ($$) {
 
         my $lnk = $ctx->{-linkify};
         my $dctx; # {}, keys: Q, oid_a, oid_b
+        my $zfh = $ctx->zfh;
 
         while (defined(my $x = shift @top)) {
                 if (scalar(@top) >= 4 &&
@@ -202,7 +201,7 @@ sub flush_diff ($$) {
                                 $top[0] =~ $IS_OID) {
                         $dctx = diff_header(\$x, $ctx, \@top);
                 } elsif ($dctx) {
-                        my $after = '';
+                        open(my $afh, '>>', \(my $after='')) or die "open: $!";
 
                         # Quiet "Complex regular subexpression recursion limit"
                         # warning.  Perl will truncate matches upon hitting
@@ -218,25 +217,25 @@ sub flush_diff ($$) {
                                         (?:(?:^-[^\n]*\n)+)|
                                         (?:^@@ [^\n]+\n))/xsm, $x)) {
                                 if (!defined($dctx)) {
-                                        $after .= $s;
+                                        print $afh $s;
                                 } elsif ($s =~ s/\A@@ (\S+) (\S+) @@//) {
-                                        $ctx->zadd(qq(<span\nclass="hunk">) .
-                                                diff_hunk($dctx, $1, $2) .
-                                                $lnk->to_html($s) .
-                                                '</span>');
+                                        print $zfh qq(<span\nclass="hunk">),
+                                                diff_hunk($dctx, $1, $2),
+                                                $lnk->to_html($s),
+                                                '</span>';
                                 } elsif ($s =~ /\A\+/) { # $s may be huge
-                                        $ctx->zadd(qq(<span\nclass="add">),
+                                        print $zfh qq(<span\nclass="add">),
                                                         $lnk->to_html($s),
-                                                        '</span>');
+                                                        '</span>';
                                 } elsif ($s =~ /\A-- $/sm) { # email sig starts
                                         $dctx = undef;
-                                        $after .= $s;
+                                        print $afh $s;
                                 } elsif ($s =~ /\A-/) { # $s may be huge
-                                        $ctx->zadd(qq(<span\nclass="del">),
-                                                $lnk->to_html($s),
-                                                '</span>');
+                                        print $zfh qq(<span\nclass="del">),
+                                                        $lnk->to_html($s),
+                                                        '</span>';
                                 } else { # $s may be huge
-                                        $ctx->zadd($lnk->to_html($s));
+                                        print $zfh $lnk->to_html($s);
                                 }
                         }
                         diff_before_or_after($ctx, \$after) if !$dctx;
diff --git a/lib/PublicInbox/WwwAtomStream.pm b/lib/PublicInbox/WwwAtomStream.pm
index 1c7ae881..33da3244 100644
--- a/lib/PublicInbox/WwwAtomStream.pm
+++ b/lib/PublicInbox/WwwAtomStream.pm
@@ -146,15 +146,15 @@ sub feed_entry {
         my $name = ascii_html(join(', ', PublicInbox::Address::names($from)));
         $email = ascii_html($email // $ctx->{ibx}->{-primary_address});
 
-        $ctx->zadd(
-                (delete($ctx->{emit_header}) ? atom_header($ctx, $title) : '').
+        print { $ctx->zfh }
+                (delete($ctx->{emit_header}) ? atom_header($ctx, $title) : ''),
                 "<entry><author><name>$name</name><email>$email</email>" .
                 "</author>$title$updated" .
-                qq(<link\nhref="$href"/>).
+                qq(<link\nhref="$href"/>) .
                 "<id>$uuid</id>$irt" .
                 qq{<content\ntype="xhtml">} .
                 qq{<div\nxmlns="http://www.w3.org/1999/xhtml">} .
-                qq(<pre\nstyle="white-space:pre-wrap">));
+                qq(<pre\nstyle="white-space:pre-wrap">);
         $ctx->{mhref} = $href;
         $ctx->{changed_href} = "${href}#related";
         $eml->each_part(\&PublicInbox::View::add_text_body, $ctx, 1);
diff --git a/lib/PublicInbox/WwwStream.pm b/lib/PublicInbox/WwwStream.pm
index 2a318e5e..77b6f9c2 100644
--- a/lib/PublicInbox/WwwStream.pm
+++ b/lib/PublicInbox/WwwStream.pm
@@ -181,11 +181,12 @@ sub html_oneshot ($$;@) {
                 'Content-Length' => undef ];
         bless $ctx, __PACKAGE__;
         $ctx->{gz} = PublicInbox::GzipFilter::gz_or_noop($res_hdr, $ctx->{env});
+        my @top;
         $ctx->{base_url} // do {
-                $ctx->zadd(html_top($ctx));
+                @top = html_top($ctx);
                 $ctx->{base_url} = base_url($ctx);
         };
-        my $bdy = $ctx->zflush(@_[2..$#_], _html_end($ctx));
+        my $bdy = $ctx->zflush(@top, @_[2..$#_], _html_end($ctx));
         $res_hdr->[3] = length($bdy);
         [ $code, $res_hdr, [ $bdy ] ]
 }
@@ -216,7 +217,7 @@ sub html_init {
         my $h = $ctx->{-res_hdr} = ['Content-Type', 'text/html; charset=UTF-8'];
         $ctx->{gz} = PublicInbox::GzipFilter::gz_or_noop($h, $ctx->{env});
         bless $ctx, __PACKAGE__;
-        $ctx->zadd(html_top($ctx));
+        print { $ctx->zfh } html_top($ctx);
 }
 
 1;