[6255] Plaintext.pm faster byte counting

[Gavin D. Smith]

Revision: 6255
          http://svn.sv.gnu.org/viewvc/?view=rev&root=texinfo&revision=6255
Author:   gavin
Date:     2015-05-07 02:06:19 +0000 (Thu, 07 May 2015)
Log Message:
-----------
Plaintext.pm faster byte counting

Modified Paths:
--------------
    trunk/ChangeLog
    trunk/tp/DebugTexinfo/DebugCount.pm
    trunk/tp/Texinfo/Convert/Info.pm
    trunk/tp/Texinfo/Convert/Plaintext.pm

Modified: trunk/ChangeLog
===================================================================
--- trunk/ChangeLog     2015-05-06 22:03:58 UTC (rev 6254)
+++ trunk/ChangeLog     2015-05-07 02:06:19 UTC (rev 6255)
@@ -1,3 +1,22 @@
+2015-05-07  Gavin Smith  <address@hidden>
+
+       * tp/Texinfo/Convert/Plaintext.pm (_add_text_count)
+       (_count_added): Don't call Texinfo::Common::count_bytes, collect 
+       the argument instead into a string.
+       (_update_count_context): Count the text collected in 
+       _add_text_count.
+
+       * tp/Texinfo/Convert/Plaintext.pm (add_location)
+       (_update_locations_counts, _align_environment, _node_line)
+       (_printindex_formatted, _convert),
+       * tp/Texinfo/Convert/Info.pm (output),
+       * tp/DebugTexinfo/DebugCount.pm (_convert):
+       Call _update_count_context before accessing 'bytes' value of 
+       current count context.
+
+       * tp/Texinfo/Convert/Plaintext.pm (_count_added): Access
+       'end_line_count' value on hash directly instead of via function 
+       call.
 2015-05-06  Karl Berry  <address@hidden>
 
        * doc/texinfo.tex (\ensuremath): new helper macro.

Modified: trunk/tp/DebugTexinfo/DebugCount.pm
===================================================================
--- trunk/tp/DebugTexinfo/DebugCount.pm 2015-05-06 22:03:58 UTC (rev 6254)
+++ trunk/tp/DebugTexinfo/DebugCount.pm 2015-05-07 02:06:19 UTC (rev 6255)
@@ -34,6 +34,7 @@
   
   $self->{'level'}++;
   $self->{'command_type_nr'}++;
+  $self->_update_count_context();
   my $bytes_before = $self->{'count_context'}->[-1]->{'bytes'};
   my $number_before = 
"($self->{'count_context'}->[-1]->{'bytes'},$self->{'count_context'}->[-1]->{'lines'})";
   my $command_nr = '['.$self->{'command_type_nr'}.']';
@@ -88,6 +89,7 @@
   }
   #push @{$self->{'debug_count_strings'}}, ' ' x $self->{'level'}. "TEXT: 
$text|\n"
   #  if ($self->{'count_context'}->[-1]->{'bytes'} > $bytes_before);
+  $self->_update_count_context();
   my $number_after = 
"($self->{'count_context'}->[-1]->{'bytes'},$self->{'count_context'}->[-1]->{'lines'})";
   my $string_after = ' ' x $self->{'level'}. "$command_nr $number_after\n";
   $string_after .= " locations $all_locations_string"

Modified: trunk/tp/Texinfo/Convert/Info.pm
===================================================================
--- trunk/tp/Texinfo/Convert/Info.pm    2015-05-06 22:03:58 UTC (rev 6254)
+++ trunk/tp/Texinfo/Convert/Info.pm    2015-05-07 02:06:19 UTC (rev 6255)
@@ -148,6 +148,7 @@
       } else {
         $result .= $node_text;
       }
+      $self->_update_count_context();
       if (defined($self->get_conf('SPLIT_SIZE')) 
           and $self->{'count_context'}->[-1]->{'bytes'} > 
                   $out_file_nr * $self->get_conf('SPLIT_SIZE') 
@@ -205,6 +206,7 @@
            return undef;
         }
         print $fh $complete_header;
+        $self->_update_count_context();
         $self->{'count_context'}->[-1]->{'bytes'} += $complete_header_bytes;
         push @indirect_files, [$self->{'output_filename'}.'-'.$out_file_nr,
                                $self->{'count_context'}->[-1]->{'bytes'}];

Modified: trunk/tp/Texinfo/Convert/Plaintext.pm
===================================================================
--- trunk/tp/Texinfo/Convert/Plaintext.pm       2015-05-06 22:03:58 UTC (rev 
6254)
+++ trunk/tp/Texinfo/Convert/Plaintext.pm       2015-05-07 02:06:19 UTC (rev 
6255)
@@ -717,9 +717,10 @@
 {
   my $self = shift;
   my $text = shift;
-  $self->{'count_context'}->[-1]->{'bytes'}
-      += Texinfo::Common::count_bytes($self, $text, 
-                                      $self->{'output_perl_encoding'});
+  if (!$self->{'count_context'}->[-1]->{'pending_text'}) {
+    $self->{'count_context'}->[-1]->{'pending_text'} = '';
+  }
+  $self->{'count_context'}->[-1]->{'pending_text'} .= $text;
 }
 
 sub _add_lines_count($$)
@@ -729,6 +730,22 @@
   $self->{'count_context'}->[-1]->{'lines'} += $lines_count;
 }
 
+# Update $SELF->{'count_context'}->[-1]->{'bytes'} by counting the text that
+# hasn't been counted yet.  It is faster to count the text all together than
+# piece by piece in _add_text_count.
+sub _update_count_context($)
+{
+  my $self = shift;
+  if ($self->{'count_context'}->[-1]->{'pending_text'}) {
+    $self->{'count_context'}->[-1]->{'bytes'} +=
+      Texinfo::Common::count_bytes($self,
+        $self->{'count_context'}->[-1]->{'pending_text'},
+        $self->{'output_perl_encoding'});
+    $self->{'count_context'}->[-1]->{'pending_text'} = '';
+  }
+}
+
+# Save the line and byte offset of $ROOT.
 sub _add_location($$)
 {
   my $self = shift;
@@ -736,8 +753,8 @@
   my $location = { 'lines' => $self->{'count_context'}->[-1]->{'lines'} };
   push @{$self->{'count_context'}->[-1]->{'locations'}}, $location;
   if (!($root->{'extra'} and $root->{'extra'}->{'index_entry'})) {
-    $location->{'bytes'}
-      = $self->{'count_context'}->[-1]->{'bytes'};
+    _update_count_context($self);
+    $location->{'bytes'} = $self->{'count_context'}->[-1]->{'bytes'};
     $location->{'root'} = $root;
   } else {
     $location->{'index_entry'} = $root;
@@ -770,12 +787,16 @@
   my $text = shift;
 
   #$self->_add_lines_count($container->end_line_count());
-  $self->{'count_context'}->[-1]->{'lines'} += $container->end_line_count();
+  $self->{'count_context'}->[-1]->{'lines'} += $container->{'end_line_count'};
 
   #$self->_add_text_count($text);
-  $self->{'count_context'}->[-1]->{'bytes'} +=
-    Texinfo::Common::count_bytes($self, $text,
-                                 $self->{'output_perl_encoding'});
+  #$self->{'count_context'}->[-1]->{'bytes'} +=
+  #  Texinfo::Common::count_bytes($self, $text,
+  #                               $self->{'output_perl_encoding'});
+  if (!defined $self->{'count_context'}->[-1]->{'pending_text'}) {
+    $self->{'count_context'}->[-1]->{'pending_text'} = '';
+  }
+  $self->{'count_context'}->[-1]->{'pending_text'} .= $text;
   return $text;
 }
 
@@ -783,6 +804,7 @@
 {
   my $self = shift;
   my $locations = shift;
+  _update_count_context($self);
   foreach my $location (@$locations) {
     $location->{'bytes'} += $self->{'count_context'}->[-1]->{'bytes'}
        if (defined($location->{'bytes'}));
@@ -1044,6 +1066,7 @@
   my $max = shift;
   my $align = shift;
 
+  _update_count_context($self);
   my $counts = pop @{$self->{'count_context'}};
   my $bytes_count;
   ($result, $bytes_count) = $self->_align_lines($result, $max,
@@ -1169,6 +1192,7 @@
     push @{$self->{'count_context'}}, {'lines' => 0, 'bytes' => 0};
     $self->{'node_lines_text'}->{$node}->{'text'} 
        = _normalize_top_node($self->convert_line($node_text));
+    _update_count_context($self);
     my $end_context = pop @{$self->{'count_context'}};
     $self->{'node_lines_text'}->{$node}->{'count'} 
       = $end_context->{'bytes'};
@@ -1318,6 +1342,7 @@
         my $node_text = $self->gdt('(outside of any node)');
         $self->{'outside_of_any_node_text'}->{'text'} 
           = $self->convert_line($node_text);
+        _update_count_context($self);
         my $end_context = pop @{$self->{'count_context'}};
         $self->{'outside_of_any_node_text'}->{'count'} 
           = $end_context->{'bytes'};
@@ -3281,6 +3306,7 @@
       pop @{$self->{'format_context'}};
       pop @{$self->{'text_element_context'}};
       push @{$self->{'format_context'}->[-1]->{'row'}}, $result;
+      _update_count_context($self);
       my $cell_counts = pop @{$self->{'count_context'}};
       push @{$self->{'format_context'}->[-1]->{'row_counts'}}, $cell_counts;
       $result = '';