helgrind/tests/filter_xml

   1 #!/usr/bin/env perl
   2
   3 #---------------------------------------------------------------------
   4 # Quick and dirty program to filter helgrind's XML output.
   5 #
   6 # The script works line-by-line and is generally unaware of XML structure
   7 # and does not bother with issues of well-formedness.
   8 #
   9 # Consists of two parts
  10 # (1) Global match and replace (see PATTERNS below)
  11 # (2) Removal of stack frames
  12 #     Stack frames whose associated file name does not match any name in
  13 #     TOOL_FILES or in the list of files given on the command line
  14 #     will be discarded. For a sequence of one or more discarded frames
  15 #     a line   <frame>...</frame> will be inserted.
  16 #
  17 #---------------------------------------------------------------------
  18
  19 use warnings;
  20 use strict;
  21
  22 #---------------------------------------------------------------------
  23 # A list of files specific to the tool at hand. Line numbers in
  24 # these files will be removed from stack frames matching these files.
  25 #---------------------------------------------------------------------
  26 my @tool_files = ( "hg_intercepts.c", "vg_replace_malloc.c" );
  27
  28 # List of patterns and replacement strings.
  29 # Each pattern must identify a substring which will be replaced.
  30 my %patterns = (
  31     "<pid>(.*)</pid>"       => "...",
  32     "<ppid>(.*)</ppid>"     => "...",
  33     "<time>(.*)</time>"     => "...",
  34     "<obj>(.*)</obj>"       => "...",
  35     "<dir>(.*)</dir>"       => "...",
  36     "<exe>(.*)</exe>"       => "...",
  37     "<tid>(.*)</tid>"       => "...",
  38     "<unique>(.*)</unique>" => "...",
  39     "thread #([0-9]+)"      => "x",
  40     "0x([0-9a-zA-Z]+)"      => "........",
  41     "Using Valgrind-([^\\s]*)"    => "X.Y.X",
  42     "Copyright \\(C\\) ([0-9]{4}-[0-9]{4}).*" => "XXXX-YYYY",
  43     '<fn>pthread_.*(@\*)</fn>'  => ""
  44 );
  45
  46 # List of XML sections to be ignored.
  47 my %ignore_sections = (
  48     "<errorcounts>" => "</errorcounts>",
  49     "<suppcounts>"  => "</suppcounts>"
  50 );
  51
  52
  53 # If FILE matches any of the FILES return 1
  54 sub file_matches ($$) {
  55     my ($file, $files) = @_;
  56     my ($string, $qstring);
  57
  58     foreach $string (@$files) {
  59         $qstring = quotemeta($string);
  60         return 1 if ($file =~ /$qstring/);
  61     }
  62
  63     return 0;
  64 }
  65
  66
  67 my $frame_buf = "";
  68 my ($file, $lineno, $in_frame, $keep_frame, $num_discarded, $ignore_line);
  69
  70 $in_frame = $keep_frame = $num_discarded = $ignore_line = 0;
  71
  72 line:
  73 while (<STDIN>) {
  74     my $line = $_;
  75     chomp($line);
  76
  77 # Check whether we're ignoring this piece of XML..
  78     if ($ignore_line) {
  79         foreach my $tag (keys %ignore_sections) {
  80             if ($line =~ $ignore_sections{$tag}) {
  81                 print "$tag...$ignore_sections{$tag}\n";
  82                 $ignore_line = 0;
  83                 next line;
  84             }
  85         }
  86     } else {
  87         foreach my $tag (keys %ignore_sections) {
  88             if ($line =~ $tag) {
  89                 $ignore_line = 1;
  90             }
  91         }
  92     }
  93
  94     next if ($ignore_line);
  95
  96 # OK. This line is not to be ignored.
  97
  98 # Massage line by applying PATTERNS.
  99     foreach my $key (keys %patterns) {
 100         if ($line =~ $key) {
 101            my $matched = quotemeta($1);
 102            $line =~ s/$matched/$patterns{$key}/g;
 103         }
 104     }
 105
 106 # Handle frames
 107     if ($in_frame) {
 108         if ($line =~ /<\/frame>/) {
 109             $frame_buf .= "$line\n";
 110 # The end of a frame
 111             if ($keep_frame) {
 112 # First: If there were any preceding frames that were discarded
 113 #        print <frame>...</frame>
 114                 if ($num_discarded) {
 115                     print "    <frame>...</frame>\n";
 116                     $num_discarded = 0;
 117                 }
 118 # Secondly: Write out the frame itself
 119                 print "$frame_buf";
 120             } else {
 121 # We don't want to write this frame
 122                 ++$num_discarded;
 123             }
 124             $in_frame = $keep_frame = 0;
 125             $file = "";
 126         } elsif ($line =~ /<file>(.*)<\/file>/) {
 127             $frame_buf .= "$line\n";
 128             $file = $1;
 129             if (file_matches($file, \@tool_files) ||
 130                 file_matches($file, \@ARGV)) {
 131                 $keep_frame = 1;
 132             }
 133         } elsif ($line =~ /<line>(.*)<\/line>/) {
 134 # This code assumes that <file> always precedes <line>
 135             $lineno = $1;
 136             if (file_matches($file, \@tool_files)) {
 137                 $line =~ s/$1/.../;
 138             }
 139             $frame_buf .= "$line\n";
 140         } else {
 141             $frame_buf .= "$line\n";
 142         }
 143     } else {
 144 # not within frame
 145         if ($line =~ /<\/stack>/) {
 146             print "    <frame>...</frame>\n" if ($num_discarded);
 147             $num_discarded = 0;
 148         }
 149         if ($line =~ /<frame>/)  {
 150             $in_frame = 1;
 151             $frame_buf = "$line\n";
 152         } else {
 153             print "$line\n";
 154         }
 155     }
 156 }
 157
 158 exit 0;